Independent AI researcher, creator of datasette.io and llm.datasette.io, building open source tools for data journalism, writing about a lot of stuff at https://simonwillison.net/
Here's the Daily Show coverage www.tiktok.com/@thedailysho...

@simonwillison.net
Independent AI researcher, creator of datasette.io and llm.datasette.io, building open source tools for data journalism, writing about a lot of stuff at https://simonwillison.net/
Here's the Daily Show coverage www.tiktok.com/@thedailysho...
Found the YouTube music video by that pizza chain guy who was bribing Kimberley Guilfoyle to lobby the IRS and send his song to Kid Rock youtu.be/2B6uBfh7bjc
Example prompt:
I built a new feature for my blog entirely by voice with Codex Desktop, while I was cooking dinner simonwillison.net/2026/Oct/9/b...
Haiku 5.5 is SO MUCH better at drawing pelicans riding bicycles than Claude Haiku 4.5, released almost a year ago and costing 10x more. Here's Haiku 4.5's attempt:
Pelicans and a bunch of notes on the pricing of the new Claude Haiku 5.5 simonwillison.net/2026/Oct/7/c...
I released an LLM plugin for sending prompts to OpenAI's new Jev-clone decision model that's based on GPT-6 Luna simonwillison.net/2026/Oct/6/l...
I vaguely remember that! I liked being able to navigate between lines easier using the keyboard h
Oh no... I think the default answer may differ if you are talking about disks or RAM!
I've called this the 'The Kobayashi Maru' interview question for programmers. There's no "right" answer only painful trade-offs which makes it very revealing.
I need to decide if 1KB means 1,000 bytes or 1,024 bytes in all of my software I was leaning 1,024 but it looks like macOS and Windows both define KB as 1,000 these days I want whatever people are most likely to understand (I do not think most people know what KiB or MiB means)
Do you think there's any chance at all someone deep within Apple might reconsider and mark it as not-deprecated? To use a Claude-ism it's increasingly "load-bearing" these days
Just published this post about how we’re going to need default hard budget caps on pretty much everything simonwillison.net/2026/Oct/3/d...
Do you think it makes sense to coach them in learning how to build interactive prototypes?
I mean yeah, this is a stupid way to add numbers I did run this on a computer on my desk at home though, so it's not a great example of wildly irresponsible power usage
Help me understand why that is
Is your distrust of reasoning traces based on the fact that the proprietary vendors deliberately obfuscate theirs and turn them into summaries? The open weight models make it much more clear what's actually going on
Have you seen how reasoning levels for Qwen actually work? It's pretty funny, the level you select gets used to dynamically assemble a system prompt effectively telling it to "think carefully" huggingface.co/Qwen/Qwen3.8...
Honestly not sure if there's something we disagree on here
Sure. Still works though.