excellent idea. 


As long as there is light, the darkness holds no fear. And yet, even in the deepest black, there is life. - Arklan Uth Oslin
I want to leave this world the same way I came into it: backwards and on fire. - Arklan Uth Oslin

On Thu, Jan 3, 2013 at 9:47 AM, Gavin Andresen <gavinandresen@gmail.com> wrote:
On Thu, Jan 3, 2013 at 11:16 AM, Arklan Uth Oslin
<arklan.uthoslin@gmail.com> wrote:
> all this sounds reasonable to me. and i agree with the major pitfalls. ie:
> how do we know someone did what they say they did and such. that's actually
> been rattling around my head for a while. unfortunately i can't really think
> of any way to solve the problem.

Agreed; I think we need to "just do it" and expect to lose a few BTC
to scammers.

If it is a big problem, we could try a "trust but verify" system,
where we ask test executors to run bitcoin with the -logtimestamps and
-debug options and save their debug.log files.  If there is a question
of whether or not they actually executed the test plan, the debug.log
could be pretty good proof that they actually did (depending on what
is being tested, of course). For the amount of BTC we're talking about
it would probably be more work to forge a fake debug.log than just run
the tests...

Actually, come to think of it, asking testers to routinely save the
debug.log files is a good idea, because it makes it easier to track
down issues.

Gavin Andresen