What if someone discovers a Mersenne prime using not-so-legal methods?

[ixfd64]

Originally Posted by chalsall

I truly don't know who the Anonymous user on GPU72 who is doing incredible amounts of DCTF is. And even if I did know, I couldn't say.

But based on their behaviour, I suspect this is a single individual, or a very small group, who have access to a serious amount of GPGPU compute (much like you (airsquirrels)).

But, likely, not via a "botnet". After all, someone engaged in such illegal activities would probably task that compute to mine coins, or crack passwords.

This post from the M#49 thread got me thinking: suppose someone does find a new prime using... questionable... means. Let's say the person is an IT admin for a large school district who got caught installing Prime95 on several thousand school computers without permission. The program is removed, but not before one of the computer finds a prime.

On one hand, a discovery is a discovery. But at the same time, announcing the new prime would also make GIMPS look bad. How would GIMPS deal with the situation?

[Uncwilly]

Originally Posted by ixfd64

Let's say the person is an IT admin for

Aaron, care to take this one?

[aketilander]

Have we not already had such a case?

[Dubslow]

Originally Posted by aketilander

Have we not already had such a case?

Not where a prime was found by the "offender".

[ewmayer]

I suggest a similar approach as to convicted murderers who write tell-all books from prison: They can publish what they want and get proper credit for it, but they are not allowed to profit from it. Here of course we might have to broaden the definition of 'profit'. For one, no prison conjugal visits from the usual hordes of sex-crazed M-prime groupies.

[Nick]

Originally Posted by ewmayer

I suggest a similar approach as to convicted murderers who write tell-all books from prison: They can publish what they want and get proper credit for it, but they are not allowed to profit from it. Here of course we might have to broaden the definition of 'profit'. For one, no prison conjugal visits from the usual hordes of sex-crazed M-prime groupies.

See also: http://retractionwatch.com/2016/01/1...ademic-papers/

[LaurV]

Haha, you guys are wonderful! I can not stop laughing.

Now you know why I can not come forward with the five primes I discovered...

[E_tron]

It's kind of like winning the lottery. Unauthorized installs of Prime95 will suddenly become a good source of PR for the affected party (not to mention a footnote in history books).

A discovery from a botnet would be interesting... The discovery should be reported as such and leave it at that. The discovery should be attributed to the botnet (and/or the suspect and victim if identified). The 100M prize should be forfeited.

It probably happens all the time on crypto-currency. That said, a true botnet probably couldn't run individual nodes long enough to complete full LL tests. I think true botnet discoveries are unlikely, but 'unauthorized' discoveries are probable (though we will probably never know it was 'unauthorized' since the parties invoked would likely sort it out before a discovery announcement).

[airsquirrels]

Originally Posted by E_tron

It's kind of like winning the lottery. Unauthorized installs of Prime95 will suddenly become a good source of PR for the affected party (not to mention a footnote in history books).

A discovery from a botnet would be interesting... The discovery should be reported as such and leave it at that. The discovery should be attributed to the botnet (and/or the suspect and victim if identified). The 100M prize should be forfeited.

It probably happens all the time on crypto-currency. That said, a true botnet probably couldn't run individual nodes long enough to complete full LL tests. I think true botnet discoveries are unlikely, but 'unauthorized' discoveries are probable (though we will probably never know it was 'unauthorized' since the parties invoked would likely sort it out before a discovery announcement).

It is only the current architecture of Prime95 that requires all the iterations to be finished on one machine...

[Uncwilly]

Originally Posted by airsquirrels

It is only the current architecture of Prime95 that requires all the iterations to be finished on one machine...

It is possible to move assignments and work on them on different machines. Plus, if one where to use a single machine to front for all of the others, it would be possible to check for discoveries, and rerun from the save file the last few iterations on a legit machine.

[0PolarBearsHere]

Originally Posted by E_tron

That said, a true botnet probably couldn't run individual nodes long enough to complete full LL tests.

Originally Posted by Uncwilly

It is possible to move assignments and work on them on different machines. Plus, if one where to use a single machine to front for all of the others, it would be possible to check for discoveries, and rerun from the save file the last few iterations on a legit machine.

If it's smartly coded it could do it.
Mr Baddie starts the LL for a few thousand exponents and saves the backup files to a server.

1) When computer is infected
a) connect to server
b) download LL code
c) "check out" save file
d) perform Z iterations (either number or for a time period)
e) create save file
f) "check in" the save file to the server again.
2) worm removes itself and LL code from computer

As long as there is proper error checking (perhaps compare residuals between several victims) it could work.

But it would be very very naughty, so please don't do it.