Newer milestone thread

[Uncwilly]

Originally Posted by Madpoo

Speaking of the page, so I guess the next minor milestone we'll cross is that double-check of all Mp under 10M digits. Any thoughts on what could or should be next?.

I noticed some progress on the first time LL number. I think that there is room for a Responsible Party™ to do 3 "early double checks", that would move the move the 1LL number up past 2 xx,000,000 milestones.
http://www.mersenne.org/assignments/...et+Assignments
Maybe someone can monitor these for a while. If there is low progress, maybe Chris can be summoned to perform a GIMPS "Christmas Miracle".

The next xx,000,000 milestone after that would likely take until at least the end of March.

[Prime95]

Originally Posted by Uncwilly

I noticed some progress on the first time LL number.

Two points:

1) I think we need to rename the "active assignments" page to the "poach me" page.

2) For years I've wanted to eliminate the "all exponents tested once" milestones. To me, the milestone is pointless as it is a virtual certainty that there are many bad LL results. Thus some exponents are in reality untested and possibly prime. My opinion, however, is in the minority and many users feel such events are important.

[flagrantflowers]

Originally Posted by Prime95

…we need to rename the "active assignments" page to the "poach me" page.

I agree. I don't see why they shouldn't be left alone at least until the older assignment rules no longer apply.

If you really want to see those "once tested" milestones pass just think about saving up X of them and completing X in a short time period.

[Uncwilly]

Originally Posted by Prime95

Two points:

1) I think we need to rename the "active assignments" page to the "poach me" page.

Just for the record, I personally have not done an LL in at least 6 years (unless one of my borgs did some unsupervised), maybe 8. And the last LL's that I did do were DC assignments handed out by the server.

[Primeinator]

Originally Posted by Madpoo

And speaking of the 10M milestone... I found myself going back and forth on the nomenclature. What's the proper, or just best, way of phrasing it? "All exponents below 10M" isn't right, because it's not the exponent, it's the 2^(exponent)-1 that's 10M digits. I think I've used that phrasing as shorthand here and there although I know it's not correct. Is it just like "M(P)", or is there some generally accepted shorthand?

As you see, I stuck with the very literal format of "all 2^p-1 below 10M digits", it just doesn't roll off the tongue. Perhaps "all Mersenne #'s less than 10M digits" ?

What about "Mersenne candidates below 10 million decimal digits"

Originally Posted by Prime95

2) For years I've wanted to eliminate the "all exponents tested once" milestones. To me, the milestone is pointless as it is a virtual certainty that there are many bad LL results. Thus some exponents are in reality untested and possibly prime. My opinion, however, is in the minority and many users feel such events are important.

I kind of like this milestone marker even though I know some of the LLs that have been 'tested' are faulty. It provides a moderately good estimate of the progress of the project as a whole (even though it is nowhere near the LL wave front).

[Brian-E]

Originally Posted by Prime95

2) For years I've wanted to eliminate the "all exponents tested once" milestones. To me, the milestone is pointless as it is a virtual certainty that there are many bad LL results. Thus some exponents are in reality untested and possibly prime. My opinion, however, is in the minority and many users feel such events are important.

Originally Posted by Primeinator

I kind of like this milestone marker even though I know some of the LLs that have been 'tested' are faulty. It provides a moderately good estimate of the progress of the project as a whole (even though it is nowhere near the LL wave front).

Perhaps the solution is to rename this particular milestone so that casual readers see the "progress" aspect of it rather than the erroneous suggestion of completion which the word "tested" can conjure up.

Not easy to word it though. Somthing like "all exponents below XXXXXXXX have undergone their initial unconfirmed LL run" perhaps? Can anyone make it sound less clumsy?

[ATH]

Originally Posted by Prime95

1) I think we need to rename the "active assignments" page to the "poach me" page.

The 3 exponents he linked should have been recycled by now according to the rules?

Recycle rules (top 1500 exponents):
1) If expected completion date is not updated for 60 days.
2) If assignment made before 2014-02-01 and:
2a) assignment is 12 months old and < 50% complete it is recycled.
2b) assignment is 15 months old it is recycled.
3) If assignment made after 2014-02-01 and:
3a) assignment is 6 months old and not started it is recycled.
3b) assignment is 9 months old it is recycled.

