How to guess M#48's exponent to < 1% (based on qwiki-leaks)

[TheMawn]

So when do you tell us what the next one is?

If you're within 0.1% then with 1/1000 of the computing power curtisc has (a single Ivy Bridge ought to do it...) then you should be finding more of them than him.

[Prime95]

Originally Posted by TheMawn

So when do you tell us what the next one is?

You've misunderstood the context of this thread. It occurred during the time where M#48 had already been discovered, but not announced. One of the independent testers revealed a little too much information during his test (the roundoff error) which someone could have used to guess M#48 within a pretty tight range.

[TheMawn]

Lol I think I still don't understand. "Here's a way to guess which exponent is the right one but it only works if we already know what it is"

My favourite colour is green, but can you guess what my favourite colour is?

[TheMawn]

Sorry. That may have been a bit snarky. I just want to understand what this is about.

[ewmayer]

Originally Posted by TheMawn

Lol I think I still don't understand. "Here's a way to guess which exponent is the right one but it only works if we already know what it is"

Perhaps you should try to acquaint yourself with how such new-prime finds are first confirmed and only then publicly announced, rather than nattering on. George has already explained it: There is a typically several-days-to-a-week-long delay between the time the PrimeNet server first "reports" a new prime has been discovered. I put that in quotes because only George and a handful of other folks are privy to the identity of the (alleged) new prime. While independent verification proceeds, there is naturally a lot of interest in what the number is, so there is always a cat-and-mouse guessing game in which GIMPS members who are not in the "insider's list" try to glean the number either from server logs or from accidental "data leakage" by the verifiers, who try to keep the public apprised of the progress of the verification run(s) and provide some noteworthy details about the hardware used, any software issues encountered, etc. It turns out to be surprisingly difficult to say anything meaningful about the verification runs without inadvertently providing clues as to the identity of the number.

My favourite colour is green, but can you guess what my favourite colour is?

No, more analogous is "I have a favorite color which I won't reveal, but here are some interesting and true anecdotes about it which I will attempt to tell without revealing the color..."

[LaurV]

"hey, jack, I heard you bought a new car, what's its color?"
"like the sky during the sunset, except that is green"

[TheMawn]

Originally Posted by ewmayer

There is a typically several-days-to-a-week-long delay between the time the PrimeNet server first "reports" a new prime has been discovered. I put that in quotes because only George and a handful of other folks are privy to the identity of the (alleged) new prime. While independent verification proceeds, there is naturally a lot of interest in what the number is, so there is always a cat-and-mouse guessing game in which GIMPS members who are not in the "insider's list" try to glean the number either from server logs or from accidental "data leakage" by the verifiers, who try to keep the public apprised of the progress of the verification run(s) and provide some noteworthy details about the hardware used, any software issues encountered, etc. It turns out to be surprisingly difficult to say anything meaningful about the verification runs without inadvertently providing clues as to the identity of the number.

Ahhh okay. Thank you for this. This does make everything crystal clear. I don't know how I was supposed to have the clairvoyance to know that this is the way it works, but I do now. I thought the exponent would have been immediately revealed so that people could begin to double-check, but I guess this works too.

Thanks again.

[ewmayer]

My pleasure - glad we cleared things up sans serious flamage breaking out. :)

[cheesehead]

Originally Posted by TheMawn

I thought the exponent would have been immediately revealed so that people could begin to double-check,

a) The exponent _is_ revealed, by personal e-mail rather than forum broadcast, to a group of people who have access to hardware that can perform doublechecks faster than typical user systems could.

It's not required to publicly ask for new volunteers each time, because that group of people has done doublechecks for previous Mersenne discovery reports, and most will do it again when needed.

b) Consider the situation if GIMPS were to publicly announce an exponent before it was doublechecked, then doublechecks showed that the original test was faulty: GIMPS gets public embarrassment.

There's no upside to GIMPS in announcing an exponent before it's verified. The rest of the world can wait a few more days.