DeLong's Grasping Reality Weblog

Annals of Technologies of Cognitive Alertness: Qahveh Khaneh Comes to Telegraph Avenue: How Coffee’s 600-Year-Old Birthplace Has Filled a Peet’s-Shaped Hole:

There is a Peet’s-shaped hole in my memory of the corner of Dwight & Telegraph, Peet’s! The ur-roaster. The place from which the founders of Starbucks learned their trade in the early 1970s. The fountainhead of the entire American specialty-coffee revolution that the—alas! now-late but then-assistant economics professor Suzanne Scotchmer introduced me to, back when I was a undergraduate junior in 1980-1981 at George Howell’s Coffee Connection in The Garage at the corner, of Mt. Auburn St. and Kennedy St. in Cambridge.

Share

I cannot say Peet’s abandoned a prime piece of Southside retail frontage. Telegraph Avenue had better days back when bookstores were still magnets. Traffic at that Peet’s was never anything like, say, Caffé Strada which does an extraordinary land-office business all the time during term time when it is not raining. But Peet’s did abandon a piece of frontage that was, objectively, a very well-placed perching spot for people coming up to campus from the south and west.

Replacing it is something called Sana’a Café.

The Sana’a Café (or, rather, a Sana’a Café) <https://thesanaacafe.com/> has opened with a bracing very strong pull—lots of grounds in the bottom of the cup:

Give a gift subscription

Thats blurry. Here’s the text OCR’d and reformatted:

SANA’A CAFÉ

ESPRESSO DRINKS (Hot | Iced)

Espresso (2oz) — $4.10
Cortado (4oz) — $4.49
Americano — $4.39 / $4.89
Cappuccino (12oz) — $4.95 / $5.45
Latte — $5.50 / $6.00
Caramel Macchiato — $6.79 / $7.29
White Mocha Latte — $6.59 / $7.09
Spanish Latte — $6.19 / $6.69
Yemeni Latte (Cardamom and Cinnamon) — $6.19 / $6.69
Sana’a Latte (Cardamom, Cinnamon, and Ginger) — $6.19 / $6.69
Matcha — $5.89 / $6.39
Pistachio Latte — $6.99 / $7.49
Hot Chocolate — $3.99 / $4.49
Mocha Latte — $5.95 / $6.45
Rose Latte — $6.19 / $6.69
Cinnamon Dolce Latte — $5.99 / $6.49
Shaken Iced Brown Sugar — $7.49 / $7.99
Biscoff Latte — $6.45 / $6.95

YEMENI SPECIALTY DRINKS (12oz | 16oz)

Sana’ani (Coffee with Cardamom and Cinnamon) — $4.99 / $5.99
Jubani (Coffee with Cardamom and Cinnamon) — $4.99 / $5.99
Arabic Light Roast (Light Roast Coffee with Cardamom and Cinnamon) — $4.99 / $5.99
Mofawar (Medium Roast Coffee with Cardamom and Cinnamon) — $4.99 / $5.99
Qishr (Coffee with Husk of the Coffee with Cardamom and Cinnamon) — $4.99 / $5.99
Adeni Chai (Yemeni Chai with Cardamom and Cinnamon) — $4.99 / $5.99
Yemeni Black Tea (Yemeni Black Tea, Cardamom and Cloves) — $4.99 / $5.99
Turkish Coffee (Medium and Dark Roast Coffee Only) — $4.99 / $5.99

DINE-IN

Small Pot (18oz) — $9.95
Medium Pot (24oz) — $15.95
Large Pot (32oz) — $19.95
Traveler Pot (96oz) — $29.95

ICED YEMENI SPECIALTY DRINKS

Iced Adeni Chai — $5.99
Iced Mofawar — $5.99
Iced Sana’ani — $5.49
Iced Dirty Chai (with espresso shot) — $6.99

ADD-ONS

Alternative Milk — $0.75
Add Flavor — $0.75
Add Shot — $1.75
Add Honey — $0.75

REFRESHERS

Mango Passion Fruit — $6.19
Kiwi Dragon Fruit — $6.19
Mango Watermelon — $6.19
Lemonade — $6.19
Mojito (with any flavor) — $6.99 (Choice of sparkling, water, or milk base)
Strawberry Lemonade — $6.19
Dragon Fruit — $6.19
Strawberry Acai — $7.19
Pink Drink — $7.19
Lemon Ginger — $5.59

SMOOTHIES

Mango Smoothie — $6.95
Strawberry Banana — $6.95
Strawberry Smoothie — $6.95
Banana Smoothie — $6.95

TEA (Hot or Iced)

Wild Peach — $4.99
Hibiscus — $4.79
Tangerine — $4.99

Share DeLong's Grasping Reality Weblog

It is one node in a rapidly branching network that did not exist anywhere on earth four years ago.

