> LLMs... impedes the ordinary process of theory-building
As I have said on here before, I actually really enjoy LLMs for code understanding precisely because they are imperfect.
They're good enough to point you in the right direction, they might even see something you miss, but you can't trust anything they say.
This forces you to deeply understand what you're looking at, and to constantly re-evaluate your mental model against the code you're reading.
This works the same way as "writing forces clarity".
I could get to the same place on my own - I did this for many years before LLMs existed - but I feel when an LLM works well it gets you to the same place (or better) quicker.
(as idiomatic code generators I have mostly found LLMs to be poor, even Opus, but this comment isn't about new code generation)
Other engine simulators work by approximating the engine.
Ange's engine simulator works by approximating physics of air fluid dynamics through a combustion chamber and exhaust, sound propagation, etc and then putting an engine into that simulation.
I still prefer Mistral Nemo 12B for text summarisation tasks. It has a nice style. The Mistral Small 24B is also decent. I have a YouTube transcript summariser which I like these for.
However these days I usually have Qwen 3.6 27B already loaded so I mostly just use that instead.
There are many other posts here which agree with you. Filling context with what you think the model needs adds nothing and possibly just inflates context which is harmful.
A good method seems to be only make a skill or memory when the LLM gets something wrong, or if you actually observe it's always doing the same step and you can get the model to the same place with less tokens.
Depending on what container you are running, sometimes you need to fiddle with passing volumes/devices/permissions/groups/capabilities to the container.
If you are just running normal userspace programs then it's pretty seamless.
Podman should support all the same arguments as Docker, and you can install `podman-docker` on Debian which creates an alias:
An organisation which lets a person release a website without understanding how the web works is just asking for trouble. This was true before Transformer LLMs and is true after them.
LLMs probably just let those organisations make a larger and more load-bearing website faster, so that poor decisions speed the time to catastrophe.
The architecture which made me fall in love with RISC (or "load-store" if you prefer) and see the error of my ways with x86 (insert your derogatory term here).
When doing this, I particularly like that the LLM sometimes gets things wrong.
It forces me to really understand each thing deeply so that I evaluate it properly.
It is like taking an exam where the exam writer is hostile and sneaks in trick questions. You only spot that the question is wrong when you fully reason through the answer.