tarheel

- Chinese LLMs have historically not competed

- Many of the top AI researchers (inc. Chinese) seem to be at GDM, OpenAI, Claude, Anthropic

- Beating OpenAI models will require a massive talent advantage over OpenAI and GDM

- Availability of compute could help (e.g. if the CCP put all its eggs in one model basket), but even the hardware advances being made by GOOG, MSFT, and X seem to be hard to beat.

-  More likely that X beats the incumbents than a non-US company

- Chinese tech companies generally do a good job replicating, so I think they can broach the top five. But as long as OpenAI, Anthropic, and Google are competing, I find it hard to believe.

- ChatBot Arena will inevitably be biased by user experience and "design considerations" rather than purely engineering prowess. I think this favors the American incumbents even more

- Putin could lose the election. This is highly unlikely because (i) he "won" with 78% of the vote in 2018 (indicating either popularity and/or control of the ballot box), (ii) Navalny seems to be the only unifying public figure in opposition and he is barred from running (i.e. there's no other politician to rally around)

- Putin could be ousted. This is also unlikely since Prigozhin failed. For one, he had the best chance of success at ousting Putin (serves as a sort of upper bound). His failure also probably has a chilling effect (soldiers and generals more likely to think twice before disrespecting Putin). On the other hand, the war is dragging on and soldiers may be getting more antsy and agitated. Even so, it strikes me as improbable.

- Putin could die. He could be assassinated by the Ukrainians, western forces, or someone high up in the cabinet (oligarchs must be annoyed at their lost prestige and business opportunities). This strikes me as the most likely way for