Problems with Phase II!

Post here if you are having any kind of problem with the Stardust@home website.

Moderators: Stardust@home Team, DustMods

Post Reply
Posts: 2
Joined: Fri Apr 13, 2007 1:21 pm

Problems with Phase II!

Post by ElliP »

These are the problems I faced as I was dusting today! Some of them are already reported. The reason I'm posting a new topic rather than replying to an already existing one is because I faced lots of different problems all in once.
The first half hour I was dusting was ok but then I noticed that some movies were downloading very very slow while the next ones at the average speed.
Some other could not be loaded at all. The icon was white and it started downloading the next movie. When I was pressing refresh in some cases the movie loaded but in some others it couldn't. That happened very often. About every 8-10 movies.
In some cases while I was pressing "no track" and waited for the next movie to appear, the page couldn't be loaded. I had to press the refresh button. That happened about 10-15 times. The last time that happened I got a message to log in or go home and of course I couldn't log in.
I logged in three hours ago (I'm in Greece) and all these problems happened in a period of 2 hours and I think these are way too many problems for such a short period of time!! :? :?


Alessandro Freda
Posts: 20
Joined: Fri Aug 04, 2006 12:12 pm
Location: Genoa, Italy

Re: Problems with Phase II!

Post by Alessandro Freda »

I see a problem similar, just a bit different.
Sometimes while focusing on a movie, the "next movie" bar do not start at all, and when it became the current movie, continue to not start, even after many refresh. The only clean way to skip to evaluate this "empty" movie is a logoff/logon.
This append about ever 100 movies for "burst" of few near events.


Posts: 25
Joined: Sun Aug 12, 2007 11:43 am
Location: Boulder, CO

I select bad focus

Post by gregson »

The startdust team has indicated that you should select bad focus to move on.

Posts: 65
Joined: Sat Aug 05, 2006 6:55 pm
Location: Oregon, USA

Phase II

Post by Groundling »

I'm sorry that I cannot adapt to what you call Phase II. The examples that I see in one movie have no meaning in the next movie.
I click on an object and I am told that it is not an object. When I ignore the same object I told that I am still wrong. You have indeed found a way to find small "tracks" but I am unable to tell the difference.
I have met the enemy and he is us.

Posts: 25
Joined: Sun Aug 12, 2007 11:43 am
Location: Boulder, CO

Consistency is a problem

Post by gregson »

I agree, there is no consistency in the calibration movies. I see a calibration movie and the only difference between the correct answer and the incorrect answer is the focus. In another calibration movie, it may not be that at all.

Without clear criteria, and consistent calibration movies, confusion will persist as to what the startdust team wants us to look for.

Posts: 1
Joined: Sun Mar 18, 2007 1:26 am

Re: Problems with Phase II!

Post by revent »

Alessandro Freda wrote:I see a problem similar, just a bit different.
Sometimes while focusing on a movie, the "next movie" bar do not start at all, and when it became the current movie, continue to not start, even after many refresh. The only clean way to skip to evaluate this "empty" movie is a logoff/logon.
If you hit the 'back' button to reload the last movie you evaluated (useful to re-examine a missed CM) you will get a different 'next movie', which should let you skip the one that didn't load.

I'm having problems with the CMs like everyone else, tho. My problem (an example is 7458240V1) is with CM tracks that start well below the surface. Sometimes these are obvious CMs (like the 'classic' Phase I tracks that we've learned to recognize) and sometimes they look, to me at least, like inclusions.

For instance, in the CM gave the number to the 'track' doesn't show any in-focus features until about 5 frames below the surface, and is only in-focus for a couple of frames before blurring back out. All that I see between this 'track' and the surface is what looks to me like faint diffraction features, and no entry path. I actually looked at this feature very carefully before clicking on No Track, and was /very/ surprised that this was a CM.

Honestly, this is exactly the kind of feature that I was clicking on when I started in Phase I, seeing almost /no/ agreements, and then seeing described as an inclusion.

I'm also seeing CMs with the surface near the top, and then a huge 'classic' CM track buried what seems almost ridiculously far into the gel with nothing between it and the surface.

I guess my question is, are we now supposed to click on /any/ apparent void-like structure beneath the surface of the gel, whether we think it looks like a track or not, or what? It is /very/ frustrating to evaluate a movie, decide that a feature is /not/ a track, and then be counted incorrect on a CM. OTOH, the only time I have been counted wrong for clicking on a track so far was on a movie that I actually thought was a CM before clicking, and didn't think the object was a track, but thought it looked like what I had been counted wrong for NOT clicking on.

It's not that I'm only of those people obsessed with my score. I've had enough scientific training to understand the point of what the Stardust team is trying to accomplish, and think I understand what the problem with the Phase I CMs was (not a wide enough Bell curve, basically). My problem is that some of the new CMs look to me like features that were described in Phase I as inclusions that we were /not/ to click on.

Posts: 40
Joined: Mon Jul 31, 2006 8:59 pm

Post by JOC »

I also agree with the above points totally.

I've lost the understanding of what we're expected to click on.
Ignore inclusions -> fair enough, don't click.
But now, Phase II CM's often look just like inclusions -> damn, supposed to click and I didn't -> I'm being rated as useless by the scoring system -> So, what...? I'm just wasting the teams time with my crumby dusting as I'm now so out of touch???
A CM with inclusions is now a puzzle to me.

A high score isn't important to me (I don't do enough dusting to get one!!), but at least when the sens & spec scores showed you were doing the right things you felt correct in your approach. That's kinda gone. What to look for is now very ill-defined. Which would be okay if (at the risk of repeating myself) our judgements were rated with the same latitude, but that can't be the case with contradictory, troublesome CM's.

I appreciate the StarDust teams predictament... they can't point out the answers until they know the answers. Unfortunately, attempting to create accurate synthetic sample answers is almost hopeless until some real answers are understood.


I look forward to Phase III then...


I wonder if the design team ever reconsidered the VM system... giving dusters options (like a multiple choice list) to further describe why they click on that particular movie.
EG say you click a feature and then select from:
I'm clicking on... A: distinct high-angle track
...B: doubtful high-angle track
...C: distinct low-angle track
...D: doubtful low-angle track
...E: subsurface-only feature
...F: other odd feature

Okay, dusters would be tasked with a slightly more onerous process but wouldn't such information allow dusters to inject more intelligence into the data. Personally, I think it would have been a more satisfying interaction for the duster to also self-rate their clicks.

Twinkle, twinkle, little dust!
How I wonder which to trust!
From stars above the world you fell!
Buried like treasure in aerogel.

Post Reply