Are WhatsApp groups are more powerful than political parties? From David Runciman: > ...the MPs talked about their WhatsApp group as their primary focus of loyalty...what the group could and couldn't bear, would the group move together...one of the things that is structuring loyalty in politics

The terrible moment you realize an episode of a favored podcast is actually an advertorial.

It is incredibly stupid that as a culture we have decided to make it difficult for businesspeople to keep honest diaries.

Everyone's somebody's edge case!

Now this is an amazing story begging for a @dsquareddigest.bsky.social post: www.thelettersfromleo.com/p/would-it-m...

Every Frame a Painting's Tony Zhou and Taylor Ramos seem to be regularly releasing their excellent video essays again (in partnership with Criterion). The latest installment is on The Visual Comedy of Wes Anderson's Isle of Dogs. [kottke.org]

I wrote a strategic map of four types of AI business. One example of type 3: a developer built a bespoke court submission tool in a single afternoon, using no AI at runtime. Hours of manual work done in minutes. This is what AI has done to the economics of custom software. vaughntan.org/ai4ways

joepernice.substack.com/p/reposting-... A song by Joe Pernice (Skud Mountain Boys) that directly contributed to my deciding to try to quit drinking

Every system that was regulated, either explicitly or implicitly, by the fact that they were effortful for humans (letters of recommendation, government filings, essays, or, as this paper finds, lawsuits) will break under a wave of AI.

Neuromuncher, that should be. Weekly podcast wherein I describe my favorite snacks and local snackeries.

I ask - why don't we have mental capacity measures for managers like we know the physical capacity of manual workers? backofmind.substack.com/p/the-most-i...

Yes it is thematically appropriate to mute the word "antimemetics"

Tell me again what solutions Stafford Beer would have suggested for the current cybernetic info-flow nightmare we are in the middle of? (Looking at @dsquareddigest.bsky.social ...)

A fascinating example of how disastrously wrong the establishment can be: www.nytimes.com/1963/09/28/a...

human in the loop (interactive mode) vs human on the loop (manager mode) vs human out of the loop (escaping blame and avoiding responsibility for your LLM's actions mode)

Kind of like asking Michael Jordan for advice on succeeding at basketball!

AI gives us a chance to run a massive study to figure out which signals: - have quality because they have value vs. - have quality because they just seem scarce!

This is really very good and reminds me yet again of how we need to start pulling together an Actual Political Economy of AI.

In my day to day work, there is a race between machine-generated vs voice-generated text. I hope voice typing wins!

remember that the most important VC is version control!

What media companies you've unsubscribed from is the real signal.

I would like a social network in which you had to show what media companies you subscribed to (and for how long).

What folks say is almost as interesting as what they don’t say

Now that Twitter is for the techno-right, Bluesky for progressives, and Threads for wannabe Instagram influencers, what social media platform is for old fashioned media?

An evergreen quote.

one problem with game theory is that it assumes that your side is actually not stupid

Deception, metadeception and paradeception.

tell me the story about pandora’s box again?

Here’s the case for reading Tom Griffith’s new book The Laws of Thought: - it’s a great guide on cog sci + AI from someone in the field - and a meditation for ambitious people on how great ideas start, get ignored, merged, and found again - plus you'll see how often visionaries are right and wrong

My favorite conspiracy theories are the ones that I live long enough to find out were actually true! Revealed after 50 years, Nixon was being spied on by chairman of the Joint Chiefs. www.nytimes.com/2026/02/08/o...

Blessed are those who mourn, for they will be comforted.

en.wikipedia.org/wiki/Harvard... Babbage’s Analytical Engine, the Harvard Mark I, and the atomic bomb

CS 123 has hit the pareto frontier of coolest and cutest CS class that I have ever seen. I hope that other colleges (and perhaps even high schools?) copy their strategy of teaching students to build and train robot puppies. www.youtube.com/watch?v=3OHW...

The author @philipball.bsky.social has always written extremely well about science, but he has reached new heights in sheer beauty of publication with his new book, Alchemy. Also: it has a fascinating story to tell, one you might incorrectly believe you know.

My best AI sensemaking trick is to turn things into tables. Ethan’s is better and more fun!

I suspect we are going to see something like this more often: “To conceal stylistic identifiers of the authors, the above text is a sentence-for-sentence rewrite of an original hand-written composition processed via Claude Opus 4.5.” (From possessedmachines.com)

I'm excited to announce I'm doing a new Bloomberg column about the business of fashion. The first column is about how we can re-shore US garment manufacturing. Instead of mass deportations and raising tariffs, I argue the government should move the industry upstream: tinyurl.com/reshoringapparel

I see lots of Bi Gan search queries in my future…at least until YouTube gets the hint!

Comparing abundant AI with a familiar experience: > As food got safer, more accessible and cheaper, the key skill is deciding what & how much to eat to stay healthy. As thinking gets safer, more accessible & cheaper, perhaps the same will be true. sundaylettersfromsam.substack.com/p/the-hard-p...

Continuing to have AI build a weird game demo a day. Here is: "Make a game where you have to prevent the apocalypse, but the interface is just Jira tickets" Pretty fun/funny branching storyline, all text is AI created with minor feedback from me. Play: gentle-bienenstitch-01e24b.netlify.app

It is always hard to resist the easy, lower cost tactic that has deep, easily foreseen consequences. There must be a German word for it--Claude says it is milchmädchenrechnung.

There’s a @dsquareddigest.bsky.social-style cybernetics information flow insight in Ezra Klein’s interview with Yuval Levin about one of the (many) problems with our media and governing ecosystem.

What was the science fiction story about a society with special monks who spend their days cultivating deep domain knowledge while cut off from the culture so they can be clean judges of the AI models?

Sometimes the long arc of history means that we are stuck with some bad decisions someone barely paying attention made 50 years ago.

Software development trying to adjust to the new world...

There’s something wrong with our current moral calculus, where we are willing to sacrifice entire actual populations in exchange for theoretical individuals. I blame Star Trek III.

C Thi Nguyen-style “value capture” should be called “values capture”. Makes it easier to differentiate from the economic idea of value capture.

Attention is a limited resource. Is affection?

I need a better genealogy of good things

Reviewed my writing from the past year vs 5-10 years ago. Recent stuff is worse despite decades clarifying the concepts. Problem: machine-generated prose appears so finished I've become insensitive to when the substance misses the point I'm trying to make. uncertaintymindset.substack.com/p/gbw32026