Whine Time

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

Moderators: Stardust@home Team, DustMods

DanZ
Site Admin
Posts: 776
Joined: Fri Feb 27, 2009 3:44 pm
Location: Berkeley, CA
Contact:

Re: Whine Time

Post by DanZ » Mon Jul 18, 2011 2:13 pm

gebils wrote:http://stardustathome.ssl.berkeley.edu/ ... 6&theY=336

Absolutely nothing there - 53 point loss.
That is something we'll all have to get use to. If it's above 30, just about no one is going to get it! But we need the impossible ones for scientific analysis reasons. Per Dr. Westphal: ""We want to measure the detection efficiency curve all the way down to zero. In previous phases, we were able to measure detection efficiency as a function of track size only down to about 2.5 microns in diameter,
and even here dusters were amazingly good at finding tracks (better than 70%). We really want to know how small we can go and still find tracks, and that means going over the whole range of sizes, all the way down to (essentially) zero! We know that this can be frustrating, but the important thing is to do as well as you can, but don't expect to be able to get everything -- that is impossible! We can't find the smallest ones either."

Hope that clarifies why. But above all, don't get discouraged on those tracks. The HAT's are a different matter. See separate post on those (coming momentarily).

Dan

DanZ
Site Admin
Posts: 776
Joined: Fri Feb 27, 2009 3:44 pm
Location: Berkeley, CA
Contact:

Re: Whine Time

Post by DanZ » Mon Jul 18, 2011 2:19 pm

laserphil wrote:Has anyone else noticed that the "Your Skill" shown at the bottom of the stats on each movie actually drops each time you correctly identify power movies. It seems to only increases when you get a series of them correct. Phil
Yes, you may notice that the new “Skill” score will sometimes drop unexpectedly when you click correctly on a Power Movie (movie with a known track).

Here’s the reason:

The Skill Score is your Power Score divided by the number of Power Movies you’ve searched (as long as you’ve searched over 50). So, when you click on a Power Movie with a score that is lower than “average,” your score can occasionally drop. Here is an example:

At one point Duster “X” had viewed 240 Power Movies and had a Power Score of 1934. So, her Skill score should be 1934/240 = 8.058. Then on the very next movie her stats went to 241 Power Movies viewed and a Power Score of 1939, so her Skill would be 1939/241 = 8.046 and thus lower.

But note that in this example her Power Score would have dropped even more if she had missed that low score track (1934/241 = 8.025). Eventually though (after you’ve done a significant amount of searching) these little drops should not be noticeable.

Hope this clarifies things a bit.

Dan

DanZ
Site Admin
Posts: 776
Joined: Fri Feb 27, 2009 3:44 pm
Location: Berkeley, CA
Contact:

Re: Whine Time

Post by DanZ » Mon Jul 18, 2011 2:30 pm

jsmaje wrote:Ideally the supposed correct coordinates would be marked precisely on the movie itself (e.g. circled or with an arrow), as well as given in the URL. There would then be much less ambiguity in the duster's mind as to where the team expect us to click, make spotting wrongly-calibrated movies easier, and reduce the chance of error by the team in the first place. Dan, would that be technically possible? John
Except for the high angle tracks (or HATs - which we're trying to correct), the "red bars" you see are probably as good as we're going to get for now. But they're not bad; for example the ones here (a 37 pointer I missed on Friday) do a pretty good job of identifying the track I think.

By the way, we're experimenting to see if we can fix all of the HATS at once with some script. Hopefully some progress to report soon!

Dan

DanZ
Site Admin
Posts: 776
Joined: Fri Feb 27, 2009 3:44 pm
Location: Berkeley, CA
Contact:

Re: Whine Time

Post by DanZ » Mon Jul 18, 2011 2:38 pm

gebils wrote:My complaint is why do we have to be so precise by indicating the deepest point? I see our objective being identifying possible targets with tracks. These latest examples can easily be found in the lab when clicked anywhere along the track.

My second complaint is the number of no track targets. These are found on a lot of the Power Movies and not one is presented in the tutorial movies. What are the differences between these targets and inclusions?

George
Per your point #2, see my post above George. Your point #1 is correct, but the reason for point #2 precludes the easy fix of expanding the radius of what constitutes a correct click. But we're working on the HATs as fast as we can. Stay tuned and thanks for your patience.

Dan

DanZ
Site Admin
Posts: 776
Joined: Fri Feb 27, 2009 3:44 pm
Location: Berkeley, CA
Contact:

Re: Whine Time

Post by DanZ » Mon Jul 18, 2011 2:51 pm

jsmaje wrote:If the actual terminal particle is 'off-screen' in both cases, does that mean the 'blob' we've both clicked on is in fact where the particle impacted? I've certainly never seen such a clear entry point in previous phases. If so, are we not to indicate those too if the deepest part is ambiguous or out-of-view?
Clearly, in Phase 3, clicking on the prominent "dark spot" of a HAT was successful (i.e., gained a score), but you may be right. This may not have been the "deepest point" all along. However, given that so many of you were "trained" in this way in Phase 3 (including myself), I'm trying to see if we can switch HATs back to what was successful in the past, rather than trying to retrain everyone. Stay tuned!

