Tricky calibration movie

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

Moderators: Stardust@home Team, DustMods

Post Reply
don-smith
Posts: 2
Joined: Tue Aug 08, 2006 8:18 pm

Tricky calibration movie

Post by don-smith »

I encountered a calibration movie that had something in it that looked like it might be a track but wasn't since it was counted as not a track. Unfortunately, I don't record movie ID's so I can not give one. The movie showed a dark speck on the surface and a semi-circle that came into focus below the surface. It was right up against the left edge of the movie about 1/3 of the way down. My Specificity was 100% until that happened.

If you had said only click on things that look like the example movies, then this was a wrong click. But I believe the direction was to click on anything odd because you aren't really sure what the track may look like (to me, this looked like a track coming in at an angle rather than straight down.)

Anyway, either this calibration movie should be removed, or the training should be expanded to help us see why it wasn't a track.

Don

jcbc
Posts: 65
Joined: Wed Aug 02, 2006 2:10 pm
Location: Pennsylvania, USA

Post by jcbc »

I've had three negative CMs with interesting features that I have clicked on and gotten wrong. I do believe it is counter-productive to have negative CMs with such features that you would probably click on if it was in a real movie. I've posted mine in the "I answered a calibration movie right but I was scored wrong" thread in the FAQs forum. Maybe if enough people write there about problems with negative CMs with features, they'll look into it. Next time it happens to you, click the Back button, record the ID of the CM, and post it in the above thread for them to look at.

JCBC

Jwb52z
Posts: 61
Joined: Thu Aug 03, 2006 5:05 am

Post by Jwb52z »

It's not counter productive because that is the point of them, to check and see if you can tell a calibration movie with no track, in the case of the ones without a track, as we were trained to look for initially.

jcbc
Posts: 65
Joined: Wed Aug 02, 2006 2:10 pm
Location: Pennsylvania, USA

Post by jcbc »

If that's the point, then it should say "This is a calibration movie; does it have a track or not?" otherwise people may not click on the "subtle" inclusions that the team seemed to be pleased that we were finding (see this post in the Updates forum).

For me personally, I will continue to click on faint spots that begin to appear below the surface as well as other interesting features and I am sure that my specificity percentage will continue to suffer as long as negative CMs have them. Three of my total of four errors are on negative CMs. Also, the tracks in the positive CMs are getting almost as small as the inclusions, which is another reason to make sure negative CMs are clear of anything that might be mistaken for a very small track below the surface.

JCBC

Jwb52z
Posts: 61
Joined: Thu Aug 03, 2006 5:05 am

Post by Jwb52z »

Another part of the calibration movies is that they are a test mechanism, so being TOLD they are calibration movies would defeat the point.

jcbc
Posts: 65
Joined: Wed Aug 02, 2006 2:10 pm
Location: Pennsylvania, USA

Post by jcbc »

Jwb52z wrote:Another part of the calibration movies is that they are a test mechanism, so being TOLD they are calibration movies would defeat the point.
Then, IMHO, they need to make sure that negative CMs are clear of anything interesting below the surface in order for it to be a valid test.

JCBC

Jwb52z
Posts: 61
Joined: Thu Aug 03, 2006 5:05 am

Post by Jwb52z »

That's not an opinion. In this instance, it would ruin the point of the test because we are supposed to be learning how to tell a track from other "stuff" below the surface. If there's nothing except a track below the surface, it's not a real test.

jcbc
Posts: 65
Joined: Wed Aug 02, 2006 2:10 pm
Location: Pennsylvania, USA

Post by jcbc »

Hmmm, you know what, the tutorial says nothing about "other 'stuff' below the surface", except large, slow particle tracks, about which it says:
The Tutorial, page 10 wrote:This particle is so big that you can easily see it at the end of its track; it is larger than we expect to see in the Stardust Interstellar Dust Collector. But we might be surprised! If you see such a track, or anything that you find unusual, do click on it. You might have found something completely unexpected!
My opinion based on the tutorial is that they would like us to click on anything [edit]in focus[/edit] that we see below the surface, as none of their "this is not a track" directives mention "stuff" below the surface.

