SETI@home Classic: In Memoriam
The idea for SETI@home originated in a conversation between David Gedye and Craig Kasnov late in 1994 in Seattle. Gedye ran with it, and contacted UW astronomy professor Woody Sullivan, who led him to Dan Werthimer from the UC Berkeley Space Sciences Lab (SSL), whose SERENDIP SETI project provided a potential source of data from the Arecibo Observatory. Gedye also contacted Berkeley computer scientist David Anderson, a specialist in distributed computing.
This core group - Gedye, Sullivan, Werthimer and Anderson - met periodically from 1995 through 1998, working out scientific details and trying to raise money to get the project off the ground. They pitched the idea to some large companies, with no success. During this period Anderson developed a prototype of the client/server architecture (running on an SGI workstation in his house) and he and Gedye developed several variants of the screensaver graphics.
In 1997 Starwave (an Internet startup where Gedye was working) donated $10K to the project. This funded the development of data analysis code, designed by Werthimer and implemented by Jeff Cobb. Cobb also began working on the 'data recorder' machine that would reside at Arecibo and record data on tape.
In 1998, thanks to the efforts of Lou Friedman and Charlene Anderson at The Planetary Society (TPS), Paramount Pictures donated $50K in seed money, which was matched by TPS. At this point the project took form: it was to be based at SSL and led by David Anderson, with Werthimer directing the science. Two key additions were Eric Korpela, a Research Astronomer at SSL, who contributed to both science and software, and Matt Lebofsky, an SSL programmer and system administrator. This group (Anderson, Werthimer, Cobb, Korpela, and Lebofsky) have been the core of the SETI@home project ever since.
In June 1998 Anderson developed the client and server code in more or less its final form (this time based on an Informix database) and developed the graphics on Windows. Kyle Granger was hired to flesh out the Windows version (screensaver and installer) and Charlie Fenton ported the client software to Macintosh OS 7/8/9 and later OS X (Charlie eventually had a large role in all parts of the system).
In May 1999, after several months of testing, we launched the project. Ron Hipschman developed initial content for the web site. For several months, as participation soared, we fought performance problems; Sun Microsystems made numerous donations of server hardware.
We obtained a series of grants from the UC Digital Media Innovation program. This let us hire some part-time help: Eric Person, who did web development, and Bob Bankay, who managed (and continues to manage) our databases.
A demand arose for versions of SETI@home to run on architectures ranging from Atari to Cray. Hiram Clawson volunteered to manage this porting activity, and did a fantastic job; eventually SETI@home ran on about 40 platforms.
In late 1999 it was clear that we were spending too much time on infrastructure, and not enough on science. In 2000 Anderson (while continuing to lead SETI@home) joined the startup company United Devices, with the goal of eventually having SETI@home use that infrastructure. This didn't work out, and in 2002 Anderson left United Devices, returning to SSL to start work on BOINC, which was eventually supported by a grant from the National Science Foundation.
We employed a number of UC Berkeley Computer Science undergraduate students, including Eric Heien, Leonard Chung, and Peter Leiser, and Astronomy grad student Paul Demorest. In 2001 and 2002 we hired scientific programmer Steve Fulton to do back-end data analysis code, including the 'redundancy checker'.
SSL didn't have a machine room, so we put SETI@home servers in a wiring closet. This became overheated, which led to hardware malfunctions, which in turn led to database corruption. We spent a huge amount of time in 2000-2003 dealing with database issues. The original lifetime of the project (2 years) came and went, but we let it keep running since we hadn't done the back-end data analysis, and participant demand was still strong.
In May 2002 UC Berkeley's Internet connection ran out of bandwidth (due in part to the popularity of Kazaa) and SETI@home was forced to get its own dedicated connection, from Cogent.
In March 2003 we completed the back-end analysis of our results to that point, and were given 24 hours of dedicated time at Arecibo to reobserve the leading candidates. We completed the analysis of these reobservations, and failed to find any interesting signals.
By 2004 and 2005 we had more or less exhausted our funding. A skeleton crew (Cobb, Lebofsky, Korpela and Bankay) kept the project running, while simultaneously working on the new BOINC-based version. We hired another system administrator, Court Cannick, to help us make the transition.
Thanks to everyone listed above - and, of course, to the millions of dedicated, enthusiastic and patient SETI@home participants - for making SETI@home Classic what it was. It didn't find ET, but it was the largest computation in history, it revolutionized scientific computing, it brought science into the homes of millions of people around the world, and it increased global awareness of the methods and goals of SETI.
And, although SETI@home Classic has ended, we are continuing with SETI@home/BOINC, and hopefully we will soon begin a powerful new multibeam sky survey at Arecibo and - who knows? - maybe find that elusive signal we've all been looking for.
|Copyright © 2013 University of California|