We tell ourselves that coffee comes from Ethiopia. That is true, in the narrow botanical sense. Coffea arabica is native to the highland forests of Kaffa, in the southwest of Ethiopia, where it grew wild and largely unregarded for a very long time. But a wild plant is not a beverage. And a beverage is not a culture. The plant is an accident of biogeography. The drink—coffee as a thing you brew, share, argue over, and organize your social life around—was a human invention. It was invented in Yemen.

Sometime soon after 1400, coffee made the short but decisive hop across the Red Sea, and it was Yemeni Sufi mystics who first put it to systematic use: they brewed the seeds and husks of the coffee cherry to stay awake through their long nocturnal devotions. Coffee was at the get-go not a consumer pleasure but a religious productive—salvation-generating—technology of wakefulness—a productivity enhancement for the spiritually ambitious. Graduate students bent over their laptops in the new Sana’a frantically booting their am or pm wakefulness are, whether they know it or not, the direct cultural descendants of 1400s dervishes trying to squeeze more devotional hours. The application has secularized. The pharmacology and, largely, the sociology have not changed.

From the Sufi lodges, the practice diffused outward. It reached into ordinary Yemeni society. By the middle of the 1400s, we are told the qahveh khaneh had opened their doors in Yemeni cities. Everything downstream of that traces its lineage back to Yemen: the coffeehouses of Cairo and Constantinople, of Mecca and Isfahan, and eventually the London coffeehouses, out of which Lloyd’s of London and the stock exchange and a good chunk of the eighteenth-century public sphere would crystallize. When Jürgen Habermas wants to locate the birthplace of the bourgeois public sphere, he points to the coffeehouse. He might have pointed further back to Yemen and to the qahveh khaneh of Coffea arabica, with caffeine-laden coffee being (along with caffeine-laden tea and nicotine) one of the three major early modern biotechnologies of cognitive alertness.

There is the port. For roughly three centuries, from 1450 to 1650, the bulk of the coffee that reached the wider world was grown in Yemen and then left through the single Red Sea harbor of al-Makha. The beans that passed through it became known simply as “mocha”. The name has outlived the port’s near-monopoly by three hundred years. It has now attached itself, with magnificent inaccuracy, to a drink of espresso and chocolate that has nothing whatever to do with al-Makha, but is instead a weird marketing accident. (Some claim that Yemeni beans, grown at altitude on ancient terraces and dried whole in the sun, carry a genuinely cocoa-like depth.)

Yemen, of course, understood exactly what it had. For nearly two hundred years it ran one of history’s most successful commodity cartels. Green beans were boiled or roasted before export so that they could not germinate abroad in an early-modern equivalent of DRM. Smuggling live plants out of the country was punishable. It worked, more or less, until it didn’t: by 1625 Dutch traders had gotten viable plants out into the botanical gardens of Amsterdam. Then they leaped, to Ceylon, to Java, and by the mid-1700s coffee was growing across the Americas, Yemen’s monopoly was simply gone, and the fact that all of coffee’s pests and parasites were located in the horn of Africa meant that Yemen and Ethiopia faced large production disadvantages vis-à-vis producers in the rest of the world.

What survived the collapse of the monopoly was the culture.

Qishr—brewed from the dried husks of the coffee cherry rather than the bean, spiced with ginger and cardamom, naturally low in caffeine, originally, I am told, a poor-man’s drink for coffee workers who could not afford beans that had had their price boosted by export demand—is still drunk across Yemen.

Get 75% off a group subscription

It is on the menu at Dwight and Telegraph, priced at $4.99 for twelve ounces. That is a sentence that would have been unintelligible to any human being alive before about 2023. (I have not yet tried it.)

There is now a Yemeni-style coffeehouse miniboom. The Associated Press tells me that the number of cafés operated by six Yemeni-style chains in the United States grew by fifty percent in 2025 to 136 locations. The oldest of these, Qahwah House, opened its first store in Dearborn, Michigan, in 2017—Dearborn being home to one of the largest Arab American communities in the country, a legacy of the auto industry’s twentieth-century labor demand. It now runs more than twenty locations.

There is a clear niche for a non-chain, single-location, local-focused coffeehouse in a world in which there are 700 Starbuckses and 130 Peet’ses in Greater San Francisco. But is there a niche for another chain? The idea isnot “beat Starbucks at being Starbucks”, but to occupy a position Starbucks structurally cannot occupy with a favorable cost structure. I see four things at work here:

1. Product differentiation: Starbuckses could add a cardamom latte tomorrow; it could never be Yemeni. That’s a moat made of identity, and it lets them sidestep price comparison, for there is no reference price for a Sana’ani.

2. Late-night: Starbuckses close early and have spent a decade engineering out lingering, and so there is a place for serving the 9 p.m. slot Starbuckses will not occupy.

3. Co-ethnicity: community financing, lower monitoring costs, and labor supplied by cousins and friends of cousins because it’s kin and it’s “purpose, not work”; the classic immigrant-entrepreneurship advantage of sweat equity, trust networks, and trade diasporas doing work.

