Should political social media posts be captcha’d?

Since "references to politics" is an undefinable, and beyond the realm of computing to identify even if it could be defined, you're essentially asking for the impossible.

Your actual question is really -- "Should all social media posts be CAPTCHA'd?" Because that's the only way this could work on a technical or logical level.

I'm a big proponent of anti-bot measures on platforms, but you have to balance that against reasonableness. CAPTCHA-challenging at random intervals with higher frequency with new or dormant accounts would be ideal.

I don't considering trying to identify what is "political" as a worthwhile use of time or resources (least of all because no two people would agree on a definition e.g. COVID-19 political or not? Highways political or not? Wild fires political or not?).

Good idea - but why? If the purpose is to make sure someone's human, bots are pretty good at cracking captchas. If it's to slow someone down, help them think before they post, then fine, but why a captcha? They are really annoying. Why not make them listen to https://www.youtube.com/watch?v=w_DKWlrA24k all the way through?

Even a hypothetical "perfect" captcha that resists 100% of bots can still be defeated by just paying humans to solve them. If your objective is to defeat political interference by well-funded adversaries it will do absolutely nothing.