This makes a lot of sense, but it might take some getting used to. I think people are more self-critical and embarrassed about their coding ability and mistakes than they should be, and they might worry about all that being public.
this post was submitted on 27 May 2026
14 points (79.2% liked)
Opensource
6444 readers
56 users here now
A community for discussion about open source software! Ask questions, share knowledge, share news, or post interesting stuff related to it!
⠀
founded 2 years ago
MODERATORS
I don't see it catching on at all... I think forcing people to record themselves is asking way too much, and not everyone uses the command-line or the same application/machine for everything. Plus they acknowledge that it can be faked and that LLMs can/will get better at faking it.
Proving a request came from a human is a many-billion dollar task right now, and nobody seems close at all to coming up with a workable solution. The only thing even remotely feasible (IMO) that I've seen so far (and which still doesn't actually prove anything) is charging money for accounts. But tons of people still don't want to do that, even though it would lessen the amount of bots and make it easier to ban bad behavior if everyone needed an account to access a site. I don't like that option, but it is an option. Making payments on the web easier (and less tied to the whims of Visa et al) probably also needs to happen.
Making payments on the web easier (and less tied to the whims of Visa et al) probably also needs to happen.
Yeah, this has stopped me from buy stuff quite a few times. Also why I don't have a Proton account.
Isn’t this literally what facebook is requiring for their employees that even facebook workers won’t tolerate? That makes me think data like this is already valuable to AI companies, which is an additional “no” factor on top of the privacy aspect of typing behaviours being able to identify people across git forge accounts
this is so stupid. I develop in a full gui IDE. use Kate for everything else, both have a terminal embedded.
how is "recording vim" is proof... you could just have a AI agent wrapped around a stdout to another ai agent running vim... that would instantly defeat this...
this is a searching for a solution where there is no real problem... sure some AI slop is still going to be submitted. do the same thing as any other industry would. inspect the code, if it works and can be maintained, accept it. if not, say why...