4. Customer base: Arab-American and the broader “somewhere to go that isn’t a bar” crowd, plus the curious.

5. Incumbent retreat: Peet’ses and Starbuckses spent 40 years teaching Americans to pay $6. Sana’a inherits that trained demand and the physical carcasses. Low build-out cost, pre-qualified locations, demand already conditioned.

Of course, “authenticity”, “community”, and immigrant-networks and sweat equity powerfully erode with scale.

The city of Sana’a:

Sana’a is not on the seacoast. Sana’a is not a port. Sana’a is located in the western-highland geographic spine that is the terraced-slopes highland coffee-growing region. It is the traditional capital of Yemen. (It is the capital of the Houthi-controlled part of Yemen; the rest is ruled from Aden.) It is Yemen’s largest city, sitting at about 7,500 feet. Its old walled city is a UNESCO World Heritage Site.

Sana’a Café began in downtown San Francisco in 2023 as a family business. A cinnamon-cardamom combination in a strength never seen outside Yemen. Ginger, saffron, and honeycomb bread in an elaborate and very nonstandard pastry case (baklava, milk cakes, pistachio cheesecakes). Now Sana’a Café counts on the order of sixteen locations in California, plus one in Washington state; has taken over a former Starbucks in San Francisco’s Financial District at 27 Drumm Street; has three storefronts in the Sacramento area; and has one in Hollywood and one in Orange County. In the words of co-founder Sadeq Aleqel:

Sadeq Aleqel: Exploring Life & Business <https://voyagela.com/interview/exploring-life-business-with-sadeq-alaqel-of-sanaa-cafe/>: ‘The response from the community was immediate and heartfelt—people weren’t just discovering coffee, they were experiencing a culture, a story, and a centuries-old tradition in every cup…. Our focus is on Yemeni-style coffee and drinks, prepared with care and authenticity…. Serving Yemeni coffee and sharing our heritage doesn’t feel like work; it feels like purpose…. For us, the greatest risk would be not sharing Yemen’s coffee heritage at all….

The Dwight & Telegrph in Berkeley café location opened up, I have been told, in late July. The Daily Cal was soon reporting that at 11 pm you could see it full of summer-school students, heads down, studying, drawn by the midnight closing time, and by coffee mixed with amounts of cardomom, cinnamon, and other spices that you would never find in Italy or France.

And by the honeycomb pastries.

But how is it going to do for me this fall? The floors from the subbasement through four of Evans Hall have now been shut. The occupied floors are reduced to five through ten, with economics on floors five and six, and math on floors seven through ten. (You may ask: Why? Doesn’t economics have as many professors, more graduate students, and nearly aleph-null times as many undergraduates majors as math? Don’t ask.) That closing-off of the bottom half of the building does have consequences. For one thing, no drink vending machines. for another, no library on floor one to pick up things delivered from the remote book stacks. So I find myself going over to the main library, and then finding a chair there and perching with my laptop rather than going back up the seven flights of stairs. (Elevators, you say? Don’t ask.)

Could Sana’a Café this fall be, for me, a place to (a) drink good coffee, (b) hold office hours that are enjoyable for people to come to, and (c) have a chance of running into people by accident?

The dervishes knew coffee as a biotechnological device for extending the productive day, for wringing more devotional hours out of a body that wanted to sleep. That function has not gone anywhere. But three others have grown up around it, for the coffeehouse also matters. The drink narrows and sharpens attention. The room does more.

One defining pathology of intellectual life since about 2010 is that we have become extraordinarily good at finding exactly the information and exactly the people we already know we want, and correspondingly terrible at encountering the ones we did not know to look for. The search bar, the calendar invite, the scheduled Zoom—these are instruments of intention. They deliver what you asked for. They are useless for delivering what you had no way to ask for because you did not yet know it existed. As my teacher David Landes once said: you have to have a stack pass and go into the stacks of the library because the thing you want is not the thing you think you want but the thing three slots to the left.

The café is one of the last machines we have for manufacturing the unplanned collisions. Jane Jacobs had a vision of a dense and various public space that is useful precisely because you cannot predict what it will throw at you. A coffeehouse is such a sidewalk ballet with seating and a caffeine subsidy.

There is also the coffeehouse is a device for shifting frame. You think differently in a mild ambient hum than in silence. A change of physical location can unstick a stuck mental problem, that has you going in circles. Being among strangers who neither know nor care what you are working on can loosen the grip of a malevolent internal editor in a way that a familiar desk never will. The café supplies enough stimulation to keep the vigilance system idling, but not so much that it captures you.

Coffeehouses are an efficient bundling of: a legitimate reason to leave the house; a socially sanctioned excuse to be idle in public while your back-brain works; a stimulant delivered in a ritual that paces its consumption; and a nonzero probability of a useful accidental encounter.

Where else do you get all four for the price of a latte?

For me, it probably will not work for (c): Caffé Strada on eastsouthside is the champion, clearly; there is no parallel northside gathering place; and Sana’a’s location is very off of all beaten tracks for economists. (a) is very attainable. (b), perhaps as well.

