BWM and Cherry picking

This has been posted because the original thread at EVGA has been edited. It gives their justification and method for "cherry picking" -bigadv units. Read and discuss at your pleasure.....

Personally i think they should grow up and fold what everybody else folds...

"Re:Team EVGA is minus one mega-producer... 7 hrs. ago (permalink)
Hello everyone. This is Connie. I returned to clarify a couple of points as to why my husband and I left this proud team. It was over the so-called "cherry picking" problem. We understand the problem and do not condone it ourselves either. We do chose what work units we will run, and we do it for the reasons which were explained in some detail, although I suspect those among other posts have been deleted. The point is, we have ALWAYS done it in a way that is NOT in any way "damaging" to the program. Damage being defined as "dumping a work unit received to run." We have NEVER done that, yet two idiots here never allowed us the time to explain. Nor will we now because of the way we were treated. Moderators took no interest in it either, and that sealed it.

We run everything under Solaris Unix fine tuned for ONLY bigadv work units and GPU, by tailoring the unix kernal itself specifically for those tasks, and none other. We run all of the clients within a sophisticated script that chooses which work units to run based on the work unit server we are assigned to. This is well before the work unit download even begins, and thus is is never actually assigned. All it does is look at the IP of the work unit server. If it isn't the one we want, the script terminates the client, does a little cleanup, and retries.. That loop is repeated until the right IP appears, then it allows the downloading of the work unit. At that point, it has been officially assigned to us and we run it. It is literally that simple. Not downloading a work unit is NO DIFFERENT than no one showing up to ask for one. Is it?

Everyone jumped on our backsides and condemmed us for it instantly, without giving us a chance to explain ANYTHING. Shame on those who did so. Does anyone really think we would deliberately hurt the program by fouling up the sceduling of work units issuance and expected return dates, and throw that asunder by stupid uneducated cherry picking? How dare they. Typical uneducated prejudgemental responses is all we received, and THAT is why we left in disgust. The individuals responsible know who they are.

No more needs to be said except that obviously more than a few peeps have googled around, likely noting that we have had similar bad luck at other places we've tried to call home, all "enthusiast" sites, all with the same basic results. We either left in disgust, or got angry and read someone the riot act and were banned for it. Believe me, there is no fun in the sites on the strictly professional side of the net, into which the general public is not invited.

Makes no difference to these two old PhD's who only want to fold and have some fun at the same time. Maybe even help others along the way. Our experience here was the last such attempt we have made, nor will we try anywhere else on this side of the fence ever again. Is it lonely going it alone? Most definitely. But we will not tolerate being treated as we were here, the same as every other "enthusiast" site we ever visited, not respected for what we are, or how we do what we do, so we left. That, ladies and gentleman, is the truth of the matter and the complete story.

Now let us all put it to rest. To those who wished us well in our own team effort, thank you most sincerely. To those very few, who know who they are, who caused the mess, drop dead. We remain gone until they are. Should they be done away with, we may possibly return, but we aren't stupid enough to expect that. Either way, we will reach 100 million my roughly mid May next year. We would rather be here doing it with you, but it will never be as long as things remain as they still are, and two certain individuals remain. Goodbye from Brad and I once again. God bless all of you, well, minus two.

Connie"

Humm mixed feelings about this - while I suppose it's nice to use only units that your rig is optimally efficient with, it doesn't seem to be in the spirit of things...

At first i was like, its not that bad because they are not dumping units, but then i realized that essentially you could cut out 67xx's. If everyone did this lots and lots of WUs would never get completed and that pisses me off.

nitrobass24 said:

At first i was like, its not that bad because they are not dumping units, but then i realized that essentially you could cut out 67xx's. If everyone did this lots and lots of WUs would never get completed and that pisses me off.

Click to expand...

Their farm is a pile of i7 machines, 8 i think - how many 67xx units could that chew through a day?? In the meantime everyone else is left to do the stuff that they can't be bpthered to do. its just plain wrong.

nitrobass24 said:

At first i was like, its not that bad because they are not dumping units, but then i realized that essentially you could cut out 67xx's

Click to expand...

They have said before that they refuse to crunch any 67xx WUs so I am sure they have their scripts set to avoid them.

