Did we read the same article? I mean it wasn't talking about a specific line of code, but it had deep architecture details, rational for memory safety and bugs, agentic coding patterns for scale. It was full of substance, multiple times.
Why was it in their own words a real option? And why did they not go for it? This is the technical substance I'm looking for: an engineer explaining what options have been considered and what wanted and unwanted tradeoffs they present.
Also only mentioning figures for the platforms that saw an improvement is sketchy. "With ICU and code folding Windows and Linux get 20% smaller", what about macOS? Why did it not see the same gains? The fact that they don't mention it makes me think that they don't KNOW, and isn't confidence inspiring coming from the engineer who SHOULD know.
You can argue that that's a bad justification for not doing it, but that's a debate on the technical merits, not a claim that they didn't provide justification.
Why is just a wrapper type "less ergonomic"? Everybody does that all the time in every language. And then if you can convince me it is less "ergonomic": why does it matter if it solves the problem? And that is considering humans write the code manually. Why does it matter if agents write the code? This makes no sense.
> And that is considering humans write the code manually. Why does it matter if agents write the code? This makes no sense.
Because the particular form of non-ergonomic code that is demonstrated here amounts to a quadrupling of tokens. That's a substantial hit to the context window for safety that isn't even statically enforced.
Also I'm not sure the token count is very relevant for an Anthropic employee who just spent 165k worth of token rewriting the whole thing in Rust. Sprinkling a few defers here and there and swapping types surely would have been immensely cheaper in tokens.
You're correct that the former is not a concern for Anthropic, but it's also not the problem I'm pointing out.
The actual is not token count, it's context window. Your context window is finite, and every token you add to it reduces the model's accuracy during inference.
In one comment you are presuming that agents can just ignore ergonomic problems, which would be a kind of magic if it were true, and in the next you are saying that their inability to digest enormous quantities of unnecessary tokens is some kind of flaw that needs to be called out.
Enormous quantities of unnecessary tokens are bad ergonomics for anyone, agent or otherwise.
"Technical substance" simply requires that some technically substantive things are mentioned. Which was indeed the case.
If you write a blog post about "I switched from X to Y", I expect the WHY: the pros and cons of X, of Y, and of the alternatives that were considered and dismissed before considering Y.
This is more a blog post about how to use Fable to switch from X to Y, than it is about X or Y.