Strategies to Hide the NEXT Mersenne Prime

[nitro]

In order to prevent the data mining activities that took place with the discovery of M40, it appears that we need a better way to disguise the candidate.

Simple solution, create a false truly random residue with a hardware real life source.

Like this

http://www.fourmilab.ch/hotbits/

and here's one I just made

2C8B8C91B1C8CCC3

Simple and foolproof and impossible to analyse or reverse engineer.

[ixfd64]

Or how about not display the residue at all?

[GP2]

Some ideas:
- a truly random fake residue.
- don't reveal (through summary.txt) the exact hour the prime was discovered, only the day.

Or consider my option 4 (instead of creating a fake residue in cleared.txt, just leave the exponent in status.txt with its existing pre-discovery status).


Hints helped: knowing that it was not a team, knowing it was 5M - 10M digits. However, the hints only worked in conjunction with a narrow set of candidates... if the prime is better disguised next time, it should still be possible to provide the traditional teasing hints.

There was one flukey piece of knowledge that helped: knowing that the program was a recent version with safeguards against false positives. Presumably that will not be an issue next time, as the recent version becomes more widely used. But again, this clue only helped in conjunction with an already narrow set of candidates.


The partial residue match was the giveaway. An impossible coincidence. I hesitated before posting about it, knowing the implications... finally added weakly deceptive disclaimers and wording that didn't fool PrimeCruncher or roy1942. Given that it was only a day before the estimated completion of the verification run, I figured it couldn't do too much harm in any case.

[GP2]

One question for Pi_Rho... was the prime identified in any way in your individual account report? I'm guessing it appeared with the same fake residue as in cleared.txt.

If not, this could be an issue if the next prime is discovered by, say, Team_Prime_Rib, where hundreds of people could look at the individual account report.

[pakaran]

Originally posted by GP2
One question for Pi_Rho... was the prime identified in any way in your individual account report? I'm guessing it appeared with the same fake residue as in cleared.txt.

If not, this could be an issue if the next prime is discovered by, say, Team_Prime_Rib, where hundreds of people could look at the individual account report.

Not hundreds - tens of thousands. Our password (and that of any team) is public, so anyone can look at the report.

Please tel me the number didn't always show up in cleared.txt as:

20996011 67 P 0000000000000000 17-Nov-03 19:04 Pi_Rho Rhenium

What do you mean by a "partial residue match?"

[nitro]

Originally posted by ixfd64
Or how about not display the residue at all?

Just a BIT too much of a giveaway...

[GP2]

Originally posted by pakaran
What do you mean by a "partial residue match?"

Read this thread:
http://www.mersenneforum.org/showthr...&threadid=1508

[PageFault]

Your comment raises a slew of additional considerations. Those of us who do read account reports would know to keep quiet (it must be said that the report is the most frightening and mysterious entity ever seen at the dc arcana). More likely would be the novice, posting all over ars when he figured out why his box was beeping endlessly.

TPR team stats also show info on all recently completed exponents. It however would require someone with the knowledge that a 00000000 residue was indicator of a prime find.

Originally posted by GP2

If not, this could be an issue if the next prime is discovered by, say, Team_Prime_Rib, where hundreds of people could look at the individual account report. [/B]

[nfortino]

Originally posted by GP2

Or consider my option 4 (instead of creating a fake residue in cleared.txt, just leave the exponent in status.txt with its existing pre-discovery status).

That option could be used for a time, but not for the entire verification. As far as I know, when a computer reports a prime, it still continues on a new exponent. When the computer reports in on this new exponent, it will appear as if the computer is running two separate tests. Although this also has plausible explanations, it would raise some flags. However, the exponent could be delayed for n hours (n randomly selected and secret for obvious reasons), and then put into cleared.txt with an improved random residue. Using this method, even with hints, there should be 100 or so plausible exponents with nothing to distinguish between them.

[Prime Monster]

Probably the easiest would be not to alert the community in general. Add the exponent with a "false" residue to the in cleared.txt file. Hardly anyone will systematically check that file, UNLESS the UNVERIFIED prime is on the status page.

Problem solved

PM

[ixfd64]

Originally posted by nitro
Just a BIT too much of a giveaway...

I was thinking about removing residues from all exponents, not just one that would give a Mersenne prime.