Tobit said:

They have said before that they refuse to crunch any 67xx WUs so I am sure they have their scripts set to avoid them.

Click to expand...

Is it worth raising this with PG or not?? Its certainly not in the spirit of the program

PG will not do anything.

Tobit said:

PG will not do anything.

Click to expand...

That sux... I'm surprised as it pisses off a fair few people

I like how they act like this is an unheard of strategy and its some new idea.

Lots of us have the know-how and capability to pull this off, but no matter how much inefficient WUs suck, you start gaming the system and hurting the whole project.

While I think you should run what is assigned to you, this also hammers home one important issue: Get BOTH sides of a story before you start the BS slinging. This is not isolated to the EVGA forums, not by far. Far too many just love to pile on without all the facts. Always get the facts from both sides then make an informed decision.

What I find interesting is if they can, (allegedly, I don't know), program which WU's they get and this is within the rules, what is the problem? Dumping a WU I can see but if they aren't even assigned it yet, what is the problem? If you don't like it, don't do it.

The real problem is that -bigadv gives too many points. If the points rewarded was proportional to processor cycles consumed, we wouldn't have this problem. I'd be ok with someone sharing this script and everyone running it until Stanford notices and corrects the problem. As it is now, this act is encouraged i.e. they want you to pick out these work units by labeling them as disproportionately more important.

KWH said:

While I think you should run what is assigned to you, this also hammers home one important issue: Get BOTH sides of a story before you start the BS slinging. This is not isolated to the EVGA forums, not by far. Far too many just love to pile on without all the facts. Always get the facts from both sides then make an informed decision.

What I find interesting is if they can, (allegedly, I don't know), program which WU's they get and this is within the rules, what is the problem? Dumping a WU I can see but if they aren't even assigned it yet, what is the problem? If you don't like it, don't do it.

Click to expand...

They know what servers give what WUs, so all they do is basically block the IP of the servers that give out bad WUs.

Of course tho, the script has to reinstall the folding client to appear as a new client and try to get a different WU server assigned.

Its cheating. The server wanted to get that unit done but you either delay that roll out or put it on someone else.

My SR-2 rigs are far more powerful than any rig they have, yet I fold the 670X WU....

Why? Because they need done.

They have a bad case of "we are better than you."

I wish I lived in their la la land where a OC-ed i7 was "teh best rig evar!!!11!"

to lazy to read the rest of the thread.. just the cherry picking was enough for me to think that lady and her husband are a pos. i dont even want to know what their justification is for cherry picking WU's. go run boinc if you want to choose the WU's you run. you dont belong in F@H. i honestly hope vijay gets wind of this and has their ip's banned on the F@H servers. i dont care how many systems they run, how much science they do. they do not belong on F@H.

and im glad team EVGA got rid of these 2 idiots. hopefully this gets posted on all the major team forums in case they try to show up there.

Kendrak said:

My SR-2 is far more powerful than any rig they have, yet I fold the 670X WU....

Why? Because they need done.

They have a bad case of "we are better than you."

Click to expand...

While that is your opinion and you are certainly entitled to it but they are in it for the PPD as most if not all are. You can claim it's the science all you want to but if the points go away, how many will remain? Precious few I would wager.
Yes ALL the WU's need to be done but the system is broken and in need of repair. Instead of piling on the PPD lovers, why not pile on PG until this is reevaluated and the points system is fixed?

if their machine is optimized for big adv, wouldnt it be highly optimized to run lowly SMP units ?

What Brad and Connie fail to understand is you are actually assigned the WU before it is downloaded. The very first server the client connects to is the ASSIGNMENT server. It then sends you off to the work server for the download. So they are essentially dumping the WU even if they don't download it.

Whilst most would agree the points system is not perfect, we are also aware that there are far more donors to the project than there are people actually working at improving it. PG know there is a problem, when they can get around to fixing it is something else.

Tobit said:

What Brad and Connie fail to understand is you are actually assigned the WU before it is downloaded. The very first server the client connects to is the ASSIGNMENT server. It then sends you off to the work server for the download. So they are essentially dumping the WU even if they don't download it.

Click to expand...

