Review the system of stats

[manalog]

Hi all,
I started folding 3 months ago by running a GPU H24 after two years of Boinc. I love the project and by reading the projects's description I think is probably one of the best in purposes and result. Despite this, I noticed that the partecipation in Folding from the Boinc team's is rather poor: just as an example my team, BOINC.Italy, is very important in the BOINC Scenario, while has only 3 active users in Folding. The explanation is IMHO simple: Folding doesn't provide a system of stats like Boinc.
I suggest Folding's Admins to give more revelance to user stats in order to increase dramatically the number of users crunching, and I think it could be done in a very simple way:
first of all by allowing users to register with e-mail and publishing the stats for any user, with the number of points, the features of his machine and, why no, the list of projects in which he partecipated, allowing him to learn more about the work he's doing. Eventually you could also add a personal profile and some social features. Users who won't like this new feature could continue folding as Anonymous user.
The best of the best would be if users could link their boinc CPID by putting Folding in the BoincStats circuit and in this way syncronizing stats from different projects, contributing to the team's stats and creating challenges even in Folding. I think it would be no too difficult setting a conversion ratio between folding points and boinc's credits. Folding would continue to live separated from Boinc in all except stats, one of the most important think in my opinion to get new users chrunching.
What do you think about this? I'm a novice and so I don't know if there are some technical or phylosophical questions that don't allow to do this.
Sorry for my english,
Matteo.

[bruce]

See these recent topics on BOINC
Subject: BOINC Addition
Subject: Can Folding@home be added as a BOINC project?

I think it would be not too difficult setting a conversion ratio between folding points and boinc's credits. Folding would continue to live separated from Boinc in all except stats, one of the most important think in my opinion to get new users chrunching

While your ideals seem positive, I don't think folks will agree on any kind of a direct conversion ratio.

It has been many years since I participated in BOINC so my information may be obsolete, but from what I remember, BOINC awarded a specific credit for each WU completed. FAH awards a small credit for each WU completed, but it awards a sizeable bonus if you complete the same WU faster. (See my comments in the linked thread.) That bonus is much more than the points awarded for simply completing the WU and I doubt there's a way to convert those bonus points to BOINC points.

[foldy]

Before poem@home folding closed boinc app I could compare the points generated to folding@home and it was POEM 1 mill points to FAH 100 k points given.
Boinc still has the GPUGRID project which does folding similar but different to folding@home.
So if users on Boinc want to do folding they will choose GPUGRID and not switch to folding@home.
But recently GPUGRID is low on work units they have to setup more projects first so some users think about folding@home because their GPUs got idle.

[manalog]

Ok, I got it.
I didn't noticed the bonus system cos I launched folding on my GPU 3 months ago and never checked the points per WU.

Althought this I think a review of stats pages should be anyway made. First of all IMHO letting users to signup and have their own private page with detailed stats could fidelize users and giving the opportunity to check every single result, as BOINC does. Just as an example, I became a "folder" expecially because I read some projects's descriptions and found them very interesting, but I think that the way you explain users what exactly they're doing is confused: any information on the workunits get lost after completion, and users can only see their points, but not what they crunched, which computer does the work and in how many time. For example every time a scientific paper is made using folding, users who chrunched that WU's could receive a badge on their page, inciting folders to do more and informing them on the scientific background.

Moreover, after registration users should get a code, linked to their e-mail and username, representing them uniquely. In this way would be simple to check information even with scripts: my team (and I think others too) has a very nice page for stats, and it would be good if folding's stats could be added in a simple way in that page.
As a final touch, user's code could be linked to BOINC's CPID, giving the possibility to show the credits from BOINC projects on the profile page of Folding and viceversa (by agreeing with Berkeley, obvioulsy).

In conclusion, I think a modernization of the stats system would be a key point to improving partecipation of users and, adding to that a sort of integration with BOINC (only as regard stats), also to reclutate new users.
Bye.

[Nathan_P]

Ok, I got it.
Althought this I think a review of stats pages should be anyway made. First of all IMHO letting users to signup and have their own private page with detailed stats could fidelize users and giving the opportunity to check every single result, as BOINC does. Just as an example, I became a "folder" expecially because I read some projects's descriptions and found them very interesting, but I think that the way you explain users what exactly they're doing is confused: any information on the workunits get lost after completion, and users can only see their points, but not what they crunched, which computer does the work and in how many time. For example every time a scientific paper is made using folding, users who chrunched that WU's could receive a badge on their page, inciting folders to do more and informing them on the scientific background.

In conclusion, I think a modernization of the stats system would be a key point to improving partecipation of users and, adding to that a sort of integration with BOINC (only as regard stats), also to reclutate new users.
Bye.

Its a reasonable idea but a badge for the users that contribute to a particular project is not necessary, If I run a gpu, over the course of a couple of months I am extremely likely to have run WU for every project out there, you don't stick to the same project, the projects i'm currently working on are 13500 and 9838. If I go and check my logs I will have run a different project earlier today and when these WU finish I will end up with something else. Its extremely rare to end up working on only 1 or 2 projects for any length of time unless you are running the nacl client. The last time I remember it happening to any great degree was when bigadv projects were around.

I do like the idea of being told info on how long it took to complete a project; number of WU folded per run/Clone, number of gpu's used and time needed would be insightful and perhaps drive people to contribute more.

Stats may need an overhaul but at the moment i'd much rather the effort be put into the new cpu & gpu cores, the new client that's in testing and keeping the current stats server online for more than a few weeks at a time

[davidcoton]

I assume you are aware of EOC and Kakaostats? If not, find them at http://www.kakaostats.net and http://folding.extremeoverclocking.com

[bruce]

Users already have a code, linked to their account name and email -- it's called a passkey.

Unfortunately, in the early years of FAH, there was no passkey, so the old-timers earned points which were not associated with their email. While the points are probably linked to their folding name, more than one person could use the same name so there's no way to retroactively associate those points with their email (/passkey).

At one time, users could query a list of projects and WU counts that they completed. The database grew to the point that it became "expensive" to maintain the ability to query that information so it was discontinued.

As Nathan_P has implied, the FAH servers are responsible to assign specific projects to your system. Assignment priority is based on science, not user preferences. A long trajectory consisting of many sequential WUs is much more valuable that lots of short ones. This drives both the bonus points and the assignment methodology.