Manuel Matuzović has written “Context-aware headings in HTML” for the uninitiated. Jake Firefox has a good garage wall video covering the attribute too.

It solves a real problem for component libraries and basic template includes. I just can’t vibe with seeing the <h1> to <h6> elements being a different level than their names imply. Has a simple <h> element been discussed?

I’m curious if the slop machines are able to “parse” the correct heading levels. On the basis that I imagine they’ll fail spectacularly, I give headingoffset the thumbs up!

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

How do website makers ensure what we’ve built is accessible? It is impractical to test everything. Even with access to all devices we’d struggle to experience their usage like those who depend upon assistive tech daily.

Testing matters but “fixing accessibility” as an afterthought is guaranteed failure. We have web standards for a reason. Declarative HTML allows us to express web content and UI semantically. This allows browsers to understand and adapt. It’s remarkably effective. It’s criminal that we have a generation of developers who fail this simple step.

Get the basic foundations right and there will be less cracks to paper over. Nothing can be perfectly accessible but we can at least give assistive tech a chance.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

I’ve always played at the highest ascension rank possible so my winrate is around 8–10% except for The Regent at 20%. I suspect that’ll fall much lower before I beat A10. I’ve reached the final boss a few times but it’s a brutal affair.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

⚠️ Slay the Spire 2 major spoilers!

Steam says I’ve put 300+ hours into Slay the Spire 2 if that’s to be believed. I’m up to ascension level 9–10 on all characters. Except The Regent because that requires brain power. A10 has decimated my win rate.

Anyway, go away now if you want to avoid a big spoiler.

Seriously, don’t let me ruin this for you.

I heard about a secret boss and I found it!

You probably won’t believe a secret boss exists.

When I read the rumour I thought it was a silkpost.

Last warning…

So yeah I just kinda murdered the fake Merchant??? and stole all the goods.

Edit: I beat the run. The Ice Cream relic to conserve energy between turns is over-powered. The Diamond Diadem relic came it clutch on the final boss allowing me to setup a 382 damage Sovereign Blade.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

If your job doesn’t require you to burn tokens, then probably ignore MWG and, by extensions, LLM for code. In case the environmental damage, intellectual property theft, wealth redistribution to the rich, encoded biases, and built-in support of fascism weren’t enough.

Maybe Don’t Rely on Google’s “Modern Web Guidance” - Adrian Roselli

I’ve quoted the part that I find most relevant but Roselli does provide more constructive criticism.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

If you catch me online please insist I disconnect immediately.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

[Generative AI exists] to keep entry-level workers pruned all the way back to the roots, in the interest of keeping wages low, employment tenuous, staff nervous, and the unfathomably rich insulated from the potential financial repercussions of destroying countless lives. If you can never advance beyond a “press the button, generate the thing,” you’re as replaceable as the next person — if there’s nothing beyond entry-level experience, they’ll never have to pay you more than an “entry-level” salary.

LLemdashes - Mat Marquis

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

This is devastating for a guy like me who wants to champion European alternatives! Speaking of: Go European is another initiative.

So yeah I’m back to Figma until the bug is fixed and I hate myself.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

The domains in which a programming language is popular influences the ecosystem. I’m deeply grateful to Jarred and Anthropic for giving Zig communities a chance to reroll for something other than LLMs. Hoping for better than a nat 1 next time

@andrewrk@mastodon.social - Andrew Kelley

Spicy words from the Zig creator!

Relevant note for context.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

When it comes to Tailwind, there is a lot of conflation between “CSS is hard” and “design is hard”. Tailwind offers a little bit of both for those who know neither. It’s both a shortcut and a dead end. That might be good enough if you’re never going to learn CSS. Nobody should need to learn CSS to publish on an open web.

Hard truth time: Tailwind is simply not a professional tool. It’s completely irrelevant once you reach a professional level of CSS proficiency. That’s objectively a fact. Tailwind offers a limited subset of CSS with limited application.

Those who promote Tailwind aggressively diminish the time required to learn “one of the most advanced graphics, layout, and typesetting languages available in computing.” The same people are prone to belittling design talent. Tailwind has been cargo-culted to React levels of insanity. What should exist as a useful tool for amateur developers has become a bane for professionals. That didn’t need to happen.

Use Tailwind. Maybe learn CSS by using Tailwind. Other than potentially large file sizes, Tailwind should be harmless. Unfortunately, misinformation and tribalism is abound and now Tailwind just happens to extrude well through LLMs. Sigh.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

Bluesky vs Mastodon vs Threads vs Twitter?

Well if you’re still using Twitter you have a moral deficit so large there is no salvation.

I have literally never visited Threads. Never clicked or even seen a single link to it. Every time a Zuckerberg product is in the news it’s perverted and sickening invasions of privacy. I stay clear of Threads.

That leaves Bluesky vs Mastodon. Both of which are filled with meta drama about “federation” vs “decentralisation” but have some merits.

Mastodon has constant infighting across instances all done via subtext so good luck trying to understand if your instance is the bad one. I’ve found Mastodon to be more engaging in nerdy and technical topics.

