May 11, 2009, 10:31 AM — You can play a video game called Fold.it on the Web while making important contributions to science. Understanding how 3-D proteins "fold" into their optimum structures is critical to understanding disease, but it's difficult computationally because there are an astronomical number of possible folds for most proteins.
The thousands of people who play Fold.it use their pattern-recognition and puzzle-solving skills to predict protein structures in a way computers can't easily duplicate. "Teenaged gamers are beating the pants off Ph.D. biochemists," says Ed Lazowska, a computer science professor at the University of Washington, where Fold.it was created.
Fold.it is an example of two enormously important trends. One is the marriage of computer science with other sciences such as biology, and the other is the emergence of hybrid systems in which people are not just users, but integral parts of the system.
I recently visited the computer science department at Carnegie Mellon University and was briefed on a dozen or so projects, all involving at least three disciplines. For example, biologists, linguists and computer scientists have combined medical models, language analysis tools and machine learning to understand the "language" of molecules and thereby detect the chemical signals of organ transplant rejection. A researcher said she couldn't imagine making progress without all those disciplines working together.
That same week, I attended a seminar on advanced applications of computer science, sponsored by the Computing Community Consortium. Alfred Spector, research vice president at Google, spoke of his company's vast "cloud" that not only computes but also taps into the intelligence of millions of users in order to make Google applications work better. He advised, "The focus of design should not be on the computer and software itself, but rather the computer and software in conjunction with its user community."
Then Luis von Ahn from Carnegie Mellon described his reCAPTCHA project. He's the inventor of CAPTCHAs, those distorted words on Web registration forms that people can read but computers can't. He figured out how to help digitize The New York Times' paper archives by harnessing the energy that millions of people collectively spend on this security measure every day. Words that the Times' optical character recognition software can't read are sent to a free CAPTCHA engine used by various Web services. Users are now deciphering 35 million words a day as they process these CAPTCHAs. Von Ahn says the job, which would have taken years with human editors, will be finished in just a few months.
Jon Kleinberg, a professor of computer science at Cornell, told of his pioneering studies of the social networks behind Facebook, Flickr, LiveJournal and other "systems" that incorporate the knowledge and experience of millions of users. He spoke of a profound change in the way knowledge is created and shared, once mostly top-down from a select few authorities, but now increasingly bubbling up from millions of individuals with wildly varying levels of authority, credibility and underlying intent.
This calls for new tools for verifying the authority and credibility of information. And Kleinberg said designers of hybrid computer-human systems should combine social models with traditional computer tools and employ algorithms that take into account game theory, incentives, reputation, recommendations and other human factors.
Even if you are in more traditional areas of IT, there are lessons here. First, look for ideas outside of IT. Be a Renaissance person. The days of stovepipe IT -- in systems, projects and people -- are over.
Second, make your users part of your systems -- an IT asset. Mine their preferences, study their habits, learn from their mistakes, and feed their recommendations and social behaviors back into your systems. Maybe you could give them an online game to play.
