[mrbodoia]

Yes I agree. The potential for lying about your affiliation is in my opinion the biggest problem with this concept. I'll have to see how much of a problem this is in practice, and whether there are ways I could incentivize users to be truthful.

If it does become a problem, one thing I'm considering is using voting behavior as a way to try and identify and ban "false-flaggers". So if a user has classified themselves as left-leaning, but their voting behavior and the way other users vote on their posts/comments looks a lot like a right-leaning user, they could be banned or forced to change groups.

[cookiecaper]

You can give the user the option to self-identify, and then basically ignore it and use an internal index for the actual determination based on their voting patterns, etc. I think that's better than punishing people for "being open-minded", i.e., identifying as left-leaning but playing devil's advocate for the right, or finding their political leanings beginning to shift.

[laretluval]

If you're interest in incentivizing honesty, check out these papers on the Bayesian Truth Serum:

https://nel.mit.edu/bayesian-truth-serum/