Is it a 'real' or a 'CM'?

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

Moderators: Stardust@home Team, DustMods

jsmaje
Posts: 616
Joined: Tue Aug 15, 2006 8:39 am
Location: Manchester UK

Re: Right Clicking

Post by jsmaje »

fjgiie wrote:The mouse-pad has sticky syrup from eating and the mouse does not slide easily.
Eating or drinking at your computer has always been the eighth deadly sin, as I found in the early 70's when part of my sandwich gummed up the Fortran punch-card machine. Serves you right!

DustSabre
Posts: 63
Joined: Sat Dec 09, 2006 4:51 pm

Post by DustSabre »

The 9th deadly sin is to place powerful magnetic fields near your computer.

midloman
Posts: 3
Joined: Fri Jul 13, 2007 11:26 am

Post by midloman »

The right-click method works fine for checking your work. With observation, you can tell what CM's are positive and negative - and if it's a 50-micron or 100-micron. I'll leave it for you to make those determinations for yourself if you are so inclined. :)

I don't think that inclusions are desirable. Think short-penetration particles. This phase has deliberately been made to be ambigious at times - two or more reasonable (or unreasonable) candidates may be on the screen at the same time. My advice is - pick the one that spans the largest number of focus bars. This has worked well for me.

Looking for the little box or circle indicating an edited track is not always reliable - I think they've edited in some inclusions as well. Some particles I've found were hidden under much larger dark debris.

Look at other possibilites - there may be 5 inclusions and a barely visible bubble-track right at the lowest focus level.

DustBuster
DustMod
Posts: 694
Joined: Wed May 17, 2006 8:12 pm
Location: Horsetown, USA

Post by DustBuster »

midloman wrote:The right-click method works fine for checking your work. With observation, you can tell what CM's are positive and negative - and if it's a 50-micron or 100-micron. I'll leave it for you to make those determinations for yourself if you are so inclined. :)

I don't think that inclusions are desirable. Think short-penetration particles. This phase has deliberately been made to be ambigious at times - two or more reasonable (or unreasonable) candidates may be on the screen at the same time. My advice is - pick the one that spans the largest number of focus bars. This has worked well for me.

Looking for the little box or circle indicating an edited track is not always reliable - I think they've edited in some inclusions as well. Some particles I've found were hidden under much larger dark debris.

Look at other possibilites - there may be 5 inclusions and a barely visible bubble-track right at the lowest focus level.

I must attest, midloman, this is the most logical perspective I've read for some time.
My only other contribution being that I don't think there are any 'inclusions' that have been edited in... at least I don't think we are deliberately being 'tricked' into giving any specific responses (this would taint the data) but some of the CM's may be more--- subjective?

Nikita
DustMod
Posts: 994
Joined: Wed May 17, 2006 8:33 pm
Location: Indiana, USA

Post by Nikita »

I think they gang at Berkeley have a twisted sense of humor. Either that or they are assisting someone on the thesis that "You can fool all the people all the time" :lol:
Seriously, I agree, this is definately making us be more discriminating.
From dust we come

bmendez
Stardust@home Team
Stardust@home Team
Posts: 530
Joined: Wed Apr 19, 2006 11:28 am
Location: UC Berkeley Space Sciences Lab
Contact:

Post by bmendez »

Hi All,

Well, well...seems the cat really is out of the bag. The truth is we did know that you could determine if a movie was CM by checking the full URL path or by inputing the movie ID into the My Events viewer URL. We just hoped you wouldn't find out and then tell everybody about it :oops:

We did try to make the CMs less obvious where we could, such as obscuring the movie ID numbers and improving the look of them so that they were not so compressed. But some changes would require more subtle changes to the code that would take more time than we really had to spend.

We've been discussing what we can do to further obfuscate the CMs. We are considering mixing CMs into the real movie directory and adding another flag in the database to know when that has occured.

The reason we wish the CMs to be as close to indistinguishable from real movies as possible is not to prevent cheating. If someone wants to simply drive up their score just so that they can see their name in lights, well who really cares. As we've mentioned before, it doesn't effect the overall project. We want the status of CMs to be unknown, because it can introduce biases into the results if you know if a movie is a CM or not while you are making your determinations in the VM.

Recall that the purpose of the CMs is not to test or teach you what to look for, but rather to calibrate your individual responses. Each of you has a different approach to searching, and so what is a feature of interest to one person may not be to another. The CMs allow us to calibrate everyone to the same level when we examine your responses. So, if you know ahead of time that a movie is a CM, and you always answer them correctly, then you will artificially have a high sensitivity and specificity. Your artificially high score will make your clicks count for more than they really should. In scienctific lingo that is called a bias.

So, we'd ask that you try and set aside your strong desires to see 100% in your score and just go along searching as if every movie you see is a real one.

Thanks,
-Bryan
"I am made from the dust of the stars, and the oceans flow in my veins"
- RUSH

jsmaje
Posts: 616
Joined: Tue Aug 15, 2006 8:39 am
Location: Manchester UK

Post by jsmaje »

bmendez wrote:Well, well...seems the cat really is out of the bag. The truth is we did know that you could determine if a movie was CM by checking the full URL path or by inputing the movie ID into the My Events viewer URL. We just hoped you wouldn't find out and then tell everybody about it :oops:
Sorry Bryan, even though you actually recommend my post here!
First of all, I'm full of admiration for the wonderful efforts your small team have put into this project, given all the many other and more important things you have to do with limited resources in a finite time within a finite budget.

But I guess one lesson to be learned about a distributed-observer experiment such as this is that you should never underestimate what your anonymous volunteers might get up to(!), and that this should be anticipated before launch (CerealKiller also comes to mind).

Whatever, I'm sure your Chicago audience will have much to learn from your pioneering efforts.

John

Titanium
Posts: 38
Joined: Wed Jul 11, 2007 11:41 pm
Location: Nara, Japan

Post by Titanium »

fjgiie wrote:7. Using java script and HTML to write a program that will click the tracks for you while you sleep.
Is that really possible? I was gonna make a mock-post about Star Geezer being a machine or something, but ...


Also, what do you all think about "tricks" such as realizing that a huge percent of all phase 1 movies are CMs, looking for the poorly photoshopped track, etc?
A journey of 1000 miles begin with just one step.

DustSabre
Posts: 63
Joined: Sat Dec 09, 2006 4:51 pm

Post by DustSabre »

7. Using java script and HTML to write a program that will click the tracks for you while you sleep.
That's what cerealKiller was. It racked up a huge score, so it must have been fairly good.

Star Gezzer
Posts: 86
Joined: Fri Jan 26, 2007 3:24 pm
Location: Newport Oregon

Machine

Post by Star Gezzer »

I have been called a lot of things but never a machine. I am a 70 year old,single insomniac with lots of time on my hands. I also used to edit graphic design at the pixel level. I do not see tracks or CM's. I see a fault at or near max focus. It is kids play! I am on a quest for star dust. If they are there I will find one!

Post Reply