[Prime95]

The actual SQL code is below. As you can see there is no 15 month rule. I'm about to head out the door, perhaps you can calculate when these three might expire.

       ((dt_when_assigned < '2014-03-01' AND		-- Grandfathered assignment      
         exponent < @exp1 AND				-- exponent is in the most critical category      
         dt_when_assigned < DATEADD (DAY, -365, GETDATE()) AND     -- and assignment is over a year old      
         percent_done < 10 + (DATEDIFF (DAY, dt_when_assigned, GETDATE()) - 365) / 3.33) OR -- plus a grace period if close to finished

[Madpoo]

Originally Posted by Prime95

The actual SQL code is below. As you can see there is no 15 month rule. I'm about to head out the door, perhaps you can calculate when these three might expire.

       ((dt_when_assigned < '2014-03-01' AND		-- Grandfathered assignment      
         exponent < @exp1 AND				-- exponent is in the most critical category      
         dt_when_assigned < DATEADD (DAY, -365, GETDATE()) AND     -- and assignment is over a year old      
         percent_done < 10 + (DATEDIFF (DAY, dt_when_assigned, GETDATE()) - 365) / 3.33) OR -- plus a grace period if close to finished

In the case of those 3 exponents, they're not moving fast, but they are moving, and being updated every so often. I wouldn't think the server would reassign exponents as long as the original assignee was making progress on it, no matter how slowly.

I know that from a human point of view, milestones are interesting... those nice round numbers like 52 or 53 million, and having all exponents first time LL checked. But of course that's just a mental thing and in the grand scheme of things it doesn't matter much.

I can see both sides of the argument though... yeah, it doesn't really mean much... the work will get done when it gets done, but on the other hand if people are able to measure some artificial measure of progress, it gives that warm fuzzy that "things are happening" and hopefully keeps them engaged in the project as a whole.

If all we ever did was crunch numbers and never saw how that's moving the bar ever higher, it could give you that feeling that your effort doesn't matter and you don't bother keeping it going. I gather that's probably behind the whole idea of being able to form teams, or tracking and ranking individual work effort in the first place, because you can get that idea of "I'm doing something" or "we're doing something"

[ATH]

Originally Posted by Prime95

The actual SQL code is below. As you can see there is no 15 month rule. I'm about to head out the door, perhaps you can calculate when these three might expire.

       ((dt_when_assigned < '2014-03-01' AND		-- Grandfathered assignment      
         exponent < @exp1 AND				-- exponent is in the most critical category      
         dt_when_assigned < DATEADD (DAY, -365, GETDATE()) AND     -- and assignment is over a year old      
         percent_done < 10 + (DATEDIFF (DAY, dt_when_assigned, GETDATE()) - 365) / 3.33) OR -- plus a grace period if close to finished

I just quoted your rules from the "Proposed LL assignment and recycle rules"-thread. I guess they changed along the way, I did not read all the 97 post in that thread.

It was not a demand or statement, but a question why they were not recycled, by which I meant, what are the rules now? Which I got my answer to.

From the code it seems they get 3.33 days extra beyond one year for every procent that are complete above 10%. That means they are very close to being recycled, unless they fall into the unspecified grace period:
1st one: 63.10% done, so will be recycled after: (63.1-10)*3.33 + 365 days ~ 549 days (currently at 536)
2nd and 3rd ~ 81% done, so will be recycled after: (81-10)*3.33 + 365 days ~ 601 days (currently at 576)

[lycorn]

Originally Posted by Madpoo

If all we ever did was crunch numbers and never saw how that's moving the bar ever higher, it could give you that feeling that your effort doesn't matter and you don't bother keeping it going. I gather that's probably behind the whole idea of being able to form teams, or tracking and ranking individual work effort in the first place, because you can get that idea of "I'm doing something" or "we're doing something"

Very good point, nicely put. I guess it applies to many of us around here.

That said, I think that now, that new and more strict rules are in place to reassign exponents,we should also do something to prevent poaching in a more effective way. In fact, in the past we could say that the server wasn´t "doing its job properly" in that the supposed reassignments, under the old rules, were not taking place and milestones were systematically blocked. On the other hand, the server would (and still will) accept and credit any result, regardless of the assignment status of the exponent. I propose that from now on the server will not accept results for an exponent that is reserved by someone else. That would be the end of poaching. Period.

Any thoughts?