I don't believe in this search

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

Moderators: Stardust@home Team, DustMods

Posts: 5
Joined: Thu Aug 17, 2006 1:47 pm

I don't believe in this search

Post by decomite »

Hi all,
I experimented stardust@home for two days now, viewing some 1000 movies
I am beginning to realize the vacuity of this spent time :
- Which proportion of the gel did I explore ? 0.0000001 percent ? Less ?
- The number of participants is around 10.000 ? Seeing the speed I climb the scores, I guess a lot of them already resigned...
- I think a standard neural network would work faster and more accurately :
- some NN applications can follow people in a crowd, having a video as input. A lot of public domain implementation of NN are available (and it is not too difficult to write a other more specific)
- Using the computational power and the people maintaining this site, this could have be done very quickly.

It is strange you don't use data mining technics, while the first applications
were in the field of astronomy (automatic detection of galaxies on pictures, Usama Fayyad)

In fact, I wonder it the real subject of stardust@home is a psychological study of people interested in the subject, rather than the discovery of spatial dust.

There is a Philip Dick novel (Time out of joint) where the hero thinks of playing a game every day in the newspapers, when he actually helps the governement to fight aliens....

I will perhaps return sometime to play with the video, with exactly the same state of mind I have when I play tetris : emptying my mind.

the moon
Posts: 177
Joined: Thu Aug 03, 2006 12:34 am

Post by the moon »

You saw about 0.1% of the gel. And were you serious about the psychological study? These are astrophysicists and such, what do they care about psychology? And don't you think if there was any way at all to do this with software instead of with the public's help they'd jump all over it? They've said before the software detection doesn't work well. They have enough trouble just identifying the surface level automatically.

Anyway, there's already enough people doing this to view each new movie over 100 times within a few days after it's available, so if you're not enjoying it, don't worry about it.

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

Post by JOC »

The Stardust team addressed this question themselves. You're right, decomite, NN can be very successful for patern matching. However, they'd need to be trained with existing data.... which isn't available.

Hence the laborious manual approach of getting lots of people to do a little bit each.
You'll find that the Stardust team project that the number of online searchers only need to spend 1 minute(!) each per day to complete the search in a reasonable time.

So trained NN's would be wonderful but it's a non-starter for this project.

And, as the moon pointed out, you underestimated your contribution by quite a few orders of magnitude.
Which proves just how easily the project will work. :-)
Twinkle, twinkle, little dust!
How I wonder which to trust!
From stars above the world you fell!
Buried like treasure in aerogel.

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

Post by tim_yoda »

I remember a lecture about a University experiment in which the voulunteers were asked to participate in a behavioral modification study.
The participants were required to deliver an electric shock to another human participant each time the other human "guinea pig" made a wrong decision in a controlled environment.
The electronic shock was increased with each "wrong" response.
The volunteer (shocker) was annonymous to the other participant (guinea pig or shockee) via two-way mirror(?)
The real purpose of the study was not to determine if the participant responded correctly to the stimulus, but to ascertain how much pain the volunteers were willing to administer at the urging of the moderators!
No real electricity was used, the people being "shocked" were merely students (who were "in on" the real nature of the project) acting as if they were taking a test and getting zapped.
Why? I remember the project was a study to probe the attitudes of humanity relative to those in nazi Germany's "final solution".
"I was just following orders" was the reocurring excuse for those who participated in delivering the holocaust, which prompted the university study.
I do NOT believe that Stardust@Home is some ruse with an ulterior motive.
The collateral data however; decision fatigue, cheating, debate, even some mention of political alignment, over the demographic spectrum represented here would be a waste if it too were not analyzed.
The "Funny,Interesting,movie" link, although incidental, would be a boon to the Rorschach Inkblot crowd over in Psych101.
I could go on, but I too have felt the nagging feeling of the examiner being the examinee.

All the world's a stage...

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

use the collateral data

Post by sub212 »

Suggestion to Berkley: Get in cotact with a Psych group somewhere, and have them analyse the results as outlined by Tim et al: fatigue, boredom, excitement, thrill, naggers, tatters, difference in age and corresponding comments (from "I'm 16 and excited about math..." to "Well, well, my young friend you underestimate,...").
Worth some nice MPsych PhDPsych studies, I reckon.


Posts: 53
Joined: Wed Aug 02, 2006 9:42 am
Location: Kansas City

Post by endust »

In suggesting the development of an NN to review these movies your forgeting two very important aspects. Time and Money. This review will last 7 months. How long would it take to develop, test, debug, then apply a very sophisticated program and how much would it cost?

And like the other web volunteer programs (such as SETI@home), you would not have the combined computing power of 10,000 computers and the eyes and brains attached. This is a brilliant application of the net and a tool that will be used again for other projects in this and other fields.

I'm sorry your bored. Many of us are far from it.

