The operator, a man named Johann Maelzel, would assemble a paying audience, open the doors of the lower cabinet and show an impressively whirring clockwork mechanism that filled the inner compartments beneath the seated figure. Then he would close the cabinet, and invite a challenger to play chess. The automaton - the robot, as we would say now - would gaze at the opponent's move, ponder, then raise its mechanical arm and make a stiff but certain move of its own.

The thing was a sensation.

Before it was destroyed by fire in New York in the 1850s, it played games with everyone from Benjamin Franklin to, by legend at least, Napoleon Bonaparte. Artificial intelligence, the 18th Century thought, had arrived, wearing a fez and ticking away like Captain Hook's crocodile.

I should rush to say that, of course, the thing was a fraud, or rather, a trick - a clever magician's illusion. A sliding sled on well-lubricated castors had been fitted inside the lower cabinet and the only real ingenuity was that this let a hidden chess player glide easily, and silently, into a prone position inside. There was just a lot more room to hide in the cabinet than all that clockwork machinery suggested.

Now, the Turk fascinates me for several reasons. First, because it displays an odd, haunting hole in human reasoning. Common sense should have told the people who watched and challenged it that for the Turk to have really been a chess-playing machine, it would have had to have been the latest in a long sequence of such machines. For there to be a mechanical Turk who played chess, there would have had to have been, 10 years before, a mechanical Greek who played draughts.

It's true that the late 18th Century was a great age of automatons, machines that could make programmed looms weave and mechanical birds sing - although always the same song, or tapestry, over and over. But the deeper truth that chess-playing was an entirely different kind of creative activity seemed as obscure to them as it seems obvious to us now.

Suppose that you have k sorted arrays, each of size n. You would like to search for single element in each of the k arrays (or its predecessor, if it doesn't exist).

Obviously you can binary search each array individually, resulting in a runtime. But we might think we can do better that: after all, we're doing the same search k times, and maybe we can "reuse" the results of the first search for later searches.

Here's another obvious thing we can do: for every element in the first array, let's give it a pointer to the element with the same value in the second array (or if the value doesn't exist, the predecessor.) Then once we've found the item in the first array, we can just follow these pointers down in order to figure out where the item is in all the other arrays.

But there's a problem: sometimes, these pointers won't help us at all. In particular, if a later lists is completely "in between" two elements of the first list, we have to redo the entire search, since the pointer gave us no information that we didn't already know.

So what do we do? Consider the case where k = 2; everything would be better if only we could guarantee that the first list contained the right elements to give you useful information about the second array. We could just merge the arrays, but if we did this in the general case we'd end up with a totally merged array of size , which is not so good if k is large.

But we don't need all of the elements of the second array; every other item will do!

Let's repeatedly do this. Take the last array, take every other element and merge it into the second to last array. Now, with the new second to last array, do this to the next array. Rinse and repeat. How big does the first array end up being? You can solve the recurrence: , which is the geometric series . Amazingly, the new first list is only twice as large, which is only one extra step in the binary search!

What we have just implemented is fractional cascading! A fraction of any array cascades up the rest of the arrays.

There is one more detail which has to be attended to. When I follow a pointer down, I might end up on an element which is not actually a member of the current array (it was one that was cascaded up). I need to be able to efficiently find the next element which is a member of the current array (and there might be many cascaded elements jammed between it and the next member element, so doing a left-scan could take a long time); so for every cascaded element I store a pointer to the predecessor member element.

Fractional cascading is a very useful transformation, used in a variety of contexts including layered range trees and 3D orthogonal range searching. In fact, it can be generalized in several ways. The first is that we can cascade some fixed fraction α of elements, rather than the 1/2 we did here. Additionally, we don't have to limit ourselves to cascading up a list of arrays; we can cascade up an arbitrary graph, merging many lists together as long as we pick α to be less than 1/d, where d is the in-degree of the node.

Exercise. Previously, we described range trees. How can fractional cascading be used to reduce the query complexity by a factor of ?

Exercise. There is actually another way we can setup the pointers in a fractionally cascaded data structure. Rather than have downward pointers for every element, we only maintain pointers between elements which are identical (that is to say, they were cascaded up.) This turns out to be more convenient when you are constructing the data structure. However, you now need to maintain another set of pointers. What are they? (Hint: Consider the case where a search lands on a non-cascaded, member element.)

In today’s article, we’re embarking on something unprecedented! We’ll guide you step by step through the process of transforming an SSD equipped with QLC NANDs into an SLC SSD, significantly enhancing its durability and overall performance!

Especification of the DUT SSD:

The SSD I chose is a Crucial BX500, which we’ve tested numerous times both on our website and on my YouTube channel.

ATENTION: BEFORE YOU CONTINUE READING!!!

Firstly, this procedure is safer than overclocking, but it still requires caution. Only proceed if you are genuinely interested, as I cannot be held responsible if any steps are executed incorrectly. I will explain as clearly as possible to minimize any misunderstandings.

This voids the warranty of any SSD. AND REMEMBER, WHEN FLASHING THE FIRMWARE TO THE SSD, ALL DATA WILL BE ERASED, so be sure to back up your devices before proceeding with anything.

NECESSARY TOOLS

To perform this procedure, it was necessary to use an adapter SATA to USB 3.0 adapter with the Jmicron JMS578 Bridge Chip model.

In addition to, we also need a clamp to perform the short on the ROM/Safe Mode terminals on the SSD’s PCB.

Technical Specs

Before we move on to the tutorial, let’s analyze this SSD a little further.

Controller

The SSD controller is responsible for handling all data management tasks, including over-provisioning and garbage collection, among other background functions. Naturally, this contributes to the SSD’s overall performance.

In this project, the SSD utilizes the Silicon Motion controller model SM2259XT2, which is a new variant of the SM2259XT.

In this case, it’s a single-core controller, meaning it has one main core responsible for managing the NANDs, with a 32-bit ARC architecture, not ARM as we’re accustomed to. This controller has an operating frequency up to 550 MHz, but as we’ll see in the following image, in this project, it was operating at 437.5 MHz.

This controller also supports up to 2 communication channels with a bus speed of up to 800 MT/s, where each of these channels supports up to 8 Chip Enable commands, allowing the controller to communicate with up to 16 Dies simultaneously using the interleaving technique.

What was different from its predecessor, the SM2259XT, which had 4 channels and 4 C.E. supporting a maximum of 16 dies.

DRAM Cache or H.M.B.

Every top-of-the-line SSD aiming to deliver consistent high performance requires a buffer to store its mapping tables (Flash Translation Layer or Look-up table). This enables better random performance and responsiveness.

Being a DRAM-Less SATA SSD, it doesn’t support Host Memory Buffer (HMB) technology.

NAND Flash

Regarding its storage integrated circuits, the 500GB SSD has 2 NAND flash chips labeled “NY240,” which when decoded yield the NANDs “MT29F2T08GELCEJ4-QU:C” from the American manufacturer Micron, model N48R Media Grade. In this case, they are 1Tb (128GiB) dies containing 176 layers of data and a total of 195 gates, resulting in an array efficiency of 90.2%.

In this SSD, each NAND Flash contains 2 dies with 1Tb of density, totaling 256GB per NAND, resulting in a total of 500GB. They communicate with the controller using a bus speed of 262.5 MHz (525 MT/s), which is considerably below what the NANDs are capable of. These N48R dies are capable of operating at 800 MHz (1600 MT/s).

There are several reasons why they might be running so low, such as the manufacturer opting to reduce power consumption and heat. Or even this batch of NAND Flash not being able to pass Micron’s Quality Control at higher frequencies and ends up being sold cheaper or perhaps has a lower endurance as well, which generally results in lower NAND costs, enabling SSDs like this to have a very low price.

SOFTWARE UTILIZED FOR THIS PROJECT

As this is a Silicon Motion controller, we will be using a mass production tool from them, known as MPTools. It’s worth noting that these softwares are NOT provided by the manufacturers but are LEAKED by individuals with access, and posted on Russian or Chinese forums.

For this project, we will use the “SMI SM2259XT2 MPTool FIMN48 V0304A FWV0303B0“, which needs to be compatible with both the controller and the NAND Flash, and this tool allows us to do that.

Before making any modifications, we need to retrieve certain parameters from the SSD to preserve them. These values in the software are a preset from another SSD that may have different parameters. We need to obtain the following parameters:

• Flash IO Driving with it’s subivisons
• Flash Control Driving
• Flash DQS/Data Driving

These parameters use hexadecimal values and must be changed according to the desired speed that we will configure for the SSD.

We also have many more parameters such as:

• Control ODT (On-die Termination)
• Flash ODT (On-die Termination)
• Schmitt Window Trigger

To get these parameters, we need to go to the main screen of MPTools as shown below:

And then we’ll click on “Scan,” which will scan all compatible disks in the system:

After this, the SSD will be shown on port 1 if everything has gone smoothly so far, remembering that it’s not necessary to put the SSD in Safe Mode/ROM Mode yet.

Now we double-click on this Blue Name “Ready (FW: M6CR061, MN48R)“, which, when clicked twice, will open this new screen with SSD information.

Then we should click on both Card mode and CID Settings to see all the parameters that the SSD comes with from the factory.

After noting these parameters, we also see here the speed of the controller and the NAND, which for the sake of a fair comparison, we will leave at these same frequencies.

Applying Configurations

Initially, we should click on the “Edit Config” button in the top right corner, and the default password is “space 2x,” which is literally ” “.

After enabling the options to configure the SSD, let’s start by giving a name to this project. In the “Model Name:” field, we’ll enter the name that the SSD will have. This one was named “SSD SLC Test.”

Next, we’ll add a tag to this new firmware. In the red rectangle number 3, we’ll go to the “Firmware Version:” field and enter whatever we desire. I used “SSD-SLC” as an example.

Next, we arrive at one of the most crucial parts, the section on signal integrity, as all these other parameters are sensitive and must be adjusted precisely.

Let’s start with the top 2 parameters, “Flash Control Driving (hex)” and “Flash DQS/Data Driving (Hex)“. As we saw in the previous images, these parameters come with values of 66 in hexadecimal, so we will keep them. These 2 parameters can be found in the images below:

After configuring these 2, let’s move on to the frequencies. As we can see in the image below, we take these 2 values and set them. The CPU in this software came by default at 500 MHz while the NAND at 250 MHz. The NAND will increase the clock slightly and the CPU will decrease, I will not overclock here for a fair comparison. Next, we’ll leave the Output driving at 03H, which is the signal closest to 04H that the SSD had.

Next, we have the last 3 settings to resolve: Flash ODT, Control ODT, and Schmitt Window. In this case, we apply the values circled in red in each of these parameters in their respective fields.

Good, here we have reached the end of another stage of this procedure. And we begin the next following step, which is the modification of the software. Because by default, this version of MPTools would not support this modification.

Initially, we need to go to the directory of this program in the “UFD_MP” folder located in the root directory.

Inside this folder, we should look for the file named “Setting.set,” which is a configuration file of MPTools. Let’s open it using the Windows Notepad.

With the file open, we’ll make 2 modifications, the first one being in the section “[Function]“, where we have the configuration named “ENFWTAG=1,” which we should change its logical level from 1 to 0.

The other configuration is in the category “[Option]“, where we will add one more extra command line. This command is as follows: “EnSLCMode=1“. So after that, we save the file and reopen the MPTools.

With MPTools open, we can see that in the “Select Procedure” section, there is now an option called “Force SLC Mode“, which we should check. But let’s take it easy because we haven’t finished the modifications yet. There’s no point in trying to write this new firmware to the SSD if it’s still going to operate in its native mode, whether it’s TLC or QLC.

Now we’ve reached the crucial part that enables all these modifications we’ve made to become possible. We need to take the boot and firmware initialization files from a folder within MPTools and place these files in another directory of the program.

First, we return to the default directory of MPTools and open the “Firmware” folder within the software.

Inside this folder, we will find one named “2259,” which refers to the SM2259XT2 controller of this SSD. Within this folder, there should be another folder named “IMN48” along with a configuration file and parameters file.

Once again, we enter this IMN48 folder, where we will encounter numerous files and folders.

Let’s move forward and open the “00” folder, then select all the files and folders within the “00” folder.

We will “copy” (not cut) to the previous folder, the “00” folder, which should look like the following image:

And then we should enter the “XT2” folder and copy this single file inside it called “BootISP2259.bin” to this “00” directory as shown in the next image.

Next, we’ll copy all these files from the folder and paste them into the previous “2259” directory as shown in the following image:

IT IS IMPORTANT TO NOTE THAT THIS PROCEDURE WITH THESE FILES IS FOR THIS KIT OF SM2259XT2 + NANDS N48R.

OTHER SSDS WITH DIFFERENT NANDS FOLLOW THE SAME PROCEDURE, BUT WITH DIFFERENT FOLDER NAMES. THE N48 FOLDERS WILL BE NAMED ACCORDING TO THE NAND MANUFACTURER, AS SHOWN IN THE EXAMPLE BELOW OF AN SSD WITH SM2259XT2 CONTROLLER + KIOXIA BiCS5 NANDs.

P.S.: Some NAND models may not be 100% compatible. So far, I’ve only tested with Intel and Micron NANDs.

Tendo deixado isto claro, agora sim voltamos ao programa MPTools, vamos em Parameter novamente e vamos checar todas as configurações anteriores para ver se ainda estão aplicadas.

If everything is correct, let’s go to the “Test” section next to “Parameter,” which is the program’s main screen. Now we should put the SSD into ROM mode. Let’s close the software again.

HOW MUCH DID THE ENDURANCE INCREASED?

To calculate durability precisely, we need the following information:

Write Amplification Factor

NAND: Program/Erase Cycle

SSD’s Capacity

With these 3 parameters, we can have a basic understanding of TBW (Terabytes Written), but remember that it’s an approximate value. For a more precise calculation, following the JEDEC JESD218A parameters would be necessary, which includes more complicated parameters like Wear-Leveling Efficiency (W.L.E.).

Using this basic calculation with the SSD in its default mode, we see that it has a TBW of 120TB, with a Program/Erase Cycle of these Media Grade N48R NANDs around 900 P.E.C. And how do I know this? I managed to access the datasheet of the NANDs. Taking this into consideration, we can reach the conclusion below, considering the calculation:

120 TB (TBW) = (900 P.E.C. x 0.5 TB)
———————-
X (W.A.F)

X = 3.75 W.A.F.

We see that based on this, the SSD’s WAF in its native form would be quite high, in the range of 3.75, when tested in practical scenarios it was close to 3.8 WAF.

Now, in pSLC mode, the parameters change. The NAND from this Die can withstand up to 60,000 P/E cycles according to the datasheet, and its capacity drops to 0.12TB (120GB). When I randomly tested the SSD, I noticed that its WAF was below 2, which improved significantly.

X TB (TBW) = (60.000 P.E.C. x 0.12 TB)
———————-
1.8 (W.A.F)

X = 4000 TB (TBW)

We see that the TBW has increased drastically, from 120TB(500GB QLC) to 4,000TB (120GB pSLC), which is an increase of over 3333%, more than 3000 percent.

TEST BENCH
– OS: Windows 11 Pro 64-bit (Build: 23H2)
– CPU: Intel Core i7 13700K (5.7GHz all core) (E-cores e Hyper-threading desabled)
– RAM: 2 × 16 GB DDR4-3200MHz CL-16 Netac (c/ XMP)
– Motherboard: MSI Z790-P PRO WIFI D4 (Bios Ver.: 7E06v18)
– GPU: RTX 4060 Galax 1-Click OC (Drivers: 537.xx)
– (OS Drive): SSD Solidigm P44 Pro 2TB (Firmware: 001C)
– DUT SSD: SSD BX500 “SLC-Test” 2TB (Firmware: My custom firmware)
– Chipset Driver Intel Z790: 10.1.19376.8374.
– Windows: Indexing disabled to avoid affecting test results.
– Windows: Windows updates disabled to avoid affecting test results
– Windows: Most Windows applications disabled from running in the background.
– Boot Windows: Clean Image with only Drivers
– Test pSLC Cache: The SSD is cooled by fans to prevent thermal throttling, ensuring it doesn’t interfere with the test results.
– Windows: Antivirus disabled to minimize variation in each round.
– DUT SSDs: Used as a secondary drive, with 0% of space being utilized, and other tests conducted with 50% of space utilized to represent a realistic scenario.
– Quarch PPM QTL1999 – Power consumption test: conducted with three parameters—idle, where the drive is left as a secondary, and after a period of idle, a one-hour write test is performed, and the average power consumption is recorded

CONTRIBUTIONS TO PROJECT LIKE THIS IN THE FUTURE

If you enjoyed this article and would like to see more articles like this, I’ll be leaving a link below where you can contribute directly. In the future, I plan to bring a comparison showing the difference in SLC cache sizes, transforming a QLC or TLC SSD into SLC, among many other topics.

Paypal – [email protected]

CRYSTALDISKMARK

We conducted synthetic sequential and random tests with the following configurations:

Sequential: 2x 1 GiB (Blocks 1 MiB) 8 Queues 1 Thread

Random: 2x 1 GiB (Blocks 4 KiB) 1 Queue 1/2/4/8/16 Threads

In these sequential scenarios, the difference is basically nonexistent because even with the pSLC Cache, the SSD already reaches its maximum bandwidth and the manufacturer’s sequential speeds. Not to mention that this is a quick test; in a more extensive and heavy benchmark, we will see that there will indeed be a difference.

In terms of latency, there was indeed a considerable drop because when the SSD is in Idle, its NANDs, when they start to write or “read,” are in native mode, which would be QLC, and until they are reprogrammed to SLC, they have a certain latency. However, with the SSD in full pSLC mode, this latency is much lower because it always stays in pSLC mode.

The same happens with its random speeds; we can see that there was a greater difference in these benchmarks compared to sequential speed, where the difference was almost negligible.

The same happens at QD1; we can see that in reading, the SSD had an increase of over 16% in its speeds, while in writing, there was a much larger increase of over 30%.

ATTO Disk Benchmark QD1 and QD4

We conducted a test using ATTO to observe the speed of SSDs at various block sizes. In this benchmark, it was configured as follows:

Block sizes: from 512 Bytes to 8 MiB

File size: 256MB

Queue Depth: 1 and 4.

The ATTO Disk Benchmark is a software that performs a sequential speed test with compressed files. Therefore, for a simulation under a data transfer load like in Windows, we typically see block sizes ranging from 128KB to 1MB. Now, we observe that the SSD in pSLC mode outperforms the SSD in its factory mode across all block sizes, which is impressive once again.

The same pattern repeated at queue depth 1, although the difference in some block sizes was slightly lower compared to a queue depth of 4.

3DMark – Storage Benchmark

In this benchmark, various storage-related tests are conducted, including game loading tests for games like Call of Duty Black Ops 4, Overwatch, recording and streaming with OBS of a gameplay at 1080p 60 FPS, game installations, and file transfers of game folders.

In this benchmark focusing more on casual environments, we can see that even here, in a scenario fully representative of reality, there is indeed a performance difference, especially in latency. Although it may not be something entirely noticeable in everyday use in these “lighter” scenarios.

PCMARK 10 – FULL SYSTEM DRIVE BENCHMARK

In this test, the Storage Test tool was used along with the “Full System Drive Benchmark,” which performs light and heavy evaluations on the SSD.

Among these traces, we can observe tests such as:

• Boot Windows 10
• Adobe After Effects: Launching the application until it’s ready for use
• Adobe Illustrator: Launching the application until it’s ready for use
• Adobe Premiere Pro: Launching the application until it’s ready for use
• Adobe Lightroom: Launching the application until it’s ready for use
• Adobe Photoshop: Launching the application until it’s ready for use
• Battlefield V: Loading time until the start menu
• Call of Duty Black Ops 4: Loading time until the start menu
• Overwatch: Loading time until the start menu
• Using Adobe After Effects
• Using Microsoft Excel
• Using Adobe Illustrator
• Using Adobe InDesign
• Using Microsoft PowerPoint
• Using Adobe Photoshop (Intensive use)
• Using Adobe Photoshop (Lighter use)
• Copying 4 ISO files, totaling 20GB, to a secondary disk (Write test)
• Performing the ISO file copy (Read-write test)
• Copying the ISO file to a secondary disk (Read)
• Copying 339 JPEG files (Photos) to the tested disk (Write)
• Creating copies of these JPEG files (Read-write)
• Copying 339 JPEG files (Photos) to another disk (Read)

In this scenario, which is a practical benchmark with a slightly greater focus on writing than 3DMark, as it is more productivity-oriented, it’s possible to notice the practical difference in day-to-day use. The difference was striking, almost twice the performance.

Adobe Premiere Pro 2021

Next, we used Adobe Premiere to measure the average time it takes to open a project of about 16.5GB with 4K resolution, 120Mbps bitrate, and full of effects until it was ready for editing. It’s worth noting that the tested SSD is always used as a secondary drive without the operating system installed, as this could affect the results, leading to inconsistencies.

Here we can see that, as it is more of a scenario of sequential data reading from the project, the difference was almost negligible, just a variation between runs.

WINDOWS BOOT TIME AND GAME LOADING TIME

We compared the SSD with pSLC Cache and in pSLC Mode using the Final Fantasy XIV benchmark.

The same happens with game loading times because the limitation lies in the game’s API, which differs from DirectStorage. This API is not optimized for us to feel a significant difference.

The same can be said for Windows, as although it is a completely new system, it cannot take advantage of features like the one we applied to the SSD.

SLC CACHING

A large part of SSDs on the market currently utilize SLC Caching technology, where a certain percentage of their storage capacity, whether it’s MLC (2 bits per cell), TLC (3 bits per cell), or QLC (4 bits per cell), is used to store only 1 bit per cell. In this case, it’s used as a write and read buffer, where the controller starts writing, and when the buffer is depleted, it writes to the native NAND Flash (MLC/TLC/QLC).

Through IOmeter, we can get an idea of the SLC cache volume of this SSD, as manufacturers often do not provide this information. Based on the tests we conducted, it was found that it has a pSLC cache volume that appears to be dynamic, relatively small, around 45GB. It managed to maintain an average speed of approximately 493MB/s until the end of the buffer, which is a good speed considering it is a SATA SSD.

However, after writing 45GB, it begins to enter the folding process because it allocated all its capacity to work as pSLC. So now we see the true Achilles’ heel of QLC SSDs. Its sustained speed was quite low, averaging around 50 MB/s.

Now, when we transform this SSD into pSLC, we see that it writes to its full capacity of 120GB at an average of 498 MB/s. And to confirm, we wrote up to 500GB to the SSD, and even then, it continued rewriting its capacity more than 4 times at almost 500 MB/s.

As we can see in the graph above, we averaged the SSD’s write speed, combining the speed within the pSLC Cache + Folding + Native. Taking this into account, we see that the difference was striking, almost 10 times higher.

FILE COPY TEST

In this test, the ISO files and CSGO were copied from a RAM Disk to the SSD to see how it performs. The Windows 10 21H1 ISO of 6.25GB (1 file) and the CSGO installation folder of 25.2GB were used.

In a more realistic test like this, we can see that there is no difference because the SLC cache volume of the SSD natively is larger than the size of the tested file.

And even when using a larger folder, it is still smaller than the volume of the SSD’s SLC cache. I don’t test with larger files because I use a RAM Disk, and since I only have “32GB” to make a larger RAM Disk, I would need more RAM.

TEMPERATURE TEST

In this part of the analysis, we will observe the temperature of the SSD during a stress test, where the SSD receives files continuously, to determine if there was any thermal throttling with its internal components that could cause a bottleneck or loss of performance.

The SSD doesn’t even heat up because it’s a low-power consumption SSD, as we’ll see throughout the analysis, and I believe this sensor to be the NAND Flash sensor.

POWER CONSUMPTION AND EFFICIENCY

SSDs, like many other components in our system, have a certain power consumption. The most efficient ones can perform tasks quickly with relatively low power consumption, allowing them to transition back to idle power states where they consume less energy.

SPECIAL THANKS FOR QUARCH FOR SENDING THIS UNIT

In this section of the analysis, we will use the Quarch Programmable Power Module provided by Quarch Solutions (pictured above) to conduct tests and determine how efficient the SSD is. This methodology involves conducting three tests: measuring the maximum power consumption of the SSD, calculating an average power consumption in practical and casual scenarios, and measuring power consumption during idle periods.

This set of tests, especially those related to efficiency and idle power consumption, is important for users who intend to use SSDs in laptops. SSDs spend the vast majority of their time in low-power states (idle), so understanding their power consumption characteristics can significantly impact battery life and overall energy efficiency.

We can see that thanks to this modification, its efficiency has increased dramatically. This occurred because, although the difference in power consumption was not as significant as we will see shortly, the speed in MB/s was extremely high.

Due to the benchmark exceeding the SSD’s 45GB cache by a large margin, it spent a significant portion of the test at a very low speed of less than 55 MB/s, resulting in low efficiency. In pSLC mode, it was able to write twice its capacity at even lower power consumption than in QLC mode, and its bandwidth did not drop at any point. This led to the significant difference in power consumption.

Although this SSD naturally has low power consumption, we do observe a decrease when transforming it into pSLC mode. This occurs because SLC NANDs have only 2 logical levels, meaning the threshold voltage required to allow electrons to flow in the gate channel of each cell is lower since there are fewer levels needed to represent binary 1 or 0. In contrast, QLC NANDs have 16 logical levels, requiring a higher threshold voltage. This explains the reduction in power consumption.

Once again, we see this in the average of both SSDs.

Last but not least, the Idle test, which represents the scenario where the vast majority of SSDs are in everyday use. Here we can see that it had even lower consumption in Idle. Another positive point.

What can we conclude from this?

Once again, I stress the importance of caution with this procedure as it can indeed go wrong if not done correctly. However, we see that the differences are significant in some scenarios while more subtle in others. But now, in terms of durability, the difference is immense!

Pexels

This story was originally published by Undark and is reproduced here as part of the Climate Desk collaboration.

In 2020, the Wisconsin city of Appleton brought No Mow May to the United States. Each spring, by temporarily forgoing mowing and letting residential lawns grow with flowers, residents could support bees and other insect pollinators, proponents claimed. Indeed, they even said the practice was backed by their own published scientific study.

But a little more than two years later, critics turned up severe flaws in the work conducted in Appleton, and the study was formally retracted.

To Sara Stricker, a plant pathologist, it was another sign of anti-grass zealotry gone awry. In spring 2023, she launched a campaign criticizing No Mow May, saying the movement was misguided and delivered little on its intended value. But, she said, people don’t typically want to digest the science behind such a thing: “They just want the quick hashtag and a little sign with a cute little icon in it that makes them feel good.”

Before long, Stricker’s inbox filled with hate mail. For instance, she said she was accused of being in the pocket of industry. Stricker works as a communications and outreach coordinator for the Guelph Turfgrass Institute, outside of Toronto, which has outdoor research fields for the express purposes of studying the green stuff that covers lawns and sports fields and golf fairways. She maintains that she was just trying to start a conversation and acknowledges that her institution draws support from chemical manufactures and the lawn care industry—which is not uncommon in agricultural and horticultural research. But, she said, “None of us are walking around with devil horns saying, ‘Woohaha, I want to kill something today.’ We’re all in it because we like being outside, we like doing sports, we like nature.”

Stricker had not exactly called for a Blitzkrieg on biodiversity. In fact, she encouraged people to plant a wider range of species if they wanted to help support wildlife. (Other experts have proposed similar measures.) In an experiment described in GreenMaster, a trade publication for Canadian golf course managers, her institute found that No Mow May was unlikely to make much of a difference: Over the course of the one-month study, unmowed plots of Kentucky bluegrass did not substantially support more flowers compared to plots mowed at two-and-a-half inches.

But why had experts on grass, at a premier grass-growing institute in North America, come out swinging against something as seemingly innocuous as No Mow May? If something as simple as not mowing for a month doesn’t do much for biodiversity, then what lawn-care practices—from tearing up sod entirely to planting lush wildflower meadows—are supported by data and research?

Lawns represent one of the largest, fastest growing landscapes in the US. These ecosystems—water-hungry, energy-intensive monocultures—extend far beyond the picket fence, including highway medians, cul-de-sacs, and corporate office parks. De facto lawns also exist under solar arrays, on soccer fields, and in city parks. In the US, it’s a landmass that, by some estimates, covers an area about the size of Iowa.

The concept of well-manicured, tightly shorn spaces originated as a status symbol in Europe. Colonizers introduced grasses in an effort to replicate the places they left. By the 18th century, wealthy Americans began tending formal lawns, and much later, particularly in the wake of post-war industrialization and suburban development, the masses followed suit.

But by the 1980s, the lawn began to lose its luster as nonprofit organizations and individuals increasingly called for its eradication. In 1989, the writer Michael Pollan published a landmark essay “Why Mow?” which indelibly described lawns as “nature purged of sex and death.” In some circles, lawns have come to exemplify a consumptive sensibility. “Over the past 50 years, we’ve slowly fallen out of love with lawns,” The Washington Post reported in 2022. “They began to signal waste, disregard, disharmony, homogeneity, gentrification, zombie Boomerism.”

The imperatives to remove lawns span a range of concerns: managing scarce water, building climate resiliency, boosting biodiversity, and even supporting mental health. As J. Leighton Reid, a restoration ecologist at Virginia Tech, put it, any change is probably better than the status quo. “It almost goes without needing evidence to say that a lawn has lower biodiversity than a native ecosystem,” he said. “I mean, that’s just inevitably going to be the case.”

But what actually works? Quick-fix lawn-conversion solutions encounter several problems in practice. For one, contemporary agricultural and horticultural research often focuses on commercial production on a large scale. (Private firms, such as those marketing chemical fertilizers and herbicides, overwhelmingly bankroll R&D in the US.) In contrast, ornamental lawn maintenance falls to practitioners—DIYers and hired professionals alike—who learn from experience and tend to focus on what looks best. Advocates also take matters into their own hands, although no single foolproof method for replacing the lawn and transforming it into a more functional ecosystem exists.

Perhaps the most obvious and most enduring problem rests with grass itself. Grasses generally have a number of reproductive methods and survival mechanisms that make them hard to wipe out. As Laura Jackson, director of the University of Northern Iowa’s Tallgrass Prairie Center, which conducts large-scale prairie restoration, put it: “You can drive a stake through its heart, you can put on the garlic, you can get out the silver cross or whatever, and you’re still going to have that stuff.”

Old habits die hard, too. When it comes to lawns—to walk barefoot on a soft carpet of grass, the sensibility, the institution of lawn care, the whole way of thinking—the practice seems indestructible. And so, some people fire up gas-powered mowers while their neighbors stake No Mow May signs in their lawns gone “wild.” Surely, there has to be a middle ground.

In ecological circles, there is a consensus for why the traditional lawn is perhaps long overdue for disruption, as monoculture grasses do not support much life. Simply put, lawns are deceptively lush: They look green, but are not figuratively “green” in terms of supporting a functional ecosystem.

One study on 159 residential yards in the Washington, DC, area found that, even in the leafy suburbs, an area rich with flowering plants, the landscape filled with nonnative vegetation lacked resources for birds. These nonnative plant species supported fewer insects regularly eaten by one beloved native bird species, the Carolina chickadee. In essence, the suburban landscape doubled as a food desert for the birds.

Hence, the impetus to convert. The best way to do so, however, is unclear. Looking at grasslands in Virginia, for instance, one master’s thesis study suggests that native wildflowers established more consistently when herbicide killed the existing grass completely (rather than seeding the wildflowers with no grass removal). Another PhD student found evidence that the resulting level of native plant diversity in restored grasslands represents a pale facsimile compared to grassland that remained relatively undisturbed.

Reid, the Virginia Tech restoration ecologist who oversaw these studies, said he’s not sure how well the results translate since the studies focused on large-scale post-agricultural production systems. But one takeaway, given that native species fare better with less competition from grasses, is that it’s a case for clearly delineated strips that spatially separate grass from native plantings.

So, it’s not that there is disagreement over the why. Lawns could stand to benefit from greater biodiversity. But it’s the how that should happen—from grass-smothering mulches to sod-busting mechanical tools to chemical herbicides.

The lack of one-size-fits-all prescriptive advice isn’t just a matter of the heterogeneity of the landscape—the average rainfall and climate in Virginia is not the same as, say, southern California—but also of favoritism on an institutional scale. Financial incentives focus on research related to the commercial production of food; Reid, meanwhile says there should be no less imperative to study lawn conversion with an eye towards biodiversity.

That’s unlikely, though. His research, in part, grew out of an effort to create pastures that helped prevent cows from getting sick from eating nonnative grasses in the summer. “If we pitched that as a lawn transformation study,” he said, “I don’t think they would have received it as well as pitching it as a cattle production study.”

Out in the Corn Belt, Jackson, at the Tallgrass Prairie Center, which provides seeds and technical assistance for plantings along public roadsides in Iowa, contends that urban lawns probably amount to a rounding error in terms of, say, providing sustenance for migratory butterflies. “It’s absolutely of no significance whatsoever from the point of view of monarch populations,” she said. “From a political point of view, from the point of view of people caring about monarchs, if that were to translate into political action, that’d be great, that would be significant. But from the point of view of water quality or number of stems of milkweed, it doesn’t move a needle at all.”

Still, the institute’s research has some potentially useful findings. Jackson’s research on seeding milkweed (the host plant for monarch caterpillars) in road ditches with existing vegetation showed abysmal results without applying herbicides, such as Roundup, prior to planting. “You get precisely no plants,” Jackson said. “You use Roundup once, you get a few more, and, if you round it up a couple of times, you get a few more.”

While Jackson added that she personally prefers not to use chemical controls—which are mired in controversy because of their deleterious effects on soil, insect, and human health—the amounts of herbicides used annually in wildflower restoration amount to little when compared against the high volumes sprayed on crops like corn and soybeans. The reality is there’s little prairie left, and it isn’t regenerating on its own. “We’ve done a lot of damage and it’s not going to be easy to undo that damage,” she said.

Jackson championed restoration, and her blanket advice for others was simply to underscore the uncertainty: “A great deal of humility is called for in this business.”

Meanwhile, Linda Chalker-Scott, a professor and extension urban horticulturist at Washington State University, suggested grass-killing herbicides only be used as a last resort. Chalker-Scott has something of a reputation for debunking “horticultural myths.” Her initial skepticism piqued when she was reading a textbook on landscaping and came across some fishy citations: “So that’s when I started thinking, ‘Well, where is this information coming from?’ And then tracing it back and saying, ‘Well, this isn’t science. This is just someone’s anecdotal experience or some guru, self-proclaimed expert telling people what to do.’”

She generally cautions against digging up grass (it removes organic matter and disturbs the soil), and has co-authored one study that found flaws with covering a yard with cardboard and other sheet mulches in order to smother and kill off existing vegetation. In terms of reducing the movement of gasses from the soil into the atmosphere and vice versa, she said that black plastic (which kills unwanted plants by solarization and is used by organic practitioners) is worse than using 12 inches of coarse wood chips. Mulching grasses when dormant, she said, “is the single best way to prepare soil for planting, to get rid of weeds, and to pretty much improve and retain soil tilth.”

Chalker-Scott also made it clear she wasn’t calling for the wholesale removal of lawns. “I don’t hate lawns. I like functional lawns. And sod lawns,” she said, describing those planted with a roll of turf, “are not functional.”

Perhaps there is no one true path carved out of the grass. About 10 years ago, Nancy Shackelford, an ecologist at the University of Victoria, co-authored a paper outlining nine principles for restoration. “We talked a lot in that 2013 paper about resilience and self-sustaining ecosystems, as in, like, ‘Take the humans out of it.’ And I don’t really aspire to that anymore.”

Today, she said, the best laid plans involve restored ecosystems that people want to maintain. “Restoration is always about values. Always,” she said. “So, you can’t take that human piece out of restoration as much as you may try. You can’t pretend it’s just for nature.”

Ecology fundamentally lacks grand unifying theories, and her view reflects a larger shift. When the main restoration group, the Society for Ecological Restoration, updated its primer, for instance, the international conservation organization described restoration taking many shapes. Any step seemed like a good first step, Shackelford said, and, even what might appear like mere symbolic restoration of lawns can be beneficial. Still, she said, the international community sometimes implied that a restored ecosystem could and should take care of itself.

But she was less and less sure: “I actually think humans have a role to play in taking care of all ecosystems. It’s kind of our job at this point.”

LET’S TALK ABOUT OPTIMISM FOR A CHANGE

Democracy and journalism are in crisis mode—and have been for a while. So how about doing something different?

