> I don't really understand the urgency to get these exponents
> finished, either.
> So what if it takes 2 years to finish that check? Is one year to
> long? Six
> months? Six weeks? Should we tell people with Pentium 100s that
> we don't
> want their help? Considering we have an unlimited supply of
> exponents to test
> (barring a unexpected revelation that there are a finite number
> of Mersenne
> primes), we are better off (in terms of total work done per day)
> with older
> machines in addition to your new PIII-550's than we would be without them.
In one year, we just might be able to do first time checks on every exponent
under the currently unknown M38. Wouldn't it be nice if we could verify
that it is indeed the 38th Mersenne Prime, and not just the 38th *known*
Mersenne Prime?
I *gaurantee* that if EVERY OTHER NUMBER except this little group of a dozen
or so were tested, and only these few held us back from verifying the order
of the list, there would be very strong argument to get them tested right
away. If anything else, I'm just guilty of jumping the gun by a few months,
but considering that some of these exponents have been checked out for a
year and still have a year to go, I find the arguments that "well, I only
connect once a year" to be a bit tenuous at the outset.
> To make a long story short, just *ask* people... Mr. Woltman has
> a policy of
> not releasing email addresses, which I respect; since there are
> only a dozen
> or so of these suspected of abandoning their exponents, I would imagine he
> wouldn't be too busy to send them a quick email for you and let
> you know what
> they have to say. It just seems rude not to ask if you have
> contact information
> available.
I think a BETTER policy would be to automatically identify exponents which
meet some unusual criteria and automatically send them emails, asking them
"are you still running the software, how far along is it, do you have the
latest version, etc" and if they don't respond, then BOOM, back into the
pool of available exponents it goes.
Criteria for such an emailer might include:
- hasn't reported results since it was first checked out
- expected completion date at least 1 year in the future
- running time exceeds 1 year with no checkins
etc. If an exponent met 2 or more criteria, they get an auto-email.
I don't know how feasible this is, but I think it'd help weed out those rare
numbers that *have* been abandoned and which, unfortunately, had long
completion times to begin with.
For other abandoned numbers checked out with the latest versions, they'll
expire 60 days after the last checkin date was missed (if I understood
Scott's comments right), so the problem will eventually go away. And that's
why we might as well clean out the problem exponents now anyway since no
more new problem exponents are likely to show up with the current system.
Aaron
________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