Subscribe now

Leave a comment

If reading this gets you Value Above Replacement, then become a free subscriber to this newsletter. And forward it! And if your VAR from this newsletter is in the three digits or more each year, please become a paid subscriber! I am trying to make you readers—and myself—smarter. Please tell me if I succeed, or how I fail…##sanaa-cafe-dwight-telegraph-friday-fog-flatlands
##enlarging-the-bounds-of-human-empire
##public-reason
#yemeni-coffee
#sanaa-cafe
#coffee-history
#qahveh-khaneh
#coffee-connection
#coffea-arabica
#qishr
#cardamom-cinnamon-latte
#honeycomb-pastries

My interpretation of what is going on here: Gerard Baker saying “I am anti-Trump too!”, in the hope of shifting to firmer ground on which he can argue that he and his friends should get to keep all their ill-gotten gains from the Trump tax cuts. After all, aren’t the only things that matter (a) that, (b) making liberals sad and (c ) keeping non-”white” non-male non-elderly people in their place?:

So how come Google Maps now calls the Gulf of Mexico and Lake Ontario the Gulf of and Lake America, but does not yet show New Mexico relabled as “New America” anyway?

Share

Trump calling for the renaming of New Mexico

Give a gift subscription

We have the absolutely remarkable, in so many ways:

Gerard Baker: Help—I Feel a Bad Case of TDS Coming On <https://www.wsj.com/opinion/helpi-feel-a-bad-case-of-tds-coming-on-0d2515aa>: ‘A signal achievement [of Trump] has been getting control of the border and removing some of the people who have no legal right to be here. But even with this proper and overwhelmingly popular policy, the administration has managed to harm the country and the GOP’s standing. The unnecessarily cruel and indiscriminate approach has caught many in its dragnet who shouldn’t be treated that way and is fundamentally un-American. The political effect has been to devastate support for the GOP among Hispanics, more than reversing the gains Mr. Trump had made with the fastest-growing ethnic population in the country…

Share DeLong's Grasping Reality Weblog

In what sense can this possibly be a “signal achievement” if it is “cruel and indiscriminate… caught many… who shouldn’t be treated that way… fundamentally un-American?”

Baker’s closing:

The looming Republican calamity… will be the result of presidential hubris, vanity, recklessness and the wider party’s abdication of accountability. Instead of seizing an unrivaled opportunity to shift America off the path of progressive endangerment, the GOP risks handing it back to the Democrats and their own obsessions. It’s enough to give anyone a case of Trump derangement syndrome.

Get 75% off a group subscription

Given Baker’s earlier—I am not sure what to call it? admission?—the fact that it is “the wider party’s” rather than my own “abdication of responsibility” is an interesting attempted evasion to see at this late date. What was the earlier “admission” you ask? This:

Jan. 6… removed any doubt about the threat Donald Trump might pose to our constitutional norms and rules, [but] I thought the institutions of the U.S. would… constrain his worst instincts and intentions. And by November 2024, after four years of a Democratic presidency had revived the prospect of a restoration of the progressive left’s crippling hegemony, the risks he posed seemed smaller than theirs…

Refer a friend

But is anyone other than Gerard Baker really surprised by any of:

a self-indulgent tariff fixation based on a profound misunderstanding of international economics, an insouciance about public finances that has widened the deficit, and a reckless military adventure that has choked off global energy supplies,.. alienat[ing] everyone in the world who could be helpful to us (perhaps excepting Israel)… embolden[ing] and even embrac[ing] every country in the world that threatens us… the grotesque self-enrichment, the self-satisfying prosecutorial pursuit of political enemies, the endless projects of self-grandiosity… ballroom… Kennedy Center… Memorial Arch. …the buffoonish symbolic acts… the Gulf of America, Lake America, New America, the 51st state [of Canada]...

?

No. The only surprising thing is not any of Trump’s lunacies but, rather, that it is without a peep that the Republican congressional majorities have gone along with it all, on the Gingrich Principle that above all one must and the only thing that matters is to make the president of one’s own party look good—not actually be good, but look good—and the president of the other party look bad, no matter what the cost to the country. Thus it is not any individual one of these. It is that all of them have happened because there is no block of Republican senators and representatives willing to stand up and even threaten to, temporarily, cross the aisle to make Jeffries and Schumer Speaker and Majority Leader for one single day.

And, by the way, how the f*** does Gerard Baker imagine that Binyamin Netanyahu could possibly be “useful” in any way to anyone?

Perhaps if Baker had written this in the winter of 2025 it might have done a little bit of good, and one might be obliged to at least think of him as being, very partially, and incompletely, something one might consider calling “a man”. But at this late date?

This smells very much like somebody who wants to still be in the game after November if there is in fact a decisively Democratic congress. Thus he is now trying to join the anti-Trump parade that he has spent so much of the past eight years belittling and trying to destroy.

