SETI@home

» see setiathome.berkeley.edu as it lived, on the wayback machine

~ the obituary ~

i ran seti@home on the family pc for a stretch around 2000 and 2001. we had a pentium iii on ttnet dial-up in ankara and i had read about the berkeley project in a turkish computer magazine. the pitch was irresistible for a sixteen year old with a modem: install the screensaver, leave the pc running overnight, and your idle cpu cycles were being used to search radio telescope data for signals from other civilizations. i had no idea whether it worked. i just liked the little status window that told you it had processed a work unit.

seti@home launched on 17 may 1999 out of the berkeley seti research center. david anderson ran the software, david gedye had the original idea in 1995. it was one of the first mass-scale volunteer computing projects, arguably the first that made distributed compute famous. the concept was very simple: real seti observations at arecibo produced more radio data than the tiny berkeley team could ever analyze on their own hardware. break the data into small work units, ship them to home pcs over the internet, have the client scan for narrow-band signals, ship the results back.

the download was small, the install was trivial, and the screensaver was legibly nerdy. within its first year seti@home had more than a million registered users. within a few years it was the largest distributed computing project in the world, running on more machines than any single supercomputer of the era. the whole boinc framework, which now runs folding@home and dozens of other citizen-science projects, was built later at berkeley as the generalized version of the seti@home infrastructure.

~ the case file ~

~ the climb ~

the appeal was partly the aliens, partly the leaderboard. seti@home had teams. you could join a team, watch your work unit count tick up, watch your team climb the ranks. by 2001 there were team wars going on inside universities and companies, with sysadmins racking up work units on hundreds of machines during off hours. the project ended up with over 5.2 million people participating across its lifespan.

on the science side the yield was mostly negative. no confirmed extraterrestrial signal ever came out of the client scans, which was the expected result. what did come out was a demonstration that volunteer compute worked, that ordinary internet users would install a science client if the story was good enough, and that berkeley could federate a global supercomputer out of household hardware.

~ what got buried with it ~

on 31 march 2020 the project posted a note on its own forum saying it would stop sending out new work units. the client stayed installed on millions of pcs but did nothing new. the team's reasoning was honest. they had accumulated more raw results than they had the humans to analyze, and the science was in danger of being buried under its own backlog. hibernation gave them room to catch up.

the timing was accidental. the announcement went out two weeks into the covid-19 lockdown, when a lot of people were suddenly at home with idle machines and looking for something meaningful to run on them. folding@home picked up much of that traffic, since it had pivoted to sars-cov-2 protein simulation the same month. seti@home was the icon that folding@home took its story from. berkeley has kept the client servers up in case they ever want to restart, but no restart has come.

~ the long tail ~

the twenty-year run left behind a boinc ecosystem that other projects still run on, a generation of programmers who learned about distributed systems from watching a screensaver, and a set of results still being written up as papers years after the client went quiet. what remains for me personally is the memory of leaving that pentium iii running overnight through the ankara winter of 2001 so it could report back with another completed work unit in the morning. the aliens never called. the screensaver was the point.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »