scoring on beta client

[Brian the Fist]

Perhaps I m opening a can of worms again, but for the Beta testers:

How are you with the scoring system that is currently in place now? Aside from the obvious problem with AMD_is_logical pummeling everyone else with his cheap clock trick that is :D

Is it more or less reasonable? Are you encouraged to stick in until generation 250 since its worth many more points?

Before answering, keep in mind, the timeout delays for when the protein gets stuck will still be changing significantly before a final release, so expect each generation to take approx. the same amount of time. Although later generations might take a wee bit longer.

Constructive input is welcome but don't just say 'the current scoring sucks'.
:p

[Brian the Roman]

It'll be the same for everyone so I don't really care. And I'm primarily interested the science end anyway...

ms

[shortfinal]

Personally, I'm with Brian the Roman, I don't care as long as it is consistent.

If it's really important for the project that people let the client run to generation 250, maybe aware 'bonus' points for that generation. Something like double the points.

[m0ti]

I think that the scoring is ok.

At the very least, I did notice a larger jump up in my scores as the generations progressed.

Though giving an extra bonus for the final generation sounds like an awesome idea to me. Though you'd have to make sure that people wouldn't just back up a gen 249 and then proceed forward to gen 250.

Though I believe that most of the people in the project are more concerned with the science than getting a large number of folds.

[FoBoT]

Originally posted by m0ti
...you'd have to make sure that people wouldn't just back up a gen 249 and then proceed forward to gen 250.

Though I believe that most of the people in the project are more concerned with the science than getting a large number of folds.

if there is any way to manipulate the system to get extra points, somebody will figure it out and exploit it

even though most people wouldn't , somebody will try it and if its possible, do it on a large enough scale to mess things up

[tpdooley]

I have a teammate that's running the beta client on one of his crunching systems - and he managed in a few days to almost catch up with my score on an Athlon 600. So I moved to an AXP 1700+ (a little closer to what he was running) - and I've really got a lead now.
The new scoring technique really favors those in the lead. ( I'll refrain from taunting my teammate since he has a much larger collection of machines than I do. ;)
My only question would be whether the folks with lower powered machines would see the lower scores they get with the new client as being too small and switch to something else.
(And I realize that not everyone can go and replace the older hardware with newer 1.3-2.8Ghz (or faster) systems).

[Aegion]

Originally posted by tpdooley

My only question would be whether the folks with lower powered machines would see the lower scores they get with the new client as being too small and switch to something else.
(And I realize that not everyone can go and replace the older hardware with newer 1.3-2.8Ghz (or faster) systems).
On the other hand, with this new client, how productive are really older machines anyways? If the machine is fast enough to complete 250 structures before we switch proteins he's not at any real porportional disadvantage, and the really slow machines probably won't help much anyways. The primary end goal of the project is the scientific results, not necessarily getting as many people to participate as possible.

[Brian the Fist]

Ok, sounds great then. I will add a bonus for completing Gen. 250 as well. As for exploiting the scoring system, that's part of your jobs as beta testers :D

AMD_is_logical found a loophole (actually a bug) a while back which I since fixed - but please do try to cheat - it shouldn't let you. There should be no way you can upload a particular generation more than once, because we track what your previous generation is in each case. Thus once you've uploaded generation X, it will accept nothing from you except generation X+1. And so on. But again, a loophole could certainly still exist.

P.S. I will be wiping the results and starting a new bets on Monday or Tuesday - this will attempt to rectify the lack of helices in the structures with the current beta and should hopefulyl get us below the 5A wall we are hitting right now. If this approach works, we should be almost good to go... Keep up the great work

[Mikus]

With 2002 scoring, people got credit for building structures. Since a participant looking at the screen could see structures being built, that in my opinion gave positive feeedback.

With an earlier beta client, what I saw when looking at the screen was __not__ structures being built, but rather the client ALWAYS being "stuck". This gave me a negative feeling.

My suggestion is that the scoring you use ought to reward people for "putting in the effort" -- i.e., they ought to be made to "feel good" even while the screen is "stuck".

mikus

[Aegion]

Originally posted by Mikus
With 2002 scoring, people got credit for building structures. Since a participant looking at the screen could see structures being built, that in my opinion gave positive feeedback.

With an earlier beta client, what I saw when looking at the screen was __not__ structures being built, but rather the client ALWAYS being "stuck". This gave me a negative feeling.

My suggestion is that the scoring you use ought to reward people for "putting in the effort" -- i.e., they ought to be made to "feel good" even while the screen is "stuck".
mikus
Have you downloaded the latest version of beta 4? It pretty much addresses this issue.

[tpdooley]

Is the following considered an exploit? Or is it doable?
You run 1 fast machine and 9 slow ones. The fast machine hits generation 100, you stop everything, and copy it to all the others. likewise at 150, 200. (depending on the speed disparity of the machines).

[Brian the Fist]

why not try it and see what happens :D

[tpdooley]

'cuz I can't catch the beta system at the beginning of a new fold lately. :)