mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > No Prime Left Behind

Reply
 
Thread Tools
Old 2009-05-04, 01:36   #23
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

3×2,083 Posts
Default

I see what you're saying. However, there is one major problem that still can't be remedied by having the stats not default to any one method of sorting: which set of stats gets exported to Free-DC and other stats sites like theirs? I'm in favor of having combined stats be the ones that are exported, though obviously that's somewhat contested.

The reason why I feel this way is because I believe it is the fullest and most accurate representation of the true amount of work done by everyone on the project. Some users have done a lot of manual work, some have done a lot of LLRnet, and others have done both--so, to pick solely one or the other to export to stats sites (which users will interpret, no matter what our "official" position is regarding there being no particular default, as the primary stats metric), would be unfair to whichever group of users is biased towards the "less prominent" side. At least the way I see it, using the combined score leaves no room for unfairness either way.

Of course, as stated earlier, regardless of which one stat we end up exporting to Free-DC et al., users can still feel free to follow their own progress on whichever stat they would like. I for one find it rather interesting to sometimes see just how the users stack up when counted by raw k/n pairs (though generally I find the score most useful since it represents work done rather than raw k/n pairs), and I imagine it's similar for other users.

Max
mdettweiler is offline   Reply With Quote
Old 2009-05-04, 02:21   #24
AMDave
 
AMDave's Avatar
 
Jan 2006
deep in a while-loop

2×7×47 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
AMDave, could such a method of stats display be implemented?
Anything is possible.
They are all good ideas.
Most columns are currently re-sortable where the column heading is underlined.
I intend to cover all bases.
AMDave is offline   Reply With Quote
Old 2009-05-04, 02:52   #25
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

2×72×103 Posts
Default

Perhaps I skimmed a post or two somewhere but I didn't know that there was going to be an effort underway to export the stats elsewhere. I thought our stats were stand alone where AMDave, Bok, and IronBits have put them as they are currently linked to from our "LLRnet servers for NPLB" thread.

Based on that, here is another suggestion:

Do it the way I suggested in "our" stats with separate links for each type of stats. For sites where the stats are exported, let them decide WHICH stats they want or if they want stats shown all 3 ways, they can export all 3 types of stats, LLRnet, manual, and combined.

This means that if a majority of folks at the Free-DC site wants only LLRnet stats shown there, then so be it.

Perhaps a "disclaimer" could be put at the bottom of the LLRnet only stats that the totals do not include manually completed files. And at the bottom of the manual only stats, it could be stated that the totals do not include the LLRnet stats.

This allows the various DC folks with the huge # of machines to compete in the manner that they or their leader(s) prefer.


Gary

Last fiddled with by gd_barnes on 2009-05-04 at 02:53
gd_barnes is offline   Reply With Quote
Old 2009-05-04, 03:03   #26
IronBits
I ♥ BOINC!
 
IronBits's Avatar
 
Oct 2002
Glendale, AZ. (USA)

3·7·53 Posts
Default

It has nothing to do with Free-DC.
If you want your project membership to grow, you have to make it as simple as possible for the majority of folks, the newbies, to start someplace so they can help.
For that, we have online llrnet servers and it's current stats system that only shows the online participation levels.
We need to get the automatted team and registration setup and get it into the DC Vault to increase membership into the project.

For everything else we will have a 'fill in the blank' stats system.

I'm talking about the ONLINE system for newbie folks, which the whole stats system was developed for, to entice users to participate and be competitive.

If you can't see that or understand that, then I'll never be able to explain it any better than that, so, I'm done.

Gary, ultimately it's your project, do as you see fit.
IronBits is offline   Reply With Quote
Old 2009-05-04, 03:21   #27
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

3×2,083 Posts
Default

Quote:
Originally Posted by IronBits View Post
It has nothing to do with Free-DC.
If you want your project membership to grow, you have to make it as simple as possible for the majority of folks, the newbies, to start someplace so they can help.
For that, we have online llrnet servers and it's current stats system that only shows the online participation levels.
We need to get the automatted team and registration setup and get it into the DC Vault to increase membership into the project.

For everything else we will have a 'fill in the blank' stats system.

I'm talking about the ONLINE system for newbie folks, which the whole stats system was developed for, to entice users to participate and be competitive.

If you can't see that or understand that, then I'll never be able to explain it any better than that, so, I'm done.

Gary, ultimately it's your project, do as you see fit.
Okay, I see your point. If that's what the majority of our users feel is a wise way to go, then I can be flexible.

