What does stardust have to do with curing Alzheimer’s disease?

With Guest Blogger, Pietro Michelucci

We would not exist without particles that were created during the violent death throes of aging stars. Indeed, “We are made of star-stuff.” 26 years after Carl Sagan delivered this mind-blowing message to millions of PBS viewers, NASA’s Stardust mission brought back to Earth the first-ever samples of cosmic dust, which contained secrets about the origin of the universe. The NASA scientists had one small problem: finding the dust particles would require sorting through one million microscope images taken from the satellite’s collector grid. They estimated it would take approximately 100 years to find the particles, unless they could somehow get a supercomputer to do the work.

Stardust Mission: After collecting interstellar dust, the collector grid was returned to Earth for microscopic analysis.

THINK ABOUT IT: Supercomputers might seem like rocket science, but the basic idea behind building a supercomputer is actually quite simple: find a bunch of “everyday” computers and string them together. The fastest supercomputer in the world today combines 41,000 smaller computers and consumes “only” 15 million watts, making it one of the most efficient of its kind. It performs 90 quadrillion (million-billion) numerical calculations per second.

…the basic idea behind building a supercomputer is actually quite simple: find a bunch of “everyday” computers and string them together.

One of these space scientists, Andrew Westphal, knew that his UC Berkeley colleague, David Anderson, had managed to build a supercomputer by writing software (SETI@home) that millions of people downloaded so their personal computers would help crunch data when they were sitting idle. This supercomputer was used to analyze radio telescope signals from the SETI project to look for signs of extraterrestrial intelligence.

The SETI@home screensaver software, running on millions of Internet-connected PCs formed a supercomputer.

THINK ABOUT IT: Most people think information processing began 75 years ago, when digital computers were invented. It actually began about three billion years earlier with the emergence of life. Bacteria, the oldest known life forms, could respond to their environment and reproduce — both involving computation. Since then, life has evolved to include humans, which employ massive networks of specialized brain cells called neurons, which process information. The human brain is arguably the most complex computing system in the known universe. So how does nature’s computers stack up against a supercomputer? It can processes information at about the same speed as the fastest supercomputer, while consuming one million times less energy.

Most people think information processing began 75 years ago… It actually began about three billion years earlier…

Westphal knew that machines would not be able to find dust trails in the aerogel images. So he asked Anderson if they could somehow build a supercomputer out of people to do this job quickly. The answer was “yes” and they built an online virtual microscope called Stardust@home that engaged 30,000 volunteers, including me, to analyze those images while learning about space science. Problem solved. In just a few years instead of a century, the human-powered supercomputer found the stardust from which people arose and changed our understanding of the cosmos.

The Stardust@home “virtual microscope” enabled 30,000 people to work together as a cognitive supercomputer.

THINK ABOUT IT: If the human brain is so powerful, why do we need supercomputers at all? The simple answer is that natural and artificial computers think in very different ways. For example, humans are good at predicting what will happen next, while machines are better at crunching numbers. Although today’s fastest supercomputer may be as quick as a human brain, it is basically just a very fast calculator. Sometimes we need a very fast calculator.

Indeed, while finding dust trails in aerogel is easy for people, distributing a million images to 30,000 people and keeping track of who has seen which image is a job not well suited to humans. But machines connected to the Internet do this sort of thing very easily. This kind of partnership between humans and machines works so well that we now use it to solve lots of problems. We even have a name for it: crowdsourcing or, when it helps research, “Citizen Science.”

If the human brain is so powerful, why do we need supercomputers at all?

When I met Chris Schaffer two and a half years ago, I was a hammer looking for a nail. I had been studying this human-machine partnership for years and had decided it was time to put the science to good use. Schaffer described his promising Alzheimer’s treatment research but explained that the painstaking data analysis would extend their efforts for decades. I asked him to show me exactly what was taking so long. What I saw next gave me chills.

The search for interstellar dust particles was similar in important ways to the analysis needed for Alzheimer’s research.

THINK ABOUT IT: We are now entering into a second generation of citizen science projects that can benefit from the lessons learned in pioneering projects like Stardust@home. Now that we have direct evidence of ground-breaking scientific results that could not have been achieved any other way than harnessing the power of the crowd, we can pay close attention to what made these first generation platforms work and try to replicate their successes.

…results … could not have been achieved any other way than harnessing the power of the crowd…

The laboratory analysis Schaffer showed me — the one slowing down his research — reminded me exactly of the Stardust@home project, which was very familiar as I had spent hours using it as a volunteer “duster” in 2006. A few phone calls later, Andrew Westphal graciously joined us in a new “EyesOnALZ” initiative to help retrofit his Stardust@home platform to investigate dementia and help find a treatment orders of magnitude faster. Today, this new online game called “Stall Catchers” already has over 5,000 registered volunteers and has been validated to provide a year’s worth of accurate scientific analysis in only one short month.

By playing Stall Catchers, public volunteers accomplished a year’s worth of Alzheimer’s data analysis in one short month.

It seems fitting that we have collectively leveraged the most complex information processors in the known universe to find the first original bits of the known universe. To me, it is marvelous that these same methods are now being used to salvage the very integrity of those same information processors — aging human brains. As EyesOnALZ grows, this will be an important piece of the human story.