The latest project to move to BOINC was announced last month at UC Berkeley. Climateprediction.net, which runs models of global climate change, joins SETI@home, which searches for radio signals indicative of extraterrestrial intelligence, and Predictor@home, which predicts how proteins fold. Climateprediction.net was launched a year ago and now boasts 75,000 participants in 130 countries.

"With BOINC, you can participate in many different public computing projects and control the amount of time each gets", stated David Anderson, the director of both BOINC and of the largest public distributed computing project, SETI@home. "This will be good for everybody, because it uses computing resources more efficiently. It's the dawn of a new era of computers and computational science."

David Frame, the project co-ordinator for climateprediction.net, which is based in the physics department at Oxford University in England, agreed. "This will really help us quantify the uncertainty in our climate prediction model", stated David Frame. "But we also hope people will learn about climate and why global climate change is important."

The BOINC platform makes it easier for science projects to develop a distributed version of their software to take advantage of the many idle computers in the world, which together exceed the combined computing power in the business and academic world. Public distributed computing is open to the whole world, as opposed to Grid computing, which harnesses only computers within organisations.

In the next year, the three BOINC-based projects should be joined by several others, according to David Anderson. BOINC was developed by David Anderson and his colleagues at UC Berkeley's Space Sciences Laboratory with assistance from the National Science Foundation, and they have posted the open-source software on the Web. It's available for Windows, Mac OS X 10.3, Linux and Solaris.

David Anderson, who developed the software that allows SETI@home to run on some 5 million computers worldwide, launched the BOINC project not only to save others the hassle of reinventing the wheel, but also because SETI@home was attracting more users than the project knew what to do with.

"SETI@home has an embarrassment of riches - too much computing power", he stated. Even when he and scientific director Dan Werthimer took advantage of the vast computing resources to reanalyse each packet of data, to look for new features in the data or to plug in new sources of data, they still had computers needlessly repeating calculations. So David Anderson decided to develop a "general purpose system that lots of groups could plug into, in areas such as biology and earth science as well as astronomy".

A beta version of BOINC was introduced last December, with support from Sun Microsystems Inc., as SETI@home prepared to move its 5 million users to the platform. Predictor@home, based at The Scripps Research Institute, moved over shortly afterward, and David Anderson is working with several other projects that should transition to BOINC within a few months: LHC@home, a project based at Geneva, Switzerland's European Organization for Nuclear Research (CERN) that will simulate its Large Hadron Collider, which is now being built; and Einstein@home, a collaboration with the gravity wave experiment LIGO (Laser Interferometer Gravitational Wave Observatory) based at Caltech.

David Anderson hopes each project will generate a community around itself, much like the community of 500,000 dedicated users around SETI@home. "We hope BOINC will lead to an environment where people can learn about lots of projects , and sign up for the ones they think are the most worthwhile", he stated.

Climateprediction.net, for example, could do much to educate the public about global warming, David Frame stated. The detailed model of the Earth's climate used by this project - one of a dozen climate models used commonly in labs around the world - has been developed over decades but is so complex that the researchers themselves don't understand how sensitive it is to uncertainties in its various parameters, such as how clouds are taken into account. With a distributed computing project, they will be able to run the model with various values of the parameters and look for wild swings resulting from tiny changes in the parameters - a signal that a parameter is highly sensitive to the modelers' lack of knowledge.