It doesn't "pick" anything. It produces the most likely number after this question based on the data it has been trained with! Reasoning models might pick in a sense that they will come up the the rules (like the grand parent post shows), but still it will produce the "most likely" number after the reasoning.

They can't be random, that's not how a stochastic model produces tokens. Unless the models in question are using a tool call for it, the result will very likely carry bias

You just went and created the worst example. The model knows how to create an rng, that's not it weakness. In fact, if you give it a random mcp it won't do that.

Well, yeah! It's a probalistic model, and extremely biased - it has to be, so that it can predict the correct token.

There's no "just" in RL. Fine tuning is very important and could make a lot of difference.

Indeed, this is quite obvious on Claude models vs Gemini. I fully believe Gemini is more powerful model, but the post training process is nowhere near what Anthropic does, which results in Gemini being horrible at coding sessions, while Claude is excellent.

apparently GPT-5 uses the same pretrain as 4o did, hah

Same here. My take is that the codebase is too large and complex for it to find the right patterns.

It does work sometimes. The smaller the task, the better.

Isn’t that fixed by having it create a plan, then you review it and say “x should do y instead”, it updates the plan, iterate then “build the plan”?

Same argument(s) can be applied to age verification.

Same in Brazil. Economically and politically not nearly as important, but 250 million people affected by the same discoursem

Why not all of the above? Reducing costs and improving quality.

> no matter how small a component already is, the single-responsibility principle can still be applied: every line of code can be assigned its own responsibility

The definition of SRP is to have each class (or module) to have a single reason to change. I don't see how that has anything to do with having each line be assigned a responsibility. If the line changes for the same reasons as it surrounding lines, then, they are part of the same component (to use the author's wording). My guess is that the principle is being taken literally from its name/acronym.

"Reason" is an "in the eye of the beholder" type human thing. They're taking it in the most tortured sense, because under sufficient pressure that's "exactly" what happens anyways. It sounds silly until everything you touch is 20 indirections away.

> It sounds silly until everything you touch is 20 indirections away.

Which is how your standard Uncle Bob inspired codebase winds up looking

More to the point the definition of responsibility is ambiguous and rarely shared.

The SRP is a bit like the original agile principles: the intent in writing them down was good and they definitely alluded to something real and valuable but the actual wording is vague enough to allow almost anything - including the exact opposite of the original intent.

SRP doesnt need to be tossed away, just redefined more tightly.