Dan

Dr.Ronald C. Spencer
Posts: 2
Joined: Tue Apr 19, 2011 7:31 am
Location: Massachusetts

Re: Whine Time

Post by Dr.Ronald C. Spencer » Tue Jul 19, 2011 10:08 pm


I pick up points of various sorts from those frames ranging from 5-15 points consistently by clicking on the dark dot at the end of the connecting lighter shaded line. I move the microscope up and down to confirm it's what I am looking for, click it and on to the next frame. I have been very successful at it. Sometimes there may be one or two that are not what they appear to be but this is where discernment and trial and error trains you for what's ahead.

Dr. Ronald C. Spencer (aka) Ronald C. Spencer
Emeritus Astronomer
Astronomer's have Star's in their Eye's.

DanZ
Site Admin
Posts: 776
Joined: Fri Feb 27, 2009 3:44 pm
Location: Berkeley, CA
Contact:

Re: Whine Time

Post by DanZ » Wed Jul 20, 2011 10:40 am

Dr.Ronald C. Spencer wrote:I pick up points of various sorts from those frames ranging from 5-15 points consistently by clicking on the dark dot at the end of the connecting lighter shaded line. I move the microscope up and down to confirm it's what I am looking for, click it and on to the next frame. I have been very successful at it. Sometimes there may be one or two that are not what they appear to be but this is where discernment and trial and error trains you for what's ahead. Dr. Ronald C. Spencer (aka) Ronald C. Spencer Emeritus Astronomer
This is fascinating Dr. Spencer, that you haven't had the same problems.

Note to everyone - We are looking to temporarily shut down the VM for a half day (at most) sometime this week in an attempt to globally fix the reported problems with the HATs. We apologize in advance for any inconvenience and sincerely want to thank your for your extraordinary patience with us during this most recent phase launch.

All the best,

Dan
On behalf of the entire Stardust@home team

tiburd
Posts: 24
Joined: Thu Aug 03, 2006 7:47 pm
Location: Wyoming

Re: Whine Time

Post by tiburd » Sun Jul 24, 2011 5:10 pm

I assume the Team is aware that, as of today, the Skill ratings have suddenly dropped about half of the previous dusters, including some of the very top-ranked ones. I assume this change is inadvertent and is being addressed?
tiburd

DanZ
Site Admin
Posts: 776
Joined: Fri Feb 27, 2009 3:44 pm
Location: Berkeley, CA
Contact:

Re: Whine Time

Post by DanZ » Mon Jul 25, 2011 11:31 am

tiburd wrote:I assume the Team is aware that, as of today, the Skill ratings have suddenly dropped about half of the previous dusters, including some of the very top-ranked ones. I assume this change is inadvertent and is being addressed?
Yep, we're trying to figure that one out now.

Dan

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

Re: Whine Time

Post by fjgiie » Mon Jul 25, 2011 7:06 pm

The Skill Rank score is showing up incorrectly on the Skill Rank page and also the other pages like My Events and Home Page.
For instance my Skill is showing as 2.23 presently and it should be shown as 6892 divided by 534 which is around 12.90.

fjgiie

DanZ
Site Admin
Posts: 776
Joined: Fri Feb 27, 2009 3:44 pm
Location: Berkeley, CA
Contact:

Re: Whine Time

Post by DanZ » Mon Jul 25, 2011 7:18 pm

fjgiie wrote:The Skill Rank score is showing up incorrectly on the Skill Rank page and also the other pages like My Events and Home Page.
For instance my Skill is showing as 2.23 presently and it should be shown as 6892 divided by 534 which is around 12.90. fjgiie
Just "negotiated" a deal to fix that - everyone should see the higher numbers fairly soon.

Dan

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

Re: Whine Time

Post by fjgiie » Tue Jul 26, 2011 7:08 am

Congratulations to DanZ for getting the Skill scores fixed and
Congratulazioni a enokh - Numero uno di abilità


f

nitram819
Posts: 90
Joined: Mon Mar 02, 2009 9:38 am
Location: Quebec, Canada

Re: Whine Time

Post by nitram819 » Tue Jul 26, 2011 7:30 am

DanZ wrote: Just "negotiated" a deal to fix that - everyone should see the higher numbers fairly soon.

Dan
Great negotiating skills there DanZ, this problem is resolved too! (Thanks to Will?)

Cheers!
Martin

enokh
Posts: 54
Joined: Thu Aug 02, 2007 9:34 am
Location: Italia Reggio Emilia

Re: Whine Time

Post by enokh » Tue Jul 26, 2011 9:02 am

fjgiie wrote:Congratulations to DanZ for getting the Skill scores fixed and
Congratulazioni a enokh - Numero uno di abilità


f
thanks fjgiie

enokh

laserphil
Posts: 81
Joined: Sat Dec 02, 2006 9:28 pm
Location: Sydney Australia

Re: Whine Time

Post by laserphil » Fri Jul 29, 2011 6:13 am

Moral for dusters.

When your skill score goes up log off before it starts to go down.

I just finished a couple of hours dusting. I started with a skill level of 10.6 got up to 10.8 then the low score power movies set in. When my skill score was at 10.55 I was so P.....ed off that I missed a 23 and a 32 pointer in close succession. Saw them both but still clicked "no track" - Skill score at the end of the night 10.46. Not bad considering apart from the last two misses I had only missed one other - a negative co-ordinate movie of 6 points value! I could not even find a piece of track on the top left corner where it is supposed to be.

http://stardustathome.ssl.berkeley.edu/ ... 1&theY=-24



The "your skill level" is not really a measure of your skill its a measure of the (non)/randomness of supply of low score value power movies to you!

Phil

Locked