But once OCR is improved to the point where it can read poorer scans, won't spammers be able to use that new technology to eventually defeat CAPTCHA?
If you look at CAPTCHAs as a method for improving artificial intelligence, then the issue you're raising gets turned into a benefit of ever-improving Turing tests.
Find a problem that AI cannot readily solve, yet humans can (obfuscated word recognition).
Develop a CAPTCHA out of the problem (reCAPTCHA).
Either the problem remains unsolved and the CAPTCHA must be solved by humans...
... or we develop stronger AI capable of solving the original problem.
In either outcome, we get something useful, either better Turing tests or more robust AI. The hardest part of this loop is developing novel CAPTCHAs, but at least the cycle results in useful outcomes every time.
That's very a interesting angle on the whole thing that helps explain some broader trends in the contest. I went to CMU, and even though I was excited about their entry, I always had nagging doubts about the general design of our entry. I was surprised there were so many automobile-based designs. Since there weren't any restrictions on the vehicle shape per, you'd think there would be designs with more stability than a four wheel vehicle. I mean, even though a Hummer is ridiculously stable when it comes to cars, they flipped the thing over during the pre-race testing the first year resulting in last minute wound tending. Where were all the track based vehicles? Where were all the non-traditional form factors that can't fall over, like tracked spheres or other Platonic solids?
You've got the economics all backwards. With digital distribution, the normal laws of supply and demand don't work because supply is infinite. While there is some cost associated with digitizing and delivering each song, it is basically constant whether you're providing a thousand or a billion songs. That's one sticking point for why everything in the analogue catalog isn't provided digitally. If there's only going a couple people who ever buy some obscure song, why waste time digitizing it when there's so many more popular ones to offer instead?
What you need is some incentive to offer the unpopular songs, and that's a pricing scheme that is opposite to the one you proposed. By charging $3 for the obscure stuff, the vendor knows that if anyone ever buys it, they can recoup the expense of hosting it. Conversely, the insanely popular stuff should only cost pennies, because they know the volume of sales will make up for the low price.
This is the site administrator for midnightspaghetti.com. Thanks for all the traffic! The site's back up (we purchased more bandwidth). Try not to go too crazy with the hits...
Slashdot Mirror
But once OCR is improved to the point where it can read poorer scans, won't spammers be able to use that new technology to eventually defeat CAPTCHA?
If you look at CAPTCHAs as a method for improving artificial intelligence, then the issue you're raising gets turned into a benefit of ever-improving Turing tests.
In either outcome, we get something useful, either better Turing tests or more robust AI. The hardest part of this loop is developing novel CAPTCHAs, but at least the cycle results in useful outcomes every time.
Hi, I'm selling these fine leather Wiimote covers.
Okay, I think we can close Slashdot now.
That's very a interesting angle on the whole thing that helps explain some broader trends in the contest. I went to CMU, and even though I was excited about their entry, I always had nagging doubts about the general design of our entry. I was surprised there were so many automobile-based designs. Since there weren't any restrictions on the vehicle shape per, you'd think there would be designs with more stability than a four wheel vehicle. I mean, even though a Hummer is ridiculously stable when it comes to cars, they flipped the thing over during the pre-race testing the first year resulting in last minute wound tending. Where were all the track based vehicles? Where were all the non-traditional form factors that can't fall over, like tracked spheres or other Platonic solids?
You've got the economics all backwards. With digital distribution, the normal laws of supply and demand don't work because supply is infinite. While there is some cost associated with digitizing and delivering each song, it is basically constant whether you're providing a thousand or a billion songs. That's one sticking point for why everything in the analogue catalog isn't provided digitally. If there's only going a couple people who ever buy some obscure song, why waste time digitizing it when there's so many more popular ones to offer instead?
What you need is some incentive to offer the unpopular songs, and that's a pricing scheme that is opposite to the one you proposed. By charging $3 for the obscure stuff, the vendor knows that if anyone ever buys it, they can recoup the expense of hosting it. Conversely, the insanely popular stuff should only cost pennies, because they know the volume of sales will make up for the low price.
This is the site administrator for midnightspaghetti.com. Thanks for all the traffic! The site's back up (we purchased more bandwidth). Try not to go too crazy with the hits...