Looks like I am not alone in this. A youtube playlist consisting of the entire soundtrack of the game used to be my go-to for programming/working/studying during undergrad.
Do not recommend main bus design to new players! The most fun I've in Factorio had was getting to a rocket blind and dealing with my spaghetti factory. Even mentioning any kind of pattern to new players should be avoided.
I avoided a main bus after noticing nearly all the copper went to producing circuits and low density structure, and similarly most iron for steel and green circuits - so just site those close together linked via short buses [1]. Then ship the remaining copper, iron and steel to the Science factory using single wagon trains. (That’s just for a simple one rocket per hour factory)
I always wonder how often people in charge of massive compute tasks like LLM pre-training have messed up some detail that invalidates or fails to persist the results and only realise afterwards.
Berlin has hundreds of clubs, half of which are constantly about to close. But berghain has sufficient clout it will remain for decades even if it continues to be mediocre and hostile to its customers.
Yes, ChatGPT 5.2 Pro absolutely still does this. Just ask it for a pivot table using Polars and it will probably spit out code with Pandas arguments that doesn’t work.
The fact they pivoted away from their very compelling core offering (AI stack overflow) to complete with loveable etc in the "AI generated apps" giant fight continues to baffle me. Though I guess model updates ate their lunch.
My guess is that their pivot came after distress, and was not the cause of it. It'd be great to have @rushingcreek write a post-mortem. I think it'd benefit a lot of people because I honestly don't have a monday morning playbook of what could have saved them.
Like you said, perhaps the demise of phind was inevitable, with large models displacing them kind of like how Spotify displaced music piracy.
I did look at instructor and probably for structured output pydantic-ai and instructor are about the same, but pydantic-ai supports a ton of other stuff that isn't part of instructor's feature set. For me killer apps were the ability to serialize/deserialize conversations as json; frictionless tool-calling; and ability to mock the LLM client for testing.
Have you considered whether this is because of the local weather tendencies? I visited the Netherlands once in summer near Amsterdam, and there were short bursts of rain frequently throughout the day. I'd imagine weather patterns like that are not well represented in global weather apps lowering the prediction accuracy in such regions.
As far as I can tell, it's largely a data or modeling problem.
Weather radars exist in most of Europe, including all of Western Europe. What's missing is somebody actually getting that data (which is not open in all countries) and feeding it into a short-term precipitation prediction model like Dark Sky (now Apple) does in a few regions.
Currently, in regions where it's not available, I just do that manually: Look the rain radar data for the last hour or so and extrapolate linearly based on wind direction. You'll miss rain that's just starting or stopping or sudden changes in wind direction, but it's much better than what many apps offer.
reply