Stardust@home

Draw it mild, George!

We know the current form would give a good picture of our ability to find real tracks..., though more or less, because of some Power Movie distinctive features... a good picture, if all would steady click on every Power Movie they get in a session. But how will you be sure they do it every time when the frame of the system is so porous? By appealing to theirs conscience the whole time? Forget it.

We need a form that gives a reproducible result to take it for real. Currently, to me, it is just for fun and a challenge to go as high as possible by finding these "bloody" tiny tracks in the movies.

gebils wrote:I would advise that when you find yourself in a hole - quit digging. You already showed your colors. But then people like you never are in tune to advice - you are firmly convinced that doing anything and everything to get ahead is proper and to heck with everyone else.

George,

We welcome constructive, polite criticism and suggestions. Indeed, these have been vitally important to us in finding faulty power movies, in finding bugs in our code, and in improving the project. Impolite, sarcastic or abusive postings, however, are not helpful. If the discussion becomes unproductive, we may be forced to exercise editorial control over the forum. I hope it doesn't come to that.

Finally, I know many long-time dusters here would agree with me that fjgiie's participation in and contributions to the Stardust project have been exemplary. To accuse him of "cheating" (which makes little sense anyway) is quite simply, ridiculous. I understand your frustration with the Skill score and the problems that have arisen with Phase 4. If you need to discuss these issues with someone one on one, feel free to send me a PM. But do not take it out on those that are only here to help improve the situation for everyone involved, it is simply unwarranted and unappreciated.

Thank you.

Dan

jsmaje wrote:Re: 2019555V1 Dan. Sorry to not have made it clear. No, I didn't click anywhere on that particular movie, having missed it altogether! But on reviewing it in the 'missed' viewer I considered it to be outside your 25 micron-square red box coordinates and, given the others I mentioned where I'd clicked well within the old 60 micron circular diameter yet had been marked down because I was outside the red box rectangle, it seems I would have missed it anyway. So I repeat my question: what are the current pixel tolerance limits for an acceptable click? John

According to the code, it's still 30 John (in "browser pixels"). But one of the techs is going to look into this further when he gets back from vacation. It's in my keep tabs on file!

Dan

DanZ

Regarding your observation: "Finally, I know many long-time dusters here would agree with me that fjgiie's participation in and contributions to the Stardust project have been exemplary."

Let me be first in line to echo that statement! ALL of the dust-mods, in fact, have been more than helpful and friendly. In fact, they've gone above, over, and beyond expectations in being patient with dusters, and our comments. Our posts are often written out of pure frustration (I know mine often have been) but I've always been treated with respect. And I think we should show the same respect and appreciation toward them. I for one would like to see the postings monitored, to weed out the "nasties." They aren't pleasant to read, and are not profitable to the program. Evelyn ERSTRS

Okay, let's look at George's issues and suggestions since many of you share the same frustrations:

gebils wrote:1. Drop the Power Movies Missed figure and instead provide a running total of Power Movies Correct.

I don't see why this can't be done and will look into it.

gebils wrote:2. Drop your skill scoring of averages and substitute a simple relationship of percentage of correct results compared to total of power movies viewed. Total of correct power movies divided by total of power movies viewed. You would now have Power Rank which reflects the total movies viewed with the scores of the movies and the Skill Score relating to success with the power movies viewed.

A little tougher than it might seem. In Phase 4 we needed a way to incentivize dusters to look REALLY CAREFULLY for all possible tracks. In fact, we want you to look for the real hard ones and BE REWARDED when you find them. Power Score does this to a large degree, but it equally rewards those dusters who simply dust a lot. So then, how do we just as much reward those who really take the time to find the difficult tracks? That’s how we came up with the current Skill Score formula.

But pretend for a minute we used your suggested formula and let’s say I’m lucky enough to be served up on a consistent basis really easy power movies (since power movies come up randomly). My Skill Score will be around 1.0, the highest score you can get, and now I walk away and ride this wonderful score. No one, even those with the ability to find the real difficult tracks, will ever be able to get a much higher score than me.

But we still hear you – i.e., yes, we agree, it’s not fun to watch your score go down when you get a track right. Here’s one thing we are therefore doing, at least for now. Very soon, you will be able to opt out of Skill Score altogether, meaning if you don’t like this particular “tracking system,” you don’t have to be tracked by it. I know this is not the answer you expected, but believe me, it’s the best I can do at this time given the science team’s desires and the general nature of how the VM works. Nevertheless, I will continue to pursue other possible solutions with the team and keep you apprised of our decisions. Thank you for your patience and understanding.

gebils wrote:3. Spend the time to go through each Power Movie and adjust all the errors found.

There are over 5,000 power movies, i.e., it would be a long shut down. We are therefore trying to find what they have in common and do global fixes. For instance, we recently discovered there was a bug in the 'where clause' letting in the multiples of otherwise excluded power movies. This was a good catch thanks to all of you! Therefore, please continue to report the power movies that you think are problematic.

gebils wrote:4. I don't know what is going on with the testing for new people but it needs to be addressed.

It's been fixed already.

fjgiie wrote:When I click Bad Focus on a bad focus power movie the total of power movies viewed goes up by one. This counts against you in the Skill Score. So, the Bad Focus button has a lower value to dusters. We cannot continue to use the Bad Focus button as a bail out as we did in phases I, II & III.  Remember when jsmaje tested the bad focus button and there were no effects for clicking bad focus on a calibration movie?

The team consensus (on this side anyway) is that we'd rather you report power movies that you think are bad focus. Please do that here.

Thank you.

Dan

For sanity's sake, I'm locking this topic (which had an unappreciated negative title to begin with).

For future discussion, please post comments on pertinent topics as follows:

Comments on Skill Score here

Comments on Problematic Power Movies here

Thank you!

Dan