See, this is the kind of info we need to make a good decision, not conjecture and feelings. I think too many of us just don't know how the whole thing works to make a proper decision.

The way I see it, ie my opinion, is this:
1- you do it for the science. Then who cares about the points.
2 - you care about the points as they are part of the fun of friendly competition to stimulate points. Then you play by the rules of PG and not the rules made up by the community at large.

/opinion

SpeedyVV said:

The way I see it, ie my opinion, is this:
1- you do it for the science. Then who cares about the points.
2 - you care about the points as they are part of the fun of friendly competition to stimulate points. Then you play by the rules of PG and not the rules made up by the community at large.

/opinion

Click to expand...

Basically how I see things. "you play by the rules of PG"

Tobit said:

What Brad and Connie fail to understand is you are actually assigned the WU before it is downloaded. The very first server the client connects to is the ASSIGNMENT server. It then sends you off to the work server for the download. So they are essentially dumping the WU even if they don't download it.

Click to expand...

Therefore PG should do something about it, do they not realise that it is hurting and slowing down the project? That WU that gets dumped by one of their OC i7's could end up running on an athlon x4 taking god knows how much longer to complete

KWH said:

While that is your opinion and you are certainly entitled to it but they are in it for the PPD as most if not all are. You can claim it's the science all you want to but if the points go away, how many will remain? Precious few I would wager.
Yes ALL the WU's need to be done but the system is broken and in need of repair. Instead of piling on the PPD lovers, why not pile on PG until this is reevaluated and the points system is fixed?

Click to expand...

tobit just explained below exactly why cherry picking is bad and why blocking ip's does absolutely nothing. they are harming the science and everyone else involved. its not benefiting them because its still effecting their 80% completion rate. so even the PPD makes absolutely no difference.

i know your trying to compare this to what we were doing running GPU3 fermi work units on g92 cards. but theres a huge difference between trying to run science on cards that can do the science and get extra points compared to cherry picking WU's that effects everyone else and the science involved in it.

Tobit said:

What Brad and Connie fail to understand is you are actually assigned the WU before it is downloaded. The very first server the client connects to is the ASSIGNMENT server. It then sends you off to the work server for the download. So they are essentially dumping the WU even if they don't download it.

Click to expand...

Nathan_P said:

Therefore PG should do something about it, do they not realise that it is hurting and slowing down the project? That WU that gets dumped by one of their OC i7's could end up running on an athlon x4 taking god knows how much longer to complete

Click to expand...

I totally agree.

sirmonkey1985 said:

tobit just explained below exactly why cherry picking is bad and why blocking ip's does absolutely nothing. they are harming the science and everyone else involved. its not benefiting them because its still effecting their 80% completion rate. so even the PPD makes absolutely no difference.

Click to expand...

I wonder if their PPD would be higher if they just let the clients get on with it, there must be times when they have clients sitting idle for hours....

gee thanks for letting me fold the WU's you don't want. Assholes.

sirmonkey1985 said:

tobit just explained below exactly why cherry picking is bad and why blocking ip's does absolutely nothing. they are harming the science and everyone else involved. its not benefiting them because its still effecting their 80% completion rate. so even the PPD makes absolutely no difference.






i know your trying to compare this to what we were doing running GPU3 fermi work units on g92 cards. but theres a huge difference between trying to run science on cards that can do the science and get extra points compared to cherry picking WU's that effects everyone else and the science involved in it.

Click to expand...

You are partially correct. What I have been saying is things need to be addressed by PG. If things are not black and white, right or wrong, these things will persist.
You obviously missed where I acknowledged Tobit's reply.

They have raised as many questions as they have claimed to 'explain' with this response. It is inadequate to the degree it is condemnatory. I welcome them to debate this here. The mods won't censure them providing they maintain a modicum of respect for both the individuals here as well as the forum's TOS. They claim to be older, well they are in good company here, so let's all conduct ourselves like adults and discuss this civilly if they so desire.

Nathan_P said:

Their farm is a pile of i7 machines, 8 i think - how many 67xx units could that chew through a day?? In the meantime everyone else is left to do the stuff that they can't be bpthered to do. its just plain wrong.

Click to expand...

PG have stated on numerous occasions in the past that speed trumps volume. In such a paradigm, their i7 farm (likely composed of quads but not certain) classify as marginal systems, thus not meeting project organizers' preferent hardware specs to run -bigadv. By selecting WUs, they continuously receive more productive units to the detriment of everyone else, some of whom own far better and far more suited hardware, at the 'mercy' of the server.

nitrobass24 said:

At first i was like, its not that bad because they are not dumping units, but then i realized that essentially you could cut out 67xx's. If everyone did this lots and lots of WUs would never get completed and that pisses me off.

Click to expand...

Yes, witness what is an ongoing issue with uniprocessor clients and the backlog of WUs due to an an ever increasing emphasis in client production that has shifted heavily towards nVidia GPU and SMP. The new bonus system in place for one year now, greatly exacerbates the problem.

Tobit said:

They have said before that they refuse to crunch any 67xx WUs so I am sure they have their scripts set to avoid them.

Click to expand...

I am sure there is much that has not been stated that is occurring to no one's knowledge. If they are capable (in skill as well as intent) to do one thing in their selection script, they most assuredly can do more...

Tobit said:

PG will not do anything.

Click to expand...

Why am I not surprised? In truth and defense of PG, maybe it isn't that simple to rectify the situation. Perhaps their script is too 'slippery' for Sanford's servers to do anything in defense of it...?

KWH said:

While I think you should run what is assigned to you, this also hammers home one important issue: Get BOTH sides of a story before you start the BS slinging. This is not isolated to the EVGA forums, not by far. Far too many just love to pile on without all the facts. Always get the facts from both sides then make an informed decision.

Click to expand...

They were given ample opportunity in several threads. How long can it take to post a few sentences that they were employing a specialized selection script? They claimed they were deleting WUs; that's why they were flamed and rightly so.

Dumping a WU I can see but if they aren't even assigned it yet, what is the problem? If you don't like it, don't do it.

Click to expand...

The problem, as is too often the case with this project both on the organizational side and participatory side of things the way I see it, is that precedents are established that open the gates for all sorts of nasty things come what may. That isn't even venturing on the ethical side of the debate which is another can of worms.

Nathan_P said:

I wonder if their PPD would be higher if they just let the clients get on with it, there must be times when they have clients sitting idle for hours....

Click to expand...

it probably would be since since the WS would just be sending WU after WU waiting for the connection to time out until it finally assigns a WU that their script allows to download. so it could be 10 seconds or it could be an entire day til they get a WU. this is what happens when you have 2 idiots that try to act smarter then their own good and dont realize how F@H works. so screw em.

Another thing to consider is the amount of extra load they are putting on PG's servers. If the assignment server hands out 100 P6701s before it gives them a -bigadv, that means a lot of time and resources are wasted. If everyone did this kind of thing then it's highly unlikely the servers would be able to keep up with the load, which would result in more crashes and downtime, which in turn hurts the project significantly since work does not get done.

KWH said:

While that is your opinion and you are certainly entitled to it but they are in it for the PPD as most if not all are. You can claim it's the science all you want to but if the points go away, how many will remain? Precious few I would wager.
Yes ALL the WU's need to be done but the system is broken and in need of repair. Instead of piling on the PPD lovers, why not pile on PG until this is reevaluated and the points system is fixed?

Click to expand...

Regardless of how broken the points system is, that doesn't make their actions justifiable.

Zero82z said:

Regardless of how broken the points system is, that doesn't make their actions justifiable.

Click to expand...

I'd rather that they selectively donate than not donate at all, and they are still donating hundreds in electricity and thousands in hardware to the cause. I find the outrage a little humerous, since we are basically a philanthropy group nit picking each other.

Zero82z said:

Regardless of how broken the points system is, that doesn't make their actions justifiable.

Click to expand...

Obviously to them, it does. If it does not violate the PG rules, then they can/do justify it.
Personally, I would run what is assigned to me but that is me.

Nightbird said:

I'd rather that they selectively donate than not donate at all, and they are still donating hundreds in electricity and thousands in hardware to the cause. I find the outrage a little humerous, since we are basically a philanthropy group nit picking each other.

Click to expand...

True

Nightbird said:

The real problem is that -bigadv gives too many points. If the points rewarded was proportional to processor cycles consumed, we wouldn't have this problem.

Click to expand...

The solution is rather simple but doubt PG will change it any time soon. What essentially needs to be done is make the bonus system less lucrative to prevent exaggeration of highest performance hardware. Yes, it will impact -bigadv research but at least some of the absurd imbalance we have seen since A3 was released won't be plaguing the project. If things remain the way they are, it will become untenable once the next generation architecture is available. People will just see little reason to run even the last generation hardware, which is the current crop of processors. It is unsustainable as an exponential scale. There is a way out of this spiral, actually more than one possible solution.

I'd be ok with someone sharing this script and everyone running it until Stanford notices and corrects the problem.

Click to expand...

Who knows, maybe others are already running it...

I think I'm going to be echoing everyone's sentiments that the last thing the project needs is for this script's dissemination. That would be disastrous and prove nothing other than the folding community's lack of a sense of honor. If PG cannot improve the way things are maintained at their end, there probably exists a good reason. Where I think action should be made is towards the culprits, in like manner with the 'team' that dispersed the folding client unawares to computers across the Internet, or those who folded on work computers without permission, etc.

As it is now, this act is encouraged i.e. they want you to pick out these work units by labeling them as disproportionately more important.

Click to expand...

PG fell into this quagmire for several reasons, not least of which is the bonus sytem itself. They needn't have introduced it; we did fine for nearly a decade without it. They hinted it was coming years ago and Evil even alluded about it to which if memory serves, I replied it would be a bad move of PG's part becaue it would further tip the balance in favor of those who are more financially endowed relegating the rest of us to even lower ranks (won't use a more contentions word). For anyone who sharply disagrees with my opinion and argues that participants with greater resources always had an edge in DC, I agree, but why grant further rewards to the privileged few, and if it is necessary, why to this level?? As I have stated elsewhere, there is always some bad that comes with a good thing, however indiscernible it might be at first. It is the price for its admission.

APOLLO said:

The solution is rather simple but doubt PG will change it any time soon. What essentially needs to be done is make the bonus system less lucrative to prevent exaggeration of highest performance hardware. Yes, it will impact -bigadv research but at least some of the absurd imbalance we have seen since A3 was released won't be plaguing the project. If things remain the way they are, it will become untenable once the next generation architecture is available. People will just see little reason to run even the last generation hardware, which is the current crop of processors. It is unsustainable as an exponential scale. There is a way out of this spiral, actually more than one possible solution.

Who knows, maybe others are already running it...

I think I'm going to be echoing everyone's sentiments that the last thing the project needs is for this script's dissemination. That would be disastrous and prove nothing other than the folding community's lack of a sense of honor. If PG cannot improve the way things are maintained at their end, there probably exists a good reason. Where I think action should be made is towards the culprits, in like manner with the 'team' that dispersed the folding client unawares to computers across the Internet, or those who folded on work computers without permission, etc.

PG fell into this quagmire for several reasons, not least of which is the bonus sytem itself. They needn't have introduced it; we did fine for nearly a decade without it. They hinted it was coming years ago and Evil even alluded about it to which if memory serves, I replied it would be a bad move of PG's part becaue it would further tip the balance in favor of those who are more financially endowed relegating the rest of us to even lower ranks (won't use a more contentions word). For anyone who sharply disagrees with my opinion and argues that participants with greater resources always had an edge in DC, I agree, but why grant further rewards to the privileged few, and if it is necessary, why to this level?? As I have stated elsewhere, there is always some bad that comes with a good thing, however indiscernible it might be at first. It is the price for its admission.

Click to expand...

Well said.

Zero82z said:

Another thing to consider is the amount of extra load they are putting on PG's servers. If the assignment server hands out 100 P6701s before it gives them a -bigadv, that means a lot of time and resources are wasted. If everyone did this kind of thing then it's highly unlikely the servers would be able to keep up with the load, which would result in more crashes and downtime, which in turn hurts the project significantly since work does not get done.

Regardless of how broken the points system is, that doesn't make their actions justifiable.

Click to expand...