Subscribe now

Leave a comment

If reading this gets you Value Above Replacement, then become a free subscriber to this newsletter. And forward it! And if your VAR from this newsletter is in the three digits or more each year, please become a paid subscriber! I am trying to make you readers—and myself—smarter. Please tell me if I succeed, or how I fail…##gerard-baker-bullshit-artist-of-the-day
##public-reason
##neofascism
##journamalism
##bullshit-artist-of-the-day
#gerard-baker
#trump-derangement-syndrome
#wall-street-journal
#trump-buyers-remorse

But two observations: (1) the spike is not that big, and (2) from an “affordability” perspective, this is the last thing one would want to generate from one’s foreign policy:

A real price series that I have been watching:

And we have a related observation from Matt Yglesias:

Matt Yglesias: The Republican Urge to Start Wars in the Middle East <https://www.slowboring.com/p/the-republican-urge-to-start-wars>: ‘Despite Trump’s anti-war rhetoric on the campaign trail, the national security team he announced shortly after the election was “an extremely hawkish group that screams ‘hot war with Iran’” <https://www.slowboring.com/p/how-donald-trump-could-succeed>. In retrospect, I wish I’d made a bigger deal about this….

There’s a weird finding in the economics literature that Democratic Party presidents enjoy systematically better economic performance even though they don’t pursue systematically better economic policy…. Alan Blinder and Mark Watson… [say] “the Democratic edge stems mainly from more benign oil shocks, superior total-factor-productivity (T.F.P.) performance, a more favorable international environment, and perhaps more optimistic consumer expectations about the near-term future.” They seem to primarily understand this as a question of good luck. But… Republicans have systematically bad foreign policy that is inattentive to the downside risk of saber-rattling and hawkishness…. The… problem with Bush-era foreign policy wasn’t belief in democracy; it was excessive belief in the potential upside of coercive military force. Trump has retained the error while reconfiguring the goals…

And the problems with Nixon-era foreign policy as it affected the economy was (a) Kissinger thought high oil prices were a good thing because they enabled the then-Shah of Iran to arm himself to the teeth and thus annoy the USSR on its southern border without Kissinger having to get congress to pass any military-aid bill, and (b) Nixon’s all-out rearmament of Israel in the 1973 war (against the wishes of his own Pentagon, which feared it would “blow any image we may have as an honest broker" and alienate the Arab states, which it did), thus provoking an OAPEC reaction in the form of the oil embargo.

MEMO:

• Dine, Thomas A. 2022. “The Limits of Détente: The United States, the Soviet Union, & the Arab-Israeli Conflict”. Journal of Cold War Studies. 24:3. <https://direct.mit.edu/jcws/article-abstract/24/3/255/112894/The-Limits-of-Detente-The-United-States-the-Soviet>.

• Nixon Foundation. 2023. “America’s Airlift in the 1973 Yom Kippur War.” Richard Nixon Foundation, October 10. <https://www.nixonfoundation.org/2023/10/americas-airlift-in-the-1973-yom-kippur-war/>.

• Office of the Historian. n.d. “The 1973 Arab-Israeli War.” . U.S. Department of State <https://history.state.gov/milestones/1969-1976/arab-israeli-war-1973>.

• Quandt, William B. 2001. Peace Process: American Diplomacy and the Arab-Israeli Conflict Since 1967. Washington, DC: Brookings Institution. <ISBN 9780520225152>.

• Samet, Daniel J. n.d. “October 1973: Nixon’s Decision to Resupply Israel.” Not Even Past. University of Texas at Austin. <https://notevenpast.org/october-1973-nixons-decision-to-resupply-israel/>.

• Tal, David. 2014. “A Tested Alliance: The American Airlift to Israel in the 1973 Yom Kippur War”. Israel Studies. 19:3 (Fall). <https://doi.org/10.2979/israelstudies.19.3.29>.

• Wikipedia. n.d. “Operation Nickel Grass.” Wikipedia: The Free Encyclopedia. <https://en.wikipedia.org/wiki/Operation_Nickel_Grass>.

• Yglesias, Matt. 2026. “The Republican Urge to Start Wars in the Middle East”. Slow Boring. September 23. <https://www.slowboring.com/p/the-republican-urge-to-start-wars>.

• Yglesias, Matt. 2024. “How Donald Trump Could Succeed: Or, More Likely, Fail”. Slow Boring. November 14. <https://www.slowboring.com/p/how-donald-trump-could-succeed>.

