Jon Minton
jonminton.bsky.social
Jon Minton
@jonminton.bsky.social

Senior Statistician. Eclectic public and population health background. Very reluctant Twitter refugee. Distant memories of academia. “Evidence informed contrarian”

Public Health 41%
Medicine 25%

If it’s like Among Us, Werewolf, etc, I’m guessing almost everyone is little better than chance in their guesses - because lying and lie-detecting were stuck in an evolutionary arms race for millions of years - but comes up with a lot of reasonable sounding explanations for their decisions?

Reposted by Jon Minton

pkgdown 2.2.0 now out — tidyverse.org/blog/2025/11.... Main feature is new build_llm_docs() which makes your websites easier to understand by LLMs. (Which you can easily turn off if desired) #rstats
pkgdown 2.2.0
The latest version of pkgdown automatically builds markdown files that make it easy for LLMs to use your website.
tidyverse.org

Reposted by Jon Minton

purrr 1.2.0 out now — mostly removing long deprecated functions but a few small performance nad parallel processing improvements. Learn more at www.tidyverse.org/blog/2025/11... #rstats
purrr 1.2.0
This release tightens up the package by removing long-deprecated functions, making `map_chr()` and predicate functions more type-safe, and requiring a newer version of carrier to make `in_parallel()`...
www.tidyverse.org

Someone on ‘the other place’ just unearthed and reposted this data vis I made a couple of years ago. Such is the decline in engagement- there or here- that even 8 likes now looks impressive!

New blog post: Claude as a dialectical engine?

jonminton.github.io/jon-blog/pos...

(Have LLMs now solved their sycophancy problem?)
Jon Minton’s Blog - Claude as a Dialectical Engine?
Accounts of two recent exchanges
jonminton.github.io

Long stolen antiques roadshow: set in the British Museum?

D’oh!
This in FT this morning 👇
This in FT this morning 👇

I’ve now experienced an LLM (Claude) start to ask me questions and push back on specific suggestions and assumptions I’ve made either implicitly or explicitly through the prompt. To me this suggests LLMs are moving from ‘Affirmation Engines’ to ‘Dialectical Engines’, a change I wasn’t expecting.

Reposted by Jon Minton

Would any fluent speakers of Korean or Chinese have a couple of minutes to review some updated pkgdown translations at github.com/r-lib/pkgdow... ? These were generated by claude code, so they're probably ok, but I'd love a human to double check.
✨ Proofread translations ✨ by hadley · Pull Request #2926 · r-lib/pkgdown
with Claude code cc @jayhesselberth @maelle
github.com

Climbing with Claude: understanding git/GitHub for collaborative version control:

jonminton.github.io/jon-blog/pos...
Jon Minton’s Blog - Climbing with Claude: Understanding Collaborative Version Control
An Artificially Hand-drawn Analogy
jonminton.github.io

9️⃣ Even nice guys "trade up" - Despite his idealism, TBL's personal life follows familiar patterns of successful men through serial monogamy

8️⃣ Apps as walled gardens - Mobile apps became profit-driven gatekeepers,

7️⃣ Data ownership revolution - His current passion: SOLID pods that would give individuals ownership and control over their personal data

6️⃣ Household microcultures matter - TBL's eccentric brilliance flourished in a family that encouraged unconventional pursuits without judgment

5️⃣ The semantic web's legacy - While TBL's vision for machine-readable data didn't materialize as hoped, it may have prevented today's eloquent but unreliable LLMs

4️⃣ Information, not engagement - TBL designed the web for information discovery (like Wikipedia), not the addictive engagement models dominating today's commercial web

3️⃣ Dancing between structures - The US capitalized on Europe's invention because it balanced top-down internet infrastructure with bottom-up university experimentation

2️⃣ The myth of efficiency - CERN's "inefficient" tolerance of TBL's experimentation gave birth to the web, suggesting rigid organizational efficiency can kill innovation

1️⃣ Purpose over profit - TBL chose to keep the web open and free rather than monetize it, prioritizing universal access over personal wealth

Here’s Claude’s scarily accurate summary of my 9 ideas:
Nine thoughts and ideas inspired by reading Tim Berners-Lee's This is for Everyone: (From Monetisation to Marriage Markets)

jonminton.github.io/jon-blog/pos...
LinkedIn
This link will take you to a page that’s not on LinkedIn
lnkd.in

Nine thoughts and ideas inspired by reading Tim Berners-Lee's This is for Everyone: (From Monetisation to Marriage Markets)

jonminton.github.io/jon-blog/pos...
LinkedIn
This link will take you to a page that’s not on LinkedIn
lnkd.in

This seems borderline religious thinking, essentially posing Antifa as a kind of Satanic presence

Has the child prodigy who made the game Theme Park in 1994 ‘solved intelligence’?

en.wikipedia.org/wiki/Theme_P...

jonminton.github.io/jon-blog/pos...
Theme Park (video game) - Wikipedia
en.wikipedia.org

New blog post: the man who solved intelligence - Demis Hassabis and the Thinking Game:

jonminton.github.io/jon-blog/pos...

(I’m not even sure the title is hyperbole…)
Jon Minton’s Blog - The Man Who Solved Intelligence
Some thoughts on Demis Hassabis and The Thinking Game
jonminton.github.io

Hasn’t it been a centuries-long convention that writers for the Economist tend to be anonymous or pseudonymous? (Which in a way seems oddly collectivist/socialist given the magazine’s title!)

Reposted by Tim Bale, Jon Minton

I’ve written about this game as a way of thinking about modern gaming conventions :

jonminton.github.io/jon-blog/pos...
Jon Minton’s Blog - Remembering KGB: 1992’s subtly terrifying social poison simulator
jonminton.github.io

Reposted by Richard Shaw

KGB: 1992’s subtly terrifying social poison simulator:

jonminton.github.io/jon-blog/pos...
Jon Minton’s Blog - Remembering KGB: 1992’s subtly terrifying social poison simulator
jonminton.github.io

Trump speaks with mismatching parentheses [}