bogzz parent
compilers aren't probabilistic models though
True. The question is whether that's relevant to the trajectory described or not.
Successful compiler optimizations are probabilistic though, from the programmer's point of view. LLMs are internally deterministic too.
What? Do you even know how compilers work?
Are you able to predict with 100% accuracy when a loop will successfully unroll, or various interprocedural or intraprocedural analyses will succeed? They are applied deterministically inside a compiler, but often based on heuristics, and the complex interplay of optimizations in complex programs means that sometimes they will not do what you expect them to do. Sometimes they work better than expected, and sometimes worse. Sounds familiar...
> Are you able to predict with 100% accuracy when a loop will successfully unroll, or various interprocedural or intraprocedural analyses will succeed?
Yes, because:
> They are applied deterministically inside a compiler
Sorry, but an LLM randomly generating the next token isn't even comparable.
Deterministic complexity =/= randomness.
> Yes, because:
Unless you wrote the compiler, you are 100% full of it. Even as the compiler writer you'd be wrong sometimes.
> Deterministic complexity =/= randomness.
LLMs are also deterministically complex, not random.