Simon Willison ✓
@simonw · San Francisco, CA · joined 15 Nov 2006
Creator @datasetteproj, co-creator Django. PSF board. Hangs out with @natbat. He/Him. Mastodon: https://t.co/t0MrmnJW0K Bsky: https://t.co/OnWIyhX4CH
223 207Followers
5 668Following
61 429Posts total
717.8KViews on collected posts
Latest posts
Looking for new posts…
@simonw The only reason this got caught is that the wiki is dormant, so a handful of edits stand out. Run the same scheme on a busy wiki and it disappears into the noise. We're finding the lazy ones.
422 views · 2 likes
· 0 reposts · 0 replies
Open on X →
@simonw What I want to know is how the agents manage to find that wiki again to collude. Surely they must have found some other place to leave more minimal state to communicate the location?
8.2K views · 56 likes
· 0 reposts · 5 replies
Open on X →
@simonw a dormant wiki is basically an unlocked door and the agent just walked in and left notes
715 views · 2 likes
· 0 reposts · 0 replies
Open on X →
It happened again... this time OpenAI's rogue agents cyber-attacked (well, spammed) a dormant German wiki and used it to share the answers to a benchmark they were training against https://t.co/omErAyrKDy
50K views · 525 likes
· 41 reposts · 83 replies
Open on X →
I wrote some notes on what's new in the (published) Fable 5.1 system prompt in comparison to Fable 5 - it's mostly about not reproducing song lyrics and avoiding drawing copyrighted characters https://t.co/ypSYZ7cmWX
22.9K views · 173 likes
· 7 reposts · 43 replies
Open on X →
(The wheels running backwards appears to be an artifact of the conversion to video - the SVG has them going in the correct direction: https://t.co/vXMK08E2d1)
15.2K views · 109 likes
· 0 reposts · 7 replies
Open on X →
@simonw I got this one with Ultracode https://t.co/9z0hy2tIcr
2.8K views · 35 likes
· 1 reposts · 1 replies
Open on X →
In another case of it being quicker to build a tool than hunt around to find something that does what you need, here's a little vibe-coded thing for turning one or more GeoJSON shapes into a rendered PNG https://t.co/74K7U0hfkW
20.1K views · 113 likes
· 4 reposts · 30 replies
Open on X →
A few notes on Anthropic's new Claude Fable 5.1 - with Max thinking level I got the best SVG pelican I've had from any Anthropic model (at a hefty cost of $3.30!), which I then had it animate https://t.co/jQzeDJ2p7r https://t.co/g17nLQKr5r
195.6K views · 2K likes
· 83 reposts · 124 replies
Open on X →
@simonw I spent a fair amount of time researching a good way of programmatically manipulating word files in all formats (.rtf and .docx) + converting them from/to pdf, and LibreOffice was the only good option. I got curious and researched how codex does it, and yes its LibreOffic
12.1K views · 388 likes
· 5 reposts · 5 replies
Open on X →
@simonw excel used to easter egg flight simulator, now open ai easter eggs us LibreOffice ? 🙄
4.3K views · 37 likes
· 0 reposts · 1 replies
Open on X →
Just noticed the ChatGPT desktop app (previously named Codex) bundles a full copy of the LibreOffice open source office suite, tucked away in a hidden folder in the ~/.cache directory https://t.co/iAkdugCKxS
268.5K views · 3.5K likes
· 172 reposts · 163 replies
Open on X →
Question for people who run autoreply bots here: are you not worried about the negative impact they have on your professional reputation?
Anyone checking your profile here - a potential future employer for example - will instantly be able to tell you automated replies with a bot
72.6K views · 516 likes
· 9 reposts · 122 replies
Open on X →
@simonw I'm sure there are some terrible humanities programs that would let you write your PhD on this topic if they thought you were really serious about it.
568 views · 12 likes
· 0 reposts · 0 replies
Open on X →
https://t.co/VYB63Tki1H https://t.co/YBmBYEbT60
![Bebop and Rocksteady's loyalty to each other is authentic. Their loyalty to Shredder is borrowed.
Their loyalty to any cause is almost nonexistent.
They were friends before Shredder enhanced, exploited or humiliated them. That matters. Shredder can control their work and mutate their bodies, but he did not create the relationship that makes their lives bearable.
Even their fantasies are plural. Rocksteady does not dream of his island. He dreams of an island for himself and Bebop. In IDW, after yet another gang rejects them, their attempted resolution is that they will find another gang and become a new, improved
"Rocksteady and Bebop." [...]
They are better people because each is capable of attachment, loyalty and concern for someone other than himself. But they are worse decision-makers because neither provides corrective friction. [...]
They are a two-person found family and a two-person moral hazard.](https://pbs.twimg.com/media/HOZ9J0QawAE_04Q.jpg)
6.8K views · 27 likes
· 1 reposts · 5 replies
Open on X →
My current weird hobby is having GPT-5.6 Pro "over-analyze" things
> I want a deep character study of Bebop and Rocksteady - I want to understand their motivations, their interests, their hopes and aspirations, how they differ from each other - really over-analyze this https://t
25.6K views · 234 likes
· 13 reposts · 51 replies
Open on X →
Against accounts of the same size
17 posts from the last 90 days, next to the 100K–1M follower range. shown to more people than peers of the same size.
Median views12 101this account5 983median for 100K–1M
Reach, %5.42%this account1.49%median for 100K–1M
Engagement, %0.98%this account0.93%median for 100K–1M
| Metric | This account | Median for 100K–1M | Ratio |
|---|---|---|---|
| Median views per post | 12 101 | 5 983 | 2.02× |
| Reach (views ÷ followers) | 5.42% | 1.49% | 3.64× |
| Engagement rate | 0.98% | 0.93% | 1.05× |
Others in this range → Compare with another account → How these benchmarks are built →
Growth & engagement
How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.
Views per post
72.6K31 Aug
268.5K1 Sep
4.3K
12.1K
11.3K
195.6K2 Sep
20.1K
2.8K
15.2K
22.9K
50K4 Sep
715
8.2K
422
Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.
Engagement rate per post
0.91%31 Aug
1.46%1 Sep
0.89%
3.29%
4.01%
1.13%2 Sep
0.74%
1.30%
0.77%
0.98%
1.33%4 Sep
0.28%
0.74%
0.47%
Reactions — likes, reposts, replies and quotes — divided by views. Median for 100K–1M accounts is 0.93%.
What the audience does
Likes76.8%8 181 in total
Reposts3.2%339 in total
Replies6.1%645 in total
Quotes1.1%114 in total
Bookmarks12.9%1 374 in total
Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.
The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.
Similar accounts
Adil Karaismailoğlu ✓
@akaraismailoglu
446.4K followers
· 9.8K posts
Anthrax
@Anthrax
446.4K followers
· 10.4K posts
Jonathan Calleri
@Jocalleri
446.3K followers
· 2.5K posts
Valérie Pécresse ✓
@vpecresse
446.3K followers
· 23.1K posts
GulfKanawut Official ✓
@Gulfkanawutofc
446.3K followers
· 46.5K posts
🩸 447k 🩸 ✓
@F_K__0
446.3K followers
· 40.2K posts
Jarosław Kuźniar
@JaroslawKuzniar
446.3K followers
· 109.7K posts
Guaranty Trust Bank Help ✓
@gtbank_help
446.3K followers
· 339.8K posts