> TL;DR trick bots into revealing that they're not humans, instead of
> treating users as guilty until proven human.

I read the article, and the ideas do seem like they could replace CAPTCHAs in 
sign-up forms indented to prevent spam, but these are the least annoying kind 
of CAPTCHAs. Barring accessibility issues, the CAPTCHAs in forms aren't much 
more annoying than the form istelf.

The worst CAPTCHAs are the "select every square containing a sign. okay, now do 
vehicles. whoops, you didn't notice those tiny cars in the corner; let's go 
back to signs." bullshit. They can take up to a minute if they make you solve 
three of them, and I have no idea what vision-impaired users are supposed to 
do. I've never encountered one of these in a sign-up form, but the number of 
sites that require them just to *access* the site seems to be increasing (I 
don't have a statistic for that, it's just my impression and could be 
inaccurate). Since these CAPTCHAs don't prevent spam, my guess is that they are 
to improve the credibiity of analticks* data.

I suspect that a human-centric Internet wouldn't be as profitable as an 
Internet in which AI are the users and humans are the data served to the AI.

*stealing that term from you

Attachment: signature.asc
Description: PGP signature

Reply via email to