Hello, again.

In my eye's it is not really cheating, the software is there, the project allows it (though I doubt they could stop it), BOINC encourages it (they link to the clients on the site). It much easier to control if it's not open source.

But the reason I brought it up and in here was in response to David's journal. Where it was going to be used as a measure, maybe with prizes. I am more for having a decent credit system for the pure fact I can see and the project get a better idea of what crunching power there really is.

Is my slow computers cunching worth it, not for the credit they produce, but I still run them for the project. The cumulation of the slower computer adds up. BUT they do do less work. Since the the monitoring is on 'work done' even failed work done then it need to be at least have some reasonable accuracy to it.

The last project I came from and was quite heavily involved in as some people should be able to testify was Find-a-Drug.org Theire stats again was points/credit/work-done based, it had its problems but had a much better benchmarking system. AND for anyone who didn't like 'points', you could reorder by 'time' or 'molecules done' It was up to you.


Do we have slow computers on my team, Yes, will we ever get to number one, No, do we really care, not really credits are for fun and a way to entice people and get people to build/buy a newer computer.
But the top teams certainly wouldn't be there if it wasn't for the competition.


tio, It is NOT changing the system, just improving the current one we already have. I reall yhave no idea what the fuss is all about.

---

Team mauisun.org

lol, for a 10,523.94 sec. of work, he got 438.42 credit
and for the same time, most people have 10.60 credit???

be proud to take part in the projet... well, maybe Seti is not so out space with is granted credit...

---

Hello, again.

In my eye's it is not really cheating, the software is there, the project allows it (though I doubt they could stop it), BOINC encourages it (they link to the clients on the site). It much easier to control if it's not open source.

But the reason I brought it up and in here was in response to David's journal. Where it was going to be used as a measure, maybe with prizes. I am more for having a decent credit system for the pure fact I can see and the project get a better idea of what crunching power there really is.

Is my slow computers cunching worth it, not for the credit they produce, but I still run them for the project. The cumulation of the slower computer adds up. BUT they do do less work. Since the the monitoring is on 'work done' even failed work done then it need to be at least have some reasonable accuracy to it.

The last project I came from and was quite heavily involved in as some people should be able to testify was Find-a-Drug.org Theire stats again was points/credit/work-done based, it had its problems but had a much better benchmarking system. AND for anyone who didn't like 'points', you could reorder by 'time' or 'molecules done' It was up to you.


Do we have slow computers on my team, Yes, will we ever get to number one, No, do we really care, not really credits are for fun and a way to entice people and get people to build/buy a newer computer.
But the top teams certainly wouldn't be there if it wasn't for the competition.


tio, It is NOT changing the system, just improving the current one we already have. I reall yhave no idea what the fuss is all about.

This discussion has gotten way off topic from my point of view so I will be moving it to it's own thread soon. But I cannot help but mention a few things that seem to have once again become lost in the fray.

First, there are no prizes or money being offered for more credit. What Dr. Baker was suggesting was recognition for certain levels of work. Perhaps a mention in published papers of the production of a team, or recognition for the best models found by individuals. Some of these forms of recognition would not require even looking at the users credits. But, the credit system on ANY project is sufficient to determine the production level of participants and teams irrespective of the credit system used. Moreover, the determination of the best model produced is certainly possible as that is the whole point of Rosetta. These types of recognition would be based on more than a simple count of credit scores. So your original premise for raising this discussion is incorrect.

The Rosetta project uses the STANDARD BOINC system for credit awards. Comparisons of other credit systems used by non-BOINC projects or projects that use modified BOINC packages are not relevant to this discussion or Rosetta. The project simply does not have the resources to make significant changes to the BOINC architecture. Altering the BOINC credit system is a significant change. Rosetta is a "Team Player" in the BOINC community, and is doing their part by communicating their needs to the BOINC developers. There is no requirement for the use of redundancy for BOINC to work properly despite argument to the contrary. Redundancy is only necessary for two purposes. Scientific validation of the results where this is possible and necessary or to prevent some users in the BOINC community from exploiting weaknesses in the BOINC credit system for their own advantage.

For Rosetta the validation of results is handled differently because the Work Units are designed to produce unique results. There is no reasonable expectation that even the same Work unit processed on identical systems should produce the same results unless the protein structure is solved correctly. So checking of the results is handled in other ways.

When optimized clients or other exploitation of the BOINC system are an issue, some projects choose to use redundancy to average out the credit imbalances. In many ways this is more unfair than the system used at Rosetta. But for Rosetta there is no way to compare the results from two systems for validation purposes by use of redundancy because they will never match up.

