mersenneforum.org  

Go Back   mersenneforum.org > Fun Stuff > Lounge

Reply
 
Thread Tools
Old 2003-01-17, 04:50   #1
Kevin
 
Kevin's Avatar
 
Aug 2002
Ann Arbor, MI

1B116 Posts
Default Poaching

After hearing that many of the small exponents were finally cleared, I was sad to find out that 20+ were completed by "Malco". It seems to me that this is poaching at it's worst. Not only does it make all the months those computers spent crunching worthless, it also goes against the "every computer counts" philosophy that makes DC successful. My guess is this person wasn't aware of the projects anti-poaching stance. Maybe we should add a disclaimer about poaching when someone agrees to use primenet. I hope action is being/will be taken to fix the current issue if possible, and to prevent it in the future.
Kevin is offline   Reply With Quote
Old 2003-01-17, 08:01   #2
outlnder
 
outlnder's Avatar
 
Aug 2002

2·3·53 Posts
Default

The simplist way to take care of this problem is to give the credit to the machine the exponent was checked out under.

This way poaching can go on and the only person who suffers is the poacher.
outlnder is offline   Reply With Quote
Old 2003-01-17, 12:12   #3
eepiccolo
 
eepiccolo's Avatar
 
Dec 2002
Frederick County, MD

2·5·37 Posts
Default

Kevin, are you sure it is Malco? I figured I'd take a gander, and saw that it was the user Malfoy. Anyway, it might be nice if eventually, along with replacing the server, some sort of policy is made to help prevent poaching. Poaching BAD!!!
eepiccolo is offline   Reply With Quote
Old 2003-01-17, 14:20   #4
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

5·23·71 Posts
Default

Quote:
Originally Posted by outlnder
The simplist way to take care of this problem is to give the credit to the machine the exponent was checked out under.
Disclaimer: I'm playing Devil's Advocate here...

If that were the case, I could check out 10,000 exponents and just sit on them and eventually get credit for all of them... (Yes, it is possible to check out 10,000 exponents with just a little work!)
Xyzzy is offline   Reply With Quote
Old 2003-01-17, 16:59   #5
lycorn
 
lycorn's Avatar
 
Sep 2002
Oeiras, Portugal

101101101002 Posts
Default

As I have already stated in this forum, I´m not in favour of poaching, by the very same reasons several people have mentioned. I agree with Outlnder´s opinion about the credit, but I also think that in a future version of the server application some care should be taken to avoid exponents being assigned to users anticipating a long testing time. For example LLs or DCs would not be checked out for users when the computed time for testing was over 1 year (with perhaps the exception of 10M digits numbers). Users would receive a TF assignment instead, with a message stating the reason why. Of course one may argue that this is not 100% effective, for people can for instance enter 24h for the time the computer is on, and then leave it on only 1 or 2 hours a day. That´s right, but I think that people do not have the deliberate intention of delaying or in any way harming the project, it is just carelessness or lack of information, so if the program was to warn the user about this time limit during the installation phase, the majority of the situations could be avoided, I am sure. Another measure could be to force the people to manually enter the number of hours the computer is on, instead of having this default value of 24 h, which is impossible to achieve even leaving the computer on all the time, and which I suspect many people leave "as it is".
lycorn is offline   Reply With Quote
Old 2003-01-17, 20:19   #6
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

749210 Posts
Default

I don't think giving CPU credit to the original reservation comes anywhere close to the undoing the damage done by a poacher. Part of the thrill of an LL test is the small chance that you could make a big discovery. The poacher has completely eliminated this thrill.

I think we have at least three good ideas on the table that will reduce poaching significantly:
1) A new server that will give the smallest double-checks and first time tests to computers that have returned an LL result and should complete this assignment within a couple of months. This will reduce the number of exponents that get assigned and abandoned repeatedly.
2) The 6 or 9 month warning. Basically, stating that the server will make every effort at guaranteeing your reservation for that period. After that, the server may reassign the exponent. This is to crack down on "lost machines" that are making little to no progress and have probably been forgotten about by the user. Obviously, 10,000,000 digit numbers will be guaranteed for a much longer period.
3) Some kind of server sanctions for returning a result you were not assigned. A polite email warning, no CPU credit, etc.
Prime95 is online now   Reply With Quote
Old 2003-01-17, 21:14   #7
QuintLeo
 
QuintLeo's Avatar
 
Oct 2002
Lost in the hills of Iowa

26·7 Posts
Default

I think that the exponents should be checked out to a *user*, not to a specific machine.

