Grid computing is rapidly becoming a dominant computational paradigm. Our paper shows how large scale grid (or "distributed" computing) can be used is a new way to tackle long timescale dynamics. Our methods should be directly applicable to other researchers, especially as the available computational power grows.

Understanding how a chain of amino acids forms a distinct three-dimensional structure, or fold, is a long-standing challenge. To understand this process in atomic detail is a goal of simulation efforts. To improve and test such simulations it is crucial to compare and experiment. The most simple protein systems fold very rapidly, in millionths of a second. However, even with today's computers it is difficult to generate detailed simulations that exceed billionths of a second. Indeed, simulating folding on the microsecond timescale would take anywhere from several years to several centuries. Using tens of thousands of computers and a novel method for examining long-timescale dynamics, we can tackle this problem directly. In looking at an entire ensemble of individual molecules, we can examine the early folding events by relying upon the statistical nature of the folding process.