60 Comments
User's avatar
Aris C's avatar

It is extremely difficult to approach Dominic Cummings with goodwill, given his selfishness and hypocrisy, and you should at least acknowledge these before boosting him.

Once you remember his 2001 portrait photoshoot you'll never take this rambles against politicians who only care about their image seriously again. And you'll never believe his has an ounce of integrity and belief in democracy when you remember that he planned to backstab Boris as soon as he got him into no. 10.

https://x.com/BeardedGenius/status/1264936171517489160

Expand full comment
Jonathan Nankivell's avatar

I used to be pretty pro-Dominic Cummings back in the day. This is because I was young, because he is so often right, because of the entertaining writing style he employed and because all the books he recommended that I went on to read (*The Secret of Apollo*, *From Third World to First* and *Britain Against Napoleon*) turned out to be very high quality and really interesting.

But I've recently read a lot of the history-of-the-Covid-pandemic books that have come out and the extra details and perspectives have negatively affected my view of him.

First I want to establish that the books to portrayed him more favourably than I'd expected. There is no guilty by implication: "[gasp] he mentioned IQ". There is no complaining about Brexit (apart from, rightly, the prorogation of Parliament). Most authors emphasise his earliness in taking Covid seriously and his historic role in getting the PM to take it seriously too. In the histories he comes across as incredibly sharp, politically wily and having an unusually strong grasp of the big ideas that explain all things. His early action saved many many lives (the NHS was mid-collapse in April 2020) and he is rightly lauded as one of the few insiders who understood the situation and who was willing and able to speak up. I don't mean to dismiss his achievement; many would have done worse. But I'd expected him to be good at the projecting-lines-on-a-log-scale-graph part of the job and his success at that wasn't a surprise.

The negative update I had to make was that, having spent a decade studying highly productive cultures and organisations - Apollo, Bell Labs, Skunkworks, Xerox PARC, etc. - he got himself into a position of power (de facto chief of staff at Number 10) and immediately created a culture of toxicity, conflict and insults. Not the slightly abrasive, truth-speaking culture that leads to effective decision making, but the caricature of it where few feel safe to speak their mind and decisions flip-flop as events that someone in the room could definitely have seen coming force each U-turn into a S-turn then into a full slalom course. Is that the culture George Mueller would have nurtured? Is that the place Douglas Engelbart would have wanted to work?

Dominic remains as crafty as they come. He is still definitely worth reading, listening to and taking seriously. But the gap between his understanding of effective organisations and the actual culture he helped create is vast. The combative writing style I'd so enjoyed might have been a warning sign. The attitude behind it might have been the cause of dysfunction. Either way, the self-image he projects through his writing has now been tempered, for me at least, by numerous outside perspectives.

Expand full comment
Aris C's avatar

How is he incredibly sharp? I'll grant you he often has common sense, which is often lacking in politics... but his posts are long-winded and rambling, and devoid of original insights.

(To be clear, I don't expect original insights from most people - especially in government, where I think most big problems have solutions that are easy technically, but incredibly tough politically.)

But even re common sense... he is one of those people who really need to understand Chesterton's fence, especially when he criticises orgs for not moving fast enough. The fact is, things are often slow due to checks and balances that are there because the reality is many people making decisions are incompetent or corrupt. The fact that these are accusations he levels at the people he himself got elected tells you everything you need to know.

Finally, I really cannot stand people who lower the bar by doing things like lying when it's expedient, or leaking information, etc etc, but then accuse others who do the same.

Expand full comment
Jonathan Nankivell's avatar

Ignoring Chesterton's fence rings true as a criticism, particularly with regard to the role of parliament during the pandemic. Quoting from *Emergency State* by Adam Wagner: "When Parliament amended the Public Health Act in 2008, MPs and members of the Lords did not anticipate that regulations made under it would be used to lock down the entire country for months, even less that there would stay-at-home or gathering restrictions for 489 days, and that the emergency procedure would be used for 763 days to prevent Parliament from having a prior vote or debating on nearly all of the over one hundred changes to the regulations from 2020 to 2022." This is dangerous precedent. Unless you understand the role of parliament, don't sideline it. Unless you can find an alternative method for scrutinising proposals, suggesting amendments, implementing them into law, don't replace it with a rule-by-decree model and hope things work out. That fence is there for a reason; it's to keep out demagogues and tyrants.

The lying, leaking and rule breaking culture while obviously very bad is something I attribute more to Johnson and Martin Reynolds than to Cummings. Cummings was complicit in this culture - he did his share of leaking and the "eye test" broke rules - and he had influence and maybe could have stopped it but he wasn't the ringleader and nor was this the thing he'd spent the previous decade preparing to fix. For the lying, leaking and rule breaking, Johnson should be blamed.