Bluesky makes it easier to own your data. In theory it’s decentralised but it remains heavily reliant on infrastructure run by Bluesky the company. That company is constantly embroiled in controversies. I’ve found better connections with the UK web scene on Bluesky which is fun.

If we learnt anything from Twitter it is this: don’t invest yourself too far into any social media platform. Get out when they enshittify. I suspect Bluesky will be next to fall due to those running it. Can independent AT protocol developers truly decentralise in time?

Bonus link: W Social uncovered: the reality behind the hype. Turns out that mysterious EU social network is a fork of Bluesky. It’s not a European Union initiative either. Just some rich people doing rich people things.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

An ignorant trend I’ve seen is pointing at smart quotes and other typographic flair as a slop signal. Absolute nonsense because many word processors do this automatically. Apple Notes for example. I’m sure MS Word and Google Docs probably do it. Programmatically, I use a port of smartypants which predates “AI” as we know it. WordPress has a wptexturize function that could well be older. LLMs predict smart quotes because their training data is rife with ‘em. It ain’t magic.

I’ve shared a couple of posts by Den Odell recently.

1. Browsers Treat Big Sites Differently
2. You’re Looking at the Wrong Pretext Demo

I was surprised by the Lobste.rs comments on the first post that vibe-cused hard.

Unfortunately, the writing is so bad that I couldn’t find it. The article may as well have stopped after the headline.

comment by @orib

Oof. That’s harsh! I kinda skim-read stuff but I found the info on browser quirks interesting enough to share on social media. After seeing the vibecusations I re-read it thoroughly. The negative parallelisms do stand out. If Den does use LLMs it’s far from the most egregious use. You can’t prompt these type of posts from thin air.

Now I’m wondering though, what if facts were mangled in an LLM extrusion? Den is obviously a smart guy I don’t want this lingering doubt. A big reason I have a strict no-AI policy is that all we have is trust. AI is turning us against one another and it plays into their hands.

I’m far more certain Den uses generative AI art which saddens me. The brand logos sit in the continvoucly morged uncanny valley. Den’s article on Pretext had a six-fingered illustration until it was changed to an eery hyper-realistic version with people staring in the wrong direction. Adam Argyle uses slop art too. I just don’t understand it. Why descrate your blog with garbage everyone can see?

I’m not trying to name-and-shame anyone here. I’ve chosen two examples I still respect despite questionable AI use. There are other high-profile people in this industry I’ve unfollowed and unsubscribed from that have turned straight up AI grifters. They’re trying to flog their hogwash to peers who are too busy getting fired. Great business model ya got there. I don’t plan to call them out. When the slopocalypse fallout blows over and they crawl out of the woodwork we’ll remember.

Ahh! I hate how negative I’ve become! But I can’t just ignore what AI is destroying. Feels like a losing fight but I won’t submit because I’ve got nowhere else to go. Craig Cook, Ky Decker, and just recently Axel Rauschmayer have shared harrowing stories. It makes me angry. So I blog and I probably upset people with some of the stuff I write. I won’t apologise for that.

Please for the love of humanity, stop using AI.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

This whole thread is an overreaction. 302 comments about code that does not work. We haven’t committed to rewriting. There’s a very high chance all this code gets thrown out completely.

Zig → Rust porting guide - Jarred

In regards to a vibe-coded port of Bun to Rust.

Today the “code that does not work” was merged in a PR so big it makes GitHub buckle (which isn’t hard to be fair). Over ONE MILLION lines of code changed. All ‘reviewed’ in a week. Absolute slopware.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

These postmortems read like this instant classic:

Root Cause

A dog named Kubernetes ate a YubiKey.

Incident Report: CVE-2024-YIKES - Andrew Nesbitt

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

The agentic era affords GitLab the largest opportunity in our history as a company, and we’re making the structural and strategic decisions to meet it.

GitLab Act 2 - Bill Staples

Oh dear, GitLab has contracted the rot. Right when the competitor is sinking.

It’s rather disgusting to call firing “up to 30%” in some areas “the largest opportunity in our history”. The phrase “Role right-sizing” is used too, excuse me I need to vomit.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

1. Why I object to and reject generative AI
2. Building for the future

In the first, Prof. Deborah Lupton reminds us of the real harm.

In the second, Clownflare reminds us what systemic incompetence looks like: 1,100 employees fired. I guess they failed to impress with their performative slop.

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.

There’s also the question of what the browser is even for. A browser is a user agent – it’s supposed to act on behalf of the user, not the vendor. Silently downloading 4 GB of AI model to your machine, re-downloading it if you remove it, and then letting any website access it without your permission – that’s Chrome acting as Google’s agent, not yours. And it’s hard not to notice how many Google products stand to benefit from having an LLM pre-installed on every Chrome user’s machine. With opt-out, of course.

Lazy and Prompt - Matthias Ott

Thanks for reading! Follow me on Mastodon and Bluesky. Subscribe to my Blog and Notes or Combined feeds.