>> Today, late optimization is just as bad as premature optimization, if not more so.

You are right about the origin of and the circumstances surrounding the quote, but I disagree with the conclusion you've drawn.

I've seen engineers waste days, even weeks, reaching for microservices before product-market fit is even found, adding caching layers without measuring and validating bottlenecks, adding sharding pre-emptively, adding materialized views when regular tables suffice, paying for edge-rendering for a dashboard used almost entirely by users in a single state, standing up Kubernetes for an internal application used by just two departments, or building custom in-house rate limiters and job queues when Sidekiq or similar solutions would cover the next two years.

One company I consulted for designed and optimized for an order of magnitude more users than were in the total addressable market for their industry! Of that, they ultimately managed to hit only 3.5%.

All of this was driven by imagined scale rather than real measurements. And every one of those choices carried a long tail: cache invalidation bugs, distributed transactions, deployment orchestration, hydration mismatches, dependency array footguns, and a codebase that became permanently harder to change. Meanwhile the actual bottlenecks were things like N+1 queries or missing indexes that nobody looked at because attention went elsewhere.

I have seen some people complain about a new tendency where it can suggest wrapping up the current task even though it isn't done yet. I haven't seen it myself though.

Usually this gets worse if you have a phrase like "wrap it up" earlier in the output, or if you're at a few hundred thousand tokens without compacting.

In both cases the fix is really simple, just compact.

It is AI-generated, which is why it got killed.

Hmmm. I see some possible signs of that.

I think the sentiment is still a valid one, and I think it's an accurate assessment.

How have you determined that? Are you basing it solely on the em-dash (which is trivial to avoid if you want to generate AI comments)?

Nah, AI always has its em dashes with no spaces around them.

It's in research preview. I suspect limits are low on purpose. FWIW, I gave it twelve screenshots of different pages in my app and it did a really excellent job fixing them up. Consumed just 40% of weekly quota - still too high but it's probably a YMMV situation.

Yeah, agreed. And realistically they are correct. I’d argue that “good enough” is how most things are done.

You are not the only one. I also dislike it immensely. For a framework that established itself as "for developers who don't know or want to learn CSS", polluting the HTML in the manner you describe makes no sense. And no company I've worked at figured out how to prevent it from becoming a bloated, impossible-to-maintain mess.

>> Of course, there is indeed no shame. There is also no pride.

I disagree completely. The pride should come from the value that is delivered. Specifically, this:

>> Useful, probably optimal and will be around for decades to come.

Is something to be proud of, full stop.

I think there's something nice about the idea of a store owner which has unnecessarily decorated the store with love, even with the liability of a cat; it's not delivering the product better and the cat may actually make things worse because of allergies.

A cold American convenience store may be delivering the fundamental value at American prices, but there's something to be said about that "extra" human or creative element. One might say the same thing about the changing nature of the web over time, less individual CSS chaos and more Facebook aesthetics.

There's nothing stopping people from decorating their boutique stores (or personal blogs, portfolios, and fan websites) the way they want. And that's fun and delightful for me, as a visitor, just like boutique shops are IRL.

But I really don't need that quirkiness at Home Depot, the DMV or my bank (or Amazon, or government websites, or my banking site). I'm there to purchase some screws, register my car or pick up some checks. I just need a storefront (or a website) that lets me do that as fast and homogenously as possible.

99.9% of stores (and UIs) are the latter, not the former.

There's now an entire cottage industry that is based attempted take-downs or refutations of claims made by AI providers. Lots of people and companies are trying to make a name for themselves, and others are motivated by partisan bias (e.g. they prefer OpenAI models) or just anti-LLM bias. It's wild.

Great, it can compete with the cottage industry dedicated solely to hyping and exaggerating AI performance.

I call it a pro-human bias, personally.

Side note: after drowning in LLM-generated content, it's pretty refreshing to read something written by a human. They're a pretty good writer, too!

Who needs LLM garbage when you have MLM garbage?

>> are we being spammed? great. annoying.

Yeah, very. Every single time this happens here, where there's a thread about an Anthropic model and people spam the comments with how Codex is better, I go and try it by giving the exact same prompt to Codex and Opus and comparing the output. And every single time the result is the same: Opus crushes it and Codex really struggles.

I feel like people like me are being gaslit at this point.

this is exactly how the other side feels