Will the testing/real status bar cheat be fixed?
Moderators: Stardust@home Team, DustMods
Will the testing/real status bar cheat be fixed?
I was just wondering, as many people know, just hovering over one of the buttons will tell you if the movie is real or not (from the target of the link showing up in the status bar), opening up to the system to major abuse.
You could just click no track, no track, no track untill a testing comes up, and then do that one properly.
This obviously is bad science, and manners
So, I was just wondering if a fix for this is in the pipeline.
You could just click no track, no track, no track untill a testing comes up, and then do that one properly.
This obviously is bad science, and manners
So, I was just wondering if a fix for this is in the pipeline.
Re: Will the testing/real status bar cheat be fixed?
Read this post::Etalon wrote:I was just wondering, as many people know, just hovering over one of the buttons will tell you if the movie is real or not (from the target of the link showing up in the status bar), opening up to the system to major abuse.
You could just click no track, no track, no track untill a testing comes up, and then do that one properly.
This obviously is bad science, and manners
So, I was just wondering if a fix for this is in the pipeline.
http://stardustathome.ssl.berkeley.edu/ ... 9&start=15
Its not bad science. It doesn't matter at all. If having a high score is important to you go for it.
No, it DOES matter. The movies are only viewed a certain number of times, by only a limited number of people. If all the people viewing the ONE movie with a track in cheat.. how is that not bad science?
How about if 50% of them cheat? That might not be enough to make nasa look at that movie, even though the other 50% saw the track.
How about if 50% of them cheat? That might not be enough to make nasa look at that movie, even though the other 50% saw the track.
Really: Quote a source please!Etalon wrote:No, it DOES matter. The movies are only viewed a certain number of times, by only a limited number of people.
Who sees what and how often is NOT PUBLISHED.
Anything flagged by one person is likely automatically routed to others for review. If it only gets a few clicks because everybody is just clicking "No Track" it will have a low score.
But so will every other image in that situation.
In the end, the Stardust team will take the 100 highest scores, look at them themselves, and decide if they want to start slicing an dicing areogel to get at them.
Maybe the 100 highest scores are 400 to 500. Maybe only 20 or 30. It doesn't matter. What ever rises to the top is looked at first.
After that, they go after the lower scoring particles. Etc. Etc.
Not clicking a track does not make it disappear. Others will find it, and click it. Conversly, clicking a random place does not create a dust particle.
"Cheaters" don't care how they cheat. They are not methodical enough to sway the results. They just want high points. They can't create a dust particle, nor can they destroy them. If they get to a gazillion points but have very few REAL tracks clicked (less than the average number found per 1000 movies) their entire results can be backed out of the data base to reduce the "noise".
Those with the Earliest+Highest score could JUST AS LIKELY have their total effort backed out of the database. After all, if you review 5 movies in 5 seconds, it doesn't take a rocket scientist to figure you were not paying attention. But, by definition, NASA has a lot of rocket scientists at hand. So let the cheaters try.
But ultimately, the noise in the click data does not matter. The signal will be SO MUCH STRONGER than the noise, that the real tracks will stand out like sore thumbs. All it takes is 10 to 20 percent of the participants to be contentious, and the Random Contributions of the cheats will be buried, washed down the drain like so much sewage.
Who ever designed this search criteria is one smart cookie. You can't scam it. You can't create or destroy a dust particle. It doesn't matter if all you want is a score. The contentious users data will lead the scientists to the dust particles, and the cheats will have no effect at all.
It really is ingenious.
Rocket Scientists!! Ya gotta love em!
Just don't think you can scam them.
Okay, here ya go.
"According to the Stardust @ Home plan, if two out of four volunteers claim to find a dust track, the corresponding image will be sent to 100 more volunteers for verification. Should at least one-fifth of those reviewers affirm the find, the image will be kicked up to a team of Berkeley undergraduates trained to spot aerogel dust tracks."
http://www.msnbc.msn.com/id/10792614/
So, again, if theres ONE movie with a track, and 3 of 4 people scam the system by clicking no track on the real movies.. the dust will not be found.
This real/test thing should be fixed.
"According to the Stardust @ Home plan, if two out of four volunteers claim to find a dust track, the corresponding image will be sent to 100 more volunteers for verification. Should at least one-fifth of those reviewers affirm the find, the image will be kicked up to a team of Berkeley undergraduates trained to spot aerogel dust tracks."
http://www.msnbc.msn.com/id/10792614/
So, again, if theres ONE movie with a track, and 3 of 4 people scam the system by clicking no track on the real movies.. the dust will not be found.
This real/test thing should be fixed.
-
- Stardust@home Team
- Posts: 530
- Joined: Wed Apr 19, 2006 11:28 am
- Location: UC Berkeley Space Sciences Lab
- Contact:
We are actually not using that method anymore and icebike is 100% correct. Cheaters are really only wasting their own time, and our bandwidth. More will be posted on this issue later...Etalon wrote:Okay, here ya go.
"According to the Stardust @ Home plan, if two out of four volunteers claim to find a dust track, the corresponding image will be sent to 100 more volunteers for verification. Should at least one-fifth of those reviewers affirm the find, the image will be kicked up to a team of Berkeley undergraduates trained to spot aerogel dust tracks."
http://www.msnbc.msn.com/id/10792614/
So, again, if theres ONE movie with a track, and 3 of 4 people scam the system by clicking no track on the real movies.. the dust will not be found.
This real/test thing should be fixed.
-Bryan
"I am made from the dust of the stars, and the oceans flow in my veins"
- RUSH
- RUSH
You are reading more into that than was said. Further, I wouldn't believe anything on msnbc till i see it on the stardust site.Etalon wrote:Okay, here ya go.
"According to the Stardust @ Home plan, if two out of four volunteers claim to find a dust track, the corresponding image will be sent to 100 more volunteers for verification. Should at least one-fifth of those reviewers affirm the find, the image will be kicked up to a team of Berkeley undergraduates trained to spot aerogel dust tracks."
http://www.msnbc.msn.com/id/10792614/
So, again, if theres ONE movie with a track, and 3 of 4 people scam the system by clicking no track on the real movies.. the dust will not be found.
This real/test thing should be fixed.
No where does it say that an image will ONLY be sent to 4 volunteers on the first time around. It just says that if half of the INITIAL group find something, it will be sent for more viewings.
Furhter, these rules can be changed on the fly. The click data does not go away, it can be re-analized over and over. They may decide that if they are getting a lot of scammers to re-send any movie with 2 out of 8 clicks back to 100 viewers. Or maybe 2 our of 30.
Click data arrives back at the server with a time stamp. Your click speed could actually work against you. If they find that a particular volunteer is clicking no track 5 times a second they can just dump all his data.
I can pretty much guarantee they aren't being sent to only four. I recently clicked something rather iffy and subtle, to see that I was the first to click it, but 35 others had seen that movie before me! So I might be wrong about that one, but they're definitely being looked at by more than four.icebike wrote:No where does it say that an image will ONLY be sent to 4 volunteers on the first time around. It just says that if half of the INITIAL group find something, it will be sent for more viewings.
Let it never be said that your **** retentive attention to detail never yielded positive results. - Loki, Dogma
i'm also worried that some people may not be paying attention... but icebike is absolutely right. kudos to being so persistent icebike while the servers were running slowly i discovered that you can run the viewer in multiple windows of IE without any apparent negligible effect. thus, it seems possible that some of the REALLY fast accounts have more then one person working on them, just to condense clicks and compile the highest score. scores are nice... science is better.
-Peter
-Peter
"The Earth is the cradle of mankind, but one cannot live in the cradle forever."
~Konstantin Tsiolkovsky
~Konstantin Tsiolkovsky
-
- Posts: 153
- Joined: Wed Aug 02, 2006 8:17 pm
-
- Posts: 12
- Joined: Fri May 19, 2006 10:17 am
Seems to me the easiest fix is to not track the top scores on the website.
If there's no glory in a high score there is no reason to try to fudge your score. If that's why they were here they will leave and those who actually care will stay.
Although I find it hard to believe someone would waste their time just trying to get a good score.
If there's no glory in a high score there is no reason to try to fudge your score. If that's why they were here they will leave and those who actually care will stay.
Although I find it hard to believe someone would waste their time just trying to get a good score.
-
- Posts: 1
- Joined: Wed Aug 09, 2006 8:26 pm
Ya know...
...on this whole cheating thing. As much as this is a noble endeavor, I can think of a hell of a lot better things to cheat on that Stardust at home (all due respect of course).
I actually like the ranking idea, a nice motivation that adds some competitive spirt to what is otherwise not the most intellectually stimulating work. But to cheat at it? You'd have to be a pretty hard core loser to care enough to 'cheat' during your Stardust at home session.
Well it looks like it's fixed.
Cheers!
SCB
I actually like the ranking idea, a nice motivation that adds some competitive spirt to what is otherwise not the most intellectually stimulating work. But to cheat at it? You'd have to be a pretty hard core loser to care enough to 'cheat' during your Stardust at home session.
Well it looks like it's fixed.
Cheers!
SCB