Sugestion to the calibration movies

[Anders Lund]

Hi

This have been discussed a couple of times, but I think its very importain. I would love to see some changes to the calibration movies:

- You should get some feedback, when you do something wrong. This way we learn from or mistakes and get even better at dustspotting. It doesen't have to be explained more than it have been done in the exsample movies.

- There should be other kind of "training" in the calibration movies like: Did the user see that it was a bad focus? A user might be able to focus on the surface, but was the movie focus bad and did not allow him form focus below the surface?

- A way to get the number of calibration movies down. I can see that this would effect the scoringsystem (?), but currently, I have 85 calibration movies and 214 real movies. That is a very large percentage of the movies, however, I do have 4 calibration errors (before I figured out what I did wrong), wich might be the reason I have gotten that many calibration movies?

- A history of my (failed) calibration movies would be nice, so I could look at them later and maybe study some of my errors one time more. I know that the amount of time we use should not be that much, because the number of users that see though these movies is the way Berkeley gets good feedback. But I, as the individual, want to get better and deliver the best results! It's one of the things that drives me - the rankings.

These are the four main things I would like to see done better, The first og the last one should be easy to implement (I guess). The second one require some work, but not that much. Berkeley staff/students could review some of the movies and mark them as calibration movies and give a feedback-description and the correct answer for that movie.

What do you think?

[CellarDoor]

This may be harder to implement, I dunno, but hindsight being 50/50...a few of the times I've hit on No Track just to notice there was, in fact, a track in the movie, which is endlessly frustrating. I accept my error with good grace (and realize I need to stop rushing through and slow down), however, it makes me think...if there was a way to hit back, that would be awesome. Not so much to click on the track and improve your score, because your score is your score, but if you never saw a track and received an error, it would enable you to see what you initially missed. This would be endlessly beneficial, particularly early in the game when I know for myself, I was a little too vigilant in picking the tracks...if anything looked vaguely unlike the ones in the training I would have passed them over, only to find they were, in fact, tracks.

I would also like to know what's an acceptable error percentage. Initially I was making an error 1 out of 10 times....I felt this was unacceptable, and have since attempted to improve my score so that I miss less than 5% of the time. I would love to know what the percentage would be to be considered a real eagle eye in this endeavour!

[Anders Lund]

Actualy, you can hit "back" in your browser, to review the last movie (or go even further back). But if you don't notice that you made an error, by cheking the callibration error score, you might not be able to find the one you made an error in.

But I would like to see the thing get implemented into the webpage itself - and it should be fairly easy, because you can view old movies, just by changing the URL in your browser. So if the system records the callibration movies, that you did wrong (and I would be amazed, if they did not do this allready now), you could get an list with failed callibration movies.

About the "acceptable" error percentage. I think that all error percentages are "acceptable", but the higher your level is, the more "score" do you get in Berkeleys list of "possible stardust"-list. This will mean that Berkeley will check the results you deliver (togther with many others) faster, than results that come from less "accurate" members. It's not a question about your result wont get to the lab but rather about how fast it will get though the system.