Only YOU can help find exoplanets!
December 17, 2010 2:15 PM Subscribe
Planet Hunters lets users comb through data from the Kepler mission in search of exoplanets. [via Bad Astronomy]
Just posted this on the nacent forums. It sums up the frustration I've had with all these sites for the last two years:
What we need here is a browseable library of examples that have already been identified. Ideally there should many typical stars, so we can get a feel for what is normal. There should also be the outliers: known binaries and transiting planets, with stats for how frequent each type is.posted by clarknova at 3:57 PM on December 17, 2010
A tutorial with a single idealized example, with no discussion of the other features you're looking for, is pitiful. It wastes our time while we make learning mistakes, and wastes the science team's time when they have to follow up on them.
All of the GalaxyZoo spinoff sites are weak in this regard.
It sums up the frustration I've had with all these sites for the last two years:
Problem: So far, Kepler has discovered exactly 6 confirmed stars with planets (and one had three of them!) So, the "curves" they want are, really, not well known.
So, they're not biasing you by giving you dozens of theoretical examples. They give you one, and then let you at it. Of course, they'll salt in the few confirmed hits they have -- and use that as a gauge of how good user X is at spotting transits -- as well as a few ones they know are false.
So, if, over time, you "spot" nine candidates, and three of them have already proven to be planets, then your other six reports gain credibility -- you, without much training, are spotting known transits, so when you "spot" an unknown one, it's more credible.
Esp if they know (and they will know!) that they've only shown you three real transits, and have shown you four known bad ones that you didn't mark. By mixing known data in, you can judge the validity of the reports on the unknown data.
posted by eriko at 4:37 PM on December 17, 2010
Problem: So far, Kepler has discovered exactly 6 confirmed stars with planets (and one had three of them!) So, the "curves" they want are, really, not well known.
So, they're not biasing you by giving you dozens of theoretical examples. They give you one, and then let you at it. Of course, they'll salt in the few confirmed hits they have -- and use that as a gauge of how good user X is at spotting transits -- as well as a few ones they know are false.
So, if, over time, you "spot" nine candidates, and three of them have already proven to be planets, then your other six reports gain credibility -- you, without much training, are spotting known transits, so when you "spot" an unknown one, it's more credible.
Esp if they know (and they will know!) that they've only shown you three real transits, and have shown you four known bad ones that you didn't mark. By mixing known data in, you can judge the validity of the reports on the unknown data.
posted by eriko at 4:37 PM on December 17, 2010
I agree that their methodology is somewhat flawed, but what it amounts to is a PR op.
This is not necessarily a bad thing! Projects like Zooniverse and Seti@Home may not be doing significant amounts of science, but they get people interested.
posted by Dearastronomer at 10:11 AM on December 18, 2010
This is not necessarily a bad thing! Projects like Zooniverse and Seti@Home may not be doing significant amounts of science, but they get people interested.
posted by Dearastronomer at 10:11 AM on December 18, 2010
« Older Volcanos on Titan and Oceans on Pluto | Captain Beefheart, RIP Newer »
This thread has been archived and is closed to new comments
#possible planetary body____ _ Stable Orbit ### Closer Larger Object
Possible Moon Found #_____ _ _ Processing
#Processing#
_
_
_
_
_
_
#That's No Moon.
posted by The Whelk at 2:40 PM on December 17, 2010 [2 favorites]