Mother Jones did. We just merged with the Center for Investigative Reporting, bringing the radio show Reveal, the documentary film team CIR Studios, and Mother Jones together as one bigger, bolder investigative journalism nonprofit.

And this is the first time we’re asking you to support the new organization we’re building. In “Less Dreading, More Doing,” we lay it all out for you: why we merged, how we’re stronger together, why we’re optimistic about the work ahead, and why we need to raise the First $500,000 in online donations by June 22.

It won’t be easy. There are many exciting new things to share with you, but spoiler: Wiggle room in our budget is not among them. We can’t afford missing these goals. We need this to be a big one. Falling flat would be utterly devastating right now.

A First $500,000 donation of $500, $50, or $5 would mean the world to us—a signal that you believe in the power of independent investigative reporting like we do. And whether you can pitch in or not, we have a free Strengthen Journalism sticker for you so you can help us spread the word and make the most of this huge moment.

LET’S TALK ABOUT OPTIMISM FOR A CHANGE

Democracy and journalism are in crisis mode—and have been for a while. So how about doing something different?

Mother Jones did. We just merged with the Center for Investigative Reporting, bringing the radio show Reveal, the documentary film team CIR Studios, and Mother Jones together as one bigger, bolder investigative journalism nonprofit.

And this is the first time we’re asking you to support the new organization we’re building. In “Less Dreading, More Doing,” we lay it all out for you: why we merged, how we’re stronger together, why we’re optimistic about the work ahead, and why we need to raise the First $500,000 in online donations by June 22.

It won’t be easy. There are many exciting new things to share with you, but spoiler: Wiggle room in our budget is not among them. We can’t afford missing these goals. We need this to be a big one. Falling flat would be utterly devastating right now.

A First $500,000 donation of $500, $50, or $5 would mean the world to us—a signal that you believe in the power of independent investigative reporting like we do. And whether you can pitch in or not, we have a free Strengthen Journalism sticker for you so you can help us spread the word and make the most of this huge moment.

About us

Our team

Our team is small (just 6), technically experienced, and tight-knit. We're fully remote spanning from San Diego, Seattle, Boston, Uruguay and New York. We still love to spend time in person together, so we try to have a company retreat in cool places a few times a year.

If superconductors — materials that conduct electricity without any resistance — worked at temperatures and pressures close to what we would consider normal, they would be world-changing. They could dramatically amplify power grids, levitate high-speed trains and enable more affordable medical technologies. For more than a century, physicists have tinkered with different compounds and environmental conditions in pursuit of this elusive property, but while success has sometimes been claimed, the reports were always debunked or withdrawn. What makes this challenge so tricky?

In this episode, Siddharth Shanker Saxena, a condensed-matter physicist at the University of Cambridge, gives co-host Janna Levin the details about why high-temperature superconductors remain so stubbornly out of reach..

Listen on Apple Podcasts, Spotify,  TuneIn or your favorite podcasting app, or you can stream it from Quanta.

Transcript

[Theme plays]

JANNA LEVIN: On April 8th, 1911, a Dutch scientist made a chilling discovery. Using a carefully engineered instrument filled with liquid helium, physicist Heike Kamerlingh Onnes delicately lowered the temperature of mercury closer and closer to absolute zero. Suddenly, at an unimaginably cold negative 452 Fahrenheit, the supercooled mercury conducted electricity with perfect efficiency and no energy loss to heat. It was the first superconductor.

I’m Janna Levin, and this is “The Joy of Why,” a podcast from Quanta Magazine, where I take turns with my co-host, Steve Strogatz, exploring the biggest questions in math and science today.

The discovery of superconductivity would win Kamerlingh Onnes the 1913 Nobel Prize in Physics.

[Theme ends]

But more importantly, it marked the start of an unresolved quest for material that maintains perfect conductivity at room temperature. A quest that’s recently seen many claims put forward and then retracted. Today, we ask physicist Siddharth Shanker Saxena, known to his friends as Montu: What makes room-temperature superconductivity so elusive and how might its discovery reshape society?

Montu is a principal research associate in the Cavendish Laboratory at the University of Cambridge, researching superconductors, magnets, graphite and renewable energy applications. He also teaches at Cambridge’s Center for Development Studies, and chairs the Cambridge Central Asia Forum. Welcome to the show, Montu.

SIDDHARTH SHANKER SAXENA: Thank you, Janna. Excellent intro.

LEVIN: Thank you.

SAXENA: And I think a very fitting thing to say: “chilling discovery.”

LEVIN: Yes, our script writers like puns. Before we get into the absolutely amazing and unforeseen phenomena of superconductivity, I want to talk about regular conductivity. We have fundamental elements just on our periodic table, things that are made in the universe, that are conductors, and we call them metals. Can you tell us just very briefly the basics of conductivity?

SAXENA: Sure. So there are different ways of understanding conductivity. Conductivity of water, conductivity of heat, conductivity of electricity. And these properties of elements, it’s one side commonplace, and another side extremely mystical.

Like, when you walk into a room, you flick a switch, and something happens. We can say a circuit is completed. The current starts to flow. It can only flow in a conductor. It’s a conduit through which things can flow. And if it’s not, it stops flow, it’s an insulator.

So you flick that switch, and you allow the electrons to flow all the way to your light bulb, your computer, your fan, or whatever else that is, and it starts to work. And that is conductivity and the material property needed for that to happen is metallicity. Something has to be metallic for it to have basic conducting properties. Of course, there are various caveats and that hopefully we’ll chat about as we go forward.

LEVIN: Yeah, so basically these elements in the periodic table are all lined up on one side because they allow their electrons to do what you’re describing. To run freely through the material, carrying energy with them. And insulators hang on to their electrons and don’t allow it to happen. I use a cloth to hold my kitchen pan…

SAXENA: Yes.

LEVIN: But the pan itself, I rely on allowing the conductivity. Now, how is superconductivity so much different from this naturally occurring phenomenon?

SAXENA: First of all, we often think electrons can freely move in a conductor, but they don’t really move freely. That is the reason why you have your electricity bill. When electrons travel through the metal, they fight their way through other electrons and other defects and so on, and lose energy as they move forward. And that loss is what we pay for. And the process that is functionalized when this is happening is limited by the amount of power or current or number of electrons we put into the conduit.

While superconductivity is literally the superpower of those electrons. They can travel this time freely without being inhibited by other electrons or defects and so on. So once you get the ball rolling, they continue to go until the system cannot sustain that property.

LEVIN: It’s quite amazing, because as you say, energy is money. And so it costs energy every time you waste some of your electricity into heating up your charger, or your phone gets hot, or your computer needs to cool off. That’s all loss. And superconductivity doesn’t have these losses. That’s quite miraculous. Now, what are some examples of superconductors that are actually in application now?

SAXENA: I think the most prominent one that affects our daily — hopefully not our daily lives, but we encounter it  — is a CAT scan or MRI scanner. That’s made of a superconducting coil, which is used to generate magnetic fields. And it’s a primary well-being tool that we have.

Perhaps a more fantastic one is the new era of transport. The maglev trains, for example. Levitating trains that go at speeds similar to aircraft.

There are multiple other smaller aspects. Superconductors allow for us to develop very sensitive devices, which can pick up signals which are minute, which would otherwise not be possible to pick up. And they can be applied from anywhere, in IT or healthcare or mining applications or various other things like that.

LEVIN: Now, why can’t I have these fantastic gadgets in my home and save on my energy bills?

SAXENA: That’s literally a million-dollar question, or multimillion-dollar question.

[Both laugh]

SAXENA: And two interesting aspects of this. One is the fact that superconductivity itself, as you started by telling us in the very beginning of the podcast, was when that liquid-filled special device used by Kamerlingh Onnes was used to cool down mercury.

And similarly, we have to cool down most of these metals to a superconducting state. So basically you start with a normal metal, where electrons are fighting thermal barriers, but then you take it through a transition at which it starts to work in the super way. So now, for us to have that property, we need that special device to keep superconductivity maintained at those temperatures. And that’s what has inhibited its general use.

And that is in itself a very interesting problem that, as human beings, we have conquered the heat. We sit in a car, we turn the switch on, we put a newborn baby a foot away from thousands of degrees of heat and not even think about it. At the same time, we don’t have the same kind of control yet over cooling. Superconductivity happens at lower temperatures. However, our ways of cooling still need developing.

LEVIN: Right. So now, after Kamerlingh Onnes’ initial discovery of mercury’s superconductive properties, where did the research go? Did it immediately go to “let’s try to do this at warmer temperatures”? Because, as you said, he had to supercool it to a staggeringly cold temperature to observe this phenomenon.

SAXENA: So, if you look at the history of the development of superconductivity, we’ll find two approaches. And that happens with most physical phenomena which is discovered. The instinct of a physicist, or chemist or basic scientist, is to understand why is it happening at all? Just imagine it was happening in that era when it happened. This was the first time it’s ever been seen. One wasn’t thinking necessarily about using it or controlling it, just to understand what it is. Why is it happening at all? Or is it even true? Can we find other examples of it? Is it just a flaw in the measurement?

The early period went on establishing the phenomenon itself. Understanding its parameters. When it happens, where it happens, how it can happen. And then the next step: Can it happen in things other than mercury? And that’s when the juices start to flow. You start to think we have different materials in which this can happen. Can this happen in alloys? Can it happen in compounds? And as that happens, you start to talk to engineers, you start to talk to people in other fields, and their input is what make you realize that it can possibly be used for other things. And the two efforts started to go in parallel but separate ways: the application of superconductivity versus understanding of superconductivity.

And there’s another very interesting lever here which we haven’t mentioned. It’s not only temperature, it’s also pressure. In fact, room-temperature superconductivity is already discovered. It’s not something that’s elusive anymore, except that it happens at very high pressures. Mikhail Eremets and colleagues in 2015 already produced near-room temperature superconductivity. And now there are various examples of it.

LEVIN: Now let’s talk about why it happens, because it’s extremely different from the ordinary conductivity of the naturally occurring elements. You might imagine if I supercooled something that the electrons’ motions could be inhibited, and that actually conductivity would therefore drop to zero. And so this is really a very different and surprising phenomenon. Can you talk us through the Nobel Prize-winning work of John Bardeen, Leon Cooper and John Schrieffer and their theory of superconductivity?

SAXENA: That was a landmark moment, not just in understanding superconductivity, but understanding quantum phenomena altogether. So BCS theory: The basic essence of their theory is that they were able to explain that the electrons don’t travel alone like they do in a metal. They form a coherent state, what we call the Cooper pair. The two electrons come together and, if you want to use an analogy, they are the bully on the street. They’re able to push everyone else away so they can move effortlessly in this maze of other electrons and so on.

The formation of the Cooper pair was something that was only describable and conceptually tangible through the work of Bardeen, Cooper and Schrieffer in the 1950s. And, in fact, Cooper was the one who understands that when two electrons are able to come together in this way, they form a coherent state. It’s a lot of electrons which become coherent, but they can be described as a set of pairs, or a condensate that forms.

LEVIN: So fascinating because electrons, of course, if you’re in the quantum world or you’ve studied quantum theory, are notorious for not wanting to pair up.

SAXENA: Yep.

LEVIN: And the Pauli exclusion principle famously says that there’s, in fact, a quantum pressure associated with trying to jam electrons together. They don’t like it.

So this phenomenon that they discovered is extremely counterintuitive: the collaboration with the lattice of ions left behind when the electrons start to move away. When they pair up, now they have the opposite phenomenon where they want to bunch together. So you don’t just get one pair. You now have an encouraging accumulation of these electrons and hence this runaway phenomenon of superconductivity. It’s absolutely counterintuitive and fascinating. And that theory has held up well over the years?

SAXENA: Yeah. One picture that the theory builds in your mind, is that of a lattice and the vibrations of lattice, what we call phonons. And so you can think in terms of, if this boat or ship is moving in water and it creates a wake behind it, while it looks like it’s pushing the water back while moving forward, it creates that small wake where things can get trapped, which is attractive potential.

So it held up quite well for a vast majority of superconductors. And one of the important reasons for that is that the electron has the other property, the spin. And spin is what gives it magnetism. So everything we have discussed now is the charge of an electron. And so while we can talk of electron and its Pauli exclusion principle, we now turn to Bohr and spin, and talk about the spin itself and how opposite spin electrons can attract each other. And a Cooper pair has a spin-up/spin-down configuration in the BCS theory description.

LEVIN: Now why does it require — so far, hopefully not forever — this supercooling?

SAXENA: So, the main barrier that one has to overcome is the thermal one. So you can think about how the heat inhibits that coherence to form. It keeps rattling things before they can come into a coherent state. So we need to get to the temperature in which the electrons are interacting with the other electrons rather than other vibrations.

So two things. We have to have a cooperation between the electrons themselves, and the cooperation between those electrons and the lattice. And this lattice is driven by heat. And the lattice continues to be energetically unfavorable till you get to the temperature at which it starts to become cooperative. So we need the thermal side to decrease for the quantum state to form.

LEVIN: So that suggests that achieving room-temperature superconductivity is going to be hard. And as you said, we’ve seen it only under equally extreme conditions of high pressures. Do you think it’s hopeless, our attempts to achieve room-temperature superconductivity without the high pressures?

SAXENA: Can we hold a cube of ice over open flame and it doesn’t melt? When we talk about room-temperature superconductivity, that’s what we’re trying to achieve here in terms of an analogy. And so if that happens, usefulness aside, if can you hold ice cube over fire, it’s just shockingly amazing and mind-blowing.

On the other hand, there are ways in which we can protect the state because we know that when we chemically make compounds, when we change atoms, we change the internal pressure in a material. We increase or decrease the pressure between bonds, between elements, and that changes the properties. So by playing with the material, we are able to create the same conditions as a pressure does.

LEVIN: We’ll be right back.

[Break for ad insertion]

LEVIN: Welcome back to “The Joy of Why.”

So people have really been investigating synthetic materials as a way to make a big breakthrough. Now there have also been a number of very high-profile declarations of success, published in journals as prestigious as Nature, of new materials or synthetic materials that show room-temperature superconductivity. But then they were retracted. So what’s going on with these kind of controversial claims?

SAXENA: Yes, a couple of things to say about that. The first one, I want to reiterate that it’s not the room-temperature superconductivity which is in question. The claims are about ambient pressure or near-ambient pressure, not about room temperature at all. The question is, can it be achieved in ambient conditions?

So first of all, these high-pressure experiments are extremely difficult. I am a high-pressure scientist myself. So I know the difficulty and the challenge that comes with it. At the same time, it is the most productive in this area. And combined with the difficulty of experimental endeavor, and the promise it has for applicability, we feel that people have been rushing to give big answers. And the whole sociology of academic endeavor falling in sync with this very difficult to achieve but very promising area, has produced very interesting, and in many ways destructive, tendencies. I would say that the vast majority of superconductivity researchers are very careful.

There is a very fun word. We call them USOs. So, periodically, just like what happens with UFOs, all of a sudden, somebody declares there’s big news for a flash, and then it’s gone. Many journals have gone into the commercial publishing world rather than academic publishing. So it’s a nexus of several things which have led to these powerful claims.

But as a scientist, it just whets my appetite, rather than discourage. What I am worried about is that the public interest, and thus political interest, in this wonderful phenomenon — extremely unusual, extremely promising — could get damaged if we rely on those USOs.

LEVIN: So can you give us a little bit of intuition briefly about what these synthetic materials are, how they’re arranged out of what we consider to be ordinary atoms? Are they sheets of different fundamental elements? How are they constructed?

SAXENA: Sure. They’re like houses, they come in different shapes and sizes. And similar to houses, they have different who-can-live-in-it, and how comfortable they are. Superconductors actually come in all shapes and forms. So they can come as cubic material. So think of a structure of a rock salt, you can have that cubic material that can also be superconducting. Even gold, you can look at that.

But then you can have graphite, for instance, layers of carbon [inaudible], but also other materials, and they are very weakly bound together. So what makes graphite interesting for scientists, the reason we can write with pencil, is that those layers just fall apart. They just come onto the paper. We can write with it.

That means we can put other things in between, and that can change interactions between the sheets themselves and also between what we put inside those sheets. So these are so-called 2D materials. Before the advent of the recent room temperature, high-pressure superconductors, there was a whole family called high-TC superconductors. And these were mostly two-dimensional materials. And two-dimensionality, until today, has played a very important role in finding superconductivity.

The analogy that I gave you earlier: the boat is moving, and that there’s a small wake which can trap things in it. But if you imagine that all starts of wobble, i.e., 3D. You increase the dimensionality of your motion. Then it becomes less likely that things can get attracted and trapped and become coherent.

You can have another simpler analogy, that if you take a vat of treacle or honey and you throw a pebble in it, it’ll deform. But since it’s viscous, it cannot propagate. So it starts to retract and everything near it can get attracted and coherently bound. But if you take the same thing and start to wobble it in three dimensions, it’s less likely to happen.

So two-dimensionality has played a very important role. And we don’t know if this is true for these high-pressure phases yet. They look more to be 3D. One of the avenues of research, is can we achieve that kind of condition in 2D materials? And that’s where most likely we’re going to see zero pressure room temp superconductivity. It’s likely to come from lower dimensional materials.

LEVIN: A lot of the scientists that I most admire are unafraid of failure — their curiosity outweighs their fear of failure. You mentioned that you’re in this very challenging area of this high-pressure superconductivity. Give us a little glimpse into your process. What is a working day like in your laboratory?

SAXENA: So my high-pressure lab, it’s also a low-temperature lab. High pressures, because it’s the most difficult part of the process, it gets highlighted. In my lab, first thing that happens is making the material, or working with someone who makes the material. And identifying that a material is what it is.

See, this is where the drama is. Even the discussion that we just had about the previous cases, which were overblown, we don’t fully agree because we don’t know what the materials are. So in my lab, we spend most of our time, sometimes more than a year even, trying to hone down on the material is what it is. And that’s where the resilience of the researchers, the Ph.D. students and myself comes in. Because it could turn out that, after a year, this is not something we want to measure anymore. And we look for new systems all the time in which this thing can happen.

And then, we try to cut the sample down to size, literally. Because pressure requires a very small volume. So, a few tens of microns thickness, and a hundred-micron width, and so on, is where the highest pressures experiments happen. So then it comes down to attaching those famous electrodes to the samples, and then trying to put them in the pressure cells. And then you come to the cryostat and you chill it — literally the opposite of watching water boil. When you make a material chemically, it has certain set of atoms arranged in a particular way, and certain properties. So why pressure is a really important tool is that when you compress something uniformly — hydrostatically, in technical words — you change those atomic distances.