From the psychological side, what I find interesting is the almost instant development of a community of 12000 people. Granted, there is a silent majority but as with any group there are the high acheivers, the fanatics, politics, complainers and those who just give up. Thats worth a PHD to someone

Posts: 61
Joined: Thu Aug 03, 2006 5:05 am

Post by Jwb52z »

People like decomite are often the kind that assume they are above certain things when they believe a computer can do the same job. I call it a new form of snobbery, especially when they believe they have "out thought" someone or something by example of his weird comments assuming this is not a real project.

Posts: 6
Joined: Sat Aug 12, 2006 3:27 pm
Location: Patras, Greece

Post by avourekas »

I must admit that this project looks a lot like an arcade game (high scores, credits, spotting -killing?- an extraterrestrial grain), in fact I sometimes imagine that the thumbnail on the movie is the crosshair of a sophisticated hand gun.

I understand that many people (like decomite here) conceptualize this scientific project as a game (a thousand movies in two days??!!!) but it is not. This is pure science, in the hands of academics who take their serious work very seriously, and everyone should keep that in mind.

What is more, I am sure that the people behind Stardust are aware of all alternate approaches that can be followed in combination with volunteer work for the best possible completion of this scientific project.

Posts: 3
Joined: Fri Aug 04, 2006 10:09 pm
Location: Birmingham,AL

See you later then!

Post by acurban »

I will happilly scan whatever bits you feel are beneath you. I think this is a rare and wonderful, once in a lifetime opportunity to be a part of something "smaller" than myself.
This has been a family project for my 6yo girl and my wife as well. We are having a ball possibly looking at somethig never seen by human eyes. If you think you are in a bad episode of X-Files, go back to doing whatever you were doing before this project. We will pick up your slack.
The key to life is Truth and Honesty.
Once you've learned how to fake those,
the world is your oyster.
Mucky Pup

Posts: 9
Joined: Sat Aug 05, 2006 10:42 pm

Post by Superluminal »

I tend to believe that keeping scores and rankings was a bad idea, this is science not a competition. People can be tempted to cheat for a better score and on the other hand people may be reluctant to click on an interesting object, because if its a cal movie, it will mess-up their score. That happened to me after missing a few cal movies early on and before I could recognize them. I would see something that I knew probably wasn't a track, but it wouldn't hurt to let someone better qualified than myself take a look at it. After all, they told us to expect the unexpected. But if you are more interested in your score, something may be missed.

Posts: 5
Joined: Thu Aug 17, 2006 1:47 pm

Post by decomite »

[quote="endust"]In suggesting the development of an NN to review these movies your forgeting two very important aspects. Time and Money. This review will last 7 months. How long would it take to develop, test, debug, then apply a very sophisticated program and how much would it cost?
Tools exist : two weeks would suffice to develop,test and run the program.

And like the other web volunteer programs (such as SETI@home)[/quote]
What are the results of SETI@home till now ?

Posts: 10
Joined: Wed Aug 16, 2006 11:01 am
Location: Evanston, Illinois

Stardust or Psych expt?

Post by blewis »

Like Decomite and Tim_Yoda, my space scientist husband thinks this project would be of more value as a pysch test, and that my chances of finding any of about 300 expected real events in the thousands of photos, especially since no one really knows what they look like, is miniscule. But then, he is usually skeptical of anything and everything. I will continue to search, because it is exciting to be a tiny part of a grand discovery, or of a grand attempt.

Re value of rankings, etc. I am not interested in my ranking and do not feel that I am in a competition, but I do like the fact that there is instant feedback to my clicks on the calibration movies --- I am competing against myself, and learning to recognize a specific type of spot on the screen. It is much more addicting than Sudoku.

Posts: 61
Joined: Thu Aug 03, 2006 5:05 am

Post by Jwb52z »

The problem with the presence of the skeptics such as these, among others, is that they ruin the enjoyment that the rest of us get from it with all their comments.

Posts: 94
Joined: Thu May 18, 2006 7:12 am
Location: Third stone from the Sun

Post by TimStrange »

Hey decomite, you know too much. Please stare into the flash. :shock:

The Men In Black

Posts: 16
Joined: Wed Aug 09, 2006 6:14 am
Location: Seattle, Wahington, USA, Earth

Post by speck »

Hate to bust anyone's paranoid bubble but it would be unethical to do the kind of experiment d was describing. The rules on conducting experiments with human subjects are extremely strict and include the absolute requirement that the subjects are informed of the exact nature of the study, give consent, and have access to the conclusions.

So no, this wasn't all just rigged up as an experiment.

Now that doesn't mean that some enterprising young psych student couldn't take the aggragate data about our behaviour and analyze it. That can be OK.

I do consider this an experiment of sorts tho'. We are finding out whether using the web and a huge number of volunteers can be an effective and efficient means of doing initial sceenings in situations where there is a large amount of data to be looked at. I find that very worthwhile in itself.

I have no expectation or particular desire to personally find a particle, but I get great satisfaction knowing that I will help those who do - after all, I am identifying a whole bunch of slides that DON'T have any tracks !! ;-)

I can see where this pastime wouldn't be for some but I am having a ball.

Happy Hunting

- Starlight, star bright, first star I see tonite - I wish I may, I wish I might, have the wish I wish tonite. - Anonymous

Post Reply