8 ms·
Just last night I listened to something on NPR about CAPTCHA v3, which doesn't involve checkboxes nor pointing at parking meters anymore. It samples page activi
by rhizome 7y ago
Just last night I listened to something on NPR about CAPTCHA v3, which doesn't involve checkboxes nor pointing at parking meters anymore. It samples page activity and interactions by the user and supplies a human-likelihood score to the site (h/t Twitter bot scorers).
So we live in a future where our activity is bounded by a variety of quality scores. You're mostly a human, your site's increased traffic is less legitimate...every Taylorist-Goodhartian aspect of existence is priced into the marketplaces of society.
Perhaps a decent analogy could be Black Mirror's "Nosedive" episode, but with endless dimensions instead of just one (which admittedly could be a derived score from the dimensional ratings), and ratings that aren't derived from the moods of peers and other interactors, but passively gathered from everyday behavior.
I think this is the NPR show episode: https://www.npr.org/sections/money/2019/04/24/716854013/episode-908-i-am-not-a-robot https://www.npr.org/sections/money/2019/04/24/716854013/epis...
- kunglao 7y agoI like the way Qwant search engine deals with this. It gives a small challenge to solve which is a lot less cumbersome than finding traffic lights on photos or far less invasive than the fingerprinting technologies others use.
- megablast 7y agoYou haven't actually said anything about the challenge, so we have no idea if it is actually any different.
- kunglao 7y agoYou got to be kidding. "Qwant bot detection" on Google images shows you the challenge. For a change, try doing things for yourself.
- aeyes 7y agoSearched for this, including the quotes, the only result that comes up is this HN comment. Even without quotes nothing useful comes up. Would you be so kind to leave us a useful link?
- kunglao 7y agohttps://www.google.com.sg/search?q=qwant+bot+detection&prmd=niv&source=lnms&tbm=isch&sa=X&ved=2ahUKEwjbj7DbhNrnAhWs3mEKHX9LBNYQ_AUoAnoECA4QAg&biw=360&bih=575#imgrc=PmvAc_n9DZoLfM&imgdii=IOQ778TuHC7CyM https://www.google.com.sg/search?q=qwant+bot+detection&prmd=.... Usually the challenge i get is like may be 6 squares and one circle and I need to click on the circle. Apologies, I thought its easy to look this up because I have nt used Google search in a while and when I just searched this just now it's the first result I got on my images section.
- Wowfunhappy 7y ago...surely you could write a computer to click circles relatively easily?
- kunglao 7y agoLink to the actual image: https://images.app.goo.gl/HgaapxCmCdiTrRKw5 https://images.app.goo.gl/HgaapxCmCdiTrRKw5
- vpzom 7y agoLink to the actual image: https://pbs.twimg.com/media/D0sh3EKWkAE1t-n?format=jpg https://pbs.twimg.com/media/D0sh3EKWkAE1t-n?format=jpg but actually this one is better: https://twirpz.files.wordpress.com/2018/08/qwant-grabs-ip-addresses.png https://twirpz.files.wordpress.com/2018/08/qwant-grabs-ip-ad...
- jakecopp 7y agoI wish there was a challenge that helped verify OpenStreetMap edits!
- flukus 7y ago> Just last night I listened to something on NPR about CAPTCHA v3, which doesn't involve checkboxes nor pointing at parking meters anymore. It samples page activity and interactions by the user and supplies a human-likelihood score to the site (h/t Twitter bot scorers). That turns the web into a walled garden you can't access without submitting to google's tracking. If they'd just drop the pretense of not being a walled garden this wouldn't even be a problem that needed solving.
- DrScientist 7y agoSo you train a bot to show 'human like patterns'. The problem with relying on ML on one side is it can be gamed by ML on the other.
- rhizome 7y agoIf it was easy to train a bot to show humanlike patterns, we'd have driverless cars by now.
- DrScientist 7y agoHmm... I think training a program to fool another program is a much, much, much, much easier problem than driving a car. As an example, you only have to learn how to emulate a single 'human like' pattern to fool the ML, where you target function is a simple fool or not fool ML - once you achieved fooled you are done. Whereas to drive a car you need to deal with all the infinite variation in inputs you might experience in driving a car, with complex target functions ie safety versus speed.