SETI Cheating scandal--Stats meaningless?

[GulDukat]

Check out this thread at ARS:

SETI@Home: New Brand of Cheats

 

[Smoke]

Unbelieveable. I'm very disappointed to find out this info but it is better to know than not.

 

[networkman]

:Q :Q :Q

Cheaters suck! :|

 

[Smoke]

OK, I think we need to make sure our own house is in order. "fragile" posted some guidelines to dope out the cheaters. I have copied his suggestions and here they are:

1) You trust no-one

2) You monitor the stats - you look for incrementally increasing production, 10,10,10, 20, 20,20, 30, 30, 30 etc

3) You look for twins or worse, in sync A and B producing very similar numbers and shaddowing each other.

4) You look for big hitters that never post

5) You look for big hitters who's stories do not add up - Oh yeah I'm crunching on Cray (there is no port).

6) you look out for people with similar numbers and indentical average times.

7) all of the above.....

 

[EvilWobbles]

//rant mode one

This totally sucks! :disgust: :disgust:

I am scraping together every dollar I can to build more crunchers, assimilating everything in sight so I can squeeze a couple of extra WUs a day and these cheaters come along and do this.

I've boosted my usual output by 20+ WUs in the last couple of weeks. I've earned those extra WUs but some people might look at my recent jump in production and assume I am cheating. Guilt by association. That sucks!

I certainly hope we don't have anyone on this TeAm who is using this cheat. I don't want that type of stuff taking place here at AT!!

//rant off

Thanks for bringing this to our attention GulDukat. It will be very interesting to see how this plays out in the future!

 

[mechBgon]

Scums! :|

 

[Smoke]

After reading the thread completely through, it sounds like the formation of a "committee" might be in order. And after looking back at the "tells" posted by fragile I think one thing should be added: "Everyone is considered innocent until proven guilty!"

Does anyone feel we need to form our own "compliance committee"?

 

[Evadman]

That sucks big time. I believe that anyone caught doing this should be immediately dropped from the TeAm & SETI. There is no reason to cheat on a program such as this. I would rather loose honorably than win by cheating.

I don't believe anyone on the TeAm would stoop so low. We seem to have a very high calibur group of people. I cant wait for the news that no cheaters were found in the TeAm.

 

[networkman]

I would rather loose honorably than win by cheating.

Roger that!

 

[Smoke]

Well, I'm not Sherlock Holmes but I've just looked through the "big hitters" on our team and I don't see anything that looks the least bit suspicious.

I had to do a double take on two people though:

33) Hellburner Avg WU Time: 31.83 Total WUs 18674
37) The Def One Avg WU Time: 31.83 Total WUs 14906



LOL We know this isn't a duplication of work. Ha

 

[Robor]

So if I'm reading this correctly this is a confirmed cheat and Tim (CobraICE) is a cheater? I've heard Tim called a cheater before but I thought his results were analyzed by Berkeley and checked out.

Either way, this sucks. I've been doing SETI for 3+ years and climbing the charts up at a steady pace and managed a high of #152 in the world. Since then I've been falling back down despite the fact that I've increased my production by about 25% over that period. Knowing cheats exist I can't help but wonder now.

:| :disgust:

 

[Baldy18]

Ok everyone at Ars thinks Cobra was found guilty of cheating and had his account removed because of that. Cobra says that he was not guilty of cheating and requested his account be removed to prevent furthur harrassment and can get it back at any time. Am I the only one that thinks that now is the time for Cobra to vidicate himself once and for all by asking for his account back? I think it is time that the Cobra issue gets settled by him either getting his account back or not and then we would know why it was removed. Personally I believe Cobra and would love to see the faces of those at Ars that think he was removed for cheating when his account gets restored.

As far as cheaters on this TeAm goes I don't think there are any. I could be wrong but I haven't seen anything that would lead me to believe that there are. Cheaters at S@NL is another thing though, I've suspected that for a long time. I'm sorry to those at S@NL that are decent people and visit this forum I hope that for your sakes there aren't any cheaters on your team but I fear that there are.

 

[Smoke]

I think this is the same stuff we all heard about before. Cobra was never found guilty of anything. He told us why he quit and I believe him. Ars was giving him a hard time and along with other crxx he had received from other teams that didn't like being passed led to his quitting. There are people at Ars that are now saying it is a fact but just because they are saying so doesn't make it true.

Thread on Ars Technica back when Cobra quit doing SETI

 

[toft-dk]

I am horrified to learn this

Getting friends to join our miniTeAm to do better in the competition, and then SOME people does THIS

Cheaters are Very Very poor on morale , and are trying to get credit for something they have not done.

Cheaters suck Bigtime , as they ruin it for everyone taking this Quest seriously.

I Really hope that we do NOT have any with us here at Anandtech, but if we do, we need to find out, so action can be taken.

I TOO believe Tim ( Cobra-ICE ) because he's a great guy, and believe he would NEVER Cheat. And why would he want to with all the machinepower he has at his fingertips .

 

[RaySun2Be]

:|:|:|:|:|:|:|:|:|:|:|

I loathe cheaters. :|

OK, everyone who hates cheaters line up behind me. I've got my Titanium cane in one hand, my Platinum cane in the other. This is going to be a two-fisted *whack*fest on their heads, and everyone gets a turn! This will make that "line up and slap the hysterical woman" scene in Airplane look like "Patty Cake Play Time". :|

I really loathe cheaters. :|


And yes, I remember Cobra saying that Berkeley had check is work 4 times and found it to be ok, and that he only pulled his account because he was fed up with all the accusations.

And I tend to believe him. I've seen pictures of his workplace, and there is definitely enough horsepower there to generate those kinds of numbers.

As stated in the ARS thread, this will hurt the honest crunchers of all teams.

:|:|:|:|:|:|:|:|:|:|:|

 

[CADsortaGUY]

:|:|:|:|:|

Boy O Boy - I was afraid of such a "scandal" and now it has surfaced. In my "surfing" of the stats and such I have noticed some of these "trends" on certain teams but I just kind of ignored them because I assumed that Berkely had a pretty secure process - guess not.

If there are ANY cheats here on the TeAm - PLEASE FIND THE NEAREST EXIT! We don't need cheaters here!

I hope this gets resolved swiftly and justly. I also hope that Berkely will realize the magnitude of a scandal like this and the shockwaves it might send through the whole DC community.

As for the Cobra issue - I'm all for vindication but it is really up to Tim - since he will have to put up with the whole deal again. Here's to hoping that Tim will seek justice!

CADkindaGUY

 

[GulDukat]

I think it is important to re-emphasize what Grandpa Smoke said in his third post in this thread--Everyone is innocent until proven guilty, at least in my mind.

I find two things really sad about all of this:

1) There is a shadow of suspicion that now hangs over the results/statistics of the entire project. Sure, the science probably wasn't compromised, but all of the races, the team and individual rankings, all of those things are now unreliable.

2) I could understand (though not agree with) someone cheating on income taxes or something with a monetary reward, but on a science project like this? I simply do not understand that. Cheating on a project like this is more alien to me than the reason SETI exists in the first place.

For those (like me) who mainly do DC for the stats, this is a sad, sad day...
Dukat

 

[Robor]

Just in case anyone is wondering, I'd like to say that all of my WU's are valid. Yes, I'm close to 100K and I'm nearly at an average of 200 WU's per day but I can assure you they are all a result of time (3+ years) and a large work fleet (about 65 clients), not cheats. I'm only saying this because I'm sure some people are looking around and if you aren't you should be. I know I am. Like Ray, I hate cheaters too. They destroy the fun, competition, and purpose of the project.

:| :| Pi$$ on all the cheaters who ruin it for legit members! :| :|

 

[MoFunk]

Man this just plain sucks. I know that stats can be addicting. That is why I try and assimilate all I can. I want big numbers too, but cheating is down right f-ed up! This just puts a bad taste in my mouth. We all work very hard for every WU we pump out and now we have to deal with some schmuck padding his stats. FOR WHAT?? 8 million WU's does not make you cooler than someone with 100. Plus it's not about who can do the most, it's about the TeAm, the friendships and the cause! Damn it, if you are cheating.....GET A DAMN LIFE!!!!! Your stats mean jack sh*t when you walk outside. No one comes up to you on the street and says "WOW you pump out 6000 WU's a day, man can I touch you?" I need a beer! Maybe I will cut out of work a little early and get me one. :|

 

[Baldy18]

An interesting point has been brought to my attention. SETIQUEUE OWNERS BEWARE! If you operate a team SetiQueue and rsubmit dupes you may be under suspicion. The best thing for SetiQueue owners to do would be only resubmit dupes once and if they don't go through then delete them or not try to send them at all.

 

[MoFunk]

Well now that this has been brought up I have a question that I was wanting to ask and never did and maybe could help with the fraud. Can a WU be tracked be either the name of the computer that crunched it, or the IP address of that computer. I wanted to know because I have anywhere from 4 to 12 computers cruncing for me. Some are part time, soem are full. I was always curious if there was somewhere I could go to see which of the computers where tuning in WU's and how many. If there was a way to track this we could sniff out the cheaters. If someone has 1 PC at say IP address 195.195.195.65 or computer name home-bob and we can see that they are pumping out more than 15 wu's a day then we know we have an issue. Really a high powered computer is only going to pump out 7 a day if your lucky.

Would this even be possible?

 

[Robor]

MoFunk... Just a little correction. My home P4 1.6a @ 2200 does 7-8 WU's per day and a dualie Athlon or Xeon could easily do double that. Good point though. If a ton of results are coming from a single IP it would have to be either a badarse mainframe or a cheater. Just have to wonder how much trouble it would be for Berkeley to do such checking.

As far as the dupes... I really don't understand them. About once a month I copy my dupe and dropped results back into the main SetiQ directory. In fact, that reminds me... Just did it again. I had 12 dupes and 2 dropped since 7/31. Will see if they go.

 

[Baldy18]

Originally posted by: MoFunk
Well now that this has been brought up I have a question that I was wanting to ask and never did and maybe could help with the fraud. Can a WU be tracked be either the name of the computer that crunched it, or the IP address of that computer. I wanted to know because I have anywhere from 4 to 12 computers cruncing for me. Some are part time, soem are full. I was always curious if there was somewhere I could go to see which of the computers where tuning in WU's and how many. If there was a way to track this we could sniff out the cheaters. If someone has 1 PC at say IP address 195.195.195.65 or computer name home-bob and we can see that they are pumping out more than 15 wu's a day then we know we have an issue. Really a high powered computer is only going to pump out 7 a day if your lucky.

Would this even be possible?

I know that SetiQueue allows you to keep track of your clients by IP. I used to use SetiQueue to make sure all my remote computers were still submitting properly.

I'm not sure if they have other methods at Berkely or not.

 

[CADsortaGUY]

If a ton of results are coming from a single IP it would have to be either a badarse mainframe or a cheater.

In theory you'd be correct but people running Queues have many hundreds of results coming from 1 IP(that berkely's servers see anyway)

Now the other question is that of Baldy18's point about resubmitting duplicate WUs. I have done this in the past and most end up being rejected again but I wonder what Berkely's stance on this practice is? I don't think they'd endorse it perse but I'm not sure they will/should ban that practice either. Well - until this cheat thing is resolved I will NOT be resubmitting duplicate WUs since my Queue is a "Public" Queue and I don't want to take the chance of having one of my Queue users accused of cheating because of my actions.

CADkindaGUY

 

[Baldy18]

And what if SetiQueue sends a result in but still puts a copy back in the dupes folder? I've seen SetiQueue do stranger things.