Do you run the models locally?

No local model for me manages to get function calling right.

Are you using LangGraph tool nodes? I run very small, non-RLHF instruction models with maybe a 2% failure rate on response format matching tool definition. I would also guess you do not have your orchestration and pipe configured correctly.

Doubtful it is anything to do with simplicity.

Python's success is explained by it being the language of choice for AI.

Python has been massive since the 2000s. When AI rolled around, it was already there, a bunch of people knew it, and it was Good Enough (tm).

I think its the other way around. Python became the language of choice for AI because it was already popular. Lots of things made it popular: use for systems management scripts, web apps (Django, especially), then numerical stuff,...

I think the reason is that it is easy to learn enough to get things done, but it is very flexible, very readable, and once the ecosystem started gaining momentum (which it clearly had by the time of the XKCD cartoon) that became an advantage too.

"Turkey" is what most English people would use, which makes it the defacto official name, despite what the UN might say.

Most English people aren't even able to type ü on a keyboard.

FPTP forces coalitions to form before the vote, as otherwise they never get power.

In alternative systems, you vote and then coalitions jostle to form a majority afterwards.

Do you have any information e.g. blog posts on this pattern?

It reminds me of the proposal to shake hands at the end of Goldeneye:

> Miyamoto, with a series of suggestions for the game. “One point was that there was too much close-up killing – he found it a bit too horrible. I don’t think I did anything with that input. The second point was, he felt the game was too tragic, with all the killing. He suggested that it might be nice if, at the end of the game, you got to shake hands with all your enemies in the hospital.”

https://www.theguardian.com/technology/2015/oct/26/goldeneye...

One of my favorite childhood video games is 8 Eyes for the NES - after beating each boss, the player character sits down with them at a table, and a little skeleton butler walks over and serves them tea. The little scene plays over and over and over, you and the defeated bad guy, sitting at the table, sipping tea, while a skeleton wanders over and offers a periodic refill.

I always thought that was nice.

"Ireland also has relatively little money in politics, limits on donations..."

There are a crazy amount of NGOs in Ireland, 1 for every 155 people, many pushing forward their own political policies and views.

https://unherd.com/newsroom/in-ireland-its-progressives-who-...

With virtual threads it's difficult to see WebFlux being used in new projects.

I agree. Unless you actually need back-pressure, which almost every use case I've seen does not, it's just obfuscating simple API calls.

According to mongodb it is acid compliant.

https://www.mongodb.com/resources/products/capabilities/acid...

Who can remember clippy right click "animate"?

That, wordart, and the secret flight sim in Excel 97 were the entirety of how I spent my school days.

What, no Encarta? :P

We did not have encarta at our school, that was for the rich kids.

However I have great memories of playing the sample of 'Changes' by David Bowie when I got a bit older and had access to a copy.