How can an algorithm determine whether content is political or not? Sure you can naive filter for a list of political urls but it won't catch most of them.
On the other hand, we think that we can't effectively classify the majority of posts.
can be easily done by one person, and with a fairly good level of accuracy and key-word tagging. emphasis on "fairly good", but diminishing returns and all that.
For starters, cross-referencing it with popular propaganda talking points. Smear campaigns are a dead giveaway, and adding two or three topics should be enough to cover the majority of propaganda posts.
I'd argue that just having an effective anti-bot policy is enough to render this a non-issue, but apparently Elon Musk prefers Twitter to be a Kremlin bot sandbox.