And so effectively at each pressure, you have a new material. And at each pressure point, you can then measure all those properties. The nice thing about this way of searching is that you can have very fine steps. You can increase the pressure, you can release the pressure, and you can keep searching for new and new states.

LEVIN: Would it be fair to say that you’re therefore, as you increase the pressure, looking at phase transitions?

SAXENA: We are looking at both. So we’re looking at change and uniform change in properties, and hoping for that very drastic change, and that will be the phase transition, yes.

LEVIN: If one of these USOs were to become a bona fide reproducible synthetic material that shows superconductivity at ordinary conventional conditions that you can find in someone’s apartment, this would doubtless be incredibly lucrative. And that’s because it’s going to have some very serious implications socially. Can you talk me through some of the societal ramifications of succeeding here?

SAXENA: It may sound biased coming from me, being a superconductivity researcher, but it’s obvious that the ramifications of superconductivity being available in ambient conditions are going to be transformative for all things around us. People keep talking about AI all the time. That’s nothing. Here we are talking about energy efficiencies, which are going to transform how we travel, how we communicate. We’re talking about data farms. We’re talking about power grids. We’re talking about ships, planes, everything. Just like all of those things have a conductor in it, it’ll have a superconductor in it.

LEVIN: Presumably, they’re not going to immediately be available freely across the globe. Presumably, there will be some countries that have faster access because of their investment?

SAXENA: That is absolutely correct. The infrastructure, both industrial and scientific, meaning laboratory and the manufacturing, certainly is available only in the global north. And there is obviously a great potential for it in some of the middle-income countries. But that still leaves a huge part of globe out, which cannot produce or sustain production of these kind of materials.

They tend to be in geographies of Eurasia, Australia and some other parts of the globe only. But I would say that it’s not only about where they occur. For example, South Korea, which has zero iron ore deposits, but is one of the biggest ship makers and steelmakers. So it’s more about how you’re plugged into a system, and supply and value chain. It’s not only about having natural resources. And then you have countries of central Asia having vast resources, but they don’t produce any of them in the way they can be used. So that it’s more than just having it. The discovery itself is not going to do it.

LEVIN: Right, we have to remind ourselves that we really are in this together. We have to get along to make the future work. You clearly believe that superconductivity is one day possible at more ordinary conditions. Do you think we’re close?

SAXENA: One way of talking about this: Does a UFO sighting mean that we’re close to finding other aliens? Is it a symptom of that or not? One can think in that way. But from within the field, our progress has been absolutely impressive in the last 20 years. And that has to do with our ability to make materials, and the kind of experiments we can do now using high pressure and high fields and other conditions. So the stage is now set better than ever before to be able to find these materials, or we have found these materials, so how to engineer these to be in ambient conditions.

LEVIN: Fascinating. Now, you have a background not only in physics but also anthropology and history. How do these other subjects inform your both scientific approach and your bigger picture?

SAXENA: I often describe myself as undisciplined because I’m not bound by any particular discipline. I’ve been lucky to have had the chance to study, interact and teach and work in all these areas. But fundamentally, what I’m looking at is the same thing.

For instance, if you have a bottle of water, if you hold that bottle in your hand, reflect for a second that the hand, the bottle and the water are all made of exactly the same elements. But these three things have nothing in common, absolutely nothing. Live, dead, transparent, translucent, liquid, solid, whatever you call it, there’s nothing in common. And this is where we come in, the modern-day scientists, to say: It’s not only about what things are made of. It’s a question of how things interact. And understanding the interaction gives rise to new properties. And they can be quite counterintuitive, but we must find probes to measure and understand those properties. But also find parameters to tune them, like pressure, field temperature and so on.

Human beings are like those particles. And just imagine the pressure-temperature axis in which a human being can survive. It’s extremely narrow. And yet, our interactions with each other and our interactions with the environment produce entirely different cultures, languages, ways of thinking, and science itself.

LEVIN: There’s a question we here at “The Joy of Why” like to ask our guests, and that is, what about your research brings you joy?

[Theme plays]

SAXENA: The discovery. And doing it with the team, with people, the colleagues. That moment of togetherness when we find something together. Discovering something alone is not something that I have found fun. I found that being part of a group who are working towards some aim, going through trials and tribulations, and then the discovery happens. And it’s a reward that’s shared, without having to slice it up.

LEVIN: No, it’s beautiful. At the end of the day, science is a human endeavor. We’ve been speaking with physicist Siddharth Shankar Saxena, that’s Montu to us, on superconductors and their potential to change the world. Thanks so much for joining us, Montu.

SAXENA: Thank you. It’s been a pleasure.

LEVIN: Pleasure to have you here.

“The Joy of Why” is a podcast from Quanta Magazine, an editorially independent publication supported by the Simons Foundation. Funding decisions by the Simons Foundation have no influence on the selection of topics, guests or other editorial decisions in this podcast or in Quanta Magazine.

“The Joy of Why” is produced by PRX Productions; the production team is Caitlin Faulds, Livia Brock, Genevieve Sponsler and Merritt Jacob. The executive producer of PRX Productions is Jocelyn Gonzales. Morgan Church and Edwin Ochoa provided additional assistance.

From Quanta Magazine, John Rennie and Thomas Lin provided editorial guidance, with support from Matt Carlstrom, Samuel Velasco, Nona Griffin, Arleen Santana and Madison Goldberg.

Our theme music is from APM Music. Julian Lin came up with the podcast name. The episode art is by Peter Greenwood and our logo is by Jaki King and Kristina Armitage. Special thanks to the Columbia Journalism School and Burt Odom-Reed at the Cornell Broadcast Studios

I’m your host, Janna Levin. If you have any questions or comments for us, please email us at [email protected]. Thanks for listening.

The Iberian lynx (Lynx pardinus) continues along its upward path, although it is still at risk of extinction. The last census in 2023 shows that the species has doubled its population in the last three years and has reached 2,021 individuals, with 1,299 adults or subadults and 722 cubs. Despite the good data, 750 breeding females would be needed to classify the species in a favorable conservation status, and in this latest count only 406 have been detected. According to Spain’s Ministry for the Ecological Transition, they are “gradually” getting closer to the necessary number, but are still falling short.

Captive breeding centers have played an essential role in this recovery. From 2011 to 2023, 372 lynxes born in the four existing centers have been released. The population has been expanding, and last year the reproduction of the species was verified in 14 population centers, in addition to the stable presence in new areas of the Spanish region of Murcia, and the provinces of Albacete, Badajoz, Toledo and Ciudad Real. Most of the specimens, 1,731 or 85% of the total, live in Spain and the rest, 291, are in Portugal. The stable populations are located in four Spanish regions: Andalusia with 755 specimens (43.6% of the Spanish population), closely followed by Castilla-La Mancha with 715 lynxes (41.3%), Extremadura where 253 specimens were located, and Murcia with seven.

The ministry believes that the data allow for a degree of optimism, because the feline’s trend is positive and has continued on an upward trend since 2015, the year in which the International Union for Conservation of Nature (IUCN) lowered its threat level. The species went from being “critically endangered” to simply “endangered.” It was a fundamental step for the survival of an animal that was close to extinction in 2002, when only 94 specimens remained in Andalusia. Captive breeding programs with significant European investment have allowed the creation of different nuclei, and have managed to reverse the situation. This increase now means that lynxes are approaching inhabited places, as happened in late March, when a rancher found four lynxes born in his haystack in Menasalbas, municipality of Toledo.

“We are at an average growth of 20%, a trend that has been maintained due to the creation of three new nuclei,” says Ramón Pérez de Ayala, a member of World Wildlife Fund (WWF) and a specialist in the species. To obtain the number of breeding females, it is necessary, according to their calculations, to create another five new areas with lynxes. “There were populations that grew a lot, up to 30%, but then they stabilized,” he explains. This occurred in the mountains of Toledo, one of the biggest success stories of reintroduction. “There, the first place where they became active was practically saturated, but the lynxes have moved to a neighboring area, which has kept up the growth,” he notes. The same situation has taken place in other areas such as Guarrizas (Jaén) and in Portugal, in the Guadiana valley area.

A reproductive female needs a territory of about 500 hectares, although it all depends on the amount of food available: the more food there is, the less space is needed. The rabbit is the main component of their diet and there are places where the rabbit population cannot recover, affected mainly by a deadly hemorrhagic disease. Depending on the areas, the drop is between 30% and 87% in a decade, indicates Pérez de Ayala. In this boom situation, the feline’s biggest enemy is road accidents, which has become its main cause of mortality. “In 2023 there were 144 deaths due to this cause, 7.1% of the population, and we cannot forget poaching, which goes unnoticed.”

Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition

Computer Architecture

Programming Languages

The Computer Business; Miscellaneous

Hardware

Operating Systems

Linux

What does it take to build a little 68000-based protoboard computer, and get it running Linux? In my case, about three weeks of spare time, plenty of coffee, and a strong dose of stubborness. After banging my head against the wall with problems ranging from the inductance of pushbutton switches to memory leaks in the C standard library, it finally works! I’ve built several other DIY computer systems before, but never took their software beyond simple assembly language programs. Having a full-fledged multitasking OS running on this ugly pile of chips and wires is a thrill, and opens up all kinds of interesting new possibilities. I’ve named this plucky little machine 68 Katy.

 
Hardware

Here’s a look at the final version of the hardware. It took about a week to assemble and wire up all the parts on a solderless breadboard. The heart of the system is a Motorola 68008 CPU, a low-cost variant of the more common 68000, with fewer address pins and an 8-bit data bus. The CPU has 20 address pins, allowing for 1 MB of total address space. It’s paired with a 512K 8-bit SRAM, and a 512K Flash ROM (of which 480K is addressable – the remaining 32K is memory-mapped I/O devices).

The standard 68000 CPU has a 16-bit data bus, so it normally requires at least two 8-bit RAM chips and two 8-bit ROM chips. The 68008 requires fewer memory chips thanks to its 8-bit data bus, but the trade-off is that memory bandwidth is only half that of the 68000. Neither chip has any on-board cache, so half the memory bandwidth leads to roughly half the performance. My 68008 runs at 2 MHz (it was unstable when tested at 4 MHz), providing similar performance to a 1 MHz 68000. That’s pretty slow, even in comparison to 68000 systems from the early 1980’s, which were typically 8 MHz or faster.

An FT245 USB-to-FIFO module provides a communication link to another computer. On the external PC, it appears as a virtual serial port. Windows Hyperterm or another similar terminal program can be used to communicate with it, like an old VT100 terminal. On the 68 Katy side, the FT245 appears as a byte-wide I/O register mapped into the CPU’s address space. Reading from its address fetches the next incoming byte from the PC, and writing to the address sends a byte out to the PC. The FT245 has an internal 256-byte buffer, which helps smooth out the communication. When there’s an incoming byte waiting in the buffer, it triggers a CPU interrupt.

A 555 timer provides the only other interrupt source, generating a regular series of CPU interrupts at roughly 100 Hz. The initial version of the hardware had no timer interrupt, but I later discovered it was essential in order to get Linux working correctly.

The protoboard has eight LEDs for debugging, which are driven from a memory-mapped 74LS377 register. The rest of the protoboard is filled with assorted 7400 series parts and one PAL, which are used for address decoding, interrupt arbitration, and other basic glue logic.

Schematics? Forget it. Everything was built incrementally, one wire at a time, while staring at chip datasheets. It’s an organic creation.

 
Software

Once the hardware build was done, I began writing some simple test programs in 68K assembly language. Wichit Sirichote’s zBug monitor provided a good starting point for my own ROM-based monitor/bootloader. Using the monitor program, I was able to load other programs in binary or Motorola S-record format over the FT245 link, store them in RAM, and execute them. I was even able to get Lee Davison’s ehBASIC for 68000 working, which provided a few hours of fun.

One feature I could have added to the monitor program, but didn’t, was the ability to reprogram the Flash ROM. The ROM chip has a read/write input pin just like an SRAM, but writing to the Flash ROM is more complicated. The CPU needs to first write a magic sequence of bytes in order to unlock the ROM. Then it needs to write more magic bytes to tell the ROM which blocks to erase, followed by the new bytes to be written. Finally, it must poll the output of the ROM to learn when the erase and reprogram sequence is complete.

The monitor program could have updated itself, or any other data stored in ROM, by copying itself to RAM, then running from RAM while saving new data to Flash ROM. But I was lazy and never implemented that feature, so I had to physically pull the ROM chip from the protoboard and place it in an external EPROM programmer whenever I made a change – about 100 times over the course of the project. Ugh.

 
Linux

Inspired by a similar project, I decided that a simple monitor program and BASIC weren’t interesting enough, and I needed to run Linux on this hardware. It sounded interesting and exciting, but I really had no idea where to begin. I had plenty of experience as a Linux user (as well as other UNIX versions), but I knew nothing about how the kernel worked, or how to build it from source code, or to port it to new hardware. So the real adventure began there.

The first challenge was to learn how to build a Linux image for an existing machine. It seemed simple enough in theory – just download the source code from kernel.org or some other distribution tree, and compile it. Reality was more complicated, and there were many details that confused me, and build problems I was powerless to solve. It wasn’t easy, and I discussed the process in much more detail in this earlier post.

I chose to use uClinux, a Linux distribution designed for microcontrollers and other low-end hardware, particularly CPUs without an MMU and that can’t support virtual memory. Then I chose a very old version of uClinux, based on kernel 2.0.39, that dates all the way back to 2001! I configured it to disable nearly every single optional feature, including all networking support. This ancient code was my best hope for getting a Linux that would actually run in 512K of ROM and 512K of RAM.

Starting with the uClinux configuration for another 68000-based system, I updated the code to reflect the 68 Katy memory map, changed the system initialization code, and added a driver for the FT245. Describing those steps makes them sound simple, and they might have been for someone more experienced with Linux, but for me it was a challenge just to identify which files and functions needed to modified. Google wasn’t very helpful, since I was working with such an old version of the kernel, and the resources I found on building/porting Linux mostly weren’t applicable. I primarily relied on the Linux grep command to search through the thousands of kernel source files for strings of interest, then stared at the code until I could understand how it worked.

After about a week, I had something I was ready to test. And it worked, at least a little bit! It showed the first few lines of kernel output, but hung at “calibrating delay loop”. Aha, I needed a timer interrupt. Of course! I added a 555 timer and some extra interrupt logic, and was ready to try again.

The second attempt got further into the boot process, but failed to mount the memory-resident root filesystem. I was stumped for a while. After looking more carefully, I discovered that my linker script was mapping the root filesystem and BSS to the same address in RAM, and the earily initialization code was overwriting the filesystem with zeroes. And more fundamentally, I discovered that it wasn’t possible to fit all of Linux in 512K RAM, including the kernel code, static data, root filesystem, and dynamically allocated memory. Something had to be moved to ROM, or it was never going to work.

In the third attempt, I moved the root filesystem image to ROM, freeing up about 150K of RAM. And this kind of, almost worked! It booted, mounted the filesystem, and seemed to be working OK, but then suddenly it would land back at the monitor program prompt. Huh? I eventually tracked this one to the FT245 driver I’d written. I only implemented the minimal set of required driver functions, and the other optional functions were NULL entries in a function table somewhere. One of the functions I thought was optional proved to be required. When the kernel called it, it used a NULL function pointer, causing a jump to address 0, restarting my monitor program.

The fourth attempt was better. It spawned the init process, and ran the startup script, but died with out-of-memory errors before it completed. At the time, 68 Katy’s memory map was 256K ROM, 256K I/O devices, and 512K RAM. By shrinking the I/O space to 32K, I was able to increase the usable ROM to 480K, providing enough space to store the root filesystem image and the kernel code itself! This freed up another 251K of RAM.

The fifth attempt actually booted to a shell prompt! Now it was executing the kernel code directly from ROM. I was able to run a few commands, like ls and cat, but then the system would run out of memory and die. As I investigated, it looked like memory allocated from malloc() and do_mmap() was never beeing freed. Was this some kind of free list allocator I didn’t understand? No. It turns out I’d made a typo in a function called is_in_rom(), adding too many zeroes, causing the memory allocator to think all addresses were in ROM and so didn’t need to be freed. Then after fixing that, I discovered a small memory leak in the C library setenv() function. I never did solve that one, but instead just modified the programs that used it to avoid calling it.

My debugging method was primitive: lots of printk and printf statements sprinkled everywhere. Then pull the ROM chip, reprogram it in an external EPROM programmer, replace it in the protoboard, and try again.

The sixth attempt finally worked. Two weeks after beginning my experiments with Linux, I had a working system! Here’s a screenshot of the boot sequence:

Watch the video for more details. I’m using a shell called sash, which has a few dozen common shell commands (like ls and cat) built directly into it. The root filesystem in ROM is read-only, and a small read-write filesystem is created in a RAM disk. The system supports multitasking, so it can run an LED blinker program in the background while still working in the Linux shell. It even has vi, and Colossal Cave Adventure!

It was an interesting journey. The Linux kernel still seems big and unwiedly to me, but no longer seems so scary as it did initially. It’s just an especially big program, and most of its pieces aren’t too difficult to understand if you open up the relevant source files and start reading.

 
Memory Requirements

So how much memory does it require to run a super-minimal uClinux system, with an old 2.0 kernel? If you follow my example and put as much as possible in ROM, it needs about 343K of ROM and 312K of RAM, or just 628K of RAM if you’ve got a bootloader that can fill RAM from some external source. My 68 Katy system is slightly heavier than that due to including vi and Adventure, but not by much. Here’s the breakdown:

ROM

• kernel code and read-only data (.text and .rodata segments) – 251K
• kernel initialized static data (.data segment) – 27K
• root filesystem – 189K

RAM

• kernel static data (.data and .bss segments) – 84K
• kernel dynamic allocations during boot-up – 104K
• RAM disk – 64K
• init and shell process allocations – 58K
• stack and exception vectors – 2K

 
Problems

The kernel always measures the CPU at 0.81 bogomips, regardless of the clock crystal I use. The 555 timer interrupt is independent of the CPU clock, so with a faster clock the bogomips calculation should measure more executions of the busy loop per timer interrupt. I’m not sure why it doesn’t, but it means any real-time calculations will be off.

