Stardust@home Suggestion Thread

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

Moderators: Stardust@home Team, DustMods

sub212
Posts: 48
Joined: Fri Aug 18, 2006 4:27 pm

Post by sub212 »

totallz agree.
The suggested 100 / 100 certificates I wouldnt recommend:

Getting into the top 100 now, if you started late, is near-impossible, thus insufficient good folks might pass on a milestone, on the other end, having viewed 100 movies has been done probably by 95% of the 16000 odd participants, so it's not really a milestone.
I would suggest 1000/1000, "ever been in the top1000" and "viewed and graded more than 1000 real movies and helped evaluating them".

What do you think?
goldrake
Posts: 72
Joined: Tue Aug 08, 2006 1:29 am
Location: Florence, Italy
Contact:

Post by goldrake »

sub212 wrote:totallz agree.
The suggested 100 / 100 certificates I wouldnt recommend:

Getting into the top 100 now, if you started late, is near-impossible, thus insufficient good folks might pass on a milestone, on the other end, having viewed 100 movies has been done probably by 95% of the 16000 odd participants, so it's not really a milestone.
I would suggest 1000/1000, "ever been in the top1000" and "viewed and graded more than 1000 real movies and helped evaluating them".

What do you think?
For me 1000 is a good milestone to certified!
be-one
Posts: 4
Joined: Tue Aug 08, 2006 7:23 am
Location: Lille, France

Post by be-one »

Speck wrote (in this thread, see Mon Aug 14, 2006 7:24 pm):
I think it would be useful to reset the specificity/sensitivity stats periodically. Resetting would assure that these numbers reflect the reviewer's current skill. Some people had technical problems at the outset and it is EXTREMELY difficult to compensate for those as your numbers get higher. Conversely, as the numbers get higher, one can make more errors and there won't be much change in the spec/sens stats. Resetting would keep us all on our toes all the time.
We can read through many other threads some concerns revolving around this idea. That is : "my actual scores do not reflect my current visual capability".

It's all about complex learning effects at running some repetitive activity and I think the projet should feedback some information about that.

Here are some remarks (let me apologize if already said elsewhere and for the style as I'm not used to write US) :

1) Compute stats step by step (or when logging off) but not from the beginning. Say, over the last one thousand CMs (really I don't know how many would be optimimal). Advantages : it refines the movie ratings (more volatility in the %, see Speck above) and encourages newcomers to carry on.

2) Design some overall stats about how the community performs as a whole (like histograms, average numbers and standard deviations on more qualitative items). So, everyone will be able to assess her/his contribution to the research and if she or he is currently doing well. Furthermore, it could moderate the discussions about the top 100 figure.

3) Foresee it at the individual level : what could be the matter with 30000 or more stardusters? Where is the efficiency? How much we killed during a single logg or how long we stay alive in the project at a given level of activity? To be considered with the fact that individual learning (skilling) process is still running after we're logged off the VM. So, the longer, the better.

Happy dusting for all.
luranz
Posts: 4
Joined: Sun Aug 20, 2006 2:18 pm

Re: Stardust@home Suggestion Thread

Post by luranz »

templar781 wrote:If you have any suggestions on how this site can be improved post it here. I'll start off with two suggestions.

1. Start a suggestion/improvement forum.

2. Implement scroll wheel support for the focus bar. I would like to be able to place the cursor in the focus bar, have the bar indicate the focus position, and then allow the scroll wheel on the mouse to control the up and down focus.
- download movies beginning from the deepest image (the eye can see much easily what is in focus in an unfocused field, rather than unfocus the surface and find what keeps in focus).
- manage the list of personal suspected movies. Now that I'm more "expert", I would erase two, for example...

Well, I know more that too much good ideas could create more confusion and time loss than continue this way.... so, I have to say that I find well anyway :D
Good job Berkeley!

Lorenzo
fjgiie
DustMod
Posts: 1253
Joined: Sat May 20, 2006 8:47 am
Location: Hampton, SC, US

Two windows open

Post by fjgiie »

DustBuster wrote:Just an observation, but I have to wonder;
Wouldn't loading movies be the same no matter how many you pre-loaded?
Right now, 2 movies are pre-loaded into the buffer the first time you log onto the VM: when you click to the next movie, only 1 additional movie pre-loads into the buffer to replace it, right? So even if you pre-loaded 5 or 10 movies initially, wouldn't you experience the same time lags when one movie is dismmissed and replaced with another? If so, pre-loading more images in the buffer would only serve to rob the user of resources and not produce the desired increase.

I this not how it works?
There are a couple of things you will see upon trying the two windows. Sometimes we come upon a complicated patterned focus movie. Three more movies are being loaded while we study this difficult movie. If only one window open, only one movie loads.

With my computer, I need to wait until both blue bars are fully loaded before the bottom slide will load (the 40th one). The movies in the extra window will both be loaded when needed.

It is about three times as fast as one window.
This only applies "For slow Loading - " It's almost too fast when the site is not busy.

