Thierry, what kind of unpleasant truth have you said that demanded modding?

He copied a message from Jose that I deleted.

---

no flaming jose, you'd look awful as toast.

and what of us who don't run an optimized version and don't want the points? why can't the credits awarded be based only on the workunits themselves and not how fast an optimized boinc runs them? the project team controls those values. if you're going to revamp how credits are awarded, then how about using only the size of the unit and time taken to crunch it? they all go out blind so no one can whine that they aren't getting a big one. is something like that doable?

don't hate me beacuse i have no clue!

---

From Jose:

From Thierry:

Jose,
We are standard BOINCers. We don't want any flame. We have some requests. We will defend them. There is a lot of user behind us. We have nothing to loose in the credits changes.

(deleted)

That is why Rosetta and the DC community should rethink their dependence on the BOINC platform(..).
Sorry Ethan.

(..)

I am affraid you are going to have a long dayy.

---

no flaming jose, you'd look awful as toast.

and what of us who don't run an optimized version and don't want the points? why can't the credits awarded be based onlyon the workunits themselves and not how fast an optimized boinc runs them? the project team controls those values. if you're going to revamp how credits are awarded, then how about using only the size of the unit and time taken to crunch it? they all go out blind so no one can whine that they aren't getting a big one. is something like that doable?

don't hate me beacuse i have no clue!

I dont hate you but you are trying to refight the optimized clients issue

Hey why dont we give credits by the percentage of time the computer is crunching for Rosetta only? The larger your Rosetta Share the larger the ammount credits.

Or better still: If you only crunch Rosetta you get an additional double credit bonus?

And the better the cpu and the more memory your Rosetta only cruncher has then the bonus could be as high as 10 times .

BTW if you dont want the credits. The solution is simple ask for them to be Zeroed out. :)

What is so holy about the sheer value of the credits per work done?

imho it's absolutely irrelevant if it's 2 crdits per decoy, 200, 2000 or 0.2, as long as every decoy gets the same. The value is only interesting if you want to compare it with other projects (i.e. within BOINC)

Up to now the value is a rather random number here, as everybody gets just what he claims, regardless of his or her work done. There is even a kind of standard mechanism to claim more without doing more, called "optimized" client. So the numbers of stats don't reflect the actual work done by either team or user.

Why don't you want to change this to something meaningful? Regardless of the direction, more credits for stock users, or less for opties, it will show the real picture of work done, not some caricature.

The main reason to equalize credits up instead of down, is that a lot of top numbers were achieved using optimized clients, some obviously from before february. If we equalize down, it'll be much harder for people to catch people up at the top. However, if we equalize up, it'll let people catch up on the scoreboard at the higher pace.

Just my .02, and glad to see some people making sense of the idea.

Any change up or down of the old credits is backdating and will be consdered a breach of the NO backdating pledge.

Any change up or down of the old credits is backdating and will be consdered a breach of the NO backdating pledge.

I don't remember having done a pledge here ;)

Any change up or down of the old credits is backdating and will be consdered a breach of the NO backdating pledge.

This is not the case or I would have stopped the thread. It's a close issue, but not the same. Rather than impacting users of optimized clients, it would only impact those who used the standard client instead of the optimized. . for whatever reason.

---

I like this idea - I have always said backdating is a bad idea, and going forward is the only way to go.

No backdating for those that did not use the optimized client or backdating for those that did. All we are concerned about is going forward.

However lets add another layer of complexity into the picture, something that I have been thinking about but have no knowledge of how to implement.

Other projects use a quorum to keep credits equal and to check the science done. There is no reason why this could not be done here as well. Rosetta is different and does different science. But just hear me out for a second.

Credits are granted on time spent per WU, and amount of science completed. A cross check system could be implemented to decrease those huge credit claims so that as each WU is reported back, it is compared to the other WU's in the quorum. If credit claimed per CPU cycle is way out of line compared to the other WU;s in the quorum it could be adjusted downwards or upwards as compared to the quorum.
This does not mean the every WU in the quorum is granted the same credit as some crunchers like 2 hours and some like 24 hours. It could not be difficult to calculate the cpu time to crunch a WU and adjust it accordingly to other members of the quorum. And because the way Rosetta WU's are generated, duplicate work WU's would not be necessary to crunch. Quorums could be made up of completely different WU's it would not matter. Each WU is granted credit based on it's own merit but cross checked for proper granting against other WU's. It would not even be necessary to display the quorums unless we wanted to review what those adjustments were.

I don't know if I am making this clear or not, or just adding more complexity into the system but in another life time I used to be a programmer and I know that adjustments like this can be done.

This would also make it unnecessary for the project dev's to step in and remove those that would try to cheat the system because the "anti cheating credit calculations" would take care of that.

Everyone or no one could use the optimized clients, each would be granted credits properly based upon their dedication to the project.

---

SETI.USA

Kevint, I think you just described the new work credit system. Please search for tralala's post which has a good description of it.

---

i have no problem with the optimized clients or people who use them, it was my personal choice not to. however i don't want or need to have my credits changed (for the past work i've done, or the future work) to reflect that. some others who don't use them might want them, not me.

what is needed is a fair and equitable plan that takes boinc and the optimizing out of the equation. that would be the only fair way to award credits based on work done, not how fast it's done.

by jose

Hey why dont we give credits by the percentage of time the computer is crunching for Rosetta only> The larger your Rosetta Share the larger the credits.

the only problem i can see with this is it will award more credits to a slow machine that crunches rosetta only, like my old sony P4 1.0. which i don't think if fair.

---

Any change up or down of the old credits is backdating and will be consdered a breach of the NO backdating pledge.

This is not the case or I would have stopped the thread. It's a close issue, but not the same. Rather than impacting users of optimized clients, it would only impact those who used the standard client instead of the optimized. . for whatever reason.

Ethan it is changing the old credits. ANY CHANGe is back dating.

No backdating is no changes in the old credits. If you are telling me that Bakers said that he is going to change some credits up , then we ware going to have problems. HUGE problems

People will not accept changes up of the old credits and they will consture it as Baker going back on his word.

The main reason to equalize credits up instead of down, is that a lot of top numbers were achieved using optimized clients, some obviously from before february. If we equalize down, it'll be much harder for people to catch people up at the top. However, if we equalize up, it'll let people catch up on the scoreboard at the higher pace.

Just my .02, and glad to see some people making sense of the idea.

Any change up or down of the old credits is backdating and will be consdered a breach of the NO backdating pledge.

Well forget about the backdating then. Hence the edit in my first post. I hadn't read the promise to not change. Regardless, the raising of credits to the optimized level for the future would equal the playing field for everyone on rosetta and newer users catch up. As well as any new users that sign up in the future.

---

Kevint, I think you just described the new work credit system. Please search for tralala's post which has a good description of it.

it's here

i have no problem with the optimized clients or people who use them, it was my personal choice not to. however i don't want or need to have my credits changed (for the past work i've done, or the future work) to reflect that. some others who don't use them might want them, not me.

what is needed is a fair and equitable plan that takes boinc and the optimizing out of the equation. that would be the only fair way to award credits based on work done, not how fast it's done.

by jose

Hey why dont we give credits by the percentage of time the computer is crunching for Rosetta only> The larger your Rosetta Share the larger the credits.

the only problem i can see with this is it will award more credits to a slow machine that crunches rosetta only, like my old sony P4 1.0. which i don't think if fair.

It is fair
Any computer running 100% rosetta should have a bonus to the point a Kentfield or a power mack running 100 % rosetta should have a huge credit bonus .

I'm not saying what anyone is going to do, we dont' have a batphone. I'm just saying the issue of 'backdating' was editing all scores based on the new system. The poster of this thread was suggesting another idea, my commenting on the idea doesn't mean it's likely to be implemented.

---

Why am I not allowed to ask Jose why he opposed to a fair view at the past?

Why am I not allowed to ask Jose why he opposed to a fair view at the past?

You are, you didn't word it that way.

---

I like this idea - I have always said backdating is a bad idea, and going forward is the only way to go.

No backdating for those that did not use the optimized client or backdating for those that did. All we are concerned about is going forward.

However lets add another layer of complexity into the picture, something that I have been thinking about but have no knowledge of how to implement.

Other projects use a quorum to keep credits equal and to check the science done. There is no reason why this could not be done here as well. Rosetta is different and does different science. But just hear me out for a second.

Credits are granted on time spent per WU, and amount of science completed. A cross check system could be implemented to decrease those huge credit claims so that as each WU is reported back, it is compared to the other WU's in the quorum. If credit claimed per CPU cycle is way out of line compared to the other WU;s in the quorum it could be adjusted downwards or upwards as compared to the quorum.
This does not mean the every WU in the quorum is granted the same credit as some crunchers like 2 hours and some like 24 hours. It could not be difficult to calculate the cpu time to crunch a WU and adjust it accordingly to other members of the quorum. And because the way Rosetta WU's are generated, duplicate work WU's would not be necessary to crunch. Quorums could be made up of completely different WU's it would not matter. Each WU is granted credit based on it's own merit but cross checked for proper granting against other WU's. It would not even be necessary to display the quorums unless we wanted to review what those adjustments were.

I don't know if I am making this clear or not, or just adding more complexity into the system but in another life time I used to be a programmer and I know that adjustments like this can be done.

This would also make it unnecessary for the project dev's to step in and remove those that would try to cheat the system because the "anti cheating credit calculations" would take care of that.

Everyone or no one could use the optimized clients, each would be granted credits properly based upon their dedication to the project.

Kevin: There s a way to make a cluster analysis of all work units returned per type of client that will show the extreme deviations from what the rest of the machines wihing the same client and or as a whole were producing. And those extremelly disparate machines (up or down) could be removed from the credit detereming calculation.

Before this whole mess , I was working (a long with a friend) with an analysis of the data that included the time factors you mentioned. Alas. he and I kind of got tired of the backdating flames and the calling us cheats.

So OK:

Jose,
why are you opposed to a fair view of the past?