As mod, I'd like to discuss how we can prevent another leak from happening next time. I could have edited the post with the giveaway clue but decided not to as that would have just made things unnecessarily suspicious. Still, I'd like to discuss whether what happened this time was acceptable to George and if not, what could we do.

At my current speed I should be finished my verification of M41 in less than 3 days on my 4 CPU Itanium2 (1.5GHz) system and I think Xyzzy's box will be finished around the same time.

There have been lots of improvements to Glucas in the multi-threaded part of the code in the past week so it looks like doing this verification has been very good for the software.

Jeff.

[QUOTE]There have been lots of improvements to Glucas in the multi-threaded part of the code in the past week so it looks like doing this verification has been very good for the software.[/QUOTE]

Indeed, I improved a lot the multithreaded code. I'm still improving it, but it has been impossible to finish before the verification :smile:

And congratulations everyone!! :banana:

Guillermo

[QUOTE=gbvalor]Indeed, I improved a lot the multithreaded code. I'm still improving it, but it has been impossible to finish before the verification :smile:

And congratulations everyone!! :banana:

Guillermo[/QUOTE]

Congrats to you, Guillermo... :wink:

Luigi

I ran a Google search on 24036583, and got one hit:

Item ID, 24036583. Warranty: What is it? Replacement.
If the product is defective. Money Back Guarantee: What is it?

(Actually, the answer to "What is it?" isn't "A new Mersenne prime exponent"
but a necklace on an Indian version of E-bay).

[QUOTE=garo]As mod, I'd like to discuss how we can prevent another leak from happening next time. I could have edited the post with the giveaway clue but decided not to as that would have just made things unnecessarily suspicious. Still, I'd like to discuss whether what happened this time was acceptable to George and if not, what could we do.[/QUOTE]As they say, "Information wants to be free".

One has to start at the source. I think the PrimeNet server would have to be modifed to leave no trace in the logs.

[QUOTE=M29]As they say, "Information wants to be free".

One has to start at the source. I think the PrimeNet server would have to be modifed to leave no trace in the logs.[/QUOTE]

There was some [URL=http://www.mersenneforum.org/showthread.php?t=1548]discussion[/URL] on techniques for hiding results when M40 was found. What seemed reasonable was just having the server post a fake residue and not mention that a prime had been found until it could be verified. Of course, that would limit the amount of chatter on the boards and keep the community out of the loop, which could be detrimental to the overall effort. One other thought would be for the server to not update for a 24 hour period to increase the number of potential candidates. That might give the double check team enough time to confirm the prime before it was deduced from the data mining.

Congrats to the GIMPS team for another outstanding find! I knew M40's days were marked, I just didn't think it would be so soon. :cry: On to M42! :w00t:

There were 2 problems in this instance:

1) The fake residue output by the server contained a highly nonrandom substring;

2) One of the DCers inadvertently included a percent-complete datum in a "latest update" post (easy mistake to make - I almost did similar myself at one point).

Both of these are highly preventable. The solution to (1) is obviously to do a better job randomizing the Res64. To prevent something like (2) from occurring in future, I suggest the DCers all send their day-to-day updates to George, and George posts a daily summary to the board. Moderators should be vigilant in looking for leaks and squashing them ASAP. Simply deleting the % figure from the post yesterday would have sufficed, though a handful of users would've probably still seen it before there was time to edit it.

I also suggested to George that no DC status updates aside from "completed and verified (or not)" be given after (say) 20M, but by that time the cat had already run out the door.

But none of this should detract from the joy of the new discovery - my own DC is still poking along at ~14M iterations right now, but I'd say it's not premature to say, "congratulations!"

I really don't see anything wrong with the present system as long as better fake residues are generated. I appreciated the clues that the number was bigger than the last discovered Mersenne prime and that it was less than 10 million digits. Perhaps the clue that the contributor was among the top 1000 narrowed it down just a little too much, as I am sure that Curtis Cooper must have known it wasn't one of his two exponents. But I don't see any harm in miners and speculators narrowing down the range of suspects. It certainly adds interest to the wait.

And speaking of wait, how many verifications are we waiting for before an announcement is made?

It looks like Richard Brent can fire up his primitive polynomial machine again soon!

How many decimal digits is it?

should be ceil(24036583*log(2))=7235733 digits