Why are there so many old computers in the scientific community that are still in use?

The original author | Anna Nowogrodzki

Keeping vintage computers running is an important task and a pleasure.

NASA's Solar Dynamics Observatory, which operates 22,000 miles above Earth's surface, is a solar monitoring satellite launched in 2010 and filled with high-tech sensors. But as late as 2015, there was still a real vintage device in the many ground systems that supported these sensors.

"Hollywood always paints a particularly cool picture of cutting-edge science," says James Mason, a heliophysical scientist at the University of Colorado-Boulder. His team built an instrument on the Solar Dynamics Observatory.

But Mason's first trip to the White Sands Test Facility in New Mexico— a lab that launches rockets to calibrate his team's instruments on the solar dynamics observatory — shocked him with the sight of data streaming into a boxy, 1980s custom desktop computer spitting out yellow-green pixel text. Mason said, "It's so old, I can't find any information about it on the Internet. ”

Illustration: The Project Twins

His experience is not unusual. From geochemistry to linguistics to biochemistry to microscopy, antique computers can be found everywhere in the scientific community. Often, an old computer is connected to an expensive set of scientific tools— such as a microscope or a chromatography system— whose software is incompatible with the new computer or is too expensive to upgrade. Sometimes old computers just work, or people need it so much that they don't have time to retire it for a while to upgrade it. "Some of the equipment I'm using is older than I am," said Kristin Low, a 30-something enzymologist at Agriculture and Agri-Food Canada.

As the saying goes, "If it's not broken, don't fix it." "Whatever the reason, for many scientists, keeping their old computers running is crucial — and a pleasure.

Why use old antiques?

At the Bodega Marine Lab at the University of California, Davis, some computers still run Microsoft's Windows XP operating system, released in 2001, which Gary Cherr, director of the lab, says is designed to maintain compatibility with scanning laser confocal microscopes and other imaging devices. With the current Windows operating system, the entire team would need to replace the entire microscope. Cherr believes that the potential marginal gains are not worth spending $400,000.

Sometimes it's because upgrading the software itself is too expensive. In 2008, Low used an Intel 386 processor-based computer in mcGill University's undergraduate lab, running the 16-year-old "advanced" Microsoft Windows 3.1 operating system to connect to their liquid chromatography system. Low says the chromatography software will also need to be upgraded to support the upgraded operating system, which would cost $10,000 — prohibitive.

Like many labs, Low's team is still using old computers because they don't want to mess around with what has worked. This chromatographic system "is something that everyone is going to use to purify proteins, every day, every moment," she said. Bjoern Brembs, a neuroscientist at the University of Regensburg in Germany, will use an Intel 486-based system until 2020. "It's been running almost every day since 1994." He said it was connected to a homemade fruit fly flight simulator. "When you're a new PI or a postdoc, you really need this stuff to produce a lot of data and don't have time to upgrade it."

Some laboratories use old computers to meet technical requirements. Linguist Lauren Ackerman manages a lab at Newcastle University in the United Kingdom, where there is a custom computer that is still running the 12-year-old "advanced" Windows 7 operating system for ultrasound imaging of a person's tongue. The main demand for the computer is storage: it needs to preserve large amounts of image, audio, and video data for each one-hour meeting; linguists use this data to study how different tongue shapes produce different sounds. The computer also needed a specific type of expansion slot — for communicating with imaging hardware — and it had to be able to capture audio, vision, nasal airflow, and electrical signals from laryngeal electrodes that measured whether the vocal cords were open. Ackerman estimates that a new computer that meets these requirements will cost thousands of pounds and is not yet compatible with old equipment, and her team hopes to use it for teaching purposes.

For Brembs, the old computer also had a key feature that was gone since Windows replaced the text operating system MS-DO (Microsoft Disk Operating System): MS-DOS could "process data as it entered, without buffering delays," a feature he took advantage of in his drosophila flight simulator. "There are many processes going on in the background of Windows at the same time," Brembs says. You may want to measure in precise 50-millisecond intervals, but the operating system may only be able to process at an average of 50 milliseconds, with actual intervals ranging from 20 to 80 milliseconds, depending on what else it needs to do. Brembs said: "In fruit flies, such a large delay is perceived. ”

Old and "strong"

Computers are being replaced every few years, and it seems inconceivable that the systems of decades ago are still in use. Older computers, however, can be unexpectedly durable. Brembs said his computer never went on strike in the 26 years he used it. Mason said the machine he used to launch NASA rockets continued to work for more than 30 years, until 2015. The machine was extremely simple to operate and it was impossible to connect to the Internet, even if they wanted to. Eventually, the flat rainbow-colored ribbon cables on the machine (which have been replaced by more robust tubular cables on modern computers) began to wear out due to repeated insertions. They changed one before the machine broke down.