The display in vi acts weird. Some lines appear prefixed with a capital H, and stray Unicode characters appear here and there. At first I thought this was a hardware bug, and I’m still not certain it isn’t. But I think it’s probably an issue with the way my terminal program (Tera Term) handles the ANSI escape sequences sent by vi. I tried all the different terminal settings available in Tera Term, and also tried a different terminal program, all with the same result.

 
What’s Next?

This 68008 system on a protoboard was intended to be only an experiment and proof-of-concept for the real 68 Katy, which I had planned to build on a custom PCB with a full 68000 CPU, a CPLD for glue logic, more RAM, an SD card, and ethernet. But this experiment was perhaps a bit too successful, and now I’m wondering if it really makes sense to go to the effort of building the “real” system if it’s going to be essentially the same thing, only faster. Of course the SD card and ethernet will add some interesting new elements, so maybe it’s fine. I probably need to sleep on the question for a few days.

One way of adding more spice to the next iteration of 68 Katy would be to include video output, so it could directly drive a monitor instead of being controlled through a serial terminal. I’ve done that once before, with BMOW 1, which had VGA output. It mostly worked, although the arbitration for video memory between the display hardware and the CPU was clunky and produced visible display artifacts. To take things further, I could even aim for DVI or HDMI video output, since VGA is a slowly dying standard.

The smart move is probably to stick with my original plan. Lots of extra features are cool, but also have a way of killing a project. I’d rather have something with 10 features and that works, than something with 20 features that I never finished or that collapsed under the weight of its complexity. But until then, excuse me while I go play some more Colossal Cave…

 
Files
The source code for my 68 Katy port of uClinux is available for download, as well as the toolchain I used to build it, the monitor/bootloader source, and a preconfigured VirtualBox machine image of Ubuntu 12.04 to host it all. Grab the files here. Have fun!

Credit: Pixabay/CC0 Public Domain

Stroke is the leading cause of disability worldwide and the second leading cause of death, but the right early intervention can prevent severe consequences. A new study led by investigators from Brigham and Women's Hospital and collaborators describes how the team developed a new test by combining blood-based biomarkers with a clinical score to identify patients experiencing large vessel occlusion (LVO) stroke with high accuracy.

The results are published in the journal Stroke: Vascular and Interventional Neurology.

"We have developed a game-changing, accessible tool that could help ensure that more people suffering from stroke are in the right place at the right time to receive critical, life-restoring care," said senior author Joshua Bernstock, MD, Ph.D., MPH, a clinical fellow in the Department of Neurosurgery at Brigham and Women's Hospital.

Most strokes are ischemic, in which blood flow to the brain is obstructed. LVO strokes are an aggressive type of ischemic stroke that occurs when an obstruction occurs in a major artery in the brain. When blood supply to the brain is compromised, the lack of oxygen and nutrients causes brain cells to die within minutes. LVO strokes are major medical emergencies and require the swift treatment with mechanical thrombectomy, a surgical procedure that retrieves the blockage.

"Mechanical thrombectomy has allowed people that otherwise would have died or become significantly disabled be completely restored, as if their stroke never happened," said Bernstock. "The earlier this intervention is enacted, the better the patient's outcome is going to be. This exciting new technology has the potential to allow more people globally to get this treatment faster."

The research team had previously targeted two specific proteins found in capillary blood, one called glial fibrillary acidic protein (GFAP), which is also associated with brain bleeds and traumatic brain injury; and one called D-dimer.

In this study, they demonstrated that the levels of these blood-based biomarkers combined with field assessment stroke triage for emergency destination (FAST-ED) scores could identify LVO ischemic strokes while ruling out other conditions such as bleeding in the brain. Brain bleeds cause similar symptoms to LVO stroke, making them hard to distinguish from one another in the field, yet treatment for each is vastly different.

In this prospective, observational diagnostic accuracy study, the researchers looked at data from a cohort of 323 patients coded for stroke in Florida between May 2021 and August 2022. They found that combining the levels of the biomarkers GFAP and D-dimer with FAST-ED data less than six hours from the onset of symptoms allowed the test to detect LVO strokes with 93% specificity and 81% sensitivity. Other findings included that the test ruled out all patients with brain bleeds, signaling that the technology may ultimately also be employed to detect intracerebral hemorrhage in the field.

Bernstock's team also sees promising potential future use of this accessible diagnostic tool in low- and middle-income countries, where advanced imaging is not always available. It might also be useful in assessing patients with traumatic brain injuries. Next, they will carry out another prospective trial to measure the test's performance when used in an ambulance. They have also designed an interventional trial that leverages the technology to expedite the triage of stroke patients by having them bypass standard imaging and move directly to intervention.

"In stroke care, time is brain," Bernstock said. "The sooner a patient is put on the right care pathway, the better they are going to do. Whether that means ruling out bleeds or ruling in something that needs an intervention, being able to do this in a prehospital setting with the technology that we built is going to be truly transformative."

Additional authors include Yasir Durrani, Jakob V. E. Gerstl, Danielle Murphy, Ashley Harris, Imane Saali, Toby Gropen, Shashank Shekhar, Ari D. Kappel, Nirav J. Patel, Rose Du, Rodolfo E. Alcedo Guardia, Juan C. Vicenty-Padilla, Adam A. Dmytriw, Vitor Mendes Pereira, Saef Izzy, Allauddin Khan, Mohammed A. Aziz-Sultan, David S. Liebeskind, Jason M. Davies, Adnan H. Siddiqui, and Edoardo Gaude.

Robyn and Rand Miller.

Sometimes success smacks you right in the face. More often, it sneaks up on you from behind.

In September of 1993, the brothers Rand and Robyn Miller and the few other employees of Cyan, Inc., were prototypical starving artists, living on “rice and beans and government cheese.” That month they saw Brøderbund publish their esoteric Apple Macintosh puzzle game Myst, which they and everyone else regarded as a niche product for a niche platform. There would go another year before it became abundantly clear that Myst, now available in a version for Microsoft Windows as well as for the Mac, was a genuine mass-market hit. It would turn into the gift that kept on giving, a game with more legs than your average millipede. It wouldn’t enjoy its best single month until December of 1996, when it would set a record for the most copies one game had ever sold in one month.

All of this — not just the sales figures themselves but the dozens of awards, the write-ups in glossy magazines like Rolling Stone and Newsweek, the fawningly overwritten profiles in Wired, the comparisons with Steven Spielberg’s Jurassic Park and Michael Jackson’s Thriller — happened just gradually enough that it seemed almost natural. Almost natural. “It took a while for it to hit me that millions of people were buying this game,” says Robyn Miller. “The most I could really wrap my head around would be to go to a huge concert and see all of the people there and think, ‘Okay, this is not even a portion of the people who are playing Myst.'”

The Miller brothers could have retired and lived very comfortably for the rest of their lives on the fortune they earned from Myst. They didn’t choose this path. “We took salaries that were fairly modest and just put the company’s money back into [a] new project,” says Rand.

Brøderbund was more than eager for a sequel to Myst, something that many far smaller hits than it got as a matter of course within a year. But the Miller brothers refused to be hurried, and did not need to be, a rare luxury indeed in their industry. Although they enjoyed a very good relationship with Brøderbund, whose marketing acumen had been essential to getting the Myst ball rolling, they did not wish to be beholden to their publisher in any way. Rather than accepting the traditional publisher advance, they decided that they would fund the sequel entirely on their own out of the royalties of the first game. This meant that, as Myst blew up bigger and bigger, their ambitions for the game they intended to call Riven were inflated in tandem. They refused to give Brøderbund a firm release date; it will be done when it’s done, they said. They took to talking about Myst as their Hobbit, Riven as their Lord of the Rings. It had taken J.R.R. Tolkien seventeen years to bridge the gap between his children’s adventure story and the most important fantasy epic in modern literary history. Surely Brøderbund could accept having to wait just a few years for Riven, especially with the sales figures Myst was still putting up.

Cyan’s digs reflected their rising status. They hadn’t even had a proper office when they were making Myst; everybody worked out of their separate homes in and around Spokane, Washington, sharing their output with one another using the “car net”: put it on a disk, get into your car, and drive it over to the other person. In the immediate aftermath of Myst’s release and promising early sales, they all piled into a drafty, unheated garage owned by their sound specialist Chris Brandkamp. Then, as the sales numbers continued to tick upward, they moved into an anonymous-looking former Comfort World Mattress storefront. Finally, in January of 1995, they broke ground on a grandiosely named “Cyan World Headquarters,” whose real-world architecture was to be modeled on the virtual architecture of Myst and Riven. While they were waiting for that building to be completed — the construction would take eighteen months — they junked the consumer-grade Macs which had slowly and laboriously done all of the 3D modeling necessary to create Myst’s environments in favor of Silicon Graphics workstations that cost $40,000 a pop.

Cyan breaks ground on their new “world headquarters.”

The completed building looked very much apiece with their games, both outside…

…and inside.

The machines that made Riven. Its imagery was rendered using $1 million worth of Silicon Graphics hardware: a dozen or so workstations connected to these four high-end servers that did the grunt work of the ray-tracing. It was a far cry from Myst, which had been made with ordinary consumer-grade Macs running off-the-shelf software.

And the people who made Riven…

There were attempts to drum up controversies in the press, especially after Riven missed a tentative Christmas 1996 target date which Brøderbund had (prematurely) announced, a delay that caused the publisher’s stock price to drop by 25 percent. The journalists who always seemed to be hovering around the perimeter of Cyan’s offices claimed to sniff trouble in the air, an aroma of overstretched budgets and creative tensions. But, although there were certainly arguments — what project of this magnitude doesn’t cause arguments? — there was in truth no juicy decadence or discord going on at Cyan. The Miller brothers, sons of a preacher and still devout Christians, never lost their Heartland groundedness. They never let their fluke success go to their heads in the way of, say, the minds behind Trilobyte of The 7th Guest fame, were never even seriously tempted to move their operation to some more glamorous city than Spokane. For them, it was all about the work. And luckily for them, plenty of people were more than willing to move to Spokane for a chance to work at The House That Myst Built, which by the end of 1995 had replaced Trilobyte as the most feted single games studio in the mainstream American press, the necessary contrast to all those other unscrupulous operators who were filling their games and the minds of the nation’s youth with indiscriminate sex and violence.

The most important of all the people who were suddenly willing to come to Spokane would prove to be Richard Vander Wende, a former Disney production designer — his fingerprints were all over the recent film Aladdin — who first bumped into the Miller brothers at a Digital World Expo in Los Angeles. Wende’s conceptual contribution to Riven would be as massive as that of either of the Miller brothers, such that he would be given a richly deserved co-equal billing with them at the very top of the credits listing.

Richard Vander Wende.

Needless to say, though, there were many others who contributed as well. By the time Cyan moved into their new world headquarters in the summer of 1996, more than twenty people were actively working on Riven every day. The sequel would wind up costing ten times to fifteen times as much to make as its predecessor, filling five CDs to Myst’s lone silver platter.

Given the Millers’ artistic temperament and given the rare privilege they enjoyed of being able to make exactly the game they wished to make, one might be tempted to assume that Riven was to be some radical departure from what had come before. In reality, though, this was not the case at all. Riven was to be Myst, only more so; call it Myst perfected. Once again you would be left to wander around inside a beautiful pre-rendered 3D environment, which you would view from a first-person perspective. And once again you would be expected to solve intricate puzzles there — or not, as you chose.

Cyan had long since realized that players of Myst broke down into two broad categories. There were those they called the gamers, who engaged seriously with it as a series of logical challenges to be overcome through research, experimentation, and deduction. And then there was the other group of players — a far, far larger one, if we’re being honest — whom Cyan called the tourists, who just wanted to poke around a little inside the virtual world and take in some of the sights and sounds. These were folks like the residents of a retirement home who wrote to Cyan to say that they had been playing and enjoying Myst for two years and two months, and wanted to hear if the rumors that there were locations to explore beyond the first island — an island which constitutes about 20 percent of the full game — were in fact true.

Riven was meant to cater to both groups, by giving the gamers a much deeper, richer, more complex tapestry of puzzles to unravel, whilst simultaneously being kept as deliberately “open” as possible in terms of its geography, so that you could see most of its locations without ever having to solve a single conundrum. “The two complaints about Myst,” said Rand Miller, “were that it was too hard and too easy. We’re trying to make Riven better for both kinds of players.” Whereas Myst allowed you to visit four separate “ages” — basically, alternative dimensions — after solving those early puzzles which had so stymied the retirees, Riven was to take place all in the same dimension, on a single archipelago of five islands. You would be able to travel between the islands right from the start, using vehicles whose operation should be quite straightforward even for the most puzzle-averse players. If all you wanted to do was wander around the world of Riven, it would give you a lot more spaces in which to do so than Myst.

Of course, while the world of Riven was slowly coming together, the real world wasn’t sitting still. Myst had been followed by an inevitable flood of “Myst clones” from other publishers and studios, which, in lieu of a proper sequel from Cyan, did their best to pick up the slack by offering up their own deserted, 3D-rendered environments to explore. Few of them were more than modestly successful; Activision’s Zork Nemesis, which may have done the best of them all, sold about 450,000 copies, an order of magnitude and a half less than the final numbers that Myst would put up. Meanwhile the genre of adventure games in general had peaked in the immediate aftermath of Myst and would be well into an increasingly precipitous decline by the time Riven shipped in October of 1997. The Last Express, the only other adventure that Brøderbund published that year, stiffed badly in the spring, despite sporting prominently on its box the name of Jordan Mechner, one of the few videogame auteurs with a reputation to rival that of the Miller brothers.

Yet Cyan’s own games still seemed weirdly proof against the marketplace pressures that were driving so many other game makers in the direction of real-time strategy and first-person shooters. In June of 1997, the nearly four-year-old Myst was propelled back to the top of the sales charts by the excitement over the approaching debut of Riven. And when it did appear, Riven didn’t disappoint the bean counters. It and Myst tag-teamed one another in the top two chart positions right through the Christmas buying season. Myst would return to number one a few more times in the course of 1998, while an entire industry continued to scratch its collective head, wondering why this particular game — a game that was now approaching its fifth birthday, making it roughly as aged as the plays of Shakespeare as the industry reckoned time — should continue to sell in such numbers. Even today, it’s hard to say precisely why Myst just kept selling and selling, defying all the usual gravities of its market. It seems that non-violent, non-hardcore gaming simply needed a standard bearer, and so it found one for itself.

Riven wasn’t quite as successful as Myst, but this doesn’t mean it didn’t do very well indeed by all of the standard metrics. Its biggest failing in comparison to its older sibling was ironically its very cutting-edge nature; whereas just about any computer that was capable of running other everyday software could run Myst by 1997, you needed a fairly recent, powerful machine to run Riven. Despite this, and despite the usual skepticism from the hardcore-gaming press — “With its familiar, lever-yanking gameplay, Riven emerges as the ultimate Myst clone,” scoffed Computer Gaming World magazine — Riven’s sales surpassed 1 million units in its first year, numbers of which any other adventure game could scarcely have dreamed.

Riven was a huge hit by any conventional standard, but it didn’t have the legs of Myst. Already for long stretches during 1998, it was once again being comfortably outsold by Myst. Lifetime retail sales of around 1.5 million strike me as the most likely figure — still more than enough to place Riven in the upper echelon of late 1990s computer games.

Fans and boosters of the genre naturally wanted to see a broader trend in Riven’s sales, a proof that adventures in general could still bring home the bacon with the best of them. The hard truth that the games of Cyan were always uniquely uncoupled from what was going on around them was never harder to accept than in this case. In the end, though, Riven would have no impact whatsoever on the overall trajectory of the adventure genre.

Because Riven is a sequel in such a pure sense — a game that aims to do exactly what its predecessor did, only bigger and better — your reaction to it is doomed to be dictated to a large extent by your reaction to said predecessor. It’s almost impossible for me to imagine anyone liking or loving Riven who didn’t at least like Myst.

The defining quality of both games is their thoroughgoing sense of restraint. When Myst first started to attract sales and attention, naysayers saw its minimalism through the lens of technical affordance, or rather the Miller brothers’ lack thereof: having only off-the-shelf middleware like HyperCard to work with, lacking the skill set that might have let them create better tools of their own, they just had to do the best they could with what they had. In this reading, Myst‘s static world, its almost nonexistent user interface, its lack of even such niceties as a player inventory, stemmed not so much from aesthetic intent as from the fact that it had been created with a hypertext editor that had never been meant for making games. The alternative reading is that the Miller brothers were among the few game developers who knew the value of restraint from the start, that they were by nature and inclination minimalists in an industry inclined to maximalism in all things, and this quality was their greatest strength rather than a weakness. The truth probably lies somewhere between the two extremes, as it usually does. Regardless, there’s no denying that the brothers leaned hard into the same spirit of minimalism that had defined Myst when the time came to make Riven, even though they were now no longer technologically constrained into doing so. One camp reads this as a colossal failure of vision; the other reads it as merely staying true to the unique vision that had gotten them this far.

While I don’t want to plant myself too firmly in either corner, I must say that I am surprised by some of the things that Cyan didn’t do with twice the time and ten or fifteen times the budget. The fact that Riven still relies on static, pre-rendered scenery and node-based movement isn’t the source of my surprise; that compromise was necessary in order to achieve the visual fidelity that Cyan demanded. I’m rather surprised by how little Cyan innovated even within that basic framework. Well before Riven appeared, the makers of other Myst successors had begun to experiment with ways of creating a slightly more fluid, natural-feeling experience. Zork Nemesis, for example, stores each of its nodes as a 360-degree panorama instead of a set of fixed views, letting you smoothly turn in place through a complete circle. Riven, by contrast, confines its innovations in this area to displaying a little transition animation as you rotate between its rigidly fixed views. As a result, switching from view to view does become a little less jarring than it is in Myst, but the approach is far from even the Myst-clone state of the art.

Cyan was likewise disinterested in pursuing other solutions that would have been even easier to implement than panning rotation, but that could have made their game less awkward to play. The extent of your rotation when you click on the left or right side of the screen remains inconsistent, just as it was in Myst; sometimes it’s 90 degrees, sometimes it’s less or more. This can make simple navigation much more confusing than it needs to be, introducing a layer of fake difficulty — i.e., difficulties that you would not have if you were really in this world — which seems at odd with Cyan’s stated determination to create as immersive an experience as possible. Even a compass with which to tell which way you’re facing at any given time would have helped enormously, but no such concessions to player convenience are to hand.

Again, these are solutions that the other makers of Myst clones — not a group overly celebrated for its spirit of innovation — had long since deployed. Cyan was always a strangely self-contained entity, showing little awareness of what others were doing around them, making a virtue of their complete ignorance of the competition. In cases like these, it was perhaps not so much a virtue as a failure of simple due diligence. Building upon the work of others is the way that gaming as a whole progresses.

When it comes to storytelling as well, Riven’s differences from Myst are more a matter of execution than kind. As in Myst, there is very little story at all here, if by that we mean a foreground plot driving things along. A brief bit of exposition at the beginning picks up right where Myst ended, providing an excuse for dumping you into another open-ended environment. Whereas Myst took place entirely in deserted ages, here you’re ostensibly surrounded by the Rivenese, the vaguely Native-American-like inhabitants of the archipelago. Rather conveniently for Cyan, however, the Rivenese are terrified of strangers, and scurry away into hiding whenever you enter a scene. The few named characters you meet, including the principal villain, are likewise forever just leaving when you come upon them, or showing up, giving speeches, and then going away again before you can interact with them. By 1997, this sort of thing was feeling more tired than clever.

Rand Miller, returning in the role of the patriarch Atrus from Myst, gives you your marching orders and sends you on your way in the introductory movie. Riven makes more extensive use of such scenes involving real actors than Myst, but it’s done well, and never overdone. The end result is about as un-cheesy as these techniques can possibly look to modern eyes.

The real story, in both Myst and Riven, is the backstory that caused these spaces to become the places they are, a backstory which you uncover as you explore them. And in this area, I’m happy to say, Riven actually does outdo its predecessor. Almost everything there is to find out about how the ages of Myst became as they are is conveyed in one astonishingly clumsy infodump, a set of books which you find in a library on that first island after solving the first couple of puzzles. These stop your progress dead for an hour or so as you read through them, after which you’re back to exploring, never to be troubled by much of any exposition again.

By the time of Riven, however, the Miller brothers had learned about the existence of something called dramatic pacing. Here, too, most of the real story comes in the form of books and journals, but these are scattered around the islands, providing an enticement to solve puzzles in order to acquire and read them. The Myst “universe” grew considerably in depth and coherency between Myst and Riven, thanks to a trilogy of novels written by the British science-fiction author David Wingrove in close collaboration with the Miller brothers during that interim. In Riven, then, you get some of the same sense that you get in The Lord of the Rings, that you are only scraping the surface of a world that goes much deeper than its foreground sights and sounds. “The Lord of the Rings is so satisfying because of the details,” said Rand Miller at the time. “You get the feeling that the world you’re reading about is real. Different but real. That’s how we go about designing.” Like Tolkien, the Miller brothers went so far as to make up the beginnings at least of a coherent language for their land’s inhabitants. This sense of established lore, combined with the improved pacing and better writing, makes Riven’s backstory more compelling than that of Myst, makes uncovering more of it feel like a worthwhile goal in itself. Instead of providing a mere excuse for the gameplay, as in Myst, Riven’s backstory comes to fuel its gameplay to a large extent.

And this starts to take us into the territory of the first of the two things that Riven does really, really well, does so well in fact that you might just be willing to discount all of the failings I’ve been belaboring up to this point. The archipelago is a truly intriguing, even awe-inspiring place to explore, thank not just to the cutting-edge 3D-rendering technology that was used to bring it to life, but — and even more so — the thought that went into the place.

Riven makes its priorities clear from the beginning, when it asks you to set up your screen and your speakers to provide the immersive audiovisual experience it intends for you to have.

The adjective “surreal” seems unavoidable when discussing Myst, so much so that Brøderbund built it right into their advertising tagline. (“The Surrealistic Adventure That Will Become Your World.”) Looking back on it now, though, I realize that the surrealism of Myst was as much a product of process as intention. The 3D-modeling software that was used to create the scenery of Myst couldn’t render genuinely realistic scenes; everything it churned out was too geometrical, too stiff, too uniform in color to look in any sense real. The result was surrealism, that forlorn, otherworldly, even vaguely disturbing stripe of beauty that became the hallmark of Myst and its many imitators.

But I would not call Riven surreal. The improved technology that enabled it, on both the rendering side — meaning all those Silicon Graphics servers and workstations, with their complex ray-tracing algorithms — and the consumer-facing side — meaning the latest home computers, with their capability of displaying millions of nuanced shades of color onscreen at once — led to a more believable world. The key to it all is in the textures, the patterns that are overlaid onto the frame of a 3D model in lieu of blocks of solid color to make it look like a real object made out of wood, metal, or dirt. Cyan traveled to Santa Fe, New Mexico, to capture thousands of textures. The same visual qualities that led to that state being dubbed the “Land of Enchantment” and drew artists like Georgia O’Keeffe to its high deserts suffuses the game, from the pueblo walls of the Rivenese homes to the pebbly cliff-side paths, from an old iron tower rusting in the sun to the ragged vegetation huddling around it. You can almost feel the sun on your back and the sweat on your skin.

My wife and I are inveterate hikers these days, planning most of our holidays around where we can get out and walk. Riven made me want to climb through the screen and roam its landscapes for myself. Myst has its charms, but they are nothing like this. When I compare the two games, I think about what a revelation the battered, weathered world of Tatooine was when Star Wars hit cinemas in 1977, how at odds it was with the antiseptic sleekness of the science-fiction films that preceded it. Riven is almost as much of a revelation when set beside Myst and its many clones.

The visuals both feed and are fed by the backstory and the world-building. The islands are replete with little details that have nothing to do with solving the game, that exist simply as natural, necessary parts of this place you’re exploring. In a perceptive video essay, YouTube creator VZedshows notes how “the lived-in world of Riven lets us look at a house and say, ‘Okay, that’s a house.’ And that’s it. A totally different thought than seeing a log cabin on Myst Island and saying, ‘Okay, that’s a house. But what is it for?’ The puzzles in Riven melt into the world around them.”

Which brings us neatly to the other thing that Riven does remarkably well, the one aimed at the gamers rather than the tourists. Quite simply, Riven is one of the most elegantly sophisticated puzzle games ever created. This facet of it is not for everyone. (I’m not even sure it’s for me, about which more in a moment.) But it does what it sets out to do uncompromisingly well. Riven is a puzzle game that doesn’t feel like a puzzle game. It rather feels like you really have been dropped onto this archipelago, with its foreign civilization and all of its foreign artifacts, and then left to your own devices to make sense of it all.

Many of Riven’s puzzles are as much anthropological as mechanical. For example, you have to learn to translate the different symbols of a foreign number system.

This is undoubtedly more realistic than the ages of Myst, whose puzzles stand out from their environs so plainly that they might as well be circled with a bright red Sharpie. But does it lead to a better game? As usual, the answer is in the eye of the beholder. Ironically, almost everything that can be said about Riven’s puzzles can be cast as either a positive or a negative. If you’re looking for an adventure game that’s nails-hard and yet scrupulously fair — a combination that’s rarer than it ought to be — Riven will not disappoint you. If not, however, it will put you right off just as soon as you grow bored with idle wandering and begin to ask yourself what the game expects you to actually be doing. Myst was widely perceived in the 1990s as being more difficult than it really was; Riven, by contrast, well and truly earns its reputation.

Each of Myst’s ages is a little game unto itself when it comes to its puzzles; you never need to use tools or information from one age to overcome a problem in another one. For better or for worse, Riven is not like that — not at all. Puzzles and clues are scattered willy-nilly all over the five islands; you might be expected to connect a symbol you’re looking at now to a gadget you last poked at hours and hours ago. Careful, copious note-taking is the only practical way to proceed. I daresay you might end up spending more time poring over your real-world journal, looking for ways to combine and thereby to make sense of the data therein, than you do looking at the monitor screen. Because most of the geography is open to you from the very beginning — this is arguably Riven’s one real concession to the needs of the marketplace, being the one that allows it to cater to the tourists as well as the gamers — there isn’t the gated progress you get in so many other puzzly adventure games, with new areas and new problems being introduced gradually as you solve the earlier ones. No, Riven throws it all at you from the start, in one big lump. You just have to keep plugging away at it when even your apparently successful deductions don’t seem to be yielding much in the way of concrete rewards, trusting that it will all come together in one big whoosh at the end.

All of which is to say that Riven is a slow game, the polar opposite of the instant gratification that defines the videogame medium in the eyes of so many. There are few shortcuts for moving through its sprawling, fragmented geography — something you’ll need to do a lot of, thanks to its refusal to contain its puzzles within smaller areas as Myst does. Just double-checking some observation you think you made earlier or confirming that some effect took place as expected represents a significant investment in time. Back in the day, when everyone was playing directly from CD, Riven was even slower than it is today, requiring you to swap discs every time you traveled to a different island. In his vintage 1997 review, Andrew Plotkin — a fellow who is without a doubt much, much smarter than I am, at least when it comes to stuff like this — said that he was able to solve Riven in about twenty hours, using just one hint. It will probably take more mortal intelligences some multiple of one or both of those figures.

Your reaction to Riven when approached in “gamer” mode will depend on whether you think this kind of intensive intellectual challenge is fun or not, as well as whether you have the excess intellectual and temporal bandwidth in your current life to go all-in on such a major undertaking. I must sheepishly confess that my answer to the first question is more prevaricating than definitive, while my answer to the second one is a pretty solid no. In the abstract, I do understand the appeal of what Riven is offering, understand how awesome it must feel to put all of these disparate pieces together without help. Nevertheless, when I approached the game for this article, I couldn’t quite find the motivation to persevere down that road. Riven wants you to work a little harder for your fun than the current version of myself is willing to do. I don’t futz around with my notebook too long before I start looking out the window and thinking about how nice it would be to take a walk in real nature. I take enough notes doing research for the articles I write; I’m not sure I want to do so much research inside a game.

Prompted partially by my experience with Riven, I’ve been musing a fair amount lately about the way we receive games, and especially how the commentary you read on this site and others similar to it can be out of step with the way the games in question existed for their players in their heyday. I’m subject to the tyranny of my editorial calendar, to the need to just finish things, one way or another, and move on. Riven is not well-suited to such a mindset. In my travels around the Internet, I’ve noticed that those who remember the game most fondly often took months or years to finish it, or never finished it at all. It existed for them as a tempting curiosity, to be picked up from time to time and poked at, just to see if a little more progress was possible here or there, or whether the brainstorm that came to them unbidden while driving home from work that day might bear some sort of fruit. It’s an open question whether even folks who don’t have an editorial schedule to keep can recapture that mindset here and now, in the third decade (!) of the 21st century, when more entertainment of every conceivable type than any of us could possibly consume in a lifetime is constantly luring us away from any such hard nut as Riven. As of this writing, Cyan is preparing a remake of Riven. It will be interesting to see what concessions, if any, they chose to make to our new reality.

Even in the late 1990s, there was the palpable sense that Riven represented the end of an era, that even Cyan would not be able to catch lightning in a bottle a third time with yet another cerebral, contemplative, zeitgeist-stamping single-player puzzle game. Both Richard Vander Wende and Robyn Miller quit the company as soon as the obligatory rounds of promotional interviews had been completed, leaving the Myst franchise’s future solely in the hands of Rand Miller. Robyn’s stated reason for departing brings to the fore some of the frustrations I have with Cyan’s work. He said that he was most interested in telling stories, and had concluded that computer games just weren’t any good at that: “I felt like, you know what? It’s not working. This whole story thing is not happening, and one of the reasons it’s not happening is because of the medium. It’s not what this medium is good at.” So, he said, he wanted to work in film instead.

The obvious response is that Cyan had never actually tried to tell an engaging foreground story, had rather been content to leave you always picking up the breadcrumbs of backstory. Cyan’s stubborn conservatism in terms of form and their slightly snooty insistence on living in their own hermetically sealed bubble, blissfully unaware of the innovations going on around them in their industry in both storytelling and other aspects of game making, strike me as this unquestionably talented group’s least attractive qualities by far. When asked once what his favorite games were, Richard Vander Wende said he didn’t have any: “Robyn and I are not really interested in games of any kind. We’re more interested in building worlds. To us, Myst and Riven are not ‘games’ at all.” Such scare-quoted condescension does no one any favors.

Then again, that’s only one way of looking at it. Another way is to recognize that Riven is exactly the game — okay, if you like, the world — that its creators wanted to make. It’s worth acknowledging, even celebrating, as the brave artistic statement it is. Love it or hate it, Riven knows what it wants to be, and succeeds in being exactly that — no more, no less. Rather than The Lord of the Rings, call it the Ulysses of gaming: a daunting creation by any standard, but one that can be very rewarding to those willing and able to meet it where it lives. That a game like this outsold dozens of its more visceral, immediate rivals on the store shelves of the late 1990s is surely one of the wonders of the age.

Did you enjoy this article? If so, please think about pitching in to help me make many more like it. You can pledge any amount you like.

Sources: The books The Secret History of Mac Gaming (Expanded Edition) by Richard Moss, From Myst to Riven: The Creations & Inspirations by Richard Kadrey, and Riven: The Sequel to Myst: The Official Strategy Guide by Rick Barba; Computer Gaming World of January 1998; Retro Gamer 208; Wired of September 1997; Game Developer of March 1998. Plus the “making of” documentary that was included on the original Riven CDs. The sales figures for Zork Nemesis come from the Jordan Mechner archive at the Strong Museum of Play.

Online sources include GameSpot’s old preview of Riven, Salon’s profile of the Miller brothers on the occasion of Robyn’s departure from Cyan, VZedshows’s video essay on Myst and Riven, and Andrew Plotkin’s old review of Riven.

The original version of Riven is currently available as a digital purchase on GOG.com. As noted in the article above, a remake is in the works at Cyan.

Learning to think logically and present ideas in a logical fashion has always been considered a central part of becoming a mathematician. In this paper we compare the performance of three groups: mathematics undergraduates, mathematics staff and history undergraduates (representative of a "general population"). These groups were asked to solve Wason's selection task, a seemingly straightforward logical problem. Given the assumption that logic plays a major role in mathematics, the results were surprising: less than a third of students and less than half of staff gave the correct answer. Moreover, mathematicians seem to make different mistakes from the most common mistake noted in the literature. The implications of these results for our understanding of mathematical thought are discussed with reference to the role of error checking. [For complete proceedings, see ED489538.]

International Group for the Psychology of Mathematics Education, 35 Aandwind Street, Kirstenhof, Cape Town, 7945, South Africa. Web site: http://igpme.org.

Footer

You can’t perform that action at this time.

"Overall liked living here, our landlord was great, we stayed for about 6 years in the middle unit, had some great neighbors, and some just OK neighbors. The apartment itself was a good size, good price, great location. Would certainly recommend it!"

Mon May 06 2024

Why is it so hard for entrepreneurs to get mortgages? Why did the address I send my payment to change? Why don’t most banks own mortgages anymore? Why is there a thriving economy of small specialist lenders locally who only do mortgages? Why is a 30-year fixed rate mortgage available at all?

Welcome to the wild and wooly world for mortgages in the United States: the world’s most important manufactured product that virtually no users understand. For starters, virtually no one outside of the value chain considers a mortgage a manufactured product at all. Most people who own homes think they were the primary customer of the mortgage, but that’s not true.

(Apologies in advance for this being a very specifically U.S. flavored issue of this publication. While practice in some nations—Japan comes to mind—is trending in the direction of U.S. practice, the world’s largest and most sophisticated mortgage market is also very quirky, as a result of a hundred years of policy decisions with societal-level impact. Trying to address multiple nations’ takes on this instrument would make this issue the length of a small encyclopedia. Also, for simplicity, we’ll largely be looking at residential mortgages rather than commercial.)

Mortgages are a manufactured product

Let’s get the most important bit out of the way first: civilians think that a mortgage is a loan between a bank and a customer, and think of it primarily as services work (to the extent they think of it at all). This is a materially incorrect worldview, which drives much confusion about how the industry (and mortgages) function, both among people who can’t be expected to know better (customers) and people who probably could (regulators and journalists, among many others).

A mortgage is a product, which is built by specialist workers using an immense and costly capital edifice, to be sold into a supply chain for consumers of that product. It incidentally happens to involve a house and a loan, but those two facts do not drive most behaviors of the mortgage industry. The structure of the manufacturing process, and the consuming supply chain, do.

If it helps you to visualize this, think of a widget going into some hyperspecialized and boring bit of capitalism. Electronic flow meters, for example. [0] Somewhere, in fact in many somewheres, there exists a factory of people who do nothing but assemble electronic flow meters. These do exactly what they say on the tin; they measure flows of liquids or gasses using electric means. There are a lot of them in parts of the economy which push things through physical pipes, whether that is New York’s sanitation department or a steel smelter. It would be a very bad thing for the electronic flow meter factory to sell flow meters that didn’t meter flow accurately. Their consumers would hate that. Cakes would taste bad. Buildings would explode.

If you replaced your current mental model for mortgages with “it’s like a paper electronic flow meter for money, possibly with less paper these days”, it would improve your ability to understand the mortgage industry. The analogy is less about providing visibility into the contents of pipes (though mortgages must do that) and more “highly specialized manufactured widget that the entire world sits downstream of.”

Who manufactures mortgages?

Mortgages are written by originators. The first and most thorough misconception the public has about mortgages is that they’re written by lenders. Originators can be lenders, and historically generally were, but these days they generally are not. The people who make electronic flow meters do not also run the cake-baking factory as a fun hobby. Those are entirely different skillsets. They sell the thing they make, through a fairly complex value chain, to the cake-baking factory.

Of the top 15 mortgage originators in the U.S., only four are banks (Bank of America, Chase, U.S. Bank, and Wells Fargo), and most of those banks’ mortgage volume originate-for-sale rather than originate-to-lend. The remainder are all pure-play mortgage firms, some of which you’ve probably heard of due to direct-to-consumer marketing (RocketMortgage, Guaranteed Rate, etc) and some of which are more obscure (despite writing tens of billions of dollars of mortgages). This is due to a difference in operational models for them; we’ll get to that in a later issue.

So who actually has their hands on the tools in the mortgage factory? Every mortgage is a collaboration between:

The front office: A sales professional, called a loan officer by convention (and usually referred to in the industry as “a producer”). Their job is to interface with the home buyer, educate them on the most complicated and high-stakes financial decision they’ll have to make in their lives, and project manage the financial side of a real estate close. This will include a backbreaking amount of passing documents to…

The back office: Approximately 1.5 professionals per loan officer, who are responsible for making sure electronic flow meters are properly calibrated. Oops, wrong side of the analogy. No, they are responsible for making sure that the physical instantiation of the mortgage, 700 pages of documents or so, will pass the strict acceptance tests of the value chain that sells mortgages to end-users of mortgages.