Before I completely give up my argument, however, I will present one last point for discussion: some newbies who are heavily concerned about stats may be on dial-up, or in some other similar limited-internet-access situation, yet still want to participate such that their work will be credited in the totals that are used for interteam competitions and etc.--that is, whatever is the one stat that is chosen to be used by stats aggregators such as Free-DC's. Most stats aggregation places will probably pick whichever one is

Having stated this, I will now leave this to further discussion by others and the ultimate decision to Gary. Whatever people feel to be most useful to the members of this project, both old, new, and in between, is fine with me. And, as Gary said, stats aggregation sites of course have the option of importing all three if they'd like, which, come to think of it, would offer the best of all three worlds.

Max

Last fiddled with by mdettweiler on 2009-05-04 at 03:23
mdettweiler is offline   Reply With Quote
Old 2009-05-04, 07:01   #28
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

2·72·103 Posts
Default

I just had an excellent 45 minute phone conversation with David. It's unbelievable how big of a miscommunication has occurred here. Let me give a synopsis of the discusssion:

1. David was never concerned about the stats ULTIMATELY being shown as combined manual and LLRnet. He only did not want them shown as combined during the month of May (and perhaps a few days into June).

2. We don't need separate LLRnet, manual, and combined stats. Until we give the word, AMDave will make sure that only LLRnet stats are showing up even if we start loading in manual results immediately!

3. We can start loading the manual results into the database right away. We don't need to wait. That way, we can still query the various k's and n-ranges from the manual results but the stats won't reflect the manual results/primes until we have AMDave "flip the switch".

4. The Daves are working towards automating the online reservation system at DC-Vault. Somehow that got mixed up with the online results in some of the discussion. I know I got confused.


For anyone interested: In many controversial situations online where passionate people are involved, if you think that communication can reasonably occur without actually talking to people on the phone or in person, you are sorely mistaken. IM's do not work either. David is a gracious and very hard-working guy and really wants to see the project succeed. He feels VERY strongly that ALL stats should be in the database and was never trying to stop us from the manual stats import. That is what we all want! He is a former database administrator and what he wants for the database is this: That we can query anything, any k-value, n-range, date/time, server, person, you name it, and it will give you whatever information on primes or results that you want IN TOTAL. I remember him saying this before in a phone conversation that we had. Very cool!

Karsten, this has huge implications for future Riesel prime listings. Imagine this: Being able to ask the server exactly how many primes there are for any k-value, any n-range, etc., anything that you could possibly think of, the server could spit out. How cool would that be? Well, that is what David wants; a full accounting of every prime and result that we have done here.

If anyone wants to call me at any time to discuss anything about the project, feel free to Email or PM me and I'll give you my phone #. There is no reason to have this level of miscommunication go on for as long as it did. I'm glad I wasn't on a business trip so that the situation could be rectified.

Max, time to get busy with those manual results! As you complete them, feel free to send them right along to the Daves.

Thanks again David for your usual excellent explanation of the inner workings of the database and of DC in general. Next to you, I'm a complete newb at this DC stuff. If there is anything you'd like to add or if I misunderstood anything from our conversation, please post it here.

Everyone, I hope this puts this issue to rest. Can we please get back to the way we were 2-3 days ago before the controversy surrounding the manual stats import came up? I would be greatly relieved if we could. I don't want to lose anyone. Every person is valuable to the project in his own way.

And finally: This essentially invalidated my last 2 posts. I'll delete them sometime later today. Edit: They have been deleted.


Thank you,
Gary

Last fiddled with by gd_barnes on 2009-05-04 at 11:29
gd_barnes is offline   Reply With Quote
Old 2009-05-04, 07:15   #29
IronBits
I ♥ BOINC!
 
IronBits's Avatar
 
Oct 2002
Glendale, AZ. (USA)

3·7·53 Posts
Default

It has always been my intention to round up all the work everyone has completed that is scattered everywhere and get them into the database.
In addition, I'm still trying to work up a solution in some way to help automate the reservations for those that want to do manual work, without having to use the Forum to do so.
Having a registration page, similar to what Boinc uses, so when a new user joins the project, they can join/create a team and add a second email address for the auto-notify if they would like, based off the 'username' that they used when they registered on the new NPLB Forum.

I would like to get all the offline work imported into the database first.
I want to avoid the trickle affect of having the stats jumping all around as each file is loaded so the work won't show up until it's all imported.

Once all the data is in there, and all the hard work done by all has been loaded into the database, then Gary can make an official announcement that
the totals in the stats system now reflect the total of all the contributions by everyone since the project started.
A huge milestone if you ask me and one I've been waiting for, for a very long time.

If all the importing of the offline files is completed prior to the Free-DC Project of the Month BBQ which ends on June 1st,
I have requested from Gary that the offline stats not be combined with the online stats until a few days into the month of June to let Free-DC savor and enjoy their hard earned stats run.

Once the real totals are shown, then, the totals will be what they are, and everyone should be happy, especially me.