In practical terms forced redundancy means that BOINC has imposed a minimum 50% reduction and in most cases 66% reduction in donated resources on projects as a patch for flaws in the BOINC credit system. If the BOINC system did not have these flaws, there would be nothing to exploit and the problem would not exist. While Rosetta may be one of the first projects to value the contributions of users so highly that they have chosen not to waste those contributions by using redundancy, they will certainly not be the last.

In the case of Rosetta this goes directly to the science. The project estimates it needs 150TF in computing power to actually make a dent in the protein problem. All of BOINC only has about 450TF. Rosetta has about 26TF now. If they chose to use redundancy they would only have about 13TF. While it is often stated that the project will gain resources by fixing this issue through redundancy, there is no objective evidence that those gains would come close to matching the losses. There is a lot of evidence that redundancy would prevent Rosetta from ever reaching the project computing goal.

In standing by the decision not to use redundancy, Rosetta is helping to elevate these issues with the BOINC developers and bring a real solution for the credit problem to the user community for ALL the projects.

But Rosetta has not simply ignored the issue. The project has taken action in extreme cases, to adjust or remove credit claims that have clearly been fraudulent. But this can only work for very extreme cases. The idea that the project will routinely run some process that adjusts credit claims to some as yet unidentified "Standard" is as unworkable as it is unfair. There is no standard that all users would accept, and the nature of the Rosetta work is such that it does not lend itself to fixed credit values for Work Units, models and steps.

The optimized clients that some users feel are at the bottom of this issue are widely available to ALL users. While the project does NOT recommend or support the use of such clients, many users do install and run them. Some people feel this constitutes cheating. To my knowledge Rosetta has taken no position on that view. The choice to use or not use these clients rests solely with the user. There is nothing Rosetta can do about that. Once BOINC released the code to the public the problem was inevitable. If you have a problem with optimized clients or credit calculation, then BOINC is the place to have it repaired. Redundancy is not a viable answer.

But Rosetta has gone still further in valuing and recognizing the contributions of time from every user. They provide credit awards even if the Work Unit has an error or fails. Why? Because you donated your time to the project and that deserves recognition. The credits are awarded for this contribution of time. At Rosetta even the errors are important and they want to reward people for finding them. No other project recognizes the users contribution of time this highly. When errors occur on other projects they are rarely if ever granted credit. At Rosetta people on slower systems or systems that never even finished a model are recognized for their contributions through credit awards.

One final point. If you read through the forums you will see that the single user with a slow system that runs part time for Rosetta, is treated with the same respect, dignity, and attention and provided the same support as the users with 50 quad CPU systems working 24/7 for Rosetta. This is evidence that Rosetta values every CPU cycle you donate, and every user, no matter who you are or the level of donation you are able to provide to the project. So you should not be put off by people outside the project who you feel may be dismissing your efforts. All of you are welcome at Rosetta, and Dr. Baker wants to find ways to recognize your help on as many levels as he can.

---

Moderator9
ROSETTA@home FAQ
Moderator Contact

I can live with that.
* Face it: if there weren't credits, if there weren't any statistics, there would be no team play. There would be no competition. When there wasn't any competition, there wasn't that many resources running any DC program.
So credits *are* important for the project.
* If everyone uses (the same) optimized clients, everyone gets a correct amount of points.
* If everyone disables his optimized client, and installs the original client, there would be +- the same result.
* I only use the optimized client, because otherwise I would not get the equal credits like another person (who do run the optimized client) gets.

As FluffyChicken said: The credits on FaD were more correct on that point.
But something I like more on Rosetta is the fact that the sole users (can) get an extra reward here: his name in the "Top predictions" page.
I agree this is something like the lottery: Everyone has a chance to get into this list. You can crunch one job, and find the best result. You can crunch millions of jobs, and never be in there. Of course there's more chance when you crunch more jobs....
But once you're in there, I guess you have more honour, then 100 points extra in the stats.

And the main reason: Maybe one day, you read your name on a page that says: Thanks to [your name] we've found a way to cure Cancer, HIV, Malaria, or whatsever...

---

Member of Dutch Power Cows

As FluffyChicken said: The credits on FaD were more correct on that point.
But something I like more on Rosetta is the fact that the sole users (can) get an extra reward here: his name in the "Top predictions" page.
I agree this is something like the lottery: Everyone has a chance to get into this list. You can crunch one job, and find the best result. You can crunch millions of jobs, and never be in there. Of course there's more chance when you crunch more jobs....
But once you're in there, I guess you have more honour, then 100 points extra in the stats.

