A Kind of Hiatus

Or unconditional ceasefire

I’m thinking of putting this project on hiatus, but I also think I owe my last few remaining readers some kind of explanation and farewell. So here it goes.

When I started it, my basic premise was that until 2022 many people in the West had little interest, and thus little knowledge of Ukraine. This is understandable, for similar reasons I have little knowledge of Africa or Latin America.

If Brasil invaded Uruguay, and some guy from Paraguay started a blog explaining Latin American politics from a local perspective, I would surely read it. And since I know zilch about it, I would rather not express my expert opinion on Latin America, based on what I heard on Joe Rogan podcast.

We live in the age of unreason, with the Internet as a huge disinformation machine. In theory I knew that, but I was genuinely surprised on how bad it is.

This is a screen of a Bluesky user who claims to be from Brasil. Their identity is hidden, because I don’t want to blame anyone, this is just yet another example of someone who clearly lacks basic knowledge - you really need not to watch Russian media to deny Russia bombs Ukrainian cities - and yet he is very opinionated and even given a chance to talk with someone who has more information, he will reject it, because he prefers to believe what he believes.

I guess I overestimated curiosity as a motivation factor. Most people are not curious. They have some pre-existing opinions and avoid anything that could challenge them.

This seems to be somewhat amplified by two fairly recent trends. One is podcastification of discourse. It seems that if you want to reach a larger group of people, you need to talk, not write.

I prefer written expression - both as a content creator and a content consumer - because when you write, you usually have the time to check your notes before hitting “PUBLISH”. A podcast is usually a stream of consciousness without any notes. But our memory is playing tricks on us, so even if you are a bona fide expert on something, you might still make a honest mistake.

A lot of misinformation regarding Ukraine (“the coup d’etat”, “Boris Johnson forced them to reject peace deal”, “Angela Merkel admitted she signed Minsk agreements in bad faith”), etc., is spread by podcasters. Some of them are lying on purpose (it’s a quick way to make good rouble, just ask Tim Pool!), some simply never cared to google the basic facts and dates.

The worst thing about it that people frequently develop a cult-like trust in their favorite podcasters. This is a screen from a guy apparently unable to consider the possibility that Jeffrey Sachs might be wrong about something. Apparently, he doesn’t trust Wikipedia (good! critical thinking!), but he trusts Sachs to the point he’d gladly drink any Kool Aid if his tells him to do so on his favorite podcast. You can’t argue with that.

On the top of that, people are more and more reliant on Large Language Models. Even the “smart ones”, who should know better, are “quoting AI” in discussion, which is of course even worse. LLM don’t tell you “the truth”, they try to predict what kind response you will find true. They hallucinate a lot of stuff, especially when there was not enough English language material to train them.

When experimenting with them, I discovered that while they can quite well tell one Shakespeare’s play from another, they write complete nonsense when asked about Dostoyevsky or Mayakovsky. They produce something that looks like stereotypical imagination of Russian literature drawn by a Westerner who does not speak Russian and actually never read any of it.

For a similar reason, LLM/AI might give you fairly correct answers on American political party system (they aren’t that many parties in Congress!), but don’t trust it on Ukrainian politics. You shouldn’t! It’s extremely multi-party and complex. The foreigners see only Zelensky, but then again, how many politicians from Paraguay or Uruguay I could name (without checking notes)?

Anyway, if you use LLM/AI as a “source” to answer such questions like “is Russian speaking population oppressed in Ukraine”, or “did Zelensky ban all opposition parties?” or “are Christians persecuted”, the answer will be misleading at best, but depending on the model you use - it will be more or less tailored to your belief. These models don’t want to show you “the truth”, but “what you will most likely accept as the truth”, meaning that if they have you marked as a fan of Real Madrid, they will tell you that Cristiano Ronaldo is the best that ever was.

Combine it with the fact that LLM’s are easy to fool (“groom”) - and Russia is already putting a lot of effort to poison them with pro-Russian content - and the conclusion is clear. If you use them as a “source”, you’re delusional. And of course, I’m seeing more and more people in social media, using LLM as the ultimate argument: “LOOK, AI PROVES THAT I’M RIGHT, RUSSIAN LANGUAGE IS OUTLAWED IN UKRAINE!”.

I feel powerless. Stupidity is taking over the world. I give up writing longer forms. I will try to write something within the 300 characters limit on Bluesky as eastsplaining.bsky.social - but not sure for how long. As long as I’m there, I’m happy as always to answer all your Eastern European questions, and perhaps you could finally explain me Uruguay and Paraguay (I think Rome should invade to protect the Latin speaking population…).