There should be no reason for anyone to feel miffed, and everyone should be proud and happy at how far NPLB has come since the project database was handed to me back on 15 Sep 2008 by AES. It has come a very long way in 8 months time.

AMDave has done a terrific job on the stats system and we all will finally get a true accounting of who has done what, when, where, which server and port or offline and we should be able to tell if any k/n pairs are missing from any range.

Last fiddled with by IronBits on 2009-05-04 at 07:47
IronBits is offline   Reply With Quote
Old 2009-05-04, 11:12   #30
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

235568 Posts
Default

I have recieved an Email and a PM from 2 of our regulars expressing a concern that I believe is justified from all of this and is why this has generated such controversy. Lacking DC experience, I did not initially realize a pattern that this project may be falling into with "quasi BOINC" or "BOINC-like" projects. It is this:

A project degenerates into a huge gripe fest about the way scores are calculated, the way they can be moved, credits recieved for certain work units or activities, you name it, etc. Then the project dies as most of the people take their machines elsewhere because the negativity makes it no longer any fun.

David, can you see how your 1st post in this thread has caused this concern amongst our regulars? It appeared to many of them that we were starting down the scores gripe path. We only started out with the idea of bringing in the manual stats, which had been in the "idea phase" for months. Unfortunately the timing of that was absolutely HORRIBLE. It couldn't have been worse. In no way did we intend for such (what I thought) an innocuous thing to generate such a huge cloud of controversy.

I do want to say this: I believe your concern was justified after you made a large effort to help promote our project and as much as our regulars might not like to hear this: I believe it would have been unfair to count the manual results in our stats for May. The problem was that the concern was taken completely wrong by several people here who have experienced such things on other projects. A couple of people suggested that it should have been taken offline. Regardless, what is done is done. No one is to blame here. It was just horrendous timing on something that had been in the works for months.

David, this is in no way directed towards you. But in order to honor our regulars committment to the project, we need to set some guidelines here:

About scores: At no time is anyone to engage in griping or otherwise negative comments about the way scores are kept or teams divided. If anyone does this, I/we will simply suggest that they find another project if they don't like it and could even go so far as blocking their access to a server. NPLB has a great product and we don't need nor want negative people griping about things. Griping will get you nowhere here.

If you feel a need to gripe, moan, or complain about something scores-related, please don't post it publicly. That's when the crap starts. It's can be very difficult to imply (or infer) correct tone or intent in writing, even with emoticons. Send me a PM or an Email. The admins do that amongst themselves frequently. In the past, I have frequently sent a big venting Email to Max. Ian sends them to me from time-to-time. That's how we get it out and then we're usually OK.

About teams: People can change teams but if they do so, their score up to that point stays with their old team. You will need to create a new UserID, get your Email entered for it, and tell the admins what team you want for the new UserID, just like you were creating one as a brand new person to the project. We are not set up to split scores based on a specific date for one UserID over multiple teams. We will not get into a "move around" fest. As a matter of fact, AMDave, I would like to suggest that we not implement such a thing. It should not be easy for people to change teams. They should have to jump through hoops. I want to make the moving around of teams difficult for people.

I would like to encourage more people to compete based on top-5000 primes score. There are many primes for everyone to find; even for those with few resources. When you find a prime and report it there, you are awarded a certain score that is directly based on the length of time it should take to run that specific test.

People are allowed to make kind requests such as "Can we consider doing this or that or the other thing." about scores or teams. Many times, such requests are good ones and will always be fully considered. Other comments that are good are "Have you considered how this might affect that, etc.". I want to encourage people to make constructive suggestions. In other words, if you feel you'd like something done differently, state it in a constructive way that gives different alternatives.

To be honest, David, when I stated that we were going to start the conversion process that allowed the manual stats import, I had no idea that Free-DC was coming on for a month-long rally here. For that I apologize. I would have put it off had I known. My bad because I should have kept up with what was going on at Free-DC.

For all of the regulars that have been with us for a long time, I want to say thank you. I hope this will alleviate your fears about the project degenerating into a gripe-fest about scores. It will not be tolerated.

For all of the new folks, I want to say thank you for considering our project. Whether you are in it for the scores or the primes or just for the commoradarie, welcome. We want everyone to be positive and have fun.

Let's all not forget how cool the stats/scores are. They are what help us promote the project. If people are here only for the scores, that is fine but if you find you don't like the way that they are computed or divided up, it's probably best to move elsewhere.