And the main reason: Maybe one day, you read your name on a page that says: Thanks to [your name] we've found a way to cure Cancer, HIV, Malaria, or whatsever...

I like this part as well :-)
[We also had a similar thing at FaD with the 'Validated by NiH to have to have anti-cancer or anti-hiv http://stats.find-a-drug.biz/certify.php?Nickname=FluffyChicken&CertType=6


Thanks mod for splitting it off, I only posted it for information to David about what he was actually saying.
They are still the top 3 teams, just rate of production is not as high or as accurate as may be thought.
oh and this

We will have to have prizes when CASP finishes at the end of July for the top overall team and the
top team during CASP. definitely a citation in the science paper on rosetta distributed contributing at the minimum!

infers w.r.t. the first part of the post he typed, that credit is going to be used for recognition.

---

Team mauisun.org

Hello :) I appreciate th e last 2 paragraphs of this message. I am not happy with the quibbling regarding points or credits or compition. I volunteer my computer not because of competion or credits. I thought I just want to help with researchers need for the ordinary computer users willing to donate their machine time. I almost detach my computers from the project. I just have fun donating some computer time.

Hello, again.

In my eye's it is not really cheating, the software is there, the project allows it (though I doubt they could stop it), BOINC encourages it (they link to the clients on the site). It much easier to control if it's not open source.

But the reason I brought it up and in here was in response to David's journal. Where it was going to be used as a measure, maybe with prizes. I am more for having a decent credit system for the pure fact I can see and the project get a better idea of what crunching power there really is.

Is my slow computers cunching worth it, not for the credit they produce, but I still run them for the project. The cumulation of the slower computer adds up. BUT they do do less work. Since the the monitoring is on 'work done' even failed work done then it need to be at least have some reasonable accuracy to it.

The last project I came from and was quite heavily involved in as some people should be able to testify was Find-a-Drug.org Theire stats again was points/credit/work-done based, it had its problems but had a much better benchmarking system. AND for anyone who didn't like 'points', you could reorder by 'time' or 'molecules done' It was up to you.


Do we have slow computers on my team, Yes, will we ever get to number one, No, do we really care, not really credits are for fun and a way to entice people and get people to build/buy a newer computer.
But the top teams certainly wouldn't be there if it wasn't for the competition.


tio, It is NOT changing the system, just improving the current one we already have. I reall yhave no idea what the fuss is all about.

This discussion has gotten way off topic from my point of view so I will be moving it to it's own thread soon. But I cannot help but mention a few things that seem to have once again become lost in the fray.

First, there are no prizes or money being offered for more credit. What Dr. Baker was suggesting was recognition for certain levels of work. Perhaps a mention in published papers of the production of a team, or recognition for the best models found by individuals. Some of these forms of recognition would not require even looking at the users credits. But, the credit system on ANY project is sufficient to determine the production level of participants and teams irrespective of the credit system used. Moreover, the determination of the best model produced is certainly possible as that is the whole point of Rosetta. These types of recognition would be based on more than a simple count of credit scores. So your original premise for raising this discussion is incorrect.

The Rosetta project uses the STANDARD BOINC system for credit awards. Comparisons of other credit systems used by non-BOINC projects or projects that use modified BOINC packages are not relevant to this discussion or Rosetta. The project simply does not have the resources to make significant changes to the BOINC architecture. Altering the BOINC credit system is a significant change. Rosetta is a "Team Player" in the BOINC community, and is doing their part by communicating their needs to the BOINC developers. There is no requirement for the use of redundancy for BOINC to work properly despite argument to the contrary. Redundancy is only necessary for two purposes. Scientific validation of the results where this is possible and necessary or to prevent some users in the BOINC community from exploiting weaknesses in the BOINC credit system for their own advantage.

For Rosetta the validation of results is handled differently because the Work Units are designed to produce unique results. There is no reasonable expectation that even the same Work unit processed on identical systems should produce the same results unless the protein structure is solved correctly. So checking of the results is handled in other ways.

When optimized clients or other exploitation of the BOINC system are an issue, some projects choose to use redundancy to average out the credit imbalances. In many ways this is more unfair than the system used at Rosetta. But for Rosetta there is no way to compare the results from two systems for validation purposes by use of redundancy because they will never match up.

In practical terms forced redundancy means that BOINC has imposed a minimum 50% reduction and in most cases 66% reduction in donated resources on projects as a patch for flaws in the BOINC credit system. If the BOINC system did not have these flaws, there would be nothing to exploit and the problem would not exist. While Rosetta may be one of the first projects to value the contributions of users so highly that they have chosen not to waste those contributions by using redundancy, they will certainly not be the last.