Expand full comment
Felix Hathaway's avatar

I also oscillate on Dom. He covers the same ground over and over again and I get a bit tired of it, but then something happens that he was right on years ago and I update back to 'very much worth listening to'. I do worry he doesn't have the strongest epistemic habits though, and I have seen him repeat things that are wrong in areas outside hi field.

Fwiw, the people I know who've worked with him all had positive things to say.

Expand full comment
Shaked Koplewitz's avatar

I think this is reasonable, but would extend this to Matt Yglesias (in general, I've soured on partisan political bloggers, where I mean "partisan" even in the weak sense of "you can obviously tell their party loyalty even if they're willing to criticize it").

Expand full comment
Ondrej Kubu's avatar

Is Gwern's Substack newsletter active? The latest article in archive is from 2021. I signed up at around his Dwarkesh interview and I did not get anything since.

Expand full comment
Daniel J's avatar

Wondering this too

Expand full comment
Zvi Mowshowitz's avatar

It no longer is, I have updated accordingly.

Expand full comment
Ondrej Kubu's avatar

Typo in One useful thing - Ethan Mollick.

Expand full comment
No Dumb Ideas's avatar

I’ll take you up on the offer to pitch in the comments:

I started No Dumb Ideas to explore absurd sounding ideas with economic rigor. Each post takes something that initially sounds outrageous and follows it to its logical conclusion. Topics can cut across economics, business, and history.

Some of my favorites:

Charge $1 to apply for a job: https://nodumbideas.substack.com/p/the-big-idea-charge-1-to-apply-to

Make jobs sellable and tradable: https://nodumbideas.substack.com/p/the-big-idea-a-job-you-can-buy-and

A history of gambling on the papacy: https://nodumbideas.substack.com/p/betting-on-the-pope-was-the-original

Expand full comment
Zach Stein-Perlman's avatar

Pitching my AI safety blog: I write about what AI companies are doing in terms of safety. My best recent post is https://ailabwatch.substack.com/p/ai-companies-eval-reports-mostly. See also my websites https://ailabwatch.org and https://aisafetyclaims.org collecting and analyzing public information on what companies are doing; my blog will soon be the main way to learn about new content on my sites.

Expand full comment
For the Sake of Argument's avatar

I'm trying my hand at public writing. Evaluating the potential actions of a unique geopolitical actor like Iran is an interesting challenge, and there are a lot of strange ideas about it out there. Read the whole thing for an attempt to apply rational actor theory to Iranian leaders.

https://ftsoa.substack.com/p/assessing-the-troubled-future-of

Expand full comment
Alethios's avatar

My blog focuses on issues related to institutional reform, particularly the public sector. I publish roughly one essay a month, but also have a series of podcasts recorded with guests at Lighthaven due to be released in the coming weeks.

A sample of recent essays:

https://alethios.substack.com/p/youth-and-leadership

https://alethios.substack.com/p/localism

Expand full comment
Sergio Diaz's avatar

Hi Zvi! Could you clarify what you mean when you say that sometimes Scott A frustrates you when talking about EA?

Expand full comment
Jeremy's avatar

You might check out the posts about FTX for some of Zvi's thoughts on EA

https://thezvi.substack.com/p/sadly-ftx

https://thezvi.substack.com/p/book-review-going-infinite

Expand full comment
Zvi Mowshowitz's avatar

He starts to become much more of a partisan trying to win arguments, and also lets challenges get to him, in ways that doesn't happen on other topics.

Expand full comment
avalancheGenesis's avatar

It's a weird mix of suddenly lowering the epistemic bar, but also pushing the Principle of Charity to absurd ends. The posts trying to redeem the souls of vitalists by insisting that caring about UK trafficking victims is only one step removed from starving orphans in Africa, and people could be Good if only they'd allow themselves to be Good, were...illustrative, I suppose is a polite term. When people tell you no, actually, they really are conflict theorists rather than mistake theorists - believe them!

Expand full comment
Jordan Rubin's avatar

FUTURE TOKENS: thinking strategies, working with LLMs, and the Future.

I am on garden leave after a career in quant finance. Currently advising Metaculus.

https://open.substack.com/pub/jordanmrubin

Expand full comment
Umang Jaipuria's avatar

No where near the quality of your recommendations, but my humble blog talking about tech organization culture and more recently AI (it has hijacked my brainspace) lives at https://forcemultipliers.substack.com/

Expand full comment
Stefan Kelly's avatar

Pitching my stuff on AI and psychology/culture. I mainly map the psychological consequences of technology and the 20th century onto the reactions we see to the current moment.

