Poaching

[ckdo]

Originally Posted by Uncwilly

If the person has picked up a number and has reported in 30% LL complete, yet there are 4 bit levels and P-1 that have not been done, is that not clear? The bit levels in question might take 15 GHz days for the first 2. If the 30% LL reported in ~= 5 GHz years, would it not make tremendous sense to do the TF first?

Since the "no factor" TF and P-1 results would unassign the LL test, wouldn't it make tremendous sense to only report the former after the latter has finished?

Granted, for a 5 GHza/15 GHzd ratio I'd not have split off the factoring in the first place, but the point remains: Neither does the fact that no factoring results have been reported mean that no factoring was done nor does it mean that these will not be reported after the LL test has finished.

Which timeframe would be acceptable to allow for the original assignee to report in any factoring results before picking up the exponent in question and doing the factoring part yourself is a different question. I, myself, pick up exponents which have been LL tested without P-1 being done, below 40M. I will usually pick them up when they become available, but not start the P-1 until two weeks later or some such (43 days for those running right now).

[S485122]

Originally Posted by ckdo

Since the "no factor" TF and P-1 results would unassign the LL test

It depends how one reports them : I often do P-1 factoring on double checks when the previous P-1 had been done to low bounds. I do that on a different machine by having it run the right Pfactor workdo line without the AID component. Reporting the result will then not affect the doublecheck assignment.

Jacob

[cheesehead]

Originally Posted by Uncwilly

If the person has picked up a number and has reported in 30% LL complete, yet there are 4 bit levels and P-1 that have not been done, is that not clear?

No, it's not.

In addition to a scenario I mentioned above, an eccentric assignee could have already completed the TF and P-1 before LL but temporarily set the report (from prime.spl) aside, to be reported later simultaneously with the LL result, rather than send it to PrimeNet right away. It'd be weird but legitimate, and doesn't hurt the project.

Jumping to a false and unfair conclusion on the basis of reported status like that (30% LL, but no TF or P-1 reported) is exactly why I expressed my concerns earlier!

If the 30% LL reported in ~= 5 GHz years, would it not make tremendous sense to do the TF first?

Yes, _if_ the assignee subscribes to the conventional view that it's best to do TF and P-1 before LL in order to prevent doing a useless LL on a number that has a small factor.

Or, as described above, perhaps the assignee HAS indeed done the TF and P-1 first (and had no-factor-found results), but also has chosen to (unconventionally) delay reporting the TF and (P-1) no-factor-found result until the LL report is ready. In that case, is there any reason to poach the TF or P-1 before the LL is complete, other than to enforce unnecessary conformity?

What does GIMPS really lose (be specific) if we wait until after the LL result is reported before drawing any conclusion about unreported TF and P-1 and taking any action?

Let's not punish GIMPS participants who legitimately perform their assignments, but in an unconventional manner.

[cheesehead]

Originally Posted by ckdo

Since the "no factor" TF and P-1 results would unassign the LL test,

Not necessarily. If an exponent is assigned for LL without all default TF having already been done, then the TF test and report is, by default, part of the assignment. Any LL assignment includes, automatically, performing TF up to the default limit if not already done, and performing P-1 if not already done.

I think/hope PrimeNet's smart enough not to end an LL assignment just because TF and P-1 are reported -- if not enough of those had been previously performed/reported when the LL assignment was made.

George, is that so?

Neither does the fact that no factoring results have been reported mean that no factoring was done nor does it mean that these will not be reported after the LL test has finished.

Correct, though I think PrimeNet will consider the LL report to end the LL assignment, so the user really ought to report the TF and P-1 simultaneously with the LL, if not reported earlier, to avoid misleading others and PrimeNet.

[joblack]

Actually I don't see the problem in poaching (as far as I understand the term). There are enough Mersenne numbers to check why would somebody steal another's number?

[Mini-Geek]

Originally Posted by joblack

Actually I don't see the problem in poaching (as far as I understand the term). There are enough Mersenne numbers to check why would somebody steal another's number?

Infinite, in fact (one for each prime number). More than enough no matter your tastes.
But the trouble isn't that there's nothing else for poachers to do, it's that they want to do those more than any other numbers, because they're the smallest ones available and, as they might see it, the current reservations are just holding up the completion of milestones. (unless I'm wrong, in which case I don't know why they do it!)

