Well, success could either be measured by successful detection of an ET signal or by the sheer amount of computing that has been done. In this case it's the latter. In the article they note that roughly 4.3 x 10^20 flops have been performed in the name of S@H using more than 500,000 years of CPU time, which is a major computing landmark in and of itself.
As for successfully finding a signal, they do mention that postprocessing has yet to move into full gear. Even when it does start, it usually takes several years to complete (at least based on the timeframes for other SETI projects).
One interesting idea I've contemplated is that S@H and other similar distributed computing projects have actually had a positive effect on the technology sector of the economy. Without a doubt, there are people that care only about their stats, and are willing to purchase faster machines in order to boost their stats (and hence create a bit more demand for hardware). Given that there are over 2 million participants, this effect adds up. I'd be curious to see estimates or a study on this matter if anyone knows of some.
The close call mentioned wasn't really as much of a close call as one might think, as it involved an fairly obvious RFI anomaly with spikes. When the odd spike powers was first noted, it was almost a certainty that it was caused by either RFI (Radio Frequency Interference) or equipment problems. Given that many other SETI searches check the same bandwidth, there would have most likely been a great deal of investigation into it before if it was naturally occurring.
What proves a bit more interesting is the sort of analysis done in Newsletter #4. An extraterrestrial emitter, assuming it's constantly sending a signal, would trigger a gaussian detection rather than a spike. The only problem with the analysis that was explained in Newsletter #4 is that by looking through the top 10,000 gaussians only, one is pretty much guaranteed to only find signals from satellites passing over the telescope.
The "heavy lifting" analysis (going on right now) essentially involves mapping all 125 million gaussians against the sky, filtering them based on how many times the telescope passed over that part of the sky, taking into account redundant signals and equipment problems and RFI, and finally looking for areas that have a disproportionately high number of gaussians. When one of these areas is found, it's checked more carefully, the tapes are reanalyzed, the area is checked again with other telescopes, etc, before an actual ET claim is made.
Other SETI projects (such as SERENDIP) usually continue data analysis a few years after the project is done. In other words, the proof that we're all looking for could be buried in the S@H database, we just have to wait a while for it to be found!
A distributed nuclear simulation project wouldn't work that well, since it can't be split into parallel tasks. SETI@home works quite well because each work unit is independent of the others, but I'm pretty sure that in a nuclear bomb simulation you can't see what's happening at the end before seeing what happened in the beginning. And for the record, SETI@home is usually around 12 teraflops nowadays, getting faster as PCs themselves get faster.
Slashdot Mirror
As for successfully finding a signal, they do mention that postprocessing has yet to move into full gear. Even when it does start, it usually takes several years to complete (at least based on the timeframes for other SETI projects).
One interesting idea I've contemplated is that S@H and other similar distributed computing projects have actually had a positive effect on the technology sector of the economy. Without a doubt, there are people that care only about their stats, and are willing to purchase faster machines in order to boost their stats (and hence create a bit more demand for hardware). Given that there are over 2 million participants, this effect adds up. I'd be curious to see estimates or a study on this matter if anyone knows of some.
What proves a bit more interesting is the sort of analysis done in Newsletter #4. An extraterrestrial emitter, assuming it's constantly sending a signal, would trigger a gaussian detection rather than a spike. The only problem with the analysis that was explained in Newsletter #4 is that by looking through the top 10,000 gaussians only, one is pretty much guaranteed to only find signals from satellites passing over the telescope.
The "heavy lifting" analysis (going on right now) essentially involves mapping all 125 million gaussians against the sky, filtering them based on how many times the telescope passed over that part of the sky, taking into account redundant signals and equipment problems and RFI, and finally looking for areas that have a disproportionately high number of gaussians. When one of these areas is found, it's checked more carefully, the tapes are reanalyzed, the area is checked again with other telescopes, etc, before an actual ET claim is made.
Other SETI projects (such as SERENDIP) usually continue data analysis a few years after the project is done. In other words, the proof that we're all looking for could be buried in the S@H database, we just have to wait a while for it to be found!
A distributed nuclear simulation project wouldn't work that well, since it can't be split into parallel tasks. SETI@home works quite well because each work unit is independent of the others, but I'm pretty sure that in a nuclear bomb simulation you can't see what's happening at the end before seeing what happened in the beginning. And for the record, SETI@home is usually around 12 teraflops nowadays, getting faster as PCs themselves get faster.