The industry operates, as many people who have interacted with it can attest, in a disturbingly artisanal manner for a trillion dollar manufacturing sector which props up a huge portion of the economy. As the person buying a house, you care tremendously about the professional skill levels of your loan officer in a way which you never really care about the professional skill levels of the person who assembled your phone.

Who buys mortgages?

In a typical manufacturing supply chain, the box with the thing in it passes through different hands, or the thing gets incorporated into increasingly complex assemblies. In the mortgage supply chain, the thing gets split apart and (sometimes) repackaged into other things.

The manufactured product which is a mortgage is a collection of risks. The finance industry and government in the United States, to be more efficient and accomplish a long list of social goals, has successfully disaggregated many of those risks. Different risks are bought by different entities.

The risk of non-payment

As a policy decision, the United States has effectively socialized most of the market for the risk of non-payment of mortgages, in the interests of making home ownership more predictably available. (Whether this makes it cheaper or more expensive on net is a complicated question to answer.)

The mechanism for this are the GSEs (government-sponsored entities), like Fannie Mae, Ginnie Mae, Freddy Mac, and the Federal Home Loans Bank. These are all privately owned entities who have CEOs, shareholders, etc etc, but they’re also policy arms of the U.S. federal government and everyone knows it. (If there was any ambiguity about that, and there was very little, the financial crisis dispelled it.)

The business model of the GSEs rounds to this: they publish the specifications for the mortgage industry. This lets you know that the electronic flow meter will work as advertised and not blow up the factory you install it in. A mortgage which fits their specifications is called “conforming.” Buyers of a conforming mortgage need inquire no further about the 700 pages of documents. They treat mortgages as black boxes with some observable metadata to them.

The GSEs monetize these specifications by writing insurance against all conforming mortgages. The insurance is against, specifically, non-payment risk. Originators of mortgages who desire to sell them into value chain overwhelmingly choose to get conforming mortgages guaranteed. The cost is below 60 basis points and ultimately borne by the home purchaser, in the form of points or interest.

The GSEs, operating as galactically sized insurance companies, profit or lose in a similar manner to insurance companies: they can use capital markets to invest the float (though, due to risks and their quixotic relationship with the Treasury, they cannot be as aggressive as e.g. Berkshire Hathaway) and hope to earn an underwriting profit. That is, they hope the specifications they write and rigorously enforce correctly predict and price default risk. This was resoundingly not true in the run-up to the global financial crisis, which resulted in large government bailouts and a sort of quasi-nationalization of them.

(There has been some limited experimentation with privatizing the risk of non-payment again, via mechanisms like Freddie Mac’s Credit Risk Transfer product, but the GSEs will happily tell you that most is currently borne by the taxpayer.)

The risk of failing to service a mortgage correctly

A mortgage has a quirky little subcomponent called a Mortgage Servicing Right (MSR). Every month, it needs to collect money from the borrower and send that money… somewhere. This implies, minimally, a mailbox where you can send checks, someone to open the mail, and a phone number with a CS representative who can answer questions like “What is my current balance?” and “Did you get the last check I sent you?”

The holder of an MSR has an obligation to the owner(s) of the mortgage to not screw this up, enforced both by contract and by quite a bit of regulatory supervision. In return for taking on the boring business of opening letters and answering phone calls, they’re paid a small cut of the incoming payments. This generally works out to about 25 basis points of outstanding principal.

Their obligations are actually more complicated than this quick sketch, particularly in unhappy cases like e.g. when someone defaults on their mortgage. It’s an operationally intensive business to be in.

Importantly, MSRs trade some amount of work you could build a giant human-powered machine to do for a stream of ongoing payments. The finance industry loves that sort of stuff, and has built a large system to buy and sell MSRs. This results in the servicing of mortgages migrating to large players who have huge economies of scale to open letters and answer phone calls. It also confuses many people who walked into e.g. a Citibank to get their mortgage and think that this means Citibank will own the mortgage (probably not) or take checks from them for it (probably not).

Every other risk you could imagine, of which there are many

Private capital buys all the other risks. The How of that will have to wait for a future issue.

We can get into part of the What, though, right now.

A mortgage, like any manufactured product, has a value. You can go to your friendly neighborhood seller of mortgages and buy mortgages in any size or shape. If your factory needs a flow meter, someone has a flow meter for you and can tell you what it costs; they probably didn’t build the flow meter with their own two hands.

The value of the mortgage is not the outstanding principal, any more than the value of a flow meter is the maximum rate of water flowing through it. That is an important technical feature of the artifact at issue, but is not the price tag.

The value of flow meters goes up when there are more companies wanting to buy them in a time period than factories can manufacture them in the same time period, and down when the reverse is true. That happens to mortgages, too; supply and demand is a thing.

One driver of supply and demand is the interest rate environment, which is not entirely external to the mortgage market but which you can assume is for simplicity. All existing loans, including mortgages, become less valuable when interest rates go up and more valuable when interest rates go down. (The easy way to remember this: pretend you bought a widget for $100 which spit out $5 a year. If the new model of widgets spits out $6 a year and also costs $100 new, will anyone buy your perfectly functioning used widget for $100? No, that’s silly; they’d buy the new one. You have to offer them a discount until the yield matches that of the new model.)

This is one of the reasons why society has arranged it so that banks don’t own most mortgages. For a very, very long time, banks were primarily engaged in the business of maturity transformation; they would borrow money short term from depositors at low interest rates and lend it long-term, mostly to mortgage borrowers, at a higher interest rate.

We have learned, through long and painful experience, that this is an extremely repeatable recipe for banking crises. Interest rates go up, which increases banks funding costs and decreases the value of all the mortgages they own. (It also decreases the number of re-finances they sell, because they are now mechanically more expensive, but that’s a relatively minor factor.) If the bank is not sufficiently capitalized, to maintain their reserves (and have adequate liquidity to meet responsibilities to customers and regulators) they’ll have to sell some of those now less-valuable mortgages.

All banks experience the same pressure effectively simultaneously; supply and demand hits the value for mortgages and their values start to plummet; the book value of mortgages at other banks now means they’re under-reserved and so have to sell mortgages to raise liquidity; this creates a vicious circle in the banking sector.

A widespread misconception about mortgage securitization is that it was created to make Wall Street rich. This is a popular narrative, especially due to Michael Lewis’ best book (and one of the best finance movies ever made). This narrative is… incomplete.

Mortgage securitization, and secondary sales of loans, and other mechanisms cause mortgages to migrate from the banking sector to pools of capital which are more structurally insulated against the interest rate cycle.

A pension fund turns out to be a great counterparty to a mortgage, for example. They have known needs for cash in the short term and, if interest rates rise, that is generally good news to them. (It decreases the net present value of their funding shortfall for future commitments.) Their pile of mortgages would be worth less if they sold them but they do not structurally need to be sellers.

The pension fund cannot manufacture mortgages. That requires complex technology and a team of professionals that they do not have. They do not ever desire to write mortgages. They just want the economic exposure those mortgages represent, and they want it more than banks do.

It turns out, counterintuitively, that there are many, many buyers in the world who want to own mortgages more than banks do. They end up owning almost all of the mortgages, for fundamentally the same reason that the electronic flow meter factory ends up owning very few of the world’s current stock of flow meters. Someone has a better use for them and they exist to supply that better use, not to finance it.

Scratching the tip of the iceberg

We will return to the topic of mortgage finance, and the plumbing underlying them, in future installments of Bits about Money. But if you wanted to read about it right now, here are some excellent entry points:

Digitally Transforming the Mortgage Banking Industry. If you want to know how the sausage is made, this is the best account of it. I would read it critically; the author is from the sales side of the industry and is, ahem, in a position where they have to understand software to a greater degree than they actually do. Still, it’s a great look at mortgage origination where the forged bank statement hits the underwriter.

The 30-Year Mortgage is an Intrinsically Toxic Product by Byrne Hobart (who also writes a very good newsletter). This is lay-accessible but deep on both financial wonkery and housing policy.

The Dead Pledge. A scholarly history work on the pre-war years of mortgage finance, in whose shadow much of the U.S. economy (and social organization) now sits.

[0] My quixotic level of interest in electronic flow meters is on account of my best salaryman friend working at the company which makes the best electronic flow meters. I have had many long chats with him about the science of flow meters, the challenges of being a Japanese salaryman walking through customs in the Middle East with a suitcase full of sample flow meters, and the surprising difficulty of translating flow meter marketing copy into English. (For social reasons, I'll elide naming him or the company, but statistically speaking you are overwhelmingly likely to be... downstream of their work.)

A version of this post originally appeared on Tedium, Ernie Smith’s newsletter, which hunts for the end of the long tail.

Personal computing has changed a lot in the past four decades, and one of the biggest changes, perhaps the most unheralded, comes down to compatibility. These days, you generally can’t fry a computer by plugging in a joystick that the computer doesn’t support. Simply put, standardization slowly fixed this. One of the best examples of a bedrock standard is the peripheral component interconnect, or PCI, which came about in the early 1990s and appeared in some of the decade’s earliest consumer machines three decades ago this year. To this day, PCI slots are used to connect network cards, sound cards, disc controllers, and other peripherals to computer motherboards via a bus that carries data and control signals. PCI’s lessons gradually shaped other standards, like USB, and ultimately made computers less frustrating. So how did we get it? Through a moment of canny deception.

Commercial - Intel Inside Pentium Processor (1994)www.youtube.com

Embracing standards: the computing industry’s gift to itself

In the 1980s, when you used the likes of an Apple II or a Commodore 64 or an MS-DOS machine, you were essentially locked into an ecosystem. Floppy disks often weren’t compatible. The peripherals didn’t work across platforms. If you wanted to sell hardware in the 1980s, you were stuck building multiple versions of the same device.

For example, the KoalaPad was a common drawing tool sold in the early 1980s for numerous platforms, including the Atari 800, the Apple II, the TRS-80, the Commodore 64, and the IBM PC. It was essentially the same device on every platform, and yet, KoalaPad’s manufacturer, Koala Technologies, had to make five different versions of this device, with five different manufacturing processes, five different connectors, five different software packages, and a lot of overhead. It was wasteful, made being a hardware manufacturer more costly, and added to consumer confusion.

Drawing on a 1983 KoalaPad (Apple IIe)www.youtube.com

This slowly began to change in around 1982, when the market of IBM PC clones started taking off. It was a happy accident—IBM’s decision to use a bunch of off-the-shelf components for its PC accidentally turned them into a de facto standard. Gradually, it became harder for computing platforms to become islands unto themselves. Even when IBM itself tried and failed to sell the computing world on a bunch of proprietary standards in its PS/2 line, it didn’t work. The cat was already out of the bag. It was too late.

So how did we end up with the standards that we have today, and the PCI expansion card standard specifically? PCI wasn’t the only game in town—you could argue, for example, that if things played out differently, we’d all be using NuBus or Micro Channel architecture. But it was a standard seemingly for the long haul, far beyond other competing standards of its era.

Who’s responsible for spearheading this standard? Intel. While PCI was a cross-platform technology, it proved to be an important strategy for the chipmaker to consolidate its power over the PC market at a time when IBM had taken its foot off the gas, choosing to focus on its own PowerPC architecture and narrower plays like the ThinkPad instead, and was no longer shaping the architecture of the PC.

The vision of PCI was simple: an interconnect standard that was not intended to be limited to one line of processors or one bus. But don’t mistake standardization for cooperation. PCI was a chess piece—a part of a different game than the one PC manufacturers were playing.

The PCI standard and its derivatives have endured for over three decades. Modern computers with a GPU often use a PCIe interconnect. Alamy

In the early 1990s, Intel needed a win

In the years before Intel’s Pentium chipset came out in 1993, there seemed to be some skepticism about whether Intel could maintain its status at the forefront of the desktop-computing field.

In lower-end consumer machines, players like Advanced Micro Devices (AMD) andCyrix were starting to shake their weight around. At the high end of the professional market, workstation-level computing from the likes of Sun Microsystems, Silicon Graphics, and Digital Equipment Corporation suggested there wasn’t room for Intel in the long run. And laterally, the company suddenly found itself competing with a triple threat of IBM, Motorola, and Apple, whose PowerPC chip was about to hit the market.

A Bloomberg piece from the period painted Intel as being boxed in between these various extremes:

If its rivals keep gaining, Intel could eventually lose ground all around.

This is no idle threat. Cyrix Corp. and Chips & Technologies Inc. have re-created—and improved—Intel’s 386 without, they say, violating copyrights or patents. AMD has at least temporarily won the right in court to make 386 clones under a licensing deal that Intel canceled in 1985. In the past 12 months, AMD has won 40% of a market that since 1985 has given Intel $2 billion in profits and a $2.3 billion cash hoard. The 486 may suffer next. Intel has been cutting its prices faster than for any new chip in its history. And in mid-May, it chopped 50% more from one model after Cyrix announced a chip with some similar features. Although the average price of a 486 is still four times that of a 386, analysts say Intel’s profits may grow less than 5% this year, to about $850 million.

Intel’s chips face another challenge, too. Ebbing demand for personal computers has slowed innovation in advanced PCs. This has left a gap at the top—and most profitable—end of the desktop market that Sun, Hewlett-Packard Co., and other makers of powerful workstations are working to fill. Thanks to microprocessors based on a technology known as RISC, or reduced instruction-set computing, workstations have dazzling graphics and more oomph—handy for doing complex tasks and moving data faster over networks. And some are as cheap as high-end PCs. So the workstation makers are now making inroads among such PC buyers as stock traders, banks, and airlines.

This was a deep underestimation of Intel’s market position, it turned out. The company was actually well-positioned to shape the direction of the industry through standardization. They had a direct say on what appeared on the motherboards of millions of computers, and that gave them impressive power to wield. If Intel didn’t want to support a given standard, that standard would likely be dead in the water.

How Intel crushed a standards body on the way to giving us an essential technology

The Video Electronics Standards Association, or VESA, is perhaps best known today for its mounting system for computer monitors and itsDisplayPort technology. But in the early 1990s, it was working on a video-focused successor to the Industry Standard Architecture (ISA) internal bus, widely used in IBM PC clones.

A bus, the physical wiring that lets a CPU talk to internal and external peripheral devices, is something of a bedrock of computing—and in the wrong setting, a bottleneck. The ISA expansion card slot, which had become a de facto standard in the 1980s, had given the IBM PC clone market something to build against during its first decade. But by the early 1990s, for high-bandwidth applications, particularly video, it was holding back innovation. It just wasn’t fast enough to keep up, even after it had been upgraded from being able to handle 8 bits of data at once to 16.

That’s where the VESA Local Bus (VL-Bus) came into play. Built to work only with video cards, the standard offered a faster connection, and could handle 32 bits of data. It was targeted at the Super VGA standard, which offered higher resolution (up to 1280 x 1024 pixels) and richer colors at a time when Windows was finally starting to take hold in the market. To overcome the limitations of the ISA bus, graphics card and motherboard manufacturers started collaborating on proprietary interfaces, creating an array of incompatible graphics buses. The lack of a consistent experience around Super VGA led to VESA’s formation. The new VESA slot, which extended the existing 16-bit ISA bus with an additional 32-bit video-specific connector, was an attempt to fix that.

It wasn’t a massive leap—more like a stopgap improvement on the way to better graphics.

And it looked like Intel was going to go for the VL-BUS. But there was one problem—Intel actually wasn’t feeling it, and Intel didn’t exactly make that point clear to the companies supporting the VESA standards body until it was too late for them to react.

Intel revealed its hand in an interesting way, according to TheSan Francisco Examinertech reporter Gina Smith:

Until now, virtually everyone expected VESA’s so-called VL-Bus technology to be the standard for building local bus products. But just two weeks before VESA was planning to announce what it came up with, Intel floored the VESA local bus committee by saying it won’t support the technology after all. In a letter sent to VESA local bus committee officials, Intel stated that supporting VESA’s local bus technology “was no longer in Intel’s best interest.” And sources say it went on to suggest that VESA and Intel should work together to minimize the negative press impact that might arise from the decision.

Good luck, Intel. Because now that Intel plans to announce a competing group that includes hardware heavyweights like IBM, Compaq, NCR and DEC, customers and investors (and yes, the press) are going to wonder what in the world is going on.

Not surprisingly, the people who work for VESA are hurt, confused and angry. “It’s a political nightmare. We’re extremely surprised they’re doing this,” said Ron McCabe, chairman for the committee and a product manager at VESA member Tseng Labs. “We’ll still make money and Intel will still make money, but instead of one standard, there will now be two. And it’s the customer who’s going to get hurt in the end.”

But Intel had seen an opportunity to put its imprint on the computing industry. That opportunity came in the form of PCI, a technology that the firm’s Intel Architecture Labs started developing around 1990, two years before the fateful rejection of VESA. Essentially, Intel had been playing both sides on the standards front.

Why PCI

Why make such a hard shift, screwing over a trusted industry standards body out of nowhere? Beyond wanting to put its mark on the standard, Intel also saw an opportunity to build something more future-proof; something that could benefit not just graphic cards but every expansion card in the machine.

As John R. Quinn wrote in PC Magazine in 1992:

Intel’s PCI bus specification requires more work on the part of peripheral chip-makers, but offers several theoretical advantages over the VL-Bus. In the first place, the specification allows up to ten peripherals to work on the PCI bus (including the PCI controller and an optional expansion-bus controller for ISA, EISA, or MCA). It, too, is limited to 33 MHz, but it allows the PCI controller to use a 32-bit or a 64-bit data connection to the CPU.

In addition, the PCI specification allows the CPU to run concurrently with bus-mastering peripherals—a necessary capability for future multimedia tasks. And the Intel approach allows a full burst mode for reads and writes (Intel’s 486 only allows bursts on reads).

Essentially, the PCI architecture is a CPU-to-local bus bridge with FIFO (first in, first out) buffers. Intel calls it an “intermediate” bus because it is designed to uncouple the CPU from the expansion bus while maintaining a 33-MHz 32-bit path to peripheral devices. By taking this approach, the PCI controller makes it possible to queue writes and reads between the CPU and PCI peripherals. In theory, this would enable manufacturers to use a single motherboard design for several generations of CPUs. It also means more sophisticated controller logic is necessary for the PCI interface and peripheral chips.

