Email address obfuscation in effect -- please
click here to turn it off.
[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
Maybe there is a way to combine the idea of photo recognition with low
probability of guessing correctly. Photos of animals are a good idea.
Maybe 8 photos are shown and the user has to say (by clicking on a
menu) which of 10 animals is displayed in each photo: cat, dog, cow,
horse, etc. That would give you a 1 in 100 million shot by guessing.
I registered on a system a few weeks ago where it took me three tries
to get it right. They were using lots of distortion so that I could
not be sure on some characters.
Another problem you are likely to have is that you have a limited set
of photos.
Instead of a large photo set I'd have the script take a limited number
of different pictures and mix them up a bit each time. Take two or three
pictures of kittens and blend them together into one image and then cut
out a clip of that image randomly (of a given size and shape to
guarantee human viewers can see what it's of.
Bayesian filtering could be used to spot oddities in user's entering
data. Go through a bunch of known real logons and mark them all as
passing. Then open the site up and any patterns of interaction that seem
not to match using Bayesian filtering would be a red flag to temporarily
ban that IP. Easy enough to slow a bot down so that it'd take it years
to work through a few million possible guesses.
--
Michael McGlothlin, tech monkey
Tub Monkey
http://www.tubmonkey.com/
_______________________________________________
discussion mailing list
EMAIL:PROTECTED
http://mlug.missouri.edu/mailman/listinfo/discussion
Home | FAQ | Server | Presentations | Mailing Lists/Archives | Member Tools | Links | Sponsors | Contact