Post

Laurie Voss
@seldo.com

I have always been the kind of developer who only cared about outcomes. Give me the tool, the framework, the head start. I just want my website. A totally different kind of developer exists who cares about the *craft*. Getting that function just right. Those devs are in pain right now.

Mia [data-css=poetic]
@miriam.codes

that's part of it. but there are also core aspects of the web (esp html & css) that were already too invisible: cross-browser/device support & a11y. It's not enough to care about "outcomes" when you think everyone has the same experience as you. Many care about craft **because of the outcomes**

11 reposts114 likes

Brian LeRoux
@brianleroux.bsky.social

This is an aspect of LLM codegen I'm excited about. We can prompt professional html and css just as easily as brittle framework slop. So if the outcomes are truly what we care about then we should guide the machine to prioritize stability, performance, and a11y. (Which it does wonderfully.)

Sara Joy ☠️
@sjoy.lol

Ooh please post a demo somewhere. I've yet to see a demonstration of building an accessible app frontend that is faster than doing it "by hand" for someone who knows their way around HTML/CSS/JS.

Brian LeRoux
@brianleroux.bsky.social

Yeah I've been thinking about releasing the steering stuff I use. It's personally useful for me but I'm mega opinionated about the dependencies I use and almost always that's where ppl find fault. (I build for AWS, I prefer dynamic process behind a CDN to static build step, tend to have a database)

Brian LeRoux
@brianleroux.bsky.social

TLDR, LLMs are trained on everything and browser tech being so stable means they are exceptional at doing things in a browser native way. Frameworks have had so many versions/breakages/changes I've found llms get confused a lot ...it's easy to steer any model to "always do X" and "never use Y"

Sara Joy ☠️
@sjoy.lol

For me the issue is trust, and I simply haven't had the time (or inclination, I'll admit) to try and build some kind of AI environment that I feel like I might be able to trust. I don't know where to start, it'd take me a week just to orient myself, while feeling a little bit dirty the whole time 😅

Mia [data-css=poetic]
@miriam.codes

If you write better css, you will get better outcomes on other devices you never thought to test. same with html. I'm not worried people care too much about outcomes over craft, I'm worried that when we say "outcome" we mean All My Biases Are Fine Actually, Stop Nagging.

Mastro.{js,ts}
@mastrojs.bsky.social

Exactly. While the word “craft” resonates with me, an even better word may be “care”. Like what Johny Ive said about obsessing as much about the design of what’s inside the PC’s case as its outside. Even if most people never see the inside, they still can feel the care that went into the whole.

Aaron Sterling
@aaronsterling.bsky.social

I created a several-page UX Guidelines document that Claude uses in our projects. I was able to write it in part because a company paid for me to attend in-person accessibility training. I'll probably release it publicly at some point, but perhaps WCAG-type orgs already have published their own.

conputer dipshit
@davidcrespo.bsky.social

I would be more worried about this if agents didn’t make it dramatically cheaper and easier to write accessible HTML and CSS! like, Claude knows 100x more about a11y than the average dev and is happy to apply it

Mihai
@mihaizaurus.at

I would add you can care about the outcomes and still have an idea of the gaps your experience has, which is where community comes into play imho. Would not have thought about things like a11y, i18n, etc. myself without having seen others tackle them first. I now try to put them in personal projects