dbushell.com freelance

Notes

These tools, and the companies that manufacture them, have tremendous costs — to our labor, to our environment, to our futures. And as we’ve been seeing, those costs also include actual human lives.

Propellant. - Ethan Marcotte

Before you kneel down and place the slop pail under your LLM today, take a minute to remember what you’re supporting.

Is it officially spring? After weeks of freezing grey skies it was suddenly warm and sunny. I racked up 10K steps and possibly a tan before my podcasts ran out. I returned home to start hacking on a SvelteKit + AT Protocol side quest. I’m not entirely sure what I’m building yet but I’ve turned into an atproto nerd. I’m starting to understand how cool it is after moving Bluesky PDS. I’ve even bought a project domain!

I spent an hour this weekend to move projects from Cloudflare to Bunny. If a project has too many deployments on Cloudflare the deletion attempt errors. I followed the error link to a “known issues” documentation page.

You may not be able to delete your Pages project if it has a high number (over 100) of deployments. The Cloudflare team is tracking this issue.

Delete a project with a high number of deployments

Cloudflare’s solution is a node script that you must run yourself to delete projects via the API. It took over 15 minutes hiting each deployment one by one.

Cloudflare is a serious company.

Andy Bell is unsubscribing from the AI discourse.

I’m bored of hearing about it, bored of seeing people I respect(ed) fawn over it and it’s really not doing my mental health any good.

I’m feeling this too. I feel obliged to fight and mock the mania that threatens my career but it’s relentless and exhausting. I’m preaching to the choir at this point. What do I gain other than melancholy?

My stance is clear. I really should focus my energy elsewhere. I don’t think I’ll unsubscribe entirely but I need to repeatedly ask myself: is this worth giving attention?

The care is the work.

The care creates the value.

AI doesn’t care. AI is just another tool.

Continvoucly Morged Value - Matthias Ott

Last week I wrote on the socials:

what AI reveals is just how many people have given up, or never cared to begin with. clock in, slop around, clock out

As Matthias Ott notes, by reducing professionals to reviewers of machine output they are devalued and disrespected. Who will continue to care? To get a tune out of an LLM requires a high level of domain expertise. But the work is reduced to drudgery. Are we expecting the next generation to train for a professional craft, with all the pride and enthusiasm necessary, only to switch off and become a human LLM once they’re competent enough?

This is not a minor bug trend. It is a systematic civil-rights failure that has now found its way into software as a whole, through lightning-fast adoption of AI systems that are trained on over 20 years of institutional barriers.

AI is locking people out. At Scale.

I don’t have the chops to say whether this benchmark is fair but it can’t be worse than any other “AI” benchmark. It’s shocking how many developers think you can say the magic word “accessibility”. How many are even asking? A recent Syntax podcast left me falling to my knees in Tesco.

Is it better to have an accessibility agent for completing accessibility tasks or an accessibility skill that any agent can use, or is it better to have an accessibility skill that an accessibility agent can use? So skill, agent, or agent plus skill. Like, what is the ideal combination there? And and the answer that I personally landed on is it basically depends on what you’re doing. Like, if this is a one off thing and you’re saying, I wanna I yeah. Hey. Go fix accessibility issues, then an accessibility skill is more than appropriate for that. And that for the most part, that is the answer. The answer is, hey. Just go off and fix these issues. But if this is a longer running thing where you’re, like, gonna go back and forth with it, like, hey. I want you to go and do a full audit.

AI Coding Explained - Syntax #980 - Scott Tolinski (transcript)

NPMX caught my attention. It’s a new front-end for NPM, not a new registry like JSR. That doesn’t excite me. Nor does the trendy grey-on-black tiny font. Certainly not the Nuxt bones. I’m just intrigued by its relative popularity.

NPMX has gained 2K stars and 178 contributors on GitHub† in little over one month. JSR has only attracted 2.9K stars and 86 contributors with a two year head start.

Did Deno fumble that hard? I’ve stopped following the JavaScript ecosystem churn. I don’t see the appeal of either NPMX or JSR. All I need is good package search and both fail terribly for different reasons.

† Don’t @ me, that’s how JavaScript is measured.

Jim Nielsen writes: A Few Rambling Observations on Care

In this new AI world, “taste” is the thing everyone claims is the new supreme skill.