Sorry, my feeling is that if the PG allows this to happen, then good for them for cherry picking.

If the PG did not want this, well then penalize them. Use a ID password, whatever. each unfinished WU reduces your points or something like that. It does not matter.

Unenforceble rules just creates bad feelings. As we can see, it has.

My preference would be to have results based rules, like you get bonus points if you finish in 4 days or less. But the main thing is that the PG can enforce that.

Hell if they say only People from the Kingdom of Tonga can do bigadvs, but do not check, well then too bad, I am doing them.

If they dont like cherry picking well, then set up the rules and the systems to penalize it.

The 'benchmark machine is an i5' explanation doesn't make sense to me because i5s take a nosedive in ppd on slower units like 6701. There must be something wrong with their benchmarking system IMO. If you look at the Fermi WUs - they benchmark on a GTX480 and all WUs perform pretty much exactly the same (within +/-200ppd) on both GF100 cards - GTX470 & GTX480. That could not be further from the truth for their SMP units.

Sorry, my feeling is that if the PG allows this to happen, then good for them for cherry picking.
If the PG did not want this, well then penalize them. Use a ID password, whatever. each unfinished WU reduces your points or something like that. It does not matter.

Click to expand...

Does not make sense IMO. There are MMOs that devote multi-million dollar budgets to TOS enforcement, and people always find a way to circumvent the rules anyway. There will always be a way to cheat.

"Sorry, my feeling is that if the PG allows this to happen, then good for them for cherry picking.

If the PG did not want this, well then penalize them. Use a ID password, whatever. each unfinished WU reduces your points or something like that. It does not matter.

Unenforceble rules just creates bad feelings. As we can see, it has."

Absolutely.

Nightbird said:

I'd rather that they selectively donate than not donate at all, and they are still donating hundreds in electricity and thousands in hardware to the cause. I find the outrage

Click to expand...

It has not been established to any acceptable level that there is no harm being committed to the project by their actions. When I see Vijay post that what they are doing is OK in the sense nothing actually detrimental is being done to the project, then I will agree that on the level of the research, it could be better that they fold than not fold. Even so, a statement like this would be negating the ethical ramifications, and for me at least, ethics is more important than the research as long as research is being maintained on an appreciable level - something even you can agree the withdrawal of BWM from the project would not deter.

sirmonkey1985 said:

it probably would be since since the WS would just be sending WU after WU waiting for the connection to time out until it finally assigns a WU that their script allows to download. so it could be 10 seconds or it could be an entire day til they get a WU. this is what happens when you have 2 idiots that try to act smarter then their own good and dont realize how F@H works. so screw em.

Click to expand...

While (downloading(userID,password))
{
...
WUsDownloaded[user]++;
....
}
Points := Credit / WUsDowloaded[userID];


sirmonkey1985, it looks like they DO realize how it works better than you do, and that is how they manage to get the WUs they wanted without being penalized.

Do I think this is correct? No, unless PG finds it OK to leave it that. Then, well fair game.

Geez, comeone guys. We are talking about the University of Standford. I am sure they can get someone that knows enough about programming to implement the rules they want into code.

And to use the rules to promote the behaviour they want. No need to screw anyone.

While some of you feel you have higher moral standards than others, well thats nice, but this is a game. A points game.

The science is real, true, but the points system is a game, nothing more, nothing less. Let's play by the rules. Dont get pissy because someone found a way to make more points within the rules, but outside your moral specifications.

Tall players have better chances at being good in Basketball, is that fair? does not matter, there are no rules against it. So too bad for me, no chance really.

Heavy boxers have bettter chances of creaming a lite weight dude. Well they have different weight classes. Well I still cant box, but that is besides the point.

My point is that if I dont like the rules of the NBA a can start my own league of short people basketball.

OK, I am done, but please no name calling... I think that is againts the rules of the forum

There will always be a way to cheat.

Click to expand...

There is a diff between cheating and finding loopholes around the rules. The Australians won the Americas Cup by comming up with creative designs within the rules. If the organizers dont like it well change them going formward. But you cant disqualify them once the race started and they are within the rules.

Same thing with the Katamaran the NZ used. single haul after that... but not halfway throught the race.