From SETI@home to Hominid Fossils: Citizen Cyberscience Reshapes Research Landscape
Posted by Amir Alexander
2008/01/15 11:00 CST
In the beginning was SETI@home, the first large-scale volunteer computing project, launched in 1999 with seed money from The Planetary Society. Within months the project had millions of volunteers around the world joining to form the most powerful computer network ever assembled. Other projects soon followed, focused on everything from the search for large prime numbers to protein folding.
Then, in 2002, David Anderson, SETI@home’s own director, launched BOINC, the Berkeley Online Infrastructure for Network Computing. Instead of different projects each with its own platform and volunteer community there was now a growing family of projects, all sharing the BOINC platform. Volunteers no longer had to subscribe to a single project, but could divide their computer’s time as they saw fit between different BOINC ventures.
But just down the hall from SETI@home headquarters at the Space Science Laboratory in Berkeley, a different approach was taking shape. Planetary scientist Andrew Westphal was trying to figure out an effective way of detecting the interstellar dust grains embedded in the collector of the spacecraft Stardust. Computers, he found, could not pinpoint the elusive particles, but a practiced human eye could. The only problem was that no single person could hope to go over the millions of microscopic observations required for scanning the entire collector, and retain his or her sanity. Inspired by what Anderson and SETI@home chief scientist Dan Werthimer were accomplishing a few doors away, Westphal wondered: is it possible to harness the visual capabilities of thousands of volunteers, just as SETI@home harnesses their computer CPUs?
It wasn’t easy to translate the experience gained in volunteer computing to this new model that required the active participation of volunteers. But working in close cooperation with Anderson, and with support from The Planetary Society, Stardust@home was launched in August of 2006. Along with a NASA Ames project called ClickWorkers, dedicated to cataloguing craters on Mars, Stardust@home represented the first generation of scientific projects, in which members of the public working online contributed not computer time, but brain power.
The remarkably creative feedback loop at Berkeley's Space Science Laboratory did not end there. Westphal, who was originally inspired by the success of SETI@home to launch Stardust@home, now returned the favor. Just as SETI@home spawned BOINC, Anderson wondered, could Stardust@home beget its own universal platform? Only in this case, instead of facilitating projects where personal computers do all the work, the new platform would make it possible for volunteers to use their minds and skills in support of scientific projects.
Anderson and his team swung into action and within months developed a prototype for a platform to facilitate Stardust@home-like projects. They called it "Bossa," which stands for the "Berkeley Open System for Skill Aggregation." In many ways Bossa is similar to BOINC, and it makes use of all of BOINC's community features, including the credits, the competition, the teams, and the message boards. Where it differs is in the type of projects that it facilitates – volunteer computing for BOINC, "skill aggregation" for Bossa . There are many scientific projects, Anderson reasons, that could potentially benefit from the combined intelligence and skills of members of the lay public, but only the tiniest fraction actually make use of this approach. Bossa seeks to change this, by making it easy for scientists to launch public participation projects, and also provide a ready-made pool of dedicated volunteers. The new platfom, according to Anderson, is the natural next step in what he calls "Citizen Cyberscience."
Bossa is still in the developmental stage, but its first project is already preparing to launch. AfricaMap is a joint venture of the University of Science and Technology in Kumasi, Ghana, and the University of Geneva designed to review satellite images of the African continent. Volunteers who join AfricaMap will be scanning high-resolution images of remote regions in Africa where maps either do not exist or are sadly out of date. Their job will be to mark the precise locations of roads, rivers, villages, etc., thereby assisting efforts to aid impoverished regions and documenting the effects of global warming on the continent's landscape. According to Anderson, AfricaMap will go online within a few months.
While AfricaMap will be Bossa's first big test, Anderson already has other projects lined up to join the Bossa family. He is particularly excited about a project, not yet named, that for the first time will join Citizen Cyberscience with the search for human origins. The Awash valley in Ethiopia harbors one of the richest deposits of hominid fossils in the world, dating back as much as 4 million years. It was there in 1974 that paleontologists found the most complete hominid skeleton ever unearthed, and named her "Lucy." Every year during the wet season the rains wash away the top layer of the soil in the valley, exposing new fossils that were hidden just beneath the surface. That is when paleontologists scour the rich fossil beds of the Awash basin, looking for the remains of our ancient hominid ancestors. But time is short: soil erosion from the next rain or dust storms will soon cover up the fossils once more. As a result, even the most diligent fossil hunters can search only a small fraction of the exposed fossil beds before they are covered by earth and dust, and lost to researchers forever.
