Re:$1.39 for 1000 decoded CAPTCHAs
by jimktrains2 on Friday April 22, 2011 @04:49PM
(#2475392)
Only if those partials don't overlap. If your 70% solution also blocks the same 50% then it's not .5*.3 it's just .3
The assumption was that the solutions were chained and that they would never see the things that failed before that. But I guess I only stipulated that the failures were independent, when the math I wrote meant that the successes should be independent, as well.
You can also join with others who work to stop this kind of spam. For example, Project Honeypot ( http://www.projecthoneypot.org/ ) offers honeypots to trap all kinds of spammers, including comment spammers.
If you're expecting some kind of 100% solution to spam, I doubt that will never happen. The best anyone can do is combine a bunch of decent solutions, preferably in unique combinations, and hope to reduce spam to a manageable level. If one countermeasure blocks 50% of spam and another 70% and another 90% and their failures are independent, you're down to 0.5 x 0.3 x 0.1 = 1.5% of spam getting through. Chain enough partial solutions together and you get something better than any one alone.