Blaine

I should show you the thing, lol

AT&F1&B1&H1&R2&I0&N0&K1&M4X4S10=14S25=5&W

💀

I'm just kidding, as a ux-oriented symbol, at:// is unhinged evil.

Otoh, I said the same thing about hashtags and they happily entered pop culture, so wtfdik? 😆

but jacky, at:// is lawful neutral

@doctorow.pluralistic.net Hi! Welcome. 🎉

What I mean by "code wasn't the hard part" is that all the other under-valued pieces were *always* way more important, but the power dynamics located code at the center, to all of our detriment. I have many trepidations, of course, but I'm also incredibly hopeful to see that shift.

Absolutely. I said in a separate branch of Luis' thread, but worth repeating – I didn't mean to say that coding was easy, or that it wasn't a source of barriers. I hate, so much, the "coders rule" reality we've lived in.

Extraordinary evidence requires extraordinarily legible claims.

Great thread, which I agree with wholeheartedly.

Hard agree with all the above.

When I say "code was never hard", I'm eliding a wholesale rejection of the power dynamics that meant that code had power where people and communities should have. 💜

Button blog: Minimizing user research fraud in the age of agentic AI Learn how content designers can prepare for and prevent user research fraud to glean accurate product insights.

New_ Public’s own @absp.bsky.social, a UX Researcher and sociolinguist, has a really useful piece on how researchers should be thinking about AI-based fraud, and what to do about it.

Go Abby! 💪🦾

Totally agreed. For what it's worth, on the data reporting side of my work, using genAI to *do* the reports would be a fireable offense. Building exploration tools with it is incredibly useful in our experience, though.

Oh my. For me, LLMs are 80% about brainstorming and thinking through problems in more depth than was possible before. If people are skipping that step, i.e., "the work" ... 🫣

In any event, very stoked you're experimenting with this. That it should be possible has been a conjecture I've held for a very long time, but it felt very far away until you broke down that wall!

Yeah, the trick is more on the license/data sharing permission side, I think? Maybe the fact that extractive surveillance infrastructure does this all over the place means we can use laws that govern those situations, but with better/actually consentful agreements?

I am so excited for this public domain future.

But this matches what most folks I know are already experiencing; the hard part has always been design & taste. Spec writing can be iterative and exploratory, too!

Yuuuup!

Crossing the "rebuild all the software every day" line feels wrong, there's absolutely an aspect to this where openai are leaving the token pump running. ☹️

This is so great! I've been designing roundabout services to use this pattern (& playing with variations) so that they're decoupled by default.

The ToS question would be my main concern around bits like this. Normally authn is the click through agreement point. How are you thinking about this?

Roomy Events, by OpenMeet While Roomy is now quitely operational with a growing handful of pilot spaces, before we make a grander reveal about that particular milestone we have another exciting Report from the Atmosphere to sh...

Building in the atmosphere means you don't have to go it alone.

We know organizers, and the groups-of-coordination they facilitate, require digital infrastructure for events planning. Ergo @roomy.space needs Events.

@tompscanlan.bsky.social agreed and magicked @openmeet.net straight into Roomy.

I've been doing this work for 30 years, and I'm at my most creative and productive, ever. I also don't think this is limited to senior devs; I wish I was just starting out now, learning is 1000x easier than when I was doing it.

The job losses aren't because of LLMs - that's capitalism.

I don't believe that's true. My larger claim here is that that's only true if people thought of their jobs as knowing syntax and fundamentally uncreative plumbing work, which I hope they didn't.

fwiw, I built quite a complex standalone product last year with one other engineer and 6 product-oriented people, and I'm personally responsible for all maintenance of it. I'm very happy with where we're at right now.

I appreciate that not everyone has had the same experience, and I'll be the first to say that a year ago, these tools were only just verging on useful and definitely not good.

I'm spending less time on maintenance than ever. YMMV, and that's ok. 💜

I haven't seen a semicolon in what feels like ages 😅

Mostly Claude; the superpowers plugin helps a lot, and just being strategic with design and planning up-front. In that sense, it's not different from how software engineering has always worked.

Garbage Product/Design In = Garbage Code Out

Not to diminish yours at all, but I've had a very different experience, at least lately. The newest models are very good.

It's unnecessary to do that in public/bsky feeds on atproto!

Someone should write a NIP for writing to a nostr-specific atproto record.

😂 and also 🤢

(more seriously, I have faith that maybe the Lea will be [safely] swimmable before I expire)