By | November 13, 2023
Image of a series of bar graphs in multiple colors.

Magnify / This image has a warm, nostalgic feeling for many of us.

Distributed computing burst onto the scene in 1999 with the launch of SETI@home, a fancy program and screensaver (back when people still used them) that sifted through radio telescope signals for signs of alien life.

The concept of distributed computing is simple enough: you take a very large project, break it into pieces, and send individual pieces out to computers for processing. There is no inter-PC connection or communication; everything is done via a central server. Each part of the project is independent of the others; a distributed computing project would not work if one process needed the results of a previous process to continue. SETI@home was a prime candidate for distributed computing: Each individual unit of work was a unique moment in time and space as seen by a radio telescope.

Twenty-one years later, SETI@home shut down after finding nothing. An incalculable amount of PC cycles and electricity wasted for nothing. We have no way of knowing all the reasons people quit (feel free to tell us in the comments section), but having nothing to show for it is a pretty good reason.

Rises and falls

SETI@home’s story is emblematic of the churn that characterizes the world of distributed computing. Another major effort came from IBM; its Corporate Social Responsibility division was involved in the creation of the World Community Grid, a series of life science projects looking for treatments for AIDS, cancer and Alzheimer’s. IBM donated its technology and talent to the project, which started in 2004. But in 2021, IBM transferred the World Community Grid assets to the Krembil Research Institute, part of the University Health Network (UHN) in Toronto. A UHN spokesperson declined to comment for this story.

With the outbreak of the COVID pandemic, there was a new darling of the distributed world: Folding@home, a simulator that tries to understand how proteins adopt functional structures. Folding@home had been around for more than 20 years, simulating protein folding to understand how diseases formed. And it had something to show for this work: more than 230 peer-reviewed articles on its findings over the decades. But with proteins from SARS-CoV-2 to study, Folding@home became the It project. So many people launched it on their PCs that it broke the exaFLOP barrier long before supercomputers did.

But as the pandemic subsided, so did interest in the project. Greg Bowman, director of Folding@home and professor of biochemistry at the University of Pennsylvania, said the project soared from 10,000 active users to 1 million but quickly dropped to around 45,000 active users — still a big gain compared to before . pandemic number.

Bowman believes there is a combination of reasons why interest is falling. “The pandemic provided huge motivation and a lot of time for new hobbies. Many organizations had spare computers that they redirected to Folding@home. An example: FIFA didn’t have to scan YouTube for pirated content because no games happened.” But that didn’t last. “Inflation and energy prices skyrocketed,” Bowman said.

Even, an aggregator of distributed projects, had gone a few years without an update before the January 2023 update. But the site’s operator, Kirk Pearson, says he hasn’t abandoned the project; he has only been busy with real things.

Direct engagement

Pearson suggested that the center of attention has shifted to projects that more directly involve users, primarily in a place called Zooniverse. “I agree that there have been fewer large-scale new distributed computing projects started in recent years, but look at all the new projects that have started at Zooniverse in recent years to see how much innovation and creativity is going on in distributed computing and in distributed human-driven projects,” he said.

Pearson said Zooniverse is for “distributed human” projects, where people do the work for the project that computers can’t, such as identifying the number and types of animals in a photograph. Because Zooniverse has a relatively simple framework, it is easy and cheap for anyone to create a new Zooniverse project. It has led to the current spread and range of projects within the Zooniverse.

“It’s possible that volunteers move to Zooniverse projects because they want more involvement in projects, but I don’t think there’s enough evidence to support that claim. I think volunteers are drawn to new projects.” He suspected that “if you could follow trends in levels of participation in Zooniverse projects, you would see that the majority of volunteers move from existing projects to new projects because of the novelty of the new projects.”

Other people have attributed the decline to additional factors. David Anderson is a computer scientist at the University of California, Berkeley, former director of the SETI@home project and current director of the BOINC distributed computing project. He wrote a long paper on the topic in early 2022. He said that distributed computing peaked in 2007 and has been in decline since then for a number of reasons, from a lack of mass voting to a lack of volunteers and a strong leader to no presence of Social Media. He declined to comment further.

Practical questions

The increase in energy costs is another major contributor to the decline in distributed computing. Altruism will only get you so far when your well-intentioned hobby causes the electric bill to go up. Electricity bills are generally increasing, and the new generation of desktop CPUs and GPUs are incredibly power hungry. Making matters worse, fewer people have this hardware due to the general shift from desktops to laptops. Running a CPU at full utilization will drain even the most efficient batteries. “Things have come down sharply when people are hit with energy prices and inflation,” Bowman says. “I think there is a lot of interest, but energy prices and other economic factors have hurt participation.”

Then there’s the fact that they have nothing to show for it. Many SETI@home users probably felt they had burned through all that electricity for nothing. In other cases, the project’s scientific research and individual findings have not really stood out – it has gone into a pool of findings that can be discussed as part of a research paper. And the users themselves may not even know about it.

Then there is the lack of a strong, charismatic leader, as Anderson mentioned. How do you get people fired up and excited to leave a computer running to possibly find a scientific nugget? It’s a hard sell, and in the absence of a charismatic leader, no one will sell it.

Finally, the projects are often boring. What initially got people excited about running Folding@home was the possibility of finding a cure for COVID. With SETI@home, there was the potential to find signals from another world. But how sexy is something like Einstein@home, which looks for neutron stars? Changing the world is easier to sell.

While it may have passed its peak, distributed computing hasn’t completely disappeared, and it still draws enough computing power to be useful. Pearson said that while some of the more well-known distributed computing projects have ended, most of the lesser-known projects he covers are still going strong. “Distributed computing is dying no more than chemistry or mathematics is dying,” he said.

Andy Patrizio is a freelance technology journalist based in Orange County, California (not entirely optional). He prefers building computers to buying them, has played for too many hours Angry birds on his iPhone, collecting old coins when he has some left over.

#distributed #computing #dying #fading #background #Ars #Technica

Leave a Reply

Your email address will not be published. Required fields are marked *