Slashdot Mirror
Distributed Project To Classify SDSS Galaxies
Xandu writes "Be part of a human Beowulf by helping classify millions of galaxies from the SDSS at the Galaxy Zoo. From their about page, "Those involved are directly contributing to scientific research, while getting an opportunity to view the beautiful and varied galaxies that inhabit our universe. Why do we need people to do this, rather than just using a computer? The simple answer is that the human brain is much better at recognizing patterns than a computer. Galaxies are complicated objects that vary in appearance enormously, and yet in some ways they can be very similar. We could write a computer program to classify these galaxies, and many researchers have, but so far none have really done a good enough job. We have not been able to make computers 'see past' the complexity, to reliably identify the similarities that appear obvious to our eyes and brain. For now, and probably for some time yet, people do the best job of classifying galaxies."
They're good for something after all.
Perhaps they shouldn't all be killed then.
I don't think I could resist the urge to tag everything as 'Thats no moon!'
Heh, I'll bite. Actually, this is version 2 (which came out 2 days ago). The original Galaxy Zoo was launched in July 2007, and only classified galaxies as spiral or not. This is much more fine-grained and allows for significantly better research.
And seriously, 6 jobs in the last 18 months. C'mon!
--Xandu
Is there a reason that they can't just use Amazon Mechanical Turk for this?
I also found that it took me probably 3 times as long to classify a single galaxy, because they've gone from an array of buttons to a tree of decisions for each picture. I hope it's worth it, but it's not as dead-simple to operate and it certainly takes longer.
Don't blame me, I voted for Baltar.
"Be part of a human Beowulf"
uh, you mean a team?
They will never stop until somebody makes the
After I pick a galaxy out of the lineup, how I can I be sure that it won't come after me later? Many astronomical objects get really bent when they think that they have been incorrectly labeled, and they can hold a grudge for billions of years! Just imagine being outside on a clear dark night and looking over you shoulder and seeing an angry galaxy coming after you. Do you think you can run fast enough to keep out of trouble? I pity the people who demoted Pluto from it's planetary status, I bet that they never go outside when it's dark. I'd rather just mind my own business.
The first and third links in this article take me to a blank page. Does anyone else get this effect?
God spoke to me.
I registered and tried logging in. No luck. Then I told it to send me a new password - no response. They really need to do this better.
Only his tendency toward a dazed stupor prevented him from screaming aloud.
Six months isn't all that old.
<xml><I><am><so><damn>Web 2.0</damn></so></am></I></xml>
Depending on the jobs that could be a great thing. I know plenty of freelancers who bill $100+/hr and work a couple of 1-2 month jobs each year and just slack off the rest of the time.
So. Uhm. He probably shouldn't help either.
The World Wide Web is dying. Soon, we shall have only the Internet.
Was it Google that gets people to tag images via the "ESP game"? You build up a score by guessing how another person would label the image. The more correct hits you get, the higher the score.
In grad school I remember a class where we played the same game. We had to make an estimate of the T-type (numerical hubble class) for the galaxies in the Coma cluster off of plate material. The point of the exercise in the end was to take all of these classifications made by supposedly trained and motivated people and show how much internal scatter you got (it was impressive, for some galaxies not better than random noise).
I hope these guys make you practice on known training sets first so they can "calibrate" ya!