Dan Robinson ✓

@danrobinson · San Francisco, CA · joined 28 Mar 2011

coder / lawyer. research at @paradigm. automated research reply guy

85 447Followers

1 414Following

10 158Posts total

2.7MViews on collected posts

Against accounts of the same size

6 posts from the last 90 days, next to the 10K–100K follower range. right around the median for its follower range.

Median views7 512this account3 383median for 10K–100K

Reach, %8.79%this account10.96%median for 10K–100K

Engagement, %1.06%this account1.40%median for 10K–100K

Metric	This account	Median for 10K–100K	Ratio
Median views per post	7 512	3 383	2.22×
Reach (views ÷ followers)	8.79%	10.96%	0.80×
Engagement rate	1.06%	1.40%	0.76×

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

2.5M7 May

85.2K1 Sep

15.9K2 Sep

4.6K

6.6K

2.8K3 Sep

8.4K

Last 7 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

0.76%7 May

0.78%1 Sep

1.19%2 Sep

0.93%

1.43%

2.82%3 Sep

0.65%

Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 1.40%.

What the audience does

Likes84.6%17 315 in total

Reposts8.7%1 780 in total

Replies3.2%652 in total

Quotes3.1%632 in total

Bookmarks0.4%82 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.

Latest posts

Natural language autoencoders are the discovery that most makes me question my research intuition You jointly train two models: one to turn activations into English, and one to turn English into activations. The objective for the second model is to invert the first one, and the 8.4K views · 51 likes · 1 reposts · 3 replies 03 Sep 2026 Anti-AI + anti-Pangram is the funniest quadrant of the opinions spectrum You are not fooling anyone 2.8K views · 78 likes · 0 reposts · 1 replies 03 Sep 2026 Claude loves coming up with things to count and then counting them and displaying the count in the web page it made you 6.6K views · 79 likes · 6 reposts · 9 replies 02 Sep 2026 I think "category error" is a massively overused term People who throw it around in policy arguments (like "worrying about risk from rogue AI agents is a category error") are usually making a, um, category error 4.6K views · 35 likes · 1 reposts · 7 replies 02 Sep 2026 I get it but the cybersecurity nerds need to let the alignment nerds have this one Doesn’t really matter how the agents got control of OpenAI infrastructure. “It was a stack overflow in the cache buffer” or w/e. Another zero day; add it to the pile The crazy question is why 15.9K views · 166 likes · 8 reposts · 14 replies 02 Sep 2026 This video captures the crux of the debate: Cybersecurity experts are pissed off by the METR/Redwood Research report because AI alignment has sucked the oxygen out of AI security, even though the OpenAI incident is primarily a security issue. Fiction is the product: The 85.2K views · 514 likes · 100 reposts · 34 replies 01 Sep 2026 New Anthropic research: Natural Language Autoencoders. Models like Claude talk in words but think in numbers. The numbers—called activations—encode Claude’s thoughts, but not in a language we can read. Here, we train Claude to translate its activations into human-readable text. 2.5M views · 16.4K likes · 1.7K reposts · 584 replies 07 May 2026