Karin Forney, a marine ecologist at the National Oceanic and Atmospheric Administration's Southwest Fisheries Science Center, has been recording data in the wild using a Toshiba T1000 laptop since 1987. Her research included flying in a small plane, "only the size of a winged Mini Cooper," to investigate harbor porpoises and leatherback turtles. "There are four things you wouldn't do with a computer, three of which we do in our daily work," she said. She's referring to putting it in wet, sea breezes and extreme temperatures.

Forney has experimented with the latest laptops, even those that are "supposedly indestructible." "They were both mushy in a week," she said. But that Toshiba was perfectly suited to the wild: it was small enough to be used in a narrow cockpit, Forney sat next to the pilot but had to not interfere with the flight control system; the LCD screen could read even in bright sunlight; the heat dissipation was good enough not to burn her knees; and it rarely collapsed. Forney concludes that it is "really strong."

Since there is no internal hard drive, Forney writes data to the laptop's RAM card. She thinks it's an advantage: because the vibration and heat of a small airplane can have a big impact on hard drives. She then used a 3.5-inch floppy disk to transfer data to a more modern computer. External floppy disk readers are sold everywhere for about $20. In general, computers running outdated software cannot be connected to the Internet, as any software that is no longer updated is affected by malware and newly discovered vulnerabilities. Many organizations such as Akerman's units have IT policies in this regard. She said: "You don't want bugs. ”

Bugs aside, nothing is eternal. When an old PC breaks down, researchers need to do everything they can, be creative, and get lucky.

Physical work and creativity

Forney had bought a stack of 10 Toshiba laptops from eBay that were made in 1987, each priced from $10 to $12. "We only have the last one left now," she said. When Forney bought these old computers, she estimated that half of them still had useful parts, which she used to replace the broken keyboard, battery, screen and floppy drive. I call them Fisher-Price computers (American Toy Company, whose products are disassembleable) because I took them apart and replaced them with parts. There really are only five parts in there. "Damage to the power supply was a fatal blow because it was soldered to the motherboard and could not be replaced.

When the acquisition plate for Cherr's confocal microscope broke down four years ago, he contacted Olympus, the manufacturer, to buy a new one. Olympus' staff told Cherr that the new plate didn't match the old microscope, but luckily they found an old plate that would be sent to him for free. "We all understand that the current situation is just a matter of surviving." Cherr said. "If the plate breaks down again, we'll have to go to eBay and see it" — or replace the entire microscope.

Another strategy is to customize a new computer. When the computer connected to Cherr's fluorescence spectrophotometer didn't work, he did it. The old computer had an ISA slot, a motherboard component of the 1980s that was used to connect sound cards, graphics cards, and other devices, and fell out of favor after the mid-1990s. But to plug in the acquisition board (to communicate with the spectrophotometer) you have to have this slot. Cherr bought a brand new custom computer with an ISA slot for an extra $1,000. But not every time: Li Liu, a neuroscientist at the Chinese Academy of Sciences, uses a computer he assembled in 2000 that interfaces with a flight simulator — the same one Brembs uses. But in 2018, "the computer crashed several times," he said, "and I decided to replace it." Unable to find a motherboard with an ISA slot, he bought a new computer and instead connected it to the spectrometer via USB.

In a pinch, you can always try the nearby computer graveyard. As an undergraduate, Low walked into the lab one day and found that the computer connected to the liquid chromatograph would not start. Luckily, she is a computer enthusiast and volunteers in a student body that manages computer labs. She found out that the power supply was broken, so she went to the club's dark and moldy underground storage room. In boxes and a pile of computers, she found several backup power supplies and used one of them to bring the computer back to life. The chromatography device "is the best in its class," she said, "and you can't just go out and buy a new $500,000 machine." ”

Originally published under the title Old-school computing: When your lab PC is ancient, it was published in the technical feature section of Nature on June 1, 2021

This article is reprinted with permission from Nature Portfolio (ID: nature-portfolio), the original title is "Why are there many old computers in the scientific community still in use? | Nature technology close-up", if you need to reprint it for a second time, please contact the original author.