But when the new Bossa project launches, this could all change. According to plans, researchers will use a low-flying unmanned drone to take high resolution images of the fossil beds immediately after they are exposed by rain. Using Bossa, these images will be immediately sent out to volunteers around the world who will scan them on their computers. Just as Stardust@home volunteers are trained to detect the tracks of interstellar dust particles, the new project's participants will be trained to recognize hominid fossils in the images, mark their location, and return their results to project headquarters. With these results in hand, paleontologists will be able to cut the time it takes them to locate fossils to a fraction of what it is today. With a complete overview of the fossil beds in hand, paleontologists will head directly to the locations of the most promising fossils, and extract them for further study before they are once more swallowed by the Earth. By substituting the thousands of eyes of Bossa volunteers around the world for the exhausting labor of a small team on site in Ethiopia, the new project could revolutionize the gathering of hominid fossils. There is no telling how many precious fossils will be saved and collected in this way, and what they could tell us about our origins and distant ancestors.
Projects such as AfricaMap and the hominid fossil venture are closely modeled on Stardust@home in making use of volunteers' proven ability to recognize the finest features of high-resolution images. But while planning to implement these on the experimental Bossa platform, Anderson is also thinking ahead to more complex uses of the skills and intelligence of citizen volunteers. His focus in on "Rosetta," a well-established BOINC project, run from the University of Washington, that investigates the folding of proteins on volunteer's computers. Proteins are long strings of amino acids, which are the building blocks of life as we know it. In order to carry out their functions, however, proteins must spontaneously and almost instantaneously fold themselves in very precise ways. This process, which takes place repeatedly and reliably in biological systems, is an essential but little understood feature of life. Rosetta scientists are studying it by using volunteer's computers to model the "lowest energy" folding configurations of various proteins.
It turns out, however, that computers are not the only ones capable of determining ideal folding configurations. Some humans are good at it too, so good in fact that the very best can do it better than any computer. Can such unique and sophisticated human capabilities be harnessed for scientific research through volunteer computing? Anderson and his partners at Rosetta believe that they can. Together they are designing what amounts to an online video game, in which payers compete with each other in finding the lowest-energy folding pattern of specific proteins. If the game proves popular, they hope it will attract those unique individuals with a special "knack" for this complex task. All this work on Bossa and its different applications "has gotten me very excited" said Anderson. "It's just like in the early days of BOINC," he said, "trying to find a single common platform for a range of different applications."
And what of SETI@home, the granddaddy of all of today's citizens' cyberscience projects? It too is still going strong, forever expanding its capabilities for searching out intelligent life in the universe. Over the past few years SETI@home has begun collecting data using the new multi-beam receiver at Arecibo, instead of the old single beam it had used since its launch in 1999. This means that the project can now scan the skies much faster and more efficiently than before, because it is looking at seven different points simultaneously. Furthermore, the new hardware put in place makes possible a far more sensitive search over a frequency band 40 times greater than before. The end result is that the new SETI@home is generating 500 times as much data as before, about 300 gigabytes a day, or 100 terabytes (100,000 gigabytes) per year. According to project scientist Erik Korpela, this is about the amount of data stored in the U.S. Library of Congress…
SETI@home is still the largest scientific public participation in history, with 170,000 active volunteers around the world running 320,000 computers. But with such massive amounts of data to process, the project is looking for even more volunteers to join the search. "The next generation SETI@home is 500 times more powerful than anything anyone has done before" said Werthimer. "That means we are 500 times more likely to find E.T. than with the original SETI@home."
With projects in the works ranging from SETI@home's volunteer computing to Rosetta's cutting-edge computer game, Anderson is hopeful about the future prospects of Citizen Cyberscience. "The larger question for the long-term," he said, "is how far can this idea be pushed?" Traditionally, he explained, the lead scientists would conduct the highly skilled portion of an experiment, requiring knowledge and sophisticated thinking, while their assistants were charged with the more mundane tasks. But with Citizen Cyberscience members of the lay public are undertaking increasingly sophisticated and complex tasks. "How far up the chain of scientific research can public participation go," Anderson wonders. The answer, at this point, is anyone's guess. But with new generations of volunteers pushing the frontiers of Citizen Science ever further, it is clear that the limit has yet to be reached.