Gary

Last fiddled with by gd_barnes on 2009-05-04 at 11:24
gd_barnes is offline   Reply With Quote
Old 2009-05-04, 12:28   #31
Mini-Geek
Account Deleted
 
Mini-Geek's Avatar
 
"Tim Sorbera"
Aug 2006
San Antonio, TX USA

17·251 Posts
Default

Sounds like I missed something, probably related to Gary's two posts that he deleted, but I'm kind of glad I missed it because it seems to have been angry and misinformed.
Anyway, I agree with Gary's big post, but I'd like it to be clarified that we can suggest changes to scoring or teams without a problem, but should accept the leaders' replies even when the don't agree with our suggestion (otherwise it'd be griping).
Quote:
Originally Posted by gd_barnes View Post
About teams: People can change teams but if they do so, their score up to that point stays with their old team. You will need to create a new UserID, get your Email entered for it, and tell the admins what team you want for the new UserID, just like you were creating one as a brand new person to the project. We are not set up to split scores based on a specific date for one UserID over multiple teams. We will not get into a "move around" fest. As a matter of fact, AMDave, I would like to suggest that we not implement such a thing. It should not be easy for people to change teams. They should have to jump through hoops. I want to make the moving around of teams difficult for people.
Maybe you want to make people jump through hoops to change teams just on the principle of it, but if the problem is more technical in nature consider this alternative: (though you could still have them jump through certain hoops based on principle even when the technical side of it is easy...though I'd recommend against that)
When each pair is returned by a user, it is marked in the database with the user's current team. This value is only changed when a user is changing from no team to a team. User stats are determined by the user name attached to each pair, (as they are now) while team stats are determined by the team marker on each pair (instead of by the current team of the user, as it is now).
That way, the work they completed while on Team X stays with that team, while all new work is reported as Team Y. (as it's been stated several times is how the stats should work when a user changes teams)
Mini-Geek is offline   Reply With Quote
Old 2009-05-04, 12:52   #32
mdettweiler
A Sunny Moo
 
mdettweiler's Avatar
 
Aug 2007
USA (GMT-5)

11000011010012 Posts
Default

Quote:
Originally Posted by Mini-Geek View Post
Maybe you want to make people jump through hoops to change teams just on the principle of it, but if the problem is more technical in nature consider this alternative: (though you could still have them jump through certain hoops based on principle even when the technical side of it is easy...though I'd recommend against that)
When each pair is returned by a user, it is marked in the database with the user's current team. This value is only changed when a user is changing from no team to a team. User stats are determined by the user name attached to each pair, (as they are now) while team stats are determined by the team marker on each pair (instead of by the current team of the user, as it is now).
That way, the work they completed while on Team X stays with that team, while all new work is reported as Team Y. (as it's been stated several times is how the stats should work when a user changes teams)
Well, I may not be entirely accurate on how this stuff works, but as far as I know, that's the way it *used* to work. Ever since we transitioned away from having team names in the username in "TeamName_UserName" format, I'm pretty sure that all team names were *not* stored in a DB field with each result, but instead each result was indexed by username, and the DB just looked up that user's team to determine how the stats go.

AMDave, please feel free to correct me if I'm wrong here. (I haven't taken a look at the CSV format used for a while, so I'm a little rusty on this stuff. )

Meanwhile: I'm glad all of this has been worked out and apparent misconceptions have been cleared up! As Gary said, I'll get started on preparing the manual files for import as soon as possible. I'll send the Daves a PM shortly with some info on how I was planning to distinguish manual results from automatic ones.

Max
mdettweiler is offline   Reply With Quote
Old 2009-05-04, 13:15   #33
AMDave
 
AMDave's Avatar
 
Jan 2006
deep in a while-loop

2×7×47 Posts
Default

"Play by play vs Color vs Unilateral"

Play by Play:

Automated signup is coming to a project near you!
Manual stats merge is a non-issue because we have it whipped
1.5 million DC contributers already expect things to work just-so.
You gotta roll with it or it will roll right over you.
The Dev-Team are recording 'everything' so be careful what you wish for.


Color:

Gary said the stats are "cool"
Awesome!
Heh heh.


Unilateral:

NPLB is going to blow your freekin socks off.
Oh yeah!


Oh, and btw:
PS - the stats site migration is complete
AMDave is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Oddity in P-1 data import into mersenne.ca DB Syntony PrimeNet 5 2015-12-10 12:22
CSVs for stats available + New combined stats opyrt Prime Sierpinski Project 3 2010-05-31 08:13
GMP-ECM Manual GrK GMP-ECM 2 2007-12-25 01:55
Manual ecm jasong Factoring 7 2005-08-29 19:17
P4 On Stats HiddenWarrior Hardware 2 2003-08-13 14:39

All times are UTC. The time now is 22:12.

Sat Apr 4 22:12:34 UTC 2020 up 10 days, 19:45, 0 users, load averages: 1.42, 1.50, 1.51

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, 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.