Subscribe now

Leave a comment

If reading this gets you Value Above Replacement, then become a free subscriber to this newsletter. And forward it! And if your VAR from this newsletter is in the three digits or more each year, please become a paid subscriber! I am trying to make you readers—and myself—smarter. Please tell me if I succeed, or how I fail…##real-price-of-diesel-chart-of-the-day
##neofascism
##chaos-monkey
##chart-of-the-day
#oapec-embargo
#republican-warmongering
#energy-prices
#blinder-watson
#donald-trump

Eliezer Yudkowsky and Nick Land may now be the two most influential intellectuals in America's AI debate—and that tells you how badly the conversation has gone wrong. When "machine god: good or evil?" is the only question on the table, every question actually worth asking gets thrown away. Contra Scott Aaronson, accepting the coming machine god is not now "the epistemic bare minimum" any more than is accepting Friedrich Engels’s belief 150 years ago that the SteamPower Economy was the real singularity that would in short order deliver a classless utopia of universal abundance:

I disagree with Henry Farrell’s subheadline here, which I think contains some other-side-of-the-Atlantic understatement. It is not that there are other ways of thinking about “AI” than “Digital God”, it is that it is difficult to think of a way less helpful for coherent thought to think about “AI” than to think of it as a “Digital God”.

Share

But I think he is right in the rest: That means that the important and live questions are about what metaphors—what abstraction layers, throwing away information so that the problems become things we can grasp and think about—are useful right now? Henry Farrell’s “MAMLMs are a socio-cultural technology” is one very useful one. The “‘AI’ is a normal disruptive Schumpeterian GPT” is a second. What are the others that we need to hold in our neocortices right now?":

Give a gift subscription

Programmable MutterTechnology and politics in an interdependent worldBy Henry FarrellCROSSPOST: HENRY FARRELL: Machine God Metaphors Eat Your Brain

<https://www.programmablemutter.com/p/machine-god-metaphors-eat-your-brain> <https://www.programmablemutter.com>

Programmable MutterMachine god metaphors eat your brainHere are three propositions…Read more10 days ago · 26 likes · 9 comments · Henry FarrellThere are other ways of thinking about AIHenry FarrellSep 22, 2026

Here are three propositions.

1. Metaphors are the fundamental abstraction layers in modern AI debates.

2. We are in a moment where you could plausibly argue that Eliezer Yudkowsky and Nick Land are the most influential policy intellectuals in the US conversation. Just look at the state of the two of them. This is not a good thing.

3. The collision of (1), (2), and recent events means that “machine gods” have now entered the discussion. This is a problem. Abstractions necessarily throw information away. The information that the “machine gods” metaphor/abstraction dumps is crucial to navigating the moment we are in. If public and policy debate stays centered around “machine gods: good or evil?” we are in for a bad time.

I should acknowledge that I (and co-authors) absolutely have our own dog in this fight. Indeed, I think that the “AI as social and cultural technology” approach that we have developed is a much better alternative. The final part of this newsletter will talk about a new piece by a mathematician that builds on this approach. Equally, my claim is not, actually, that our alternative approach should become the sole dominant way of thinking about AI. What I actually want, along the lines of my previous post, is a much greater degree of pluralism in the defining debates around a technology that no-one, including its creators, seems to understand particularly well or to be capable of usefully predicting.

******

As far as I can tell, the above Twitter exchange is the source of the most recent outbreak of Machine God Thought. People who know Daniel Kokotajlo tell me that I’d like him, and that we could have good conversation; I imagine they are right. I’ve found most of the rationalists I’ve met to be decent human beings; in many cases, unusually so. Equally, I think that there is something terribly limiting to the ways they talk about AI, and contend that it’s a real problem that debate is still accelerating past the event horizon towards the very heart of Singularity thought.

The two papers Kokotajlo mentions here, AI 2027 and AI 2040, lay out futures that are roughly identical in their predictions about what AI becomes (godlike; superhuman). Where they differ is in their assumptions about how well aligned AIs are with human goals. In the scenario where AI develops more or less unconstrained (thanks in part to geopolitics), AI breaks out of containment, and humanity is likely either Corgi-ized or goes extinct. In the scenario where AI development is delayed, tamed and made subject to “mutual compute destruction,”1 men and women come to have gods as servants, and start spreading out to nearby galaxies in the next two decades.

Neither future seems to be at all convincing to me, but they are most certainly convincing some people. Casey Newton and Kevin Roose are partnering with NPR to create the “Machine Gods” podcast as part of “a bold new era of content strategy for the network.” In his newsletter, Newton thanks the fates that he and his collaborator were a little too late to tech podcasting to be “beguiled” by Sam Bankman-Fried, and suggests just two paragraphs later that critics of Roose’s famous shoggoth freakout were “previewing the ways in which the commentariat would continually twist itself into pretzels to deny that anything important was happening with LLMs.” Indeed.

Things are no better elsewhere. In Washington DC, the AI policy debate is riven by a bitter dispute between those who take the Kokotajlo perspective (slow it down before we all die!) and those who want to let innovation rip so that we hit the technocapital singularity as soon as possible. The one side owes its framing to Eliezer Yudkowsky’s MIRI institute. The other builds on Nick Land Thought as filtered through the more immediate needs of Marc Andreessen. While I’m definetely more inclined toward the former than the latter, on the grounds that we do indeed need to have serious conversation about regulation, I think it would be terrible to have that conversation starting and stopping with the priorities of the labs.

