Advanced search

Message boards : Number crunching : "Users Ranking" distorted by collective contributions

Author Message
Erich56
Send message
Joined: 1 Jan 15
Posts: 372
Credit: 1,685,138,202
RAC: 2,964,912
Level
His
Scientific publications
watwatwat
Message 47930 - Posted: 29 Sep 2017 | 5:47:33 UTC

There are increasingly cases where quite a number of cruchers register under one single user name, thus destorting the credits results shown under "Users Ranking".

Such cases can be seen here: http://gpugrid.net/top_users.php?sort_by=expavg_credit&offset=0 - rank 8, 9, 10 for RAC (at the present point of time). Each of those consist of between 130 and more than 200 hosts, so no wonder that their RACs are close to 5 million and above.

Whereas this is in no way to criticise their contribution to GPUGRID, it would be fair to individual crunchers if the Users Ranking page differentiates between individuals and collectives. To have the figures shown in one single list, as it is the case now, does not make a lot of sense.
The collective contribution of such pools shouldn't be listed under individual users, thus distorting the whole picture.

Is there a way for GPUGRID to make the necessary change to the Users Ranking page?

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 589
Credit: 2,044,516,875
RAC: 1,490,618
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 47931 - Posted: 29 Sep 2017 | 9:56:52 UTC - in response to Message 47930.
Last modified: 29 Sep 2017 | 10:07:59 UTC

If as you claim these accounts are multiple users using one account there is nothing in the rules saying they cannot do so.
It is in reality the same as any user having access to multiple machines which they have permission for but don't belong to them registered under one account.
You could also say it is unfair if one user has multiple machines that they own under one account because the majority will only have one
There is also a HOST ranking whic shows each individual machines performance regardless of user.
I know what you are saying but GpuGrid is certainly not going to get involved with this or spend time worrying about it.

I think you should realise you personally make a huge contribution to this project and I'm sure it is appreciated although maybe not rewarded in the way it should be as in project updates and good communication. Don't take the metrics too seriously as each and every metric on this site is flawed in one way or another.

Erich56
Send message
Joined: 1 Jan 15
Posts: 372
Credit: 1,685,138,202
RAC: 2,964,912
Level
His
Scientific publications
watwatwat
Message 47932 - Posted: 29 Sep 2017 | 16:53:51 UTC - in response to Message 47931.

I know that this is not against any rules.

Also, to point this out very clearly, we should be glad about anyone who contributes to this valuable project.
I do not complain in any way that the Gridcoin people participate here.
In fact, as far as I am informed, in this case grcpool is "forced" by the rules of gridcoin to form pools under user accounts, as one has to join the gridcoin team to earn GRC.

What I am argueing is not more and not less than that there should simply be an extra ranking system for such cases.

Coming back to what you are saying in your above reply: if one user has several PCs he/she is dedicating to the project, then it should definitely be no problem to have this listed under "individual". Even if a user has access to multiple machines that don't belong to him but he has permission for - this would still fall under "individual", as we talk about one person.

But once dozens or hundreds of people are using only one account what subsequently strongly distorts the user statistics, then it would only be fair to have steps to differentiate between "individual" and "collective".
To say it again, there is nothing wrong whatsoever with the "collectives" - but in any ranking lists they should be shown under what they are.

Ultimately though, you are probably right when you say that we should not take the metrics too seriously, after all :-)

Erich56
Send message
Joined: 1 Jan 15
Posts: 372
Credit: 1,685,138,202
RAC: 2,964,912
Level
His
Scientific publications
watwatwat
Message 47969 - Posted: 13 Oct 2017 | 16:48:36 UTC

no one else any opinion regarding this matter?

Profile Logan Carr
Send message
Joined: 12 Aug 15
Posts: 207
Credit: 31,560,850
RAC: 91,597
Level
Val
Scientific publications
wat
Message 47973 - Posted: 15 Oct 2017 | 2:56:41 UTC - in response to Message 47969.

no one else any opinion regarding this matter?


I can see what you mean, Erich, but as Betting Said, I don't think gpugrid takes this as a priority.

Well put though.
____________
Cruncher/Learner in progress.

klepel
Send message
Joined: 23 Dec 09
Posts: 138
Credit: 1,836,769,677
RAC: 1,438,988
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 47994 - Posted: 16 Oct 2017 | 16:43:46 UTC

I do not have any problems with pools what so ever. Some of us have access to more recourses (economic, hardware) than others. I am not able to see any distortion.

Yes, the pools will occupy the first spots of the top volunteer list at some time in the future, but I will mentally subtract them from the list and compare myself to people like you and me, whom I assume are single crunchers and have more or less the same access to money and hardware.

The pools will overtake me for sure, but I pay more attention to you and PappaLitto, as you will overtake me in the foreseeable future, if I am not able to add a one or two GPUs:-)

PappaLitto
Send message
Joined: 21 Mar 16
Posts: 271
Credit: 1,332,297,781
RAC: 5,255,441
Level
Met
Scientific publications
watwat
Message 47995 - Posted: 16 Oct 2017 | 17:00:11 UTC

Theoretically this problem can be solved by an ever increasing amount of grcpools. If the credit is spread out enough, no single grcpool will be on the leaderboards. As for me, as long as the work units are flowing, and real science is being done, I don't care how it happens.

Post to thread

Message boards : Number crunching : "Users Ranking" distorted by collective contributions