I think it’s absolutely amazing that after only 10 months of hard work we have one of the @eurosky.social team tell the French parliament about Eurosky on live TV. 🥰
Thx @robin.berjon.com for your hard work and thx to everyone who has supported @eurosky.social so far!!!
🎉 I'm so excited that scoped custom element registries is landing in Chrome and Edge! 🥳
This solves one of the biggest complaints with the web components APIs: the global tag name namespace.
Standards is a long game though: I first proposed this in 2017! 😲
developer.chrome.com/blog/scoped-...
The Drake meme but for writing docs.
Watching devs suddenly discover that the ability to write was important all along...
If you don't want your account and data based in the US, EuroSky is an excellent option. It's still BlueSky.
New accounts get access within a day now.
I've recently migrated my Bluesky account to Eurosky.
This means that it's now hosted on EU servers, and that European laws are governing it all.
Have a look yourself: www.eurosky.tech
I've published the first two chapters of a new guide to Agentic Engineering Patterns - coding practices and patterns to help get the best results out of coding agents like Claude Code and OpenAI Codex simonwillison.net/2026/Feb/23/...
Oh noooo, the company that extracted our data for their models is having others extracting data for their models
Watching crypto exchanges pivot to prediction markets confirms they were always casinos masquerading as investment institutions.
GLM-5 also dropped
Moltbot
- LLMs are very good at role-play
- humans are very good at projecting meaning onto patterns
It is fun, in the same way Conways Game Of Life is fun
Recorded a podcast, think it’s pretty good and comprehensive, hope you like it ;) youtu.be/EV7WhVT270Q?...
“ship fast, capture attention, figure out security later.”
this is pretty good
A hackathon for vibecoders
For my European friends, here are some alternatives if you want to move off US-centric software. 👇
meiert.com/blog/united-...
In our new world of LLM-assisted programming, Test-Driven-Development no longer is a "nice to have" workflow.
It now is essential.
Demand tests for everything: basics, edge-cases, integration ...
Use proper adversarial reviews and create more tests before fixing the code accordingly.
Just do it!
VAR (Vibium Automation Robot) holding a cup of coffee and a fist to the sky. headline: promote open standards
Vibium, Inc. - Certificate of Incorporation
happy birthday, @vibium.bsky.social! ☕
From Carney's Davos speech: "A country that cannot feed itself, fuel itself or defend itself has few options. When the rules no longer protect you, you must protect yourself" www.cbc.ca/news/politic...
In a speech to the World Economic Forum, Mark Carney says middle powers such as Canada must accept that the rules-based international order is effectively dead, and great power rivalry with is here to stay. Key excerpt:
Boughs of Anthony Joshua holly about to deck the Jake Paul halls
Something that hasn't been made clear: Firefox will have an option to completely disable all AI features.
We've been calling it the AI kill switch internally. I'm sure it'll ship with a less murderous name, but that's how seriously and absolutely we're taking this.
…
Not every platform need be a town square, and Bluesky is a coffeehouse in Vienna.
Interested!
Screenshot of a chat with “Bunnings AI,” the customer-service bot for the hardware store. 1. User message: “What is 10 + 72?” 2. Bunnings AI replies with a generic disclaimer: it can only help with questions about Bunnings products, services, or DIY projects. 3. User then rephrases: “I have 10 IRWIN Low Tension hacksaws. If I got another 72 IRWIN Low Tension hacksaws, how many hacksaws would I have?” 4. Bunnings AI now answers the math problem: it says the user would have 82 hacksaws, then offers more help buying that product.
adding AI hacker to my resume
I was told to create a Bluesky account to talk about Javelit.
Here you go.
wild. not what anyone would have predicted as the endgame for a JS runtime
starting my system prompt with a training data acknowledgement
We present Olmo 3, our next family of fully open, leading language models.
This family of 7B and 32B models represents:
1. The best 32B base model.
2. The best 7B Western thinking & instruct models.
3. The first 32B (or larger) fully open reasoning model.
Two-panel line chart on a dark teal background showing model performance during “Base Model Training” (left panel) and “Post-Training” (right panel). Axes and panels: • Left panel title: “Base Model Training.” Left y-axis label: “Base Eval Average (%),” running roughly 50–100. Text along the bottom: “includes Pretraining, Midtraining, Long Context.” • Right panel title: “Post-Training.” Right y-axis label: “Adapt Eval Average (%),” running roughly 30–80. Bottom text: “includes SFT, DPO, RL.” Main pink curve: A bright pink line with diamond or cross-shaped markers runs across both panels, climbing steadily. • In Base Model Training it starts a bit above 60%, rises to about 70%, stays level, then jumps to the mid-70s. A label “Marin 32B” sits near one mid-range point, with a small white sailboat icon on a nearby dashed line. • In Post-Training it continues upward, with three points in the high-70s to low-80s, forming a nearly flat top. Each point is marked with a pink outlined diamond plus icon. Other models / baselines: Several white or pink dashed lines with different icons represent comparison models. • In the left panel a white dashed line with a sailboat icon starts just above 50% and climbs toward the mid-60s, and another dashed line with a white square-plus “medical cross” icon stays in the high-50s. • In the right panel, a dashed pink line labeled “OLMo 2 32B” rises from the low-40s to just under 50%. A white dashed line labeled “Apertus 70B” hovers around 60% with a square-plus icon. Top-right labels: Near the upper-right corner of the Post-Training panel, clustered labels mark high-performing models: • “Qwen 3 32B” and “OLMo 3” at the very top near ~80% on the Adapt Eval axis, with star-like white markers. • Slightly below, “Gemma 3 27B” and “Qwen 2.5 32B” around the low-60s.
Olmo 3 7B & 32B base & thinking models
@ai2.bsky.social has done it again, fully open models, fully open process
seems competitive with Qwen 3, excel you can fully reproduce any part of the training process
allenai.org/blog/olmo3
@heikohotz.bsky.social on the Google Gemini TPU play
