It’s still 20W. We have living proof what is possible within 20W. The message has always been clear - try to get silicon computer to be as power efficient as the brain is as it is obviously possible.
> The person who specified this most likely was aware that this is turing complete and it's the rules author's responsibility not to write infinite loops
Software developer’s job description: prove that this instance of the halting problem you’re being paid to work on today is decidable
> The first language I could "just write" in was C. I had internalised the language and its standard lib and didn't need the internet to work with it.
I bet $4.20 you didn’t write C. You wrote something C-like which the compiler didn’t reject because the C standard has a gigantic surface of ‘undefined behavior’ which means once your program does one thing out of spec it isn’t C anymore silently.
The industry accepted way of handling circular dependencies is to not have them and heavily lint against them in languages which permit them in compilation or runtime.
It follows directly from the bitter lesson - a frontier model can be relatively cheaply distilled into anything you need to run quickly (and a frontier model like Mythos will help you distill it quickly), decidedly not true the other way around.
> I'm now considering the architecture of the service
What you see here is a summary of thinking tokens written by some other smaller model (e.g. old sonnet). The actual thinking sometimes (rarely) leaks and is not easy to parse.
> software engineering does not involve mathematical modeling
it absolutely can, approximately nobody was doing that because it was insanely expensive. if we narrow down the definitions, modern static typing (where modern means universally accepted nowadays) is a form of mathematical modeling and proof construction that software does what it says it does.
the economic calculation is changing extremely rapidly now with LLMs though. some of my software is now proved to be correct at some levels, e.g. I heavily (that is, LLMs I pilot) use TLA+ for tricky but nowhere near foundational distributed systems work (as in, I don't work on core S3, but do distributed transaction stuff).