To put that all another way, VESA came up with a slightly faster bus standard for the next generation of graphics cards, one just fast enough to meet the needs of Intel’s recent i486 microprocessor users. Intel came up with an interface designed to reshape the next decade of computing, one that it would let its competitors use. This bus would allow people to upgrade their processor across generations without needing to upgrade their motherboard. Intel brought a gun to a knife fight, and it made the whole debate about VL-Bus seem insignificant in short order.

The result was that, no matter how miffed the VESA folks were, Intel had consolidated power for itself by creating an open standard that would eventually win the next generation of computers. Sure, Intel let other companies use the PCI standard, even companies like Apple that weren’t directly doing business with Intel on the CPU side. But Intel, by pushing forth PCI, suddenly made itself relevant to the entire next generation of the computing industry in a way that ensured it would have a second foothold in hardware. The “Intel Inside” marketing label was not limited to the processors, as it turned out.

The influence of Intel’s introduction of PCI is still felt: Thirty-two years later, and three decades after PCI became a major consumer standard, we’re still using PCI derivatives in modern computing devices.

PCI and other standards

Looking at PCI, and its successor PCI express, less as ways that we connect the peripherals we use with our computers, and more as a way for Intel to maintain its dominance over the PC industry, highlights something fascinating about standardization.

It turns out that perhaps Intel’s greatest investment in computing in the 1990s was not the Pentium chipset, but its investment in Intel Architecture Labs, which quietly made the entire computing industry better by working on the things that frustrated consumers and manufacturers alike.

Essentially, as IBM had begun to take its eye off the massive clone market it unwittingly built during this period, Intel used standardization to fill the power void. It worked pretty well, and made the company integral to computer hardware beyond the CPU. In fact, devices you use daily—that Intel played zero part in creating—have benefited greatly from the company’s standards work. If you’ve ever used a device with a USB or Bluetooth connection, you can thank Intel for that.

Craig Kinnie, the director of Intel Architecture Labs in the 1990s, said it best in 1995, upon coming to an agreement with Microsoft on a 3D graphics architecture for the PC platform. “What’s important to us is we move in the same direction,” he said. “We are working on convergent paths now.”

That was about collaborating with Microsoft. But really, it has been Intel’s modus operandi for decades—what’s good for the technology field is good for Intel. Innovations developed or invented by Intel—like Thunderbolt, Ultrabooks, and Next Unit Computers (NUCs)—have done much to shape the way we buy and use computers.

For all the talk of Moore’s Law as a driving factor behind Intel’s success, the true story might be its sheer cat-herding capabilities. The company that builds the standards builds the industry. Even as Intel faces increasing competition from alliterative processing players like ARM, Apple, and AMD, as long as it doesn’t lose sight of the roles standards played in its success, it might just hold on a few years longer.

Ironically, Intel’s standards-driving winning streak, now more than three decades old, might have all started the day it decided to walk out on a standards body.

This is a set of notes for learning calculus using the Julia language. Julia is an open-source programming language with an easy to learn syntax that is well suited for this task.

Read “Getting started with Julia” to learn how to install and customize Julia for following along with these notes. Read “Julia interfaces to review different ways to interact with a Julia installation.

Since the mid 90s there has been a push to teach calculus using many different points of view. The Harvard style rule of four says that as much as possible the conversation should include a graphical, numerical, algebraic, and verbal component. These notes use the programming language Julia to illustrate the graphical, numerical, and, at times, the algebraic aspects of calculus.

There are many examples of integrating a computer algebra system (such as Mathematica, Maple, or Sage) into the calculus conversation. Computer algebra systems can be magical. The popular WolframAlpha website calls the full power of Mathematica while allowing an informal syntax that is flexible enough to be used as a backend for Apple’s Siri feature. (“Siri what is the graph of x squared minus 4?”) For learning purposes, computer algebra systems model very well the algebraic/symbolic treatment of the material while providing means to illustrate the numeric aspects. These notes are a bit different in that Julia is primarily used for the numeric style of computing and the algebraic/symbolic treatment is added on. Doing the symbolic treatment by hand can be very beneficial while learning, and computer algebra systems make those exercises seem kind of redundant, as the finished product can be produced much easier.

Our real goal is to get at the concepts using technology as much as possible without getting bogged down in the mechanics of the computer language. We feel Julia has a very natural syntax that makes the initial start up not so much more difficult than using a calculator, but with a language that has a tremendous upside. The notes restrict themselves to a reduced set of computational concepts. This set is sufficient for working many of the problems in calculus, but do not cover thoroughly many aspects of programming. (Those who are interested can go off on their own and Julia provides a rich opportunity to do so.) Within this restricted set, are operators that make many of the computations of calculus reduce to a function call of the form action(function, arguments...). With a small collection of actions that can be composed, many of the problems associated with introductory calculus can be attacked.

These notes are presented in pages covering a fairly focused concept, in a spirit similar to a section of a book. Just like a book, there are try-it-yourself questions at the end of each page. All have a limited number of self-graded answers. These notes borrow ideas from many sources, for example Strang (n.d.), Knill (n.d.), Schey (1997), Hass, Heil, and Weir (2018), Rogawski, Adams, and Franzosa (2019), several Wikipedia pages, and other sources..

These notes are accompanied by a Julia package CalculusWithJulia that provides some simple functions to streamline some common tasks and loads some useful packages that will be used repeatedly.

These notes are presented as a Quarto book. To learn more about Quarto books visit https://quarto.org/docs/books.

These notes may be compiled into a pdf file through Quarto. As the result is rather large, we do not provide that file for download. For the interested reader, downloading the repository, instantiating the environment, and running quarto to render to pdf in the quarto subdirectory should produce that file (after some time).

To contribute – say by suggesting addition topics, correcting a mistake, or fixing a typo – click the “Edit this page” link and join the list of contributors. Thanks to all contributors and a very special thanks to @fangliu-tju for their careful and most-appreciated proofreading.

Calculus with Julia version 0.18, produced on April 26, 2024.

Footer

You can’t perform that action at this time.

For over 15 years, YARA has been growing and evolving until it became an indispensable tool in every malware researcher’s toolbox. Throughout this time YARA has seen numerous updates, with new features added and countless bugs fixed. But today, I’m excited to announce the biggest change yet: a full rewrite.

YARA-X is a completely new implementation of YARA in Rust, and it has the following goals:

• Better user experience: The new command-line interface is more modern and colorful, and error reports are now more explicative. More features aimed at improving the user’s experience will be incorporated in the future.

• Rule-level compatibility: While achieving 100% compatibility is tough, our aim is to make YARA-X 99% compatible with YARA at the rule level. Incompatibilities should be minimal and thoroughly documented.

• Improved performance: YARA is known for its speed, but certain rules, especially those utilizing regular expressions or complex loops, can slow it down. YARA-X excels with these rules, often delivering significantly faster results. Our ultimate goal is for YARA-X to outperform YARA across the board.

• Enhanced reliability and security: YARA’s complexity in C code can lead to bugs and security vulnerabilities. YARA-X is built with Rust, offering greater reliability and security.

• Developer-friendly: We’re prioritizing ease of integration into other projects and simplified maintenance. Official APIs for Python, Golang, and C are provided to facilitate seamless integration. YARA-X also addresses some of the design flaws that made YARA challenging to maintain and extend.

Why a rewrite?

Was a complete rewrite necessary to achieve such goals? This question lingered in my mind for a long time before deciding to rewrite YARA. Rewriting is risky, it introduces new bugs, backward compatibility issues, and doubles the maintenance efforts, since legacy code doesn’t disappear after launching the new system. In fact, the legacy system may be still in use for years, if not decades.

However, I believe a rewrite was the right decision for multiple reasons:

• YARA is not a large project, it’s a medium-size project that lacks subsystems or components large enough to be migrated in isolation. Incremental migration to Rust was impractical because large portions of the code are interconnected.

• The improvements I envisioned required significant design changes. Implementing these in the existing C codebase would involve extensive rewrites, carrying the same risks as starting fresh with Rust.

• After a year of working on the project, I’ve found Rust easier to maintain than C. Rust offers stronger reliability guarantees and simplifies integrating third-party code, especially for multi-platform projects.

Is YARA really dead?

Despite the dramatic title of this post, YARA is not actually dead. I’m aware that many people and organizations rely on YARA to get important work done, and I don’t want to let them down.

YARA is still being maintained, and future releases will include bug fixes and minor features. However, don’t expect new large features or modules. All efforts to enhance YARA, including the addition of new modules, will now focus on YARA-X.

What’s the current state of YARA-X?

YARA-X is still in beta, but is mature and stable enough for use, specially from the command-line interface or one-shot Python scripts. While the APIs may still undergo minor changes, the foundational aspects are already established.

At VirusTotal, we have been running YARA-X alongside YARA for a while, scanning millions of files with tens of thousands of rules, and addressing discrepancies between the two. This means that YARA-X is already battle-tested. These tests have even uncovered YARA bugs!

Please test YARA-X and don’t hesitate to open an issue if you find a bug or some feature that you want to see implemented.

What’s next?

My aim is to surpass YARA in every possible aspect with YARA-X. I want it to be so superior that existing YARA users willingly migrate to YARA-X for its undeniable advantages, not because they are forced to do so.

Publishing a beta version is only the first step towards this goal. I’ll continue to enhance YARA-X, releasing updates and sharing insights through blog posts like this one.

Stay tuned, because this journey has only just begun.

Liikennematto is a tiny simulation game for children and grown-ups alike. The gameplay is simple: you only build the roads! Lots, buildings and their residents pop up after.

I used to keep a somewhat regular devlog for Liikennematto. The previous entries appeared 6-12 months apart, but since July 2021 the devlog has been on pause. I didn't stop working on the game, though (in fact the time I could have spent writing devlogs went to development instead). This belated entry then sums up nearly 3 years of development!

Last time I described Liikennematto's transformation to a real-time traffic simulation. Here's where I left you.

New visuals and sounds

I prototyped Liikennematto using premade assets from Kenney and some crude lot placeholders of my own. In 2022 I completely overhauled the assets. I spent a lot of time on selecting a bright and lively palette that works with the children's traffic mat metaphor.

Inspired by the traffic mat art style, I've drawn buildings and objects in various perspectives. Some of them are fake 3D and some are just flat sprites. The cars and roads follow a top-down perspective for simplicity's sake.

The trees, flowers and other organic objects are hand-drawn and traced to vector format. Buildings, roads and cars are digital vector drawings. I like the contrast between geometric man-made structures and the free-flowing nature.

I started with some obvious lots (school, café, fire station, low and high density residential). I intend to keep the lots unique, so that they only appear once. Car models and decorations re-appear in various colors.

I've used Figma as the vector drawing tool. Game asset creation is not really what Figma is usually used for, but I'm familiar with the app from work and it gets the job done. The biggest drawback is a lack of a gamedev-specific asset pipeline. I have a manual process where I adjust the SVG content to work with the game. With a low amount of assets it works, despite the labor. Maybe I'll automate the process later.

I also added some sound effects. For instance, placing a road tile triggers a sound that reinforces the action. A bit later a slightly higher pitched version of the sound plays when the road has been built. Both sounds are synced with the animation. Sound effects have a playful marimba-like timbre that complements the visuals. I've created the sounds using Ableton Live. I intend to add some light music to the game later.

UI overhaul

Liikennematto doesn't need much of an UI for the core gameplay in terms of panels and widgets. Most of the UI consists of debug tools and controls for starting a new game. The previous placeholder UI with emoji icons is gone. I've drawn custom icons for the UI and added a zoom control with three levels. The UI follows the palette of the game graphics.

I spent a good while making Liikennematto fun to play on touch devices and various screen sizes. You can now effortlessly pan and zoom around the tilemap. You can build roads with a click or tap, and remove them with a right-click or long press. The latter wasn't trivial to get right, as there's a fine line between swiping and tapping the screen. The action is triggered after a small delay and is visualized with a progress bar. Once the progress bar is full, releasing the long press will remove the road tile.

Parking

Possibly the biggest new feature since 2021 is the fully-featured parking system. Lots come with parking areas that have one or more parking spots. For each spot, I've defined a path to the lot entrance using a "parking lane" that influences the curve of the path. The paths vary a lot, since parking areas have different sizes and orientations. This path connects to the road network graph, where the car's route truly begins.

I didn't want the lots to be huge and neither are the parking areas. With the limited space, only one car may (un)park at a time. The lots have a parking lock that enforces the rule. While a car is moving in the parking area, the lock remains. Any other car willing to unpark or enter the lot will have to wait for their turn. Parking in the tiny areas is pretty fast, so the queue doesn't grow too long.

Some parking spots are reserved for the lot residents, or because of the role of a vehicle (a spot for a fire truck at a fire station). The residents prefer these spots, leaving free spots for guests.

Getting all of this to work was satisfying. I built new debug tools to visualize the state of the parking area, which was time well spent, because it revealed subtle bugs in the system.

A* pathfinding

Liikennematto has a road network graph built from the road tiles. The graph is used for routing cars around the map. The original routing algorithm simply started from a point in the graph (like an intersection entry) and navigated the graph using random directions from nodes until the route was long enough. This resulted in completely random routes with no sensible goal.

Now the routes are generated with the standard A* pathfinding algorithm, which finds a short path from a graph node to another node using distance to the goal as a heuristic. I looked up an example of the algorithm in Python and manually translated it into functional Elm code. With the small maps of Liikennematto, my optimised Elm implementation of the A* is capable of creating ~18 000 routes per second on an average machine using just one thread. That's fast enough to allow dynamic rerouting of cars on tilemap change. For example, a car that is waiting for a green traffic light may turn into a different direction from the intersection that it was about to just milliseconds before! It's fun to watch the routes adjust to the updated tilemap.

The most common route is from one lot to another lot. Sometimes the cars choose to just enjoy a short drive with no destination.

Game architecture and data flow

The Liikennematto codebase has grown over time. When I started out, I didn't have a clear plan for the game architecture. Now that I can see what the main parts are, I've refactored the Elm modules into four namespaces; Tilemap, Simulation, Render and UI.

This follows the way things build on top of each other:

• the tilemap is the base that the simulation runs on
• the simulation works with the road network graph, and listens to tilemap changes to (re)build it
• the tilemap and simulated entities, like cars, are then rendered separately
• on top of all of that, the UI triggers tilemap changes and enables debug tools

With this separation of concerns the code modules found their natural place. The namespaces interact via two interfaces: the Elm event loop and the World module, which contains the game state. Render and UI only read from the World interface, while Simulation and Tilemap may update it as well.

Liikennematto game architecture. Grey blue denotes interface, bright blue denotes namescape

The data flow inside the tilemap and simulation layers has changed as well. There's a lot of state to keep track of, and entities go through various state changes. Updates to the tilemap and entities have side-effects (sounds, re/despawn, rebuilding of lookup data structures, etc.).

The Elm Architecture and its event based updates help to some extent, yet I've implemented a custom event queue for fine-grained control of the data flow. The simulation and tilemap update process accumulate events that are then flushed once per frame to change things instantly (for later steps in the update flow), or scheduled for later.

Some examples of delayed events include cars spawning on lots and updates to car paths when the road network changes. The delay is also used to spread events over time to add some randomness to the simulation. Sometimes an event cannot be processed with the current state of the simulation. The custom event queue allows retry and if enough retries fail, the event may be discarded and a recovery strategy might then despawn related entities.

I also built a finite state machine (FSM) library to formally specify how cars, traffic lights, tiles and even the game initialization phase transition from a state to another. I included some ideas from AI for Games, Third Edition, notably entry and exit actions for a state. FSMs are the backbone of many programs and games, because of their utility. Without FSM states and transitions are hidden, often accidental. Having a clear spec for these removes a whole class of potential bugs. In Liikennematto they also help to sync animations and sounds to key actions.

Itch.io release

In November 2022 most of the changes were in place. Considering the small scope of the game, core features work well and what's missing is content. I decided then that it's time to release the game to early access. itch.io felt like a good place for that. Many fledging game developers use itch.io to preview and ship their games. itch.io is especially nice for games that can be played in a browser, like Liikennematto.

The release was a tiny success. In the first few months the game was played over 500 times and many itch.io users added the game to their collections. I also received positive feedback from the users.

In March 2023 I released a patch that added some variety to the content and improved the simulation. Around that time I also composed a review of the first three years of Liikennematto development in the form of the video below.

The future

Since early 2023 the development has slowed down quite a bit.

The last 12 months have been tough for me. Work became very stressful and I eventually changed jobs, for the better. In late 2023 I lost a friend to cancer, and recently another to a rare disease. I've been exhausted and I have grieved.

Hobby game development has offered me a way to channel my creative impulses to and escape from the negative aspects of daily life. Now I often have the time but not the energy for it. Our family is also about to grow, which itself is a wonderful thing, but the time for hobbies might become scarce. For this reason updates to Liikennematto have no timeline. Whatever time it may take, I still want to keep on developing Liikennematto to fulfill the vision that I have.

...

The next item on the roadmap is improved procedural generation of the lots and decorations. Despite slow and interrupted development, I've made some progress in implementing wave function collapse (WFC), a popular algorithm that fills a (portion of a) map with shapes that conform to constraints. With a tilemap-based game like Liikennematto, this means that each tile has multiple options (superposition) that might fit. When a tile option is picked (collapsed), the neighbor tiles' options are reduced to the point that the map resolves itself after several steps. WFC is often compared to sudoku, which has a similar solving process.

I've implemented complete map generation with multi-tile shapes and error handling. I'm now in the process of integrating WFC into the gameplay to achieve "driven WFC" - the player's decisions (road placement) drive the generation. This is quite a bit harder than generating maps without the player's input. Let's see how it goes - and how long it takes.

Here's a peek into WFC for Liikennematto:

In the meantime, you can play the most recent version of Liikennematto over at itch.io.

Next development updates and release announcements will be released on itch.io, marking this as the last devlog on this site. You can follow me on Mastodon for updates inbetween.

Liikennematto Github repository

Footer

You can’t perform that action at this time.

Mathematics

A centuries-old maths problem asks what shape a circle traces out as it rolls along a line. The answer, dubbed a “cycloid”, turns out to have applications in a variety of scientific fields

By Sarah Hart

Light reflecting off the round rim creates a mathematically significant shape in this coffee cup

Sarah Hart