Finally there’s Scott Aaronson, who I used to rely on for valuable debunkings of policy and media folk when they got over-excited about quantum computing. It is fair to say that Scott is not in a debunking mood these days, now that OpenAI has solved the Navier-Stokes problem. Witness the below:

The Age of Wonders and Terrors

Twenty years ago, when the idea of AI taking over the world in our lifetimes still struck most of us as the unconstrained fantasy of those who knew too much science fiction and too little science … update on the fact that the wild prophecies have come true. … It seems to me that the Singularity has already started; it’s just wildly unevenly distributed.

… For anyone who says AI doom sounds like an apocalyptic religion, that the rationalists/Singulatarians seem like a Bay Area cult, that Eliezer Yudkowsky gives off the vibes of a messianic prophet: yes, yes, and yes. But crucially, today you’re no longer being asked to believe in arguments and extrapolations, but only in the front-page news. Accepting the reality of the coming machine god after it’s solved Navier-Stokes and dozens of other longstanding open math problems (while dramatically ramping up in capability every month), is sort of like accepting Jesus after he’s returned to earth on the gleaming cloud. It’s the epistemic bare minimum.

Yes, there’s still enormous uncertainty about what the rest of our lives will look like, but as far as I can tell, there’s no longer any real uncertainty that it’ll all mostly revolve around AI, and the extent to which we succeed or fail at directing its power toward human flourishing.

******

So why do I think that the machine god approach, far from being an “epistemic bare minimum,” is a pernicious brain-devourer? The answer is that the metaphor shrinks rather than expands your horizons of the possible, forcing you to focus on the single overwhelming question of what Machine Jesus is going to do as soon as he descends from that gleaming cloud. Is he going to usher in paradise on earth? Or is he actually a shoggoth in a Jesus-mask, condemning us to enslavement or destruction? Once you have accepted that Machine Jesus is in the process of descending, all questions besides determining his intentions for humanity seem irrelevant.

I’ve written before (with colleagues and on my own) about why I think this whole way of thinking is utterly wrong-headed. But maybe I’m wrong. Take three somewhat inter-related possibilities: p1, that humanity is not the sole beautiful flower of the cosmos and that greater than human intelligence is perfectly possible, p2, that high intelligence is effectively impossible without strong intentionality, and p3, that current AI architectures are indeed the royal road to achieving proper generalizable superintelligence. I personally rate p1 as extremely high, p2 and p3 as very low, and p1*p2*p3, which is the most obvious state in which you might be worried about the AI alignment problem, as very low indeed. But who is to say that I might not be wrong? Equally, I, or someone who is skeptical of Machine Jesus for wildly different and incompatible reasons than my own, might be right. And if that is possible, then we ought not just be obsessing about whether we “succeed or fail in directing” His Awesome Power towards human ends. We might want to worry about other questions too, or instead.

There is a discussion of the solution of Navier-Stokes, the miracle that brought Scott to embrace the true faith, in a guest-post on Terence Tao’s blog, which usefully illustrates the point. It also has the benefit of being (sort-of) pre-registered. This discussion is specifically attractive to me, because it builds on arguments about AI as a social and cultural technology. Again, I don’t think that you need to believe that it is right on the merits to accept the more limited point that I want to make, which is that perspectives other than the machine god one reveal other questions and problems that we ought very reasonably want to be concerned with.

In his guest-post, the mathematician Nestor Guillen writes that he anticipated the solution of Navier-Stokes, but that his reaction was not that human mathematics was in danger of being replaced. Instead, he argued that it validated the entire enterprise.

the bulk of the essay was done in August and shared with a few close friends for feedback. Earlier this week, Tristan Buckmaster announced a breakthrough finding, done in collaboration with Levent Alpöge, of a finite-time blow up for 3D incompressible Euler with forcing. He also made serious allegations of misconduct by OpenAI (see announcement). The next day, OpenAI announced a finite-time blow up result for 3D Navier-Stokes with forcing, obtained from their internal LLM. … I have decided not to change the opening of the essay and instead make a note of the news here.

… What would be the significance if one day we woke up to the news that output from a Large Language Model contains a definitive answer to the question of blow up for the incompressible Navier-Stokes equations, and to the question of where the zeros of the Riemann zeta function lie?

Many of my fellow mathematicians consider this a kind of nightmare situation, and I think this is a very valid position. Here I would like to make a different case, however. Personally, I have come to believe that the news of a Large Model output containing an interesting, insightful, novel mathematical idea can and should be received as a source of communal pride for mathematics and for mathematicians.

