Page 2 of 2

Re: Where are the do/// tracks?

Posted: Sat Aug 18, 2007 12:50 pm
by stardust1
fjgiie wrote:First movie rt edge middle, second movie left edge middle. Easy to see if you know where they are!
Thanks, fjgiie, for helping meanwhile. Yes, easy to find if you know where and if your screen is really clean.
Sorry marymouse. I didn't want to discourage you or anybody even more than the new CMs already do. Those two samples are by far the most difficult CMs I've seen until now. I posted them to illustrate my point of view. The two tracks can be seen almost identically on many CMs. The point is the given context e.g concurrent features (like in sample 1) or unfortunate contrast (like in sample 2). Unfortunately such conditions are given on many real movies so the only conclusion can be to look very carefully on each movie CM or real. Common to that new CMs is that they stay in focus about a wide range. Some are stationary (sample 2) some 'moving' slightly aside while focus goes up and down ('moving dots').
Hope that helps.

Posted: Sat Aug 18, 2007 1:12 pm
by marymouse
Thank you all.

In the meantime, I took a deep breath and clicked on a speck which turned out to be a track in a cm. But there was no hole in the center.


Posted: Sat Aug 18, 2007 1:48 pm
by jsmaje
marymouse wrote:Thank you all. In the meantime, I took a deep breath and clicked on a speck which turned out to be a track in a cm. But there was no hole in the center. :?
There you go! -' expect the unexpected'.
It seems that if any actual interstellar tracks have so far been picked up in the top-left third of the collector (originally predicted to number around sixteen), they are likely to be much smaller than expected, and certainly nothing like the phase 1 calibration tracks. And such small tracks may not have visible central holes nor penetrate throughout the whole depth of focus.
The new calibrations take these possibilities into account, but as fjgiie predicted, it seems the team are inevitably going to have to trawl through a great deal more vaguely "possibles".
So, heaven knows when this project will ever end (quite appropriately so, since what we're dealing with are in a sense 'heavenly' particles).

Posted: Sat Aug 18, 2007 4:08 pm
by DustSabre
In reply to stardust 1's two posted calibration movies, I would have to say that I'm really not sure you have much to complain about. No offense, but I spotted the tracks in both calibration movies almost instantly, without knowing where they were, and both those tracks are exactly the sort of things I have always tried to find and click on anyway. I think the new calibration movies are harder, but they're harder for a good reason: Serious science doesn't come to you on a platter. You have to know what you're looking for to find anything, and it seems like the new movies more accurately reflect what we're looking for. That's my two cents, anyway, and you can take it for what it's worth. :? :)

Posted: Sun Aug 19, 2007 1:34 am
by cretinbob
This CM I have an issue with ... rds=?-1,-1

FIrst of all it's an obvious cut and paste job, and second it's not really representative of what a track looks like. Had it not been buried in the fracture I'd say fine, but there would have been traces from the surface dow. Yeah, I'm tired and the pixelated box should have given it away.

Posted: Sun Aug 19, 2007 6:36 am
by DustSabre
It does look like somebody pasted a square cut from some other place onto a blank picture, and then tried to render it transparent or something. :|

Posted: Sun Aug 19, 2007 6:46 am
by stardust1
Hi, DustSabre,
You probably misunderstood my intention posting those two samples. Let me try to explain once again. First of all I didn't 'complain' about the new CMs (why should I :) ) and to foreclose I voted this poll 'better but not optimal'. Those two movies just illustrated what the discussion was about so far.
You are probably right that the new CMs are to spot 'instantly'. As said after some training and depending on dusters eyesight but this was not the point. While the type 'tiny standing spot' frequently will be at the limit of the eyesight (maybe a knock-out criterion for some of us!) the new CMs in my opinion generally have obvious analogies to millions of what we thought were so-called inclusions:
- they only differ in two or three additional bars of focus
- they can only be validated in relation to the surface which is often hard to define and
- they can even fade out at the bottom of the movie.
The sample CMs jsmaje posted are very educational in that context. Those illustrate the difference. But what's with all those real movies that do not have this reference (both types, inclusions and "inclusionlike" tracks) on one movie? My guess is those will charge dusters' event pages inflationary.
My posts were as well meant for an answer to the initial question of that poll: Are the new CMs realistic? My votes would be
1. I don't know (how could I?) and if the scienstists knew it would be a good idea to gain access to detailed information (as much as possible, e.g. high magnification views of the phase I IS candidates)
2. If so they would be very very hard to find.
That's why I wrote in the above post adressed to marymouse: the only conclusion can be to look very carefully on every movie (CM or real) (That's the opposite of science that comes to you on a platter, isn't it? :wink: ). And the fact that dusters are forced to do so is the reason why I voted 'better than the old ones but ...'.
While the old CM type represented a clear and understandable model of the 'carot-shaped track' the corresponding model of the new CMs is hard to imagine and thus to apply on real movies in my opinion.

Obvious Track? (the tiny type)
reviewed from phase I (100 micron), found here

Posted: Mon Aug 20, 2007 4:01 am
by DustSabre
Thanks for the clarification, stardust1. I'm sorry if I misunderstood your post,
but I see what you were getting at now. Thanks in any case for posting the example CM's. :D

Posted: Mon Aug 20, 2007 6:29 am
by DiamondGirl
cretinbob wrote:Yeah, I'm tired and the pixelated box should have given it away.
I have to note though, that I clicked on a track inside a box, very similar to this one, and was scored as an incorrect selection. This is where I start to have a problem, because when I come across real slides where there's a dot I would normally click on, now I'm hesitating because I'm not sure if it was a mistake of the slide, or of my selection(s). I still go with my gut, because I'd rather find a track and get credit than have a great score, but I'm definitely more confused on how well I'm spotting real tracks than I was in Phase 1.


Posted: Tue Aug 21, 2007 2:33 pm
by jules
I have to agree with DiamondGirl on this one. I'm not interested in a high score - but I do watch the scores as they are an indication of accuracy - and I can no longer tell the difference between a real particle and a "fake" particle. Clicking on them all "just is case" is not very scientific, nor is ignoring potential particles as valuable data may be overlooked. I've been back to the tutorial but it does not help I'm afraid. Any chance of a more accurate tutorial? I think it would stop a lot of us floundering out here. :)

They said it would not be easy

Posted: Thu Aug 23, 2007 6:23 pm
by Star Gezzer
I beleive it was bmendez that said 95.00% was considered 100.00%. It was also said that it would not be easy. They were right on the second half for sure. This is for the person that made it their lifes mission to hinder my accomplisment of this feat. I appreciate the challenge. I also appreciate the difficulty that was built in because I am not looking up at the rearend of some computer programers score. By the way I really did not mean all of those things I said under my breath although my neighbors probably heard some of them. I am posting this while it is still 95.00%

Statistics For Star Gezzer

Your Overall Score: 5772
Total Movies Viewed: 31009
Your Rank: 1 out of 24478
Specificity: 99.67%
Sensitivity: 95.00%