On 15 Dec 2002 at 9:44, James M. Ray wrote: > Fascinating idea, Danny. I am personally bad-enough at math > that any transmogrify-algorithm would have to be too simple > to be much good, but I've been doing some reading on the > subject. It seems that humans are relatively-bad at recalling > alphanumeric strings (I sure am!) but are good at recognizing > shapes. For example, we all have a mental catalog of dozens > of *very* similar faces of people that we "know." (If you're like me, > you don't recall the names, in many cases.) Maybe the thing to present > to users would be ten random face-pictures, three of which they "know" > appearing in random spots, in an image with random names? It would > take more bandwidth, and (like all my ideas!) actually-programming-it > would be a nightmare, and there'd probably be awful privacy-issues too > if people don't want their images appearing on the web, but some > variant of this but with shapes might be do-able. JMR
Three faces out of ten gives one chance in a hundred of getting it right, which will lead to scripts that guess until they get it right. Assume the user has to select five correct faces, out of twenty five. The chance of getting this right by random chance is two in a hundred thousand, still disturbingly high, though perhaps acceptable if combined with software that detects guess attacks, and provides longer and longer delays before rejecting guesses. --- You are currently subscribed to e-gold-list as: [email protected] To unsubscribe send a blank email to [EMAIL PROTECTED] Use e-gold's Secure Randomized Keyboard (SRK) when accessing your e-gold account(s) via the web and shopping cart interfaces to help thwart keystroke loggers and common viruses.
