Hacker News Reader

> 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.

So 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.

If it was easy to train a bot to show humanlike patterns, we'd have driverless cars by now.

Hmm... 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.

I wish there was a challenge that helped verify OpenStreetMap edits!

I 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.

You haven't actually said anything about the challenge, so we have no idea if it is actually any different.