Official Stardust@Home Issue & Status List UPDATED 10:57
Moderators: Stardust@home Team, DustMods
-
- Posts: 1
- Joined: Sun Jun 04, 2006 9:31 am
Clicking on calib. movies with tracks counted as incorrect
As someone else noted, while running VM, when clicking on a calibration movie that have an obviuos track on it, the event is counted as an incorrect answer.
This happened to me many times since I noticed that the calibration movies we are provided during VM sessions, behave the same way the tests movies with tracks on them do. It seems IT IS NECESSARY TO CLICK RIGHT ON THE TRACK OF A CALIBRATION MOVIE TO HAVE IT COUNTED AS A CORRECT ANSWER ON VM SESSIONS TOO.
Those who are having this problem could confirm that by clicking right on the track get the problem solved?
See you.
This happened to me many times since I noticed that the calibration movies we are provided during VM sessions, behave the same way the tests movies with tracks on them do. It seems IT IS NECESSARY TO CLICK RIGHT ON THE TRACK OF A CALIBRATION MOVIE TO HAVE IT COUNTED AS A CORRECT ANSWER ON VM SESSIONS TOO.
Those who are having this problem could confirm that by clicking right on the track get the problem solved?
See you.
saphojuice,
I am clicking exactly on the tracks and still being counted incorrectly in some of the calibration movies.
The problem has nothing to do with some folks not picking up on the instructions properly. At least one of my failures below is of the error type they are referring to.
Regards,
-Shar
Calibration Movies Answered Correctly 488
Calibration Movies Answered Incorrectly 5
Your Overall Score: 483
Total Real Movies Viewed: 1072
Specificity: 100%
Sensitivity: 98%
I am clicking exactly on the tracks and still being counted incorrectly in some of the calibration movies.
The problem has nothing to do with some folks not picking up on the instructions properly. At least one of my failures below is of the error type they are referring to.
Regards,
-Shar
Calibration Movies Answered Correctly 488
Calibration Movies Answered Incorrectly 5
Your Overall Score: 483
Total Real Movies Viewed: 1072
Specificity: 100%
Sensitivity: 98%
Rank weirdnesses
I'm no rank fanatic at all, so this is no big issue to me, but as live goes some people are, SO:
I noticed that my rank stays fixed at exactly 500 for ages. I would understand if it doesn't get better because I do the movie-looking only for 1-3 hours per day. But it also doesn't get worse even when I have been away from my comp for the whole day. I assume the S@H community won't pause their activities for me, so this seems a little strange
.
I noticed that my rank stays fixed at exactly 500 for ages. I would understand if it doesn't get better because I do the movie-looking only for 1-3 hours per day. But it also doesn't get worse even when I have been away from my comp for the whole day. I assume the S@H community won't pause their activities for me, so this seems a little strange

forget about it
Issue "resolved": my rank just suddenly jumped from 500 to 868. That can be a little frustrating, even if I don't care much about it
.

REPLY TO ALL ABOVE LATEST POSTS,,,WHERE ARE THE MOD'S HELP H
Win 98se on 2 computers seem to have no problems, a couple Java errors...
My totals keep changing, except the rank is stuck now for 2 days???
My statistic page is not updating my movies since at least 15 hours ago, just stuck in the same number, accept the total people does change every few hours???
I have viewed over 1000 movies, I know, but it does not show in any statistics??? There are some problems here and I feel I have lost some of my time searching for nothing???
Let alone if I found a track and it is now lost???
They got to get this thing down path a bit better!!!
Otherwise there is going to be a lot of lost info and wasted time!!!
Howie
My totals keep changing, except the rank is stuck now for 2 days???
My statistic page is not updating my movies since at least 15 hours ago, just stuck in the same number, accept the total people does change every few hours???
I have viewed over 1000 movies, I know, but it does not show in any statistics??? There are some problems here and I feel I have lost some of my time searching for nothing???
Let alone if I found a track and it is now lost???
They got to get this thing down path a bit better!!!
Otherwise there is going to be a lot of lost info and wasted time!!!
Howie
-
- Posts: 9
- Joined: Fri Aug 04, 2006 1:05 am
If you check under 'my events' you should be able to see that the important information is being recorded, and that it updates immediately. You can even see how many other people agree with you...
Sharqua - There were some posts suggesting that with laptops, the cursor position didn't always get reported accurately - not sure if this is a real bug or not.
Sean
Sharqua - There were some posts suggesting that with laptops, the cursor position didn't always get reported accurately - not sure if this is a real bug or not.
Sean
-
- Posts: 19
- Joined: Sun Aug 06, 2006 1:24 am
- Location: Belgium
What about the 3 spots you see on every movie? Those could be seen as a type of bug, I can image new people would mark them as a track.
http://xs104.xs.to/xs104/06310/stardust2.jpg
http://xs104.xs.to/xs104/06310/stardust2.jpg
Last edited by Betelgeuze on Sun Aug 06, 2006 1:58 am, edited 1 time in total.
-
- DustMod
- Posts: 694
- Joined: Wed May 17, 2006 8:12 pm
- Location: Horsetown, USA
I think this is the wright place to post a problem which I have encountered. I think it fits in category 2:
In a real movie I saw a dust track on the right side of the image. The ID of the movie is 29320. When I encountered it i looked in my events and saw that 3 people before me also saw the track. However, when at a later time I returned to my events I saw that 56 people looked at the movie but only 6 of them identified the track. When I pressed on the link to see the movie again I got a movie where there was a large "thing"' on the right of the movie! This was definitely not the movie which I marked as having a track!!!!!!
I think this is the URL: http://s3.amazonaws.com/stardustathome. ... 51-001.jpg
Is this because two movies might have the same ID? I must say that this is bad because then a possible track can be overlooked in this way because one movie will get positive hits and the other negative..
In a real movie I saw a dust track on the right side of the image. The ID of the movie is 29320. When I encountered it i looked in my events and saw that 3 people before me also saw the track. However, when at a later time I returned to my events I saw that 56 people looked at the movie but only 6 of them identified the track. When I pressed on the link to see the movie again I got a movie where there was a large "thing"' on the right of the movie! This was definitely not the movie which I marked as having a track!!!!!!

I think this is the URL: http://s3.amazonaws.com/stardustathome. ... 51-001.jpg
Is this because two movies might have the same ID? I must say that this is bad because then a possible track can be overlooked in this way because one movie will get positive hits and the other negative..
"I don't like it and I'm sorry I had anything to do with it" Schrödinger speaking of quantum mechanics
-
- Posts: 2
- Joined: Sun Aug 06, 2006 7:56 am
- Location: Tullera NSW Australia
Focus on VM
I have been finding that the focus on the surface is 3/4 of the way down on some movies. This does not allow much space to check for tracks. Numbers 3274130V1 8273757V1 4894641V1 8829375V1 682438V1 so far.
Also, some of the movies have so much goop on them, there is a fair chance of tracks being missed.
Also, some of the movies have so much goop on them, there is a fair chance of tracks being missed.
I have had the same movie twice:
http://stardustathome.ssl.berkeley.edu/ ... e_id=16408
Which I have marked as a possible track, and it appears twice in my list of data viewed. Surely movies should not reviewed more than once by the same person?
Gavin
http://stardustathome.ssl.berkeley.edu/ ... e_id=16408
Which I have marked as a possible track, and it appears twice in my list of data viewed. Surely movies should not reviewed more than once by the same person?
Gavin
-
- Posts: 153
- Joined: Wed Aug 02, 2006 8:17 pm
Two lines of thought to that. Keeping in mind... Validity: is it doing the right thing. Reliability: is it doing it well. Keeping in mind that reliability is a necessary but not sufficient criteria for validity. You can be measuring something consistently, but consistently wrong.
Reminds me of my high school music teacher who often said: Practise does not make perfect, perfect practise makes perfect.
If you practise something wrong, you don't get any better at it. You get better at doing it wrong. That I think is a dangerous issue here with the (understable) lack of feedback. Many potentially confounding habits are being reinforced here.
By checking you against your own selections (my issue is that pads the stats, but that actually works out better for REALLY good searchers, as they are more likely to find it again) it helps see if you agree with yourself. If you are shown the same movie ten times and only saw the track once, that tells us something.
Reminds me of my high school music teacher who often said: Practise does not make perfect, perfect practise makes perfect.
If you practise something wrong, you don't get any better at it. You get better at doing it wrong. That I think is a dangerous issue here with the (understable) lack of feedback. Many potentially confounding habits are being reinforced here.
By checking you against your own selections (my issue is that pads the stats, but that actually works out better for REALLY good searchers, as they are more likely to find it again) it helps see if you agree with yourself. If you are shown the same movie ten times and only saw the track once, that tells us something.
Je ne peux pas regarder la Mer sans me demander qui vit au-delà de cela.
I agree with all you your points Orion, my concern is that this may not be the way that the Stardust project was designed to work and so should be flagged as a possible bug. If its intended to give feed back to the project on how consistent a participant is being by showing them the same movie multiple times, it would also be useful to communicate this to the participant by some kind of consistency score. I do suspect it is a bug.
Gavin
Gavin
Maybe it`s a bug. But what should be the problem? "Nice to see you again, little grain." And if I was not sure on my first decision, I correct it and don`t click again. Maybe it´s a feature in the way Orion_0169 explained (makes sense, training the volunteers).pointyhat wrote:I agree with all you your points Orion, my concern is that this may not be the way that the Stardust project was designed to work and so should be flagged as a possible bug. If its intended to give feed back to the project on how consistent a participant is being by showing them the same movie multiple times, it would also be useful to communicate this to the participant by some kind of consistency score. I do suspect it is a bug.
Gavin
As I noticed now, my events show 4 of those "doubles" (so I have to correct my number of potential candidates down to 126). Maybe (my tip) it happens randomly. I think any movie picked out will be ranked according to its individual number of agreements and maybe according to the score/stats of its finder. So some movies (hopefully the best ones) are mixed in more often (wich makes sense too). The effect for all of us: we will see more and more "concentrated" movies, worth to have a closer look on. Nice.
peter (stardust1)
Last edited by stardust1 on Mon Aug 07, 2006 9:29 am, edited 1 time in total.
Wir leben in einer Zeit vollkommener Mittel und verworrener Ziele. (Albert Einstein)