Less than 10,000 left....

[Brian-E]

Originally Posted by Mini-Geek

I can definitely see the negative consequences of poaching a DC assignment, but the only real problem with 'poaching' a first-time LL assignment is that in the exceedingly rare event that you find a prime before the other user, you rob them of being the first to discover the prime. Therefore I don't see poaching LLs as a particularly unethical thing. Anyone want to say why it is?

In addition to Jacob's answer it should also be noted that it is a thrill for some people to be (almost certainly) the first person ever to test a particular exponent as part of this splendidly orchestrated project. If anyone doubts this, then I present myself as an example of a very small contributor who has carried out just 3 first-time LL tests. Those 3 exponents are very special to me. They each took around 3 months for me to test. I would have been quite upset if it had turned out that someone else had poached any of them when my machine finally reported to the server. My enthusiasm to take part would have been severely compromised.

Speaking more generally, the easy-going nature of the GIMPS project, which allows a small contributor with just one part-time machine such as myself to be a part of it, is a crucial element in its charm for me. If people made me feel that I was holding things up and should just let the big contributors do the work, I would leave straight away. And I don't think I'm too unusual to have that attitude.

[storm5510]

Originally Posted by GIMPS Home Page

On November 17, 2003 Michael Shafer found the prime, 2^{[URL="http://www.mersenne.org/various/20996011.htm"]20996011[/URL]}-1

Originally Posted by Kevin

Countdown to proving M(20996011) is the 40th Mersenne Prime: 313

This is coming up on six years! What's the deal here?

313?

[S485122]

Originally Posted by storm5510

This is coming up on six years! What's the deal here?
313?

To prove a Mersenne Prime is indeed the 40th for instance, one has to prove that all Mersennne numbers below that number and the previous one (the 39th, 38th ...) are composite. For GIMPS the proof consists of a check and a matching double check. It takes a certain time to check a range completely once, for instance there are still 21 903 exponents to be checked once below the current record prime ; but you must be sure that all those LL tests are correct. So a number is proved to be the next in sequence only when all exponents below it have been at least double checked.

Jacob

[Kevin]

Originally Posted by Brian-E

Speaking more generally, the easy-going nature of the GIMPS project, which allows a small contributor with just one part-time machine such as myself to be a part of it, is a crucial element in its charm for me. If people made me feel that I was holding things up and should just let the big contributors do the work, I would leave straight away. And I don't think I'm too unusual to have that attitude.

How do you feel about the ability to have exponents reassigned after one year? Or the fact that there are minimum P4 GHZ requirements for certain worktypes, and the fact that you need to have a certain reliability ranking to get "preferred" exponents? All of those are ways of limiting what slower machines are allowed to do.

And with the preferred exponent system, there's the assumption that work is going to fast and reliable computers. What if some of those exponents end up going to computers that hold on to the assignment for more than a year? I mean, the rules say you have to have at least a 2400MHZ P4 equivalent PC, and if you take more than a year to return a sub-33M test, that indicates you either clearly don't have that powerful of a system (glitch in the system) or are caching work for extremely long periods of time (or possibly moving preferred assignments to non-preferred machines). There's nothing against caching like that, but I personally wouldn't be opposed to some safeguard that says that preferred exponents will be unreserved if they're updated with an estimated completion time of more then 120 days and no progress. I mean, doesn't it at least somewhat defeat the purpose of having the preferred exponent system if there are still people taking excessively long and holding back the trailing edge?

Unless there's a failure with the assignment system, it at least seems to me that a computer getting a preferred assignment should not take more than a year to complete a preferred exponent without some kind of deliberate action on the part of the user. What I've been talking about in no way would apply to people that get regular assignments of any flavor, no matter how long they take as long, just the preferred ones which have standards about where they should go. But if people are manually circumventing the way the preferred exponent assignment system is intended to work and defeating its purpose, is it really that wrong to try and close the loopholes?

[Kevin]

Originally Posted by S485122

To prove a Mersenne Prime is indeed the 40th for instance, one has to prove that all Mersennne numbers below that number and the previous one (the 39th, 38th ...) are composite. For GIMPS the proof consists of a check and a matching double check. It takes a certain time to check a range completely once, for instance there are still 21 903 exponents to be checked once below the current record prime ; but you must be sure that all those LL tests are correct. So a number is proved to be the next in sequence only when all exponents below it have been at least double checked.

Jacob

But still, the fact it takes more than 6 years to get from finding a prime with a first time test to double-checking every exponent beneath it indicates there's a pretty large gap between first time tests and double-checks. George has said that he thinks we need more people doing double-checks. How long is it going to take to prove M(43112609) is the 47th (48th? 49th?) Mersenne prime with the current rate of double-checking? I mean, "proving" that something actually is the 39th Mersenne prime is kind of an important thing, mathematically, and definitely more important than just clearing all first-time tests beneath a prime.

[lfm]

Originally Posted by Kevin

One thing is that the system technically cares about "regular updates" as opposed to "regular progress". You could keep shuffling an assignment to the bottom of your queue or be running it on a laptop you only use 1 hour every few days and still make "regular updates", but not be making "regular progress". The only thing I'm really talking about is if preferred exponents that are supposed to go to fast/trusted systems end up taking this long, which would indicate some type of loophole with/abuse of the system. In the old system, some people would horde hundreds of low assignments and have expected completion dates of more then 200 days. Some people not as knowledgeable about the assignment system might notice that their main desktop is getting smaller assignments, and use them to stock up the queue on an aging PII. Even if some of the current sub-33M exponents end up taking a year to finish, I don't think George would kick them off, but it's also not the kind of thing you necessarily want happening on a consistent basis.