JCBC
Last edited by jcbc on Thu Sep 07, 2006 6:14 pm, edited 1 time in total.

Viceroy
Posts: 3
Joined: Mon Aug 14, 2006 6:33 pm

Post by Viceroy »

jcbc wrote: My opinion based on the tutorial is that they would like us to click on anything that we see below the surface, as none of their "this is not a track" directives mention "stuff" below the surface.
They want you to click on things that look like tracks. Stuff on the surface will show up out of focus when the microscope is focused below the surface. If your eyes are bad and you can't tell if something is in or out of focus, then you shouldn't be surprised that you are getting calibration movies wrong.

elainekeefe
Posts: 190
Joined: Fri Aug 04, 2006 4:38 am
Location: Massachusetts, USA

Post by elainekeefe »

After viewing over 3000 calibration movies with no errors, I, too, was just caught by a "no-track" calibration movie. (My previous 9 errors were all made in the first 800 or 900 movies.) It was one of those "busy" movies, with lots of circles. I believed I saw a circle just coming into focus at the bottom of the focus scale and clicked on it, feeling it was better to include a dud than exclude a potential track. I guess I'll think twice before I do that again. :(

jcbc
Posts: 65
Joined: Wed Aug 02, 2006 2:10 pm
Location: Pennsylvania, USA

Post by jcbc »

Viceroy wrote:They want you to click on things that look like tracks. Stuff on the surface will show up out of focus when the microscope is focused below the surface. If your eyes are bad and you can't tell if something is in or out of focus, then you shouldn't be surprised that you are getting calibration movies wrong.
I thought it was quite clear when I said "anything below the surface" that I meant "anything in focus below the surface". I've edited my post so that others like you won't make the same mistake.

FYI, please see Bryan Mendez's post about empty calibration movies for more on this topic.

Happy searching,
JCBC

tim_yoda
Posts: 39
Joined: Fri Aug 04, 2006 5:06 pm
Location: Broadview Heights, OH USA

Post by tim_yoda »

elainekeefe wrote:After viewing over 3000 calibration movies with no errors, I, too, was just caught by a "no-track" calibration movie. (My previous 9 errors were all made in the first 800 or 900 movies.) It was one of those "busy" movies, with lots of circles. I believed I saw a circle just coming into focus at the bottom of the focus scale and clicked on it, feeling it was better to include a dud than exclude a potential track. I guess I'll think twice before I do that again. :(
I think that one got me too. It formed a pronounced black doghnut at or near deep (bottom) focus.
As in other discussion threads, the issue of CMs evolving has become apparent. I write off such occurances as the cost of being dilligent!


Tim

Siegfried
Posts: 105
Joined: Wed May 17, 2006 7:18 pm
Location: La Quinta, California
Contact:

Re: Tricky calibration movie

Post by Siegfried »

don-smith wrote:I encountered a calibration movie that had something in it that looked like it might be a track but wasn't since it was counted as not a track. Unfortunately, I don't record movie ID's so I can not give one. The movie showed a dark speck on the surface and a semi-circle that came into focus below the surface. It was right up against the left edge of the movie about 1/3 of the way down. My Specificity was 100% until that happened.

If you had said only click on things that look like the example movies, then this was a wrong click. But I believe the direction was to click on anything odd because you aren't really sure what the track may look like (to me, this looked like a track coming in at an angle rather than straight down.)

Anyway, either this calibration movie should be removed, or the training should be expanded to help us see why it wasn't a track.

Don
I think I got one similar today. I clicked it, and the next thing I know, I had one more wrong answer (which is the last thing I need, even though they count for less than five percent of my total calibration movies viewed), and a now marred (formerly perfect) specifity.

It also bugs me when I get a track but computer problems (a jumpy mouse cursor, etc.) cause me to click in the wrong place, etc.
And God said: E = +mv^2 - Ze^2/r ...and there *WAS* light!

WANTED:
Dead or Alive
^-^
( ^ )
Schroedinger's Cat

Post Reply