> > You get shown 2 words.  The server knows what the first one is, the 
> > second is cut from the book it's processing.
> 
> Smart. Probably caters for greater irritations amoungst users though.
> Especially the ones that need to find the right keys on the 
> keyboard first, like dad.
> 
> The next captcha may be '/\/\33T C@t', since 'Meet' is now 
> known, but 'C@t'
> > isn't.
> 
> Still, only after 4 out of 5 turks have filled in the word 
> right. Maybe 4 out of 5 are illiterate enough to fill in m33t 
> instead of meet. This could pose to be a problem unless the 
> capcha software would recognize a 3 to be an E in the user's 
> input, which would obviously nullify any numerical input?
> 
> I love how forums have captchas on their search pages now.  Actually I
> > don't, it's bloody annoying.
> 
> 
> Now imagine having more than one capcha! Oh, what fun it is!
> 
> Sean


Those weren't meant to be numbers, just mangled text.

Eventually you'll end up with a majority of people agreeing on what a word
is, even if they spell it wrong.  If you don't get consensus after a while,
then the word gets flagged, dropped from the captcha, and referred to an
editor.  A small bonus is you get an accuracy index for each word.  Once
every word is done it gets proofread.

There was (and maybe still is) a similar method commonly used for handling
data entry years ago.  You'd get two people to type something in, then run a
diff on it.  A third person then went thru & cleaned up the errors.  To
increase accuracy you'd have more typists, as two people may make the same
error.

Tony

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist