You would-be sleuths made it much harder...

[ewmayer]

...than it needed to be in the case of the larger of the 2 newly-discovered primes.

If you had googled for

www.mersenne.org/primenet/cleared.txt 0000000000000000

any time from roughly a day after the first [larger] prime was listed as "prime unverified" on the Primenet server up until late last week, here’s the "result 1 of 1" you would have seen [this is a snip form a screenshot from the actual Google result as it appeared in my Firefox session]:

p43112609.bmp

George explains:

For three hours after the second Mersenne was reported, M43112609's zero residue was displayed. Scott's reporting code could only mask one prime and generate one bogus residue. That's why I deleted 43112609 from the database for several days.

Yesterday, I put 43112609 back in the database with a composite status, and flagged one of the other exponents that hour as prime.

Amazingly, no one noticed during that 3 hour interval.

Google did.

[retina]

Well sure, when you know exactly what to look for it all seems very easy.

Luckily for you guys the Google search engine wasn't interested enough to post here.

[ewmayer]

Originally Posted by retina

Well sure, when you know exactly what to look for it all seems very easy.

Anyone familiar with GIMPS knows what Res64 a new prime will have, and where it will appear. What's so "exact" about that, in the sense you seem to imply, i.e. that it requires insider knowledge?

[retina]

Originally Posted by ewmayer

Anyone familiar with GIMPS knows what Res64 a new prime will have, and where it will appear. What's so "exact" about that, in the sense you seem to imply, i.e. that it requires insider knowledge?

No, I mean that since you knew that the report file has an all zero residue then if you search for it is easy to find. But I expect, that like me, other people here would not expect that the report file would ever have a zero residue posted and thus we would not have bothered to search for it.

[cheesehead]

Originally Posted by ewmayer

If you had googled for

www.mersenne.org/primenet/cleared.txt 0000000000000000

any time from roughly a day after the first [larger] prime was listed as "prime unverified" on the Primenet server up until late last week, here’s

Didn't you mean "... any time during a three-hour interval after the second [smaller] prime was listed as "prime unverified" on the Primenet server, here’s ...", in accordance with George's explanation?

Originally Posted by ewmayer

What's so "exact" about that, in the sense you seem to imply, i.e. that it requires insider knowledge?

A three-hour interval, rather than the multi-week span.

[davieddy]

I found your exposure of using a 4096K FFT enough of a giveaway
to predict ~42M and ~38M for the size of the primes.

[ewmayer]

Originally Posted by davieddy

I found your exposure of using a 4096K FFT enough of a giveaway
to predict ~42M and ~38M for the size of the primes.

Well, for all anyone knew, we could have had thread-related problems at 2048K as well, forcing us to 4096K even for numbers < ~39M ... but yeah, once the ETC for the verify of the second-number verify was given, it was pretty obvious that the 2 primes would fall on either side of the 2048K FFT-length's exponent cutoff.