People might like these posts:

There's a reasonable chance that you're seriously running out of time

https://alreadyhappened.xyz/p/theres-a-reasonable-chance-that-youre

AGI isn't for happy people

https://alreadyhappened.xyz/p/agi-isnt-for-happy-people

The things you don't want to know about yourself

https://alreadyhappened.xyz/p/the-things-you-dont-want-to-know

Expand full comment
Jeffrey Soreff's avatar

Good posts! Do you have any comments on the recent Wired article on human/AI "couples" https://www.wired.com/story/couples-retreat-with-3-ai-chatbots-and-humans-who-love-them-replika-nomi-chatgpt/ ?

I commented on it myself in https://www.astralcodexten.com/p/open-thread-388/comment/130788057 . tl;dr; My key takeaway from the article is how _premature_ all of this is.

( BTW, I did your suggested "look at your last 400 searches - do they match your self-description" exercise - yeah, reasonably well, though I hadn't explicitly put "responds to ACX comments" in the self-description and that drove quite a few searches... )

Expand full comment
Hollis Robbins's avatar

Awesome! I look forward to being on this list when I start posting more than once a week! :)

Expand full comment
Chase Hasbrouck's avatar

I find curated lists like this very useful, thank you. Some other recommendations not on your list in Subject/Name/Title format:

* Apple: John Gruber, Daring Fireball

* Finance: Matt Levine, Money Stuff

* Tech/Data: Benn Stancil, benn.substack

* AI Engineering: Simon Willison, Simon Willison's Weblog

* General: Patrick McKenzie, Complex Systems (a podcast, but full transcripts, annotated with errata and further commentary, are online)

(My Substack is currently inactive, but you're welcome to subscribe to it in case I resume it in the future: https://trustedtech.substack.com)

Expand full comment
Zvi Mowshowitz's avatar

Should have mentioned Patrick McKenzie's blog and I just missed it somehow. That's definitely there.

Matt Levine is the One True Newsletter, but I stopped reading, mostly because my subscription got messed up and I don't know how to get it back (which is a lame excuse, but I think I realized that I had to make some cuts somewhere?)

Expand full comment
Chase Hasbrouck's avatar

I figured most of these were common knowledge, but you never know, someone might be one of today's lucky 10,000. (https://xkcd.com/1053/)

Money Stuff is at https://www.bloomberg.com/account/newsletters/money-stuff (and free!) though maybe if you have a paid Bloomberg account it's different/messes with it somehow.

Expand full comment
Eric R. Ward's avatar

Make sure NOT to skip the guest essay in Astral Codex Ten this week on the Alpha School in Austin, TX. First person, relatively objective account from someone who moved across the country to put his children in it. Highly relevant to your excellent posts about childhood education.

http://astralcodexten.com/p/your-review-alpha-school

Expand full comment
Andrew Miller's avatar

I write CHANGING LANES about mobility innovation: driving automation, public transit policy, anything else to do with improving how people move around. It's at

www.ChangingLanesNewsletter.com

This is a link to my 'best of' page: https://www.changinglanesnewsletter.com/p/welcome-to-changing-lanes

Expand full comment
Shaked Koplewitz's avatar

+1 that this is a good one, although it doesn't quite pass my paywall bar so I miss most of the value now.

Expand full comment
Performative Bafflement's avatar

My favorite recent blog on the AI front is theahura's 12 grams of carbon - a lot of what he does is lucidly review the foundational AI papers in a way that brings the concepts driving current foundation model progress to light - I've learned a lot from reading and commenting on his blog, and he's quite responsive in comments:

https://theahura.substack.com/p/ilyas-30-papers-to-carmack-table

In terms of self promotion, I'm a generalist who uses math, papers, and data backed sources to surface interesting or surprising things, some of which touch on Zvi's occasional roundup topics. Some of my most popular posts:

Sex - how much are people really having? Basically none, it turns out

https://performativebafflement.substack.com/p/sex-much-more-than-you-wanted-to?r=17hw9h

More than 80% of police hours are wasted not solving any crime at all:

https://performativebafflement.substack.com/p/more-than-80-of-police-hours-are?r=17hw9h

You only get three things in mate optimization:

https://performativebafflement.substack.com/p/you-only-get-3-things-in-mate-optimization?r=17hw9h

The "ten thousand Phd's" and what it tells us about AI risk and the next trillion dollar companies:

https://performativebafflement.substack.com/p/the-ten-thousand-phds-and-what-it?r=17hw9h

The maximally pessimistic obesity and weight loss argument:

https://performativebafflement.substack.com/p/the-maximally-pessimistic-obesity?r=17hw9h

I also write about the evidence and physiology behind why you should exercise, and evidence-based fitness and training stuff in general (recovery, triathlons, hypertrophy weight training, coaching, environmental and interroceptive cues for higher performance, etc):

https://performativebafflement.substack.com/p/ae8c22a6-37a3-4e42-8468-24a298219c3e

Expand full comment