In the case of Rosetta this goes directly to the science. The project estimates it needs 150TF in computing power to actually make a dent in the protein problem. All of BOINC only has about 450TF. Rosetta has about 26TF now. If they chose to use redundancy they would only have about 13TF. While it is often stated that the project will gain resources by fixing this issue through redundancy, there is no objective evidence that those gains would come close to matching the losses. There is a lot of evidence that redundancy would prevent Rosetta from ever reaching the project computing goal.

In standing by the decision not to use redundancy, Rosetta is helping to elevate these issues with the BOINC developers and bring a real solution for the credit problem to the user community for ALL the projects.

But Rosetta has not simply ignored the issue. The project has taken action in extreme cases, to adjust or remove credit claims that have clearly been fraudulent. But this can only work for very extreme cases. The idea that the project will routinely run some process that adjusts credit claims to some as yet unidentified "Standard" is as unworkable as it is unfair. There is no standard that all users would accept, and the nature of the Rosetta work is such that it does not lend itself to fixed credit values for Work Units, models and steps.

The optimized clients that some users feel are at the bottom of this issue are widely available to ALL users. While the project does NOT recommend or support the use of such clients, many users do install and run them. Some people feel this constitutes cheating. To my knowledge Rosetta has taken no position on that view. The choice to use or not use these clients rests solely with the user. There is nothing Rosetta can do about that. Once BOINC released the code to the public the problem was inevitable. If you have a problem with optimized clients or credit calculation, then BOINC is the place to have it repaired. Redundancy is not a viable answer.

But Rosetta has gone still further in valuing and recognizing the contributions of time from every user. They provide credit awards even if the Work Unit has an error or fails. Why? Because you donated your time to the project and that deserves recognition. The credits are awarded for this contribution of time. At Rosetta even the errors are important and they want to reward people for finding them. No other project recognizes the users contribution of time this highly. When errors occur on other projects they are rarely if ever granted credit. At Rosetta people on slower systems or systems that never even finished a model are recognized for their contributions through credit awards.

One final point. If you read through the forums you will see that the single user with a slow system that runs part time for Rosetta, is treated with the same respect, dignity, and attention and provided the same support as the users with 50 quad CPU systems working 24/7 for Rosetta. This is evidence that Rosetta values every CPU cycle you donate, and every user, no matter who you are or the level of donation you are able to provide to the project. So you should not be put off by people outside the project who you feel may be dismissing your efforts. All of you are welcome at Rosetta, and Dr. Baker wants to find ways to recognize your help on as many levels as he can.

---

I turned off my 5computers when I went on vacation. When I return today, I can not upload work. Need work units to run computers.
joseps

... I volunteer my computer not because of competion or credits. I thought I just want to help with researchers need for the ordinary computer users willing to donate their machine time ...

I half agree - for me the main reason is wanting to help.

It is like in slow traffic, sometimes I feel helpful and let another driver come out of a side road, just to be helpful.

But I find it makes a big difference to me if the other driver says thanks, or waves to me, or smiles, or at least something. Just so I know my help was noticed.

And for me the credits and the honour list on the front page are Rosetta's way of saying thanks. So not the reason I came, but both of these help me to feel good about staying.

And other people have said in previous discussions that the points mod9 made about credit for failed WU have in my opinion been a reason for them sticking with Rosetta througout times of over-adventurous software development and the resulting irritations of high error rates.

I almost detach my computers from the project.

oh, pleeeeeease don't. If you don't enjoy the arguments, just don't read the threads where the arguments happen. The credits are there for those who want to look at them. including the cred-heads in the top three teams who would not be doing half so much if there was no way of keeping score.

The arguments are there for those who care about how the credits work. For those who care, having a place where their suggestions are heard can in itself be part of the positive experience. Don't let them put you off, if they don't help you then just leave them to those who do worry about the issue.

Stick with your last line, it really is an excellent reason to stay:

I just have fun donating some computer time.

:-)

---

Here's something to consider on the competition of credits:

Credits to DC projects are like points in a video game. Sure, you can't spend either one and have no real value.

But, take the point systems out of video games and see what happens to the market. The same could be said about DC projects. Here on Rosetta were not only in a sort of points race, but the race itself lends itself to something far more valuable than a victory in a video game.

It would be interesting if a study group were allowed by a given arcade to go in a yank out and disable all scoreboards from all machines for one day and see how customers respond.

---