I explored a startup idea that really didn't make sense unless there was a way to ensure users were unique humans in an anonymous and privacy preserving way.
Researched it substantially and realized it's an unsolved problem. Anything that makes a dent is incomplete and comes with ugly tradeoffs. For a time I wondered if I should try and solve it myself, but I could never think any solution that hadn't already been/being tried. Years later I'm left curious if it's even possible to solve the problem.
My point is that captcha won't solve this, and solving this problem is a lot harder than it seems at first, and might not even be solvable (which I know is hard to accept).
If someone does find an elegant privacy ensuring way to solve it, I think the impact would extend far beyond HN and could make a big difference to the future of civilization as a whole.
Having to invite people in person and maintaining a network of trust could work. There would always be people ignoring friends selling accounts to bots, but ultimately I guess it would be mostly too costly.
https://eu-digital-identity-wallet.github.io/eudi-doc-archit...
It is a solved problem with ZKPs
I studied ZKPs. You and I must have very different understandings of the actual scope of the problem. Like I said, everything that exists or is being worked on that makes a dent (including ZKPs) is incomplete or has ugly trade offs.
Maybe you are thinking purely from a math / theoretical perspective, but I'm thinking of a compete solution that's practical to use to solve the problem for sites like HN and many others.
That is true, and I'm also working in the industry.
We are currently in dire need of ZKP providers / abstractions that will aggregate both eIDs and tradition IDV (scans).
However it's extremely important that the technical capability is there. The next step is to build upon it, and I think that it's a great time for something like this due to all the horrible implementations of identity verifications currently available — that is 3rd party companies collecting scans of documents and biometrics.
Yes, I'd thought about roughly the same premise before and came to the conclusion that it really is a hard thing to do.
Even if you use state ids for it, who's to say that a particular state won't be...loose with issuing ids that can then go on to be used for bots.
It's even a problem with humans as well - one human can be having a pleasant conversation with the other, not aware that that person isn't being genuine, or is lying, has ulterior motives or has been instructed on what to say by someone else.
[dead]