Review: The 2 highest-rated candidates.

Discuss your experiences with and ideas about Stardust@home here.

Moderators: Stardust@home Team, DustMods

Louis Wu
Posts: 15
Joined: Sat Aug 05, 2006 7:14 am
Location: where the shark-bubbles blow...

Post by Louis Wu »

the moon wrote:
Louis Wu wrote:Tough call, but the fact that neither of those anomolies completely disappears if you focus up, makes me think that they are likely bubbles instead of tracks. I would have clicked them, but i wouldn't give them a high probablility of paying out.

Here's my personal best to date, and it does fade out completely on up-focus...

http://stardustathome.ssl.berkeley.edu/ ... =6415745V1
I wouldn't pay too much attention to how fast something disappears when you focus up. Bottom line is, tracks have to be under the surface. Your movie never focuses under the surface and should be marked bad focus.
Correct on both. It is noted for a bad focus. It's also marked as WCO...
If I had a particle, I'd name it Marla...

GelDelve

Post by GelDelve »

I should know better than to open my big mouth! It only gets me in trouble. I have let a day or two go by and still feel that I need to make a response to some comments I read here (not necessarily in this particular thread).

In my opinion the worst comments come from those who believe that it’s not very important to do well on “calibration” movies. Those movies are meant to educate members about what is meaningful and what is not. If any member thinks that not scoring well on calibration movies is unimportant and meaningless, then they don’t really care about learning and helping this project.

Next, there are members who think that anything oval or round with a dark circle around it is a track. Some think that if the movie has several similar items in it, then none of them can be a track, but if only a few exist, then one must be a track. Logic doesn’t seem to play any part in this decision, unless logic is illogical.

Lastly, at least for now, many of these people don’t even care whether they are looking at the surface, above the surface, or below the surface. Many don’t want to learn about how to identify tracks and ignore and dismiss the training and calibration movies completely. Seems to me that it is less than helpful to think that way.

I don’t pretend to know what is a track and what is not. I’ve certainly marked my fair share of movies as possible tracks, only to look at them again and wonder why I marked them at all. Anyone who thinks that 10% or even 1% of the movies they have looked at should be in their “My Events” folder should rethink their methodology and review the training and pay more attention to their scores on “calibration” movies in my opinion.

It’s not helpful to be either extremely liberal in identifying tracks or extremely conservative. Somewhere in between of those extremes is necessary. Even more important than either extreme is to acquaint yourself and be familiar with the what the calibration movies teach us to identify.

the moon
Posts: 177
Joined: Thu Aug 03, 2006 12:34 am

Post by the moon »

GelDelve wrote: Some think that if the movie has several similar items in it, then none of them can be a track, but if only a few exist, then one must be a track. Logic doesn’t seem to play any part in this decision, unless logic is illogical.
The logic behind this that dust particules don't come in pairs. If you see two things in the movie that look the same, chances are they are the same, and therefore not dust. It's just one more way to rule out possible tracks. Of course this doesn't mean that if something looks like it could be a track and there's only one of them, that it is one.

As for the rest of your post, I'm not sure who these people are that don't care about the CMs. I don't see posts like that. Everyone I see is very concerned about their score and getting CMs correct. Problem is many many people are confused or misguided. Like you said, they miss the concept of on the surface vs. below the surface, and instead look for the size and shape from the CMs.

Also I've marked 39 out of 2335, 1.7%, and I can tell you why I marked all of them, although on 2 I've since changed my mind. So I don't think that's too many at all. Eventhough none look like the CMs and all of them will turn out to be inclusions. But when real tracks are found they will look more like the movies I've marked, then like the CMs, so I gotta keep clicking em.

starbright
Posts: 13
Joined: Sat Aug 12, 2006 9:56 pm
Location: Perth, Western Australia

Post by starbright »

I would probably have clicked bad focus for both, if I'd seen them, although alpha is starting to look very interesting, if only the focus went deeper.

Both objects are right near the edge of the focus movies. Given that the movies seem to overlap a bit, the big question that I have is - has anyone found them on the edge of another movie? - and if not, why not? Perhaps because it had better focus and the apparent anomaly went away.

Starbright

blewis
Posts: 10
Joined: Wed Aug 16, 2006 11:01 am
Location: Evanston, Illinois
Contact:

alpha and beta

Post by blewis »

Yes, I would have clicked on both these movies; I have clicked on at least one quite similar. Neither of these movies appear in my list of events.
Pooh

ToSeek
Posts: 48
Joined: Thu Aug 03, 2006 8:40 am
Contact:

Post by ToSeek »

pointyhat wrote:I think I would have classified both these focus movies as bad focus.
Exactly what I was thinking. Looks to me as if the slides are tilted down to the right, so the stuff over there only comes into focus late.
If you're going to be just like everyone else, what's the point in existing?

DaveJ
Posts: 13
Joined: Thu May 18, 2006 1:52 pm
Location: Joliet, Illinois
Contact:

Post by DaveJ »

Alpha was my 99th image, and my first possible track, so I may be a bit biased, but...

The "track" is definitely BELOW the surface, contrary to what the moon says. If you'll notice, the dark objects along the top and toward the bottom left come into focus just above and below the mid-section of the focus bar. That's where the surface is. The track in Alpha isn't in full focus yet at that point, therefore it's below the surface.

As for the particle at the 11 o'clock position, it's in focus about 4 bars above the bottom of the slider, then begins going out of focus after that, while the track continues to get sharper, indicating it's not at the same level (close, but not exactly) as the track, so the track is extending further below that point.

the moon
Posts: 177
Joined: Thu Aug 03, 2006 12:34 am

Post by the moon »

DaveJ wrote:The "track" is definitely BELOW the surface, contrary to what the moon says.
The surface is slanted, as it is in many of the movies. In most of them it's obvious because the whole surface will have a texture to it, so it easy to tell when each part of the surface is in focus. In these 2 movies it's harder to tell because the surface is smooth, but you can still figure it out by the way the other few surface spots come into focus.

Like in alpha, first the spot at the top left comes into focus, go down a little and the spots at the top right and bottom left come into focus, down a little more and the spots at the middle right come into focus. So the surface is slanted down from the top left, to bottom right.

GelDelve

Post by GelDelve »

the moon wrote:
GelDelve wrote: Some think that if the movie has several similar items in it, then none of them can be a track, but if only a few exist, then one must be a track. Logic doesn’t seem to play any part in this decision, unless logic is illogical.
The logic behind this that dust particules don't come in pairs. If you see two things in the movie that look the same, chances are they are the same, and therefore not dust. It's just one more way to rule out possible tracks. Of course this doesn't mean that if something looks like it could be a track and there's only one of them, that it is one.

As for the rest of your post, I'm not sure who these people are that don't care about the CMs. I don't see posts like that. Everyone I see is very concerned about their score and getting CMs correct. Problem is many many people are confused or misguided. Like you said, they miss the concept of on the surface vs. below the surface, and instead look for the size and shape from the CMs.

Also I've marked 39 out of 2335, 1.7%, and I can tell you why I marked all of them, although on 2 I've since changed my mind. So I don't think that's too many at all. Eventhough none look like the CMs and all of them will turn out to be inclusions. But when real tracks are found they will look more like the movies I've marked, then like the CMs, so I gotta keep clicking em.
Then if an actual track has nothing distinct or identifiable about it, we are wasting our time, aren't we? A computer program could easily find movies with one or two distinct objects in it. Similar objects that occur multiple times in one movie and only occur once in another are to be logically treated differently? I'm sorry, but that is not at all logical.

I will again state that identifying more than 1% of your reviewed movies is excessive and counterproductive, whether or not you think you can justify most of them. You are one of those members who would benefit from reviewing the tutorial and concentrate on scoring and learning from the calibration movies. I have to wonder what your scores are from your comments.

GelDelve

Post by GelDelve »

the moon wrote:As for the rest of your post, I'm not sure who these people are that don't care about the CMs.
the moon wrote:Don't worry too much about getting CMs wrong. All the people that brag about 99% accuracy, it doesn't really factor into anything. As far as I know, (they could have changed the system and not told anyone) how they select the movies with the most hits is from your total score. So if you view 1000 CMs and get 25% wrong, your score will be 500. That means your vote still counts 5 times as much as someone who views 100 CMs and gets them all right.
Yep, I also wonder who doesn't care about scoring well on the CM's.

the moon
Posts: 177
Joined: Thu Aug 03, 2006 12:34 am

Post by the moon »

GelDelve wrote:Then if an actual track has nothing distinct or identifiable about it, we are wasting our time, aren't we? A computer program could easily find movies with one or two distinct objects in it. Similar objects that occur multiple times in one movie and only occur once in another are to be logically treated differently? I'm sorry, but that is not at all logical.

I will again state that identifying more than 1% of your reviewed movies is excessive and counterproductive, whether or not you think you can justify most of them. You are one of those members who would benefit from reviewing the tutorial and concentrate on scoring and learning from the calibration movies. I have to wonder what your scores are from your comments.
Yes similiar objects in different movies do have to be treated differently. Tracks don't have anything distinct about them, but they do have identifiable features. Even the most perfect looking track could turn out to be something else, that doesn't mean you should never mark anything. So for example if you have a feature that might be a track and something similiar looking is in a different movie, it's possible that one is a track and one isn't. Both should be marked, unless there are other clues that excludes them. For example the features in these alpha and beta movies are on the surface so they can be excluded, however if they were under the surface they should be marked. Or, say they were both under the surface but in one movie there was a pair of them, that movie can be excluded and the other should be marked. What we are really here to do is to rule out all the movies that definetly don't have tracks so the researchers don't have to look through all of them. The point is not to hand them one movie and say, this one has a track, ignore all the rest. So everyone has to decide for themselves on a relative threshold of likelyness for what kind of features they'll mark and which they won't. I say if you're marking less then 1% then you're missing too many possible tracks. You might benefit from reviewing more of the information about CMs. They say that all of the CMs were made from just a handful of real impact tracks from a different collector, digitally pasted into a 1000 movies from this collector. Different aerogel, different resolution, digitally pasted in, and yet you dismiss anything that doesn't exactly match them.

As for my score, I know I'll regret trying to prove myself to someone on a forum, but I haven't missed any out of 630. I did get 6 marked wrong though. 2 when some were buged and had the wrong coordinates of the tracks. 1 after that when I wanted to test to see if it was actually checking where you click and would mark you wrong if you clicked in the wrong place (I didn't know it was a bug yet, I thought it might be my browser). 1 I clicked on an inclusion in one without a track. And 2 I was tired and hit no track out of habit eventhough I saw the tracks.

the moon
Posts: 177
Joined: Thu Aug 03, 2006 12:34 am

Post by the moon »

GelDelve wrote:Yep, I also wonder who doesn't care about scoring well on the CM's.
Actually if you put that quote in context I was replying to someone that was so worried about getting CMs wrong that when in doubt about a track, he would choose no track to avoid getting marked wrong.

But you are sort of right. Like I said in the other post, I think the CMs are kind of misleading, making some people think that all tracks will look like this without exception. Everyone should be very familiar with the properties of the tracks in the CMs, but you also have to use a little imagnination when looking for the real thing.

I care strongly about my own score simply because I'm very confident in my abilities and want my "track or no" votes to count more towards what the researches will see.

niejell
Posts: 25
Joined: Wed Aug 02, 2006 1:10 pm
Location: Maryland, USA

Post by niejell »

Clicked on ALPHA early on ... I consider the stray objects coming into focus at the same time (to me, means relatively flat) the surface and the track begins to focus just below.
Hope I would have clicked on BETA if I ever saw it.
Shakin da tree...

wol
Posts: 3
Joined: Thu Aug 10, 2006 11:43 pm

Post by wol »

I would have clicked on Alpha, as it seems to appear just at the surface and become clearer below, as expected.

I would have marked Beta bad focus, as what appears to be a track appears just on the surface of a low spot, and it is not possible to focus down further.

bulamark
Posts: 9
Joined: Mon Aug 28, 2006 10:27 am

Alpha & Beta

Post by bulamark »

I have seen a few of these but with them coming into focus at the surface and not being able to focus further below the surface, I would and may have marked them for out of focus. Hope the office staff will give us an answer but from the last update I don't believe we'll have a definitive answer for quite sometime. Happy searching.

Post Reply