[cheesehead]

As I've explained before, my analysis is that poaching has almost always been the result of impatience, poor impulse control and failure to appreciate the poachee's rights, point of view and feelings.

As a multiple-poachee, I say that it's very discouraging, exasperating and unjust.

[henryzz]

Should we really want to try to cater for people that are doing wierd things with their reservations?
They should know that if they fiddle it so that they do the work and don't report it until past some deadline then it's their own fault.

[cheesehead]

Originally Posted by henryzz

Should we really want to try to cater for people that are doing wierd things with their reservations?
They should know that if they fiddle it so that they do the work and don't report it until past some deadline then it's their own fault.

What catering?

... don't report until past what deadline?

What fault?

Above I was referring to reporting TF (no factor found), P-1 (no factor found) and LL all at the same time (i.e., as soon as the LL finishes). Of course, finding a TF or P-1 factor would be reported as soon as found.

Do you assume that a procedure would hold up GIMPS progress just because it's unconventional?

What I'm concerned about is vigilante action in unconventional cases where the vigilantes don't wait until there's real evidence of any actual harm to GIMPS.

Here's another case:

Someone's assigned two LLs. S/he processes them in an alternating fashion: a week on one, then a week on the other, then a week on the first, and so on (assuming a single-core situation), sending progress reports every two week. S/he reports each LL completion when it's finished.

Where's the real harm to GIMPS, if any? (Note that this wouldn't cause him/her to run past any days-to-go extrapolation on reports by more than the margin currently allowed to any and all assignments and that both assignments would be completed and reported within the same combined elapsed time that they'd take if run sequentially.)

In fact, where's the real harm to GIMPS if any set of assignments is performed in any unconventional manner, as long as all of them are reported in the same elapsed time (for the entire set) as if they'd all been run conventionally? If assignment A takes 25 days and assignment B takes 20 days, what harm does it do to GIMPS if neither assignment completion is reported before 45 days have elapsed, but both assignment completions are reported afterno later than 45 days?

All of my hypothetical cases require absolutely no change to current GIMPS or PrimeNet operations, assignments or reports. The only "catering" they'd require is that would-be vigilantes refrain from poaching or otherwise jumping in to interfere with exclusive PrimeNet assignments while those assignments remain in force..

[Uncwilly]

{Deleted because of Cheesehead's post facto editing of his post}

[cheesehead]

Originally Posted by Uncwilly

If their "days to go" on an LL is a negative number (i.e. no new completion date has been sent in for over 4 months), the last 5 bits of TF have not been reported, the number has been assigned at least 7 months, and the P-1 has not been reported done; I think that this is past a deadline. This is a real case.

In my hypothetical cases, all progress reports are made on a regular schedule, with nowhere near a 4-month gap. My hypothetical cases involve no violation of any GIMPS/PrimeNet rules, but they _do_ involve violations of common assumptions about the meanings of numbers on reports.

Actions or deductions based on those not-necessarily-correct assumptions about reported numbers is what I'm concerned about.

I am asking about this, because, with the new tool, we can get an indication that the number is not being worked on and reported in a timely manner.

... but not in all cases! In some cases, the new tool would report numbers that could be (and have been -- in real cases) falsely interpreted as "evidence" of foot-dragging or other undue delay.

To avoid being considered a 'vigilante' (which is a loaded term)

... but no more loaded than the terminology with which some poaching advocates have described their justifications.

"Clean-up" implies that there's evidence the assignee left or made some mess. That's loaded, if no such evidence exists.

I would hope, that there would be some ground rules, that the honorable and known GIMPSter, could do some clean-up work. George has facilited some similar actions before.

Fine, but I'm talking about cases in which people mistakenly conclude that "clean-up work" is justified because a report does not give a fair picture of an unconventional situation.

Show me the protections against vigilante action in those cases.

Not relavent to the bypassed TF's that are the focus of the current discussion.

See? There you go, failing to notice that my hypothetical situations include cases in which it appears that TFs were bypassed, when they actually were not!!

That you've just claimed that my examples were not relevant to bypassed TFs demonstrates exactly what I'm concerned about -- misuse of, or mistaken assumptions about, reported figures that lead others to falsely conclude something about the status of someone's assignment..

Your dismissal there is a real, concrete example of how easy it can be to make the misinterpretation I'm concerned about.