Snugug

I'm so gosh-darned excited! css-tricks.com/css-tricks-h...

Dublin, how do you have a Bram Stoker festival with not one Dracula? #BiteMeDublin

Dublin, y u no have jet bridge?!

Merriam Webster out here fighting for its life

"There's no gap in the rules to point to. I didn't follow your written requirements" Alignment. It's not just to prevent doing crimes.

You can't just shape a chewy kaiser roll into a long and call it a hot dog bun.

It's really incredibly frustrating that in order to use WebRTC—ostensibly a peer-to-peer API without the need for an intermediary—you need at least 3 intermediaries. Can we get a Rev 2 spec that makes signaling/ICE/Stun/TURN servers obsolete? Or can we build some of these directly into the browser?

Canada, what is this nonsense of everything being closed and people actually having off on national holidays?! It's very un-American of you.

After a decade at Cloudinary, I am looking work. I'm a web developer who's done a lot of devrel, too. I'm good at the front of the front end, great at browsers and standards and deeply understanding and explaining technical topics, and the stem of my T-shaped skills is: "images, on the internet".

If I went to Switzerland and won a Swiss watch precision contest, I would literally never talk about anything else

"Yes boss, found the bug, this thing that's hardcoded as `true` saves as `false`. Rewriting the entire codebase to work around that bug" Couldn't you just fix that bug so it saves as true? "Yes boss, smaller fix, much better, reverting my changes"

I made a dumb little thing this weekend! Dice Goblinz! A 1 page hidden role ~deck~ dice building roguelite friendslop game for 2-8 players and all the magic math rocks they can get their grubby goblin hands on. #indiegame #ttrpg snugug.github.io/play/dice-go...

Anyone have an illustrator who has open commissions for a (hopefully) quick bit of work? Would love a few sketchy (not refined, literally like quick sketch) Goblin drawings for a little 1-pager game I put together.

This is a terrible time for systems thinkers, in part because nobody wants to listen to systems thinkers about stuff they’ve been saying for years, so now systems thinkers have to watch deeply incurious people reinvent systems thinking from first principles

The "Apple Intelligence" update to Siri has turned it from useless aging abandonware into exactly the amount of useless I expect from a modern chat agent so improvement I guess?

Had to audibly talk myself out of some of my perfectionist tendencies today so I could get something launched. Or, I guess put another way, "I think my biggest flaw is that I just care too much."

My current conspiracy theory: the press release about Anthropic changing words in Claude output to make it easier to detect generated text is really a psyop to cover for the fact that Fable writes like an alien who's just discovered the concept of language is and is "trying it out" on English.

Gemini is _so much better at writing_ than Claude and it's not even close. Everything I get out of Claude reeks of AI slop. Gemini typically just needs some light editing. Writing code, on the other hand...

The Interop 2027 call for proposals is now open! Learn more about the process at github.com/web-platform..., or start a proposal via github.com/web-platform... today. You can also read about it in our news post at www.igalia.com/news/interop....

Swapped back to Opus 4.8 from 5 this morning on a personal project and the difference is night an day: 4.8 is faster, more accurate, follows instructions better, and is generally less frustrating.

Create agent-ready web apps for the WebMCP Challenge → goo.gle/4qpDQYV We're excited to see ChatGPT supporting WebMCP, the experimental standard that lets users and AI agents navigate websites together. In the @openaibot.bsky.social WebMCP Challenge you can experiment with it and win prizes.

How is it possible that ice cream shops don't put chocolate magic shell in the bottom of their cones? This feels like a no-brainer.

Canada, I've besmirched your good name unjustly. Your Ritz crackers aren't worse than American Ritz crackers, Ritz crackers have just gotten worse!

I've gotta tell y'all a secret: if you're upset about Claude's writing getting watermarked, it means you've never read anything it's written. AI writing is so painfully obvious to anyone who actually reads it _today_ that this feels like a no-op for those who care about output quality.

I'm finding I don't just want, but _need_ an LLM and agent pair that isn't designed to be a "helpful assistant" because that framing fundamentally breaks some use cases

I've got a blog post cooking in my head about how I tend to hold technology "wrong" and how that's affected how I see AI tools versus some of my peers.