Re: Folding@Home Package

[Magnus Holmgren <holmgren@lysator.liu.se>]

Sebastiaan Couwenberg wrote:
> F@H doesn't update its client regularly. The release cycle is possibly
> slower than Debian Stable. There has been talk of v6 for several years
> now, and I think that version just recently went into alpha testing if
> it made it that far (I'm no longer a beta tester with early access to
> project info, so I can't verify that).
> 
> The v4 F@H client still works eventhough v5 has been out for a long
> time. Its usefulness is somewhat reduced because there are no more
> deadlineless WUs handed out by the project, but it still works. I don't
> think this is a real problem.

Has there been any talk about moving to the BOINC infrastructure?

>> Lastly, I am not sure that closed-sourceness is the best strategy against
>> cheating. I guess that the expertise area of folding@home is structural
>> biology, wheras the expertise of cheaters is... well... cheating.
> 
> Guaranteeing the integrity of the research is indeed the primary reason
> to keep the F@H client and cores closed-source. Even though they are
> build with GPL components like Gromacs. But "Folding@Home has been
> granted a non-commercial, non-GPL license for Gromacs, so [they] are not
> required to release [the] source."
> http://folding.stanford.edu/gromacs.html

But the only way to be sure that the individual users don't cheat would
be to require Treacherous Computing. Instead, BOINC-powered projects
rely on handing out several copies of each work unit and checking that
the reported results match. Perhaps F@H does this as well.

I don't think that all BOINC projects release source, but SETI@home code
is free (as is the BOINC client itself), for example, and in that case
it's possible to package an optimized version. Otherwise the BOINC
client automatically downloads the right executable.

-- 
Magnus Holmgren               holmgren@lysator.liu.se