Thanks,

fjgiie
LCARS
Posts: 22
Joined: Thu Aug 03, 2006 1:08 am

Polarizing

Post by LCARS »

Would there be any advantage to using a polarized light source and a linear polarizing filter on the camera (if possible) to photograph the gel?

If i'm not mistaken this should result in a (rather colorful) way to spot tracks below the surface of the gel and help distinguish a true track from any other "track-like" anomalies based on the patterns of color present as well as the focal depth.

Even if it is a bit out of focus, a track may still produce color ripples at near focus that are visible and thus draw attention to them. Well, unless the RI of aerogel is too low to produce any...? I'm not an expert in the field of photonics but I have employed polarizing as a technique in visual inspection of transparent plastic films in a professional capacity before with success.

Just a thought. :idea:
~My brain hurts~
medicjon
Posts: 4
Joined: Tue Aug 01, 2006 5:48 am
Location: Pittsburgh, PA, USA

Post by medicjon »

I tried to make one using masking tape but now my monitor is all sticky! Eww!

Just Kidding. I think that would be interesting to try (a grid like you suggest, not tape).

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

Larry, Curly and Moe

Post by tim_yoda »

Please clean the imaging optics responsible for said anomolies before shooting more movies.This would enhance the searching process, decrease false "hits" and eliminate Forum entries regarding our beloved trio.

Tim
CheyenneWills
Posts: 1
Joined: Mon Aug 21, 2006 6:36 pm

A couple of suggestions

Post by CheyenneWills »

A couple of suggestions...

An option to "invert" the color of the movie. Sometimes you can see detail that you might have missed.

An option to do a review of images before they are submitted. I might be able to quickly sort out the ones that don't have anything, but I can have a list of ones that I want to spend more time looking at before I hit "submit"

Team review, if you have a couple of friends, allow a list of potential hits to be group reviewed within your own group. -- Basically this is kind of asking a friend "hey -- do you see something here too?" kind of an option.

Anyway -- just some ideas..

Cheyenne

(30 years of looking through telescopes finally comes in handy :)
Mighty Pete
Posts: 146
Joined: Thu Aug 10, 2006 7:47 am
Contact:

We need a first to find it column in My Events.

Post by Mighty Pete »

I asked this in a differnt thread I started after searching for first:


Search on these forums sucks..


Are we going to get such a column so we can see if we are indeed the first to find something. Sure you can look to see if you where first after you find it but try doing that a week later. I have tons of things marked. There is no way I can remember now which ones I marked first.

Now that some are being rechecked it would be nice if we could some how keep better track of the ones we personally found first.

We need a first to find it column in My Events.
merelake
Posts: 7
Joined: Sun Aug 06, 2006 5:54 am

Post by merelake »

minkiemink wrote:As I was dusting, it occured to me that a pale grid overlay ....
-Minkie
An easy method to implement would be to insert scales on the X,Y axes and use L shape piece of cardboard. No software required.
merelake
Posts: 7
Joined: Sun Aug 06, 2006 5:54 am

Does the other aerogel have inclusions?

Post by merelake »

What about the spare bits of aerogel that didnt make it into space?
Do they have inclusions?
Why not shine a strong light into the side to find out?
cmsouza
Posts: 6
Joined: Tue Aug 22, 2006 6:50 am

Post by cmsouza »

I, too, would like to be able to "deselect" errant selections...reviewing some of the movies on a laptop with an all-too-sensitive touchpad has so far resulted in one errant track spotted, and one wrong calibration movie answer....Since those two, I've been using a mouse instead,with far better uniformity. :)

Also, first finder is something that should be quick to spot...For example, if you're event only has one agreement (I check each time I suspect a track), voila!

I'd definitely like to see a sample of what a non-particule subsurface inclusion looks like, too...And while inverse colours would be nice, I think that might be asking too much of the hosting server's capacity...I mean, they'd need images of both standard and inverted colours, and for all the movies they've got, that's probably no small task...
LCARS
Posts: 22
Joined: Thu Aug 03, 2006 1:08 am

Post by LCARS »

I think that using only the last 500 focus movies viewed to compute the stats is really a good idea! :)

It would keep any "mistakes" made early on from tarnishing those who become much more proficient after viewing more than 500 movies.

I got on board very early on & I know some of the calibration movies had problems and I know I missed a couple smaller tracks in a couple but I feel that now I have a better grasp on what i'm looking for.

I have only goofed on 1 calibration movie in the last 350 I've looked at but I goofed on 5 in the first 100 until I got the hang of it. Buffering out the results from over 500 movies ago will definately provide a more accurate sample of someone's current skill level.
~My brain hurts~
Dorsal
Posts: 10
Joined: Wed Aug 02, 2006 9:51 pm

Post by Dorsal »

Could the Starduster team "turn" or "flip" the calibration movies so they didn't look exactly the same every time? They are calibration movies, so the true orientation doesn't matter, right?

btw - I wonder if it might have helped the imager if the aerogel had been given a fine dusting of 1 micron talc, or micro-spheres, or some other inert material. Many of the screens seem to have no discernable surface features and only a speck or two in the entire view area, making it very difficult to judge where the specks lie, either on or below the surface. Just a thought.
Post Reply