Why is this so? Guillen says that:

Humans have always lived in a complex world and their survival has hinged on their ability to process information in amounts beyond what any individual can handle — and it is likely this has been so nearly as long as humans have existed. Herbert Simon’s concept of “bounded rationality” captures this well …

Simon saw that one implication of bounded rationality is that humans have to (and do) develop systems to process information, as a group. We see that humans have built states, stories, bureaucracies, markets, libraries, and more. These are cultural and social technologies. They are means by which humans (not individually but as a group) navigate amounts of information so vast that they cannot be handled by any individual on their own. …

LLMs … are “wrappers” around an older cultural technology: the scientific literature. When you ask an LLM a question about the isoperimetric inequality, the ABP maximum principle, or a PDE, you are effectively interacting with every person who has thought about and worked on those things — provided the data the LLM was trained on contained those contributions in one way or another.

Unsurprisingly, I think that this is largely correct (I do think that you can make the cultural and social technology argument, while also being freaked out about the consequences of AI: more on that in my next post). But even if it isn’t, it highlights different, and I think, important questions than the machine god nexus.

Specifically: if you think of AI as a cultural and social technology, your first reaction will not be to prostrate yourself in anticipation of the coming Age of the Gods. Instead, you are likely to treat fields like mathematics as some combination of individual and collective human intelligence, constructed on top of technologies that guide and direct that intelligence in particular directions. So when a new technology such as AI comes along, that reshapes the ways in which mathematicians draw on their shared intelligence, you will have questions to ask. What are the current processes through which existing knowledge is disseminated and new knowledge is created? What are the incentive structures for dissemination and creation, and how do they interact with each other? How have processes and incentives been shaped by previous and existing cultural and social technologies? How might these processes and incentives be changed by what is happening now?

And so on. You can see one way in which you might extend this in Ben Recht’s recent post, which suggests, if I understand him right, that mathematicians really ought to rethink how much of their field was already based around benchmarkmaxxing (prizes and challenges) even before AI came on the scene. Tao’s argument, as quoted in the NYT, points in a similar direction.

[Tao] compared OpenAI to a wilderness guide who finds a path to a hidden waterfall. “This has some value,” he said. “But once someone shows a specific path to the waterfall, people just take that path. They don’t spend as much time looking for other paths.”

I am not even an amateur mathematician, so am very much open to correction, but I understand Tao to be saying something like the following. Automated processes of discovery have obvious advantages in pointing the direction to new things. Equally, they push the needle from ‘explore’ to ‘exploit.’ Apparently aimless noodling around by humans can have benefits when it opens up unexpected views from Serendippo . More generally:

solving problems is only a tool and proxy for achieving the primary goal of conceptual understanding and insight. Forgetting this in the world of AI may turn the tool against the primary goal. Indeed, the mass production at faster and faster pace of “true/false” statements could destroy fertile ground instead of breathing life into new ideas.

Now it could be that AI will become universally better at discovery so that humans become irrelevant to math. That is the machine god assumption. But that assumption rules out, more or less by fiat, the possibility that human curiosity might still have independent value in discovering new possibilities, and all the questions that flow from it. If AI is better at some aspects, and humans are better at others, then you need to confront messy problems of organizational design. How do you rebuild the institutions of mathematics so that they best reconcile AI and human led approaches to discovery? How do you remake the relationship between dissemination of knowledge and discovery? What individual and organizational incentives seem best suited to achieve general goals? What general goals ought you have? None of these are easy questions, and there are likely to be complex tradeoffs between the solutions to one set of problems and the solutions to another.

I don’t think that an ‘AI as cultural and social technology’ approach provides immediate answers (beyond maybe ‘you really ought to read James March’). But the point is that it makes it possible to see that these questions may be important. If you start as Guillen does, from the perspective that AI does not inevitably replace humans in the dissemination and discovery of knowledge, so much as it reorganizes relations among them, then you begin to start asking questions that are systematically obscured by Machine God Thought. In fairness to Scott Aaronson, his post does suggest towards the end that he too cares about these questions. Perhaps when the excitement dies down, he will find better ways towards thinking about them systematically.

You could, of course, make similar claims about other approaches than cultural and social technology. Cybernetics is another very obvious example. If you are stuck on the question of AI alignment, you will necessarily have to trust godlike AIs to monitor other godlike AIs and pray that things will work out. If you start from a cybernetic perspective (say: Maxim Raginsky Thought ) you’ll end up asking very different questions. As the HuggingFace incident showed, there are a lot of holes in the AI labs’ capacities for monitoring and control. “I need abstraction layers to monitor my abstraction layers” is (a) the world’s shittiest boasty bumper sticker, (b) a profound encapsulation of the modern condition, and (c) a diagnosis of the fundamental problems that the AI labs are making for themselves and not always capable of seeing, as best as one can tell from the outside. When you need AI to summarize the data that allows you to write your incident reports on AI, some of the cheese holes are likely to be closely aligned, potentially letting stuff through that you don’t want to.

A third, related but significantly different approach is cybersecurity. As Dan Davies says, “nearly all the examples of worrying agentic behaviour seem to have been seen only in one context – that of frontier research labs doing cybersecurity projects and screwing up their sandbox precautions.” I’m sure that there are great cybersecurity people in the labs, but I suspect that they have much less internal clout than the people who are pushