Are there any other topics in that list that look particularly confusing to you about their purpose? Any reason why this would be the odd one out

There is no confusion, nor any indication that Ukraine is an odd one out since we don't know the purpose of this list.

When analyzing a code base you can make some educated guesses based on naming but generally you want to check those guesses by seeing how, where and why a variable is used.

If a very specific claim of down-ranking is made I expect to see a down-ranking function, short of that it's a list of names that might be used for bot detection, for manual review of the tweets, for requiring the user to add their phone number to their account, it might come along another map factors that down or up weighs those topics, I could go on...

I mean take a look at the names.

DoNotAmplify, CoordinatedHarmfulActivityHighRecall, UntrustedUrl, MisleadingHighRecall, NsfwHighPrecision, NsfwHighRecall, CivicIntegrityMisinfo, MedicalMisinfo, GenericMisinfo, DmcaWithheld, HatefulHighRecall, ViolenceHighRecall, HighToxicityModelScore, UkraineCrisisTopic, DoNotPublicPublish

What would you suppose the list is for? Almost everything on that list is something they have publicly stated they are trying to reduce. I don't think we have to play dumb here and say "it didn't specifically use the word downvote therefore we can't possibly know what it is for"

This is reverse engineering, not divination. I just gave you a number of other algorithmic functions that would be just as likely to use such a list.

> you can make some educated guesses based on naming

This is what people are doing.

> but generally you want to check those guesses by seeing how, where and why a variable is used

Agreed, but without the full source code available, this simply isn't possible.

That's fair which is why I'm asking if a piece of code has been published supporting the claim of down-ranking.

If it hasn't the answer is to very carefully qualify the statement "Analysis of Twitter algorithm reveals Twitter down-ranks tweets about Ukraine" rather than treating the people who question it as heretics.

I agree. It shows the danger of selectively releasing code and how the "Twitter open-sources algorithm" rhetoric is deeply misleading, at best. Hopefully, any misplaced analysis of the code will result in greater transparency.

Yeah it reminds me of everyone scrambling to draw conclusions from the leaked Yandex SEO algorithm, even though all the models were missing...

> you can make some educated guesses based on naming

> This is what people are doing.

no, they're reporting on it as fact