But I think “care” is the one I want to see in the products I buy.

AI is sold as a shortcut to short-term profit, at the expense of seemingly everything else. Care is what’s required for long-term success.

That’s why I develop for those who care.

GitHub allows you to ���block” accounts (whatever that means). This will show a warning on every repository that account has morged to.

GitHub warning: 'A user you've blocked has previously contributed to this repository.' listing the account 'claude'

GitHub warning: 'A user you've blocked has previously contributed to this repository.' listing the account 'claude'

Credit to an unknown hero who shared this a few days ago.

“Building Typographic Scales in CSS with :heading(), sibling-index(), and pow()” by Stu Robson presents some really neat ideas.

I’ve reduced Stu’s example to a magic one-liner:

:heading {
  font-size: calc(1rem * pow(1.25, 6 - sibling-index()));
}

Stu offers more practical complete design system ideas, with custom properties to make sense of it all. This is a neat demo but we’d need heading-index() for the perfect setup.

I’d written off :heading has a minor convenience but I now see how powerful it becomes with other new CSS additions.

Microsoft’s “Components of the GitHub flow” (archive) training guide is absolute slop. What’s worse is the Git flow diagram appears to be straight up plagiarised from Vincent Driessen’s “A successful Git branching model”. Except for some reason, Microsoft extruded it through the AI wringer. What did they prompt: “this, but worse?”

Edit: Vincent Driessen commented:

I find this whole thing mostly very saddening. Not because some company used my diagram. As I said, it’s been everywhere for 15 years and I’ve always been fine with that. What’s dispiriting is the (lack of) process and care: take someone’s carefully crafted work, run it through a machine to wash off the fingerprints, and ship it as your own. This isn’t a case of being inspired by something and building on it. It’s the opposite of that. It’s taking something that worked and making it worse. Is there even a goal here beyond “generating content”?

15+ years later, Microsoft morged my diagram - Vincent Driessen

I found this via Mastodon and Bluesky, I’m sworn off Microsoft and GitHub myself

side-by-side comparision of the original and Microsoft's AI version

side-by-side comparision of the original and Microsoft's AI version

The hit piece has been effective. About a quarter of the comments I’ve seen across the internet are siding with the AI agent.

An AI Agent Published a Hit Piece on Me – More Things Have Happened - Scott Shambaugh

The AI attack dog sent after volunteer maintainer Scott Shambaugh is a warning sign of things to come. There are no limits to the harm AI is allowed to do. People are cheering this on. Ars Technica published a now retracted slop piece. Shambaugh’s take should give us all pause for thought.

Journalistic integrity aside, I don’t know how I can give a better example of what’s at stake here. Yesterday I wondered what another agent searching the internet would think about this. Now we already have an example of what by all accounts appears to be another AI reinterpreting this story and hallucinating false information about me. And that interpretation has already been published in a major news outlet, as part of the persistent public record.

Brandolini’s law is referenced. AI adds a multiplier we’ve yet to fully grasp.

The latest UX deceptive pattern I despise:

  1. Enter credit card details
  2. Tab to “continue” button
  3. KAPOW! auto-focus “remember for later” disguised 3rd-party “your privacy is important to us” garbage

It’s probably not new but I’ve been seeing it everywhere lately. This is bad for many reasons. It disrupts and confuses, risking accidental opt-in. It masquerades as a simple convenience when the 3rd-party terms are almost certainly not simple.

Just show the damned form to begin with.

Generated code is rather a lot like fast fashion: it looks all right at first glance but it doesn’t hold up over time, and when you look closer it’s full of holes. Just like fast fashion, it’s often ripped off other people’s designs. And it’s a scourge on the environment.

Stop generating, start thinking - Sophie Koonin

I like this analogy. The whole article is a rare level-headed take on using LLMs for programming. Nice to read someone thinking rationally beyond the hype. Orange site comments are as expected: “You’re prompting it wrong”

A totally normal and not hard quiz about JSON

jsonwtf.org - Jacob Asper

JSON WTF is a fun quiz! I scored 24 from 35 questions. I might have got a few more if I’d thought really hard. JSON is a small and almost perfect standard (should’ve allowed trailing commas). Unlike YAML which is awful. Just awful.