Yes, you have speculated about such situations before in this thread. Yes, I admit there are no doubt loopholes and opportunities for abuse in the automated system. I think it is unavoidable, something to do with Godel's incompleteness or something (ya I know I am stretching it). I agree fully that we need to watch for abuses but I don't think we can nor should make long lists of rules in advance for things that may never happen. And certainly I would caution against trying to automate such rules.

I think we are basically in agreement here. Keep our eyes open and if something actually happens then we can consider what if anything should be done.

As I said before, I'm not suggesting we somehow contact random users and ask them why they're so slow. As cheesehead suggested, maybe some kind of anonymous reporting could be built into the program. But with the current system, preferred exponents are supposed to go to people with a certain P4 GHZ level. If this isn't always happening, it would be nice to know why this is happening. The most obvious way to address this is (if it can be done in the proper manner) just ask. It was done previously in this thread between regular forum members without incident and I'm sure it could be done again without incident if the circumstances called for it.

I didn't say you were going to contact random people. I said even just your friendly inquiry to a single user could be taken the wrong way, as some sort of censure that they were too slow. If as you say they are not actually working on their assigned exponents or queuing up extraordinary backlogs, yes that may be grounds for further investigation but frankly I don't see it actually happening (any more) and the usual manual monitoring would seem to be good enough.

This line of thread is really getting off track. At this point, all we really should be doing is watching the numbers to see if any irregularities arise. I really don't think there will be, I expect the assignment system will do a satisfactory job. But if it does turn out that the system doesn't work as well as some people would like, then we start debating whether or not the issue is severe enough to warrant action and how to properly proceed. But for right now, there's really no reason to keep debating these points.

I agree completely. I think this is actually the same conclusion that was reached last time this subject came up. Given that, then perhaps you understand the frustration some of us express when the subject seems to raise itself again with out any real evidence of problems.

[lfm]

Originally Posted by Mini-Geek

I can definitely see the negative consequences of poaching a DC assignment, but the only real problem with 'poaching' a first-time LL assignment is that in the exceedingly rare event that you find a prime before the other user, you rob them of being the first to discover the prime. Therefore I don't see poaching LLs as a particularly unethical thing. Anyone want to say why it is?

So if two people poach a first time assignment (the same one) you get no credit. If the first is ok, how do you prevent the second?

[cheesehead]

Originally Posted by retina

This is a strawman argument.

You need to re-read either my post or the definition of strawman.

You wrote, "but in reality the credit thing is nothing, it's just a number."

I responded, "So is the balance of your bank account. Care to transfer it to my account, since it's nothing, just a number?"

But then I followed that with:

"Or maybe there are other aspects than 'just a number', eh? Such as what the number represents. Effort, accomplishment, time, ..."

I didn't think I needed to explicitly add "things you can buy with it" to the latter list to make the connection with bank accounts clear.

My point was in my second paragraph: there are other aspects than "just a number".

Money in bank accounts can gain real world advantages.

... which was exactly the point of my reply: there are other aspects than "just a number".

Credit in primenet gets me what?

Recognition by your fellow participants. Measures of the time and effort you've expended for the project.

Perhaps you don't value those, but others do.

I'm already on something like my 6th primenet account. The other 5 accounts have credit that I can't reclaim (I forgot the password). You can have those credits if you want. They won't do me any good so why should I bother to save them up?

Perhaps you don't value those, but others do value those credits.

I posted my position about how I perceived things.

... in the form of direct response to what I'd previously posted.

I didn't really have any points to make.

... except to dispute what I posted.

Your demanding of me to prove points is not necessary.

You disputed what I posted, and posed questions (perhaps rhetorical) apparently directed to me.

I'm just doing the same things you did.

Indeed, I am asking for your ideas of why "poaching" has harmed primnet/GIMPS/whatever.

... and I gave some!

Surely you're not contending that it's unfair for me to post the same sort of comments in response to you that you did in response to me, are you?

[cheesehead]

Originally Posted by Kevin

I don't think any action is needed as long as there's progress being made.

Then why are you proposing any change? Progress is always being made!

Do you imagine that the project administrators just ignore what's going on and never check on progress for themselves, with reports that you don't necessarily see?

If (for example) the countdown to 30402457 gets down at 10 and stays there for a few months

... which has never yet happened! The countdowns have always moved faster than that.

then we might want to look into what's going on.

"we"?

The project administrators already DO keep tabs on that stuff.

[cheesehead]

Originally Posted by Kevin

Did some preliminary data-mining in the 27M range. I noticed there were a decent amount of exponents where they've been successfully tested once with no error, and have been assigned again as an LL test (not a double-check) to ANONYMOUS. An example would be 27577387. Anybody know what's going on there?

Yes.

The server changeover from v4 to v5 (and prime95 from v24 to v25) was rushed in, because of a hardware emergency, before the mechanisms for transferring information from v4 accounts to v5 accounts were fully featured and finished. What's happening as a result is that all v4 accounts are initially listed on the v5 server as ANONYMOUS, until/unless the v4 accountholder creates a v5 account and transfers v4 credits/records to v5.

(This could have been learned from reading past forum threads.)

[storm5510]

Originally Posted by S485122

To prove a Mersenne Prime is indeed the 40th for instance, one has to prove that all Mersennne numbers below that number and the previous one (the 39th, 38th ...) are composite. For GIMPS the proof consists of a check and a matching double check. It takes a certain time to check a range completely once, for instance there are still 21 903 exponents to be checked once below the current record prime ; but you must be sure that all those LL tests are correct. So a number is proved to be the next in sequence only when all exponents below it have been at least double checked.

Jacob

I see. So, the area I asked about is 2^13466917 to 2^20996011?

If a lot this still needs to be done, how does it get assigned?