I've had some hardware changes I've made recently that caused my first LL completed to be tested on 2 different machines, about the first 40% on one the rest on another.

I've also just had a hardware failure (dead HD) on another machine - I'm probably going to end up reassigning those exponents to my fast machine, as I don't think I'm going to have a replacement HD for the down machine real soon.


I'd also be inclined to say that if an exponent has had *no* progress made on it over the course of 3-6 months, it should be reassignable - but if regular progress *is* being made, it should not be.

I'm also inclined to think that a lot of the "poaching" problems are a direct result of the *ability* to manually pick an exponent - dunno if disabling *that* capability is an "acceptable" concept or not. I see some cases where it's perfectly valid to do so (like my dead machine problem), but it's also the root capability that allows poaching....
QuintLeo is offline   Reply With Quote
Old 2003-01-17, 21:20   #8
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

1FE516 Posts
Default

Quote:
Originally Posted by QuintLeo
I'm also inclined to think that a lot of the "poaching" problems are a direct result of the *ability* to manually pick an exponent - dunno if disabling *that* capability is an "acceptable" concept or not. I see some cases where it's perfectly valid to do so (like my dead machine problem), but it's also the root capability that allows poaching....
There are a large number of people out there who don't have the ability to work with PrimeNet... Some clients (mlucas/glucas) cannot (yet!) communicate with PrimeNet and some people are in positions where they don't have direct access to the Internet...

Finally, I doubt there is anything we can do to prevent someone from just typing the exponent into a worktodo.ini file... GIMPS obviously does not "own" exponents... What happens if a person finds a MP manually and publishes the results before the person who has been assigned that exponent by PrimeNet? (Hint: Second place is always first loser!)
Xyzzy is offline   Reply With Quote
Old 2003-01-17, 22:09   #9
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

165048 Posts
Default

Quote:
Originally Posted by QuintLeo
I think that the exponents should be checked out to a *user*, not to a specific machine.
They are. Primenet server doesn't use the computer name in any important way. You can checkout the exponent on one computer name
and report it to the server on another computer name without problem
Prime95 is online now   Reply With Quote
Old 2003-01-17, 22:35   #10
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

11101010001002 Posts
Default

Quote:
Originally Posted by QuintLeo
I'd also be inclined to say that if an exponent has had *no* progress made on it over the course of 3-6 months, it should be reassignable - but if regular progress *is* being made, it should not be.
What if progress is just a few hundred iterations a month? What if the user, without understanding prime95's CPU requirements, installed prime95 on his computer and it is on only a few minutes a day?

My personal opinion is that the user has a responsibility. Today that repsponsibility is misdefined, which causes much of the fighting over how long a user should be allowed to hold on to an exponent.

Today GIMPS says you can have the exponent as long as you report in every 60 days and are making some progress. Unfortunately, some users have exponents scheduled to complete many years from now. This annoys users that like to see a nice orderly progress of milestones.

While we still want to encourage those with less than state-of-the-art machines to contribute, we must also place a burden on the user to select an appropriate work type. Defining a 6, 9, or 12 month time limit does not seem unreasonable - as long as the user is informed ahead of time. The user can still reserve an exponent that takes longer, but he should take care to reserve an exponent that isn't likely to interfere with milestones. If you get a first-time test around 18 million today, it will probably be two years before you are holding up a milestone.

GIMPS responsibility is to clearly define what is expected of the user, and putting in place the server upgrades necessary to make it easy for the user to meet those requirements.
Prime95 is online now   Reply With Quote
Old 2003-01-17, 23:37   #11
outlnder
 
outlnder's Avatar
 
Aug 2002

2×3×53 Posts
Default

If we are to place limits on how long an exponent may be checked out to any individual, I vote for 6 months MAX.

I would gladly do all the lower outstanding exponents and give any credit for anything to the person the exponent is checked out to. That includes if the exponent was a MP.

I personally think it rediculous for any exponent(except 10M digits) to be out longer than 6 months.
outlnder is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Poaching blip Data 8 2016-01-30 01:59
Poaching davieddy Lounge 6 2010-10-16 12:31
Poaching and v5 PrimeCruncher PrimeNet 6 2004-04-05 19:17
Officially poaching very old exponents Prime95 Data 17 2003-11-13 02:13
New fashion poaching (???) lycorn Lounge 6 2003-01-31 08:33

All times are UTC. The time now is 13:39.

Thu May 13 13:39:46 UTC 2021 up 35 days, 8:20, 1 user, load averages: 3.10, 3.40, 3.30

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.