Tolga

Extremely unlikely he loses office in this period

DeepSeek is impressive, and they are current 7th on the leaderboard. There are concerns that at some point this leaderboard becomes less useful as an actual measure of AI capabilities (the optimization goal is human preference), and this would make it easier for lower-resource orgs to get a higher ranking.

I think there's a reasonable chance DeepSeek or another Chinese org could get to 3rd. I'd put about 1% on getting to first.

Another reason why it's not so low is that they only need to hold it temporarily, and product release cycles for AI models can be lenghthy, so it could be that the currently leading AI companies/labs don't publicize a new SOTA model for a while, and DeepSeek sneak in.

Updating upwards based on @Revmag's analysis of the current data so far (shifting distribution right)

Didn't happen.

I think this is a bit more likely, given the fall of the Assad regime, which I think probably slightly increases the potential for conflict between Iran and the US, but with very low confidence on that.

In general I think this becomes more likely over time, so I expect to slowly update this upwards - not on any specific events that have occurred.

Xi Jinping said in his New Year's speech that "no one can stop the historical trend of reunification", which continues to be a priority of China's government. But this doesn't seem like a change in position, or a particularly surprising thing for him to say.

Looks extremely unlikely, but I think Trump's incoming presidency makes this more likely: it could plausibly be a policy priority in the Middle-East.

Furthermore, the fall of the Assad regime means the Middle-East is just higher variance - there is more possibility for outcomes that seem impossible now to happen.

Perhaps one could be signed between Israel and Syria:
https://www.npr.org/2024/12/27/g-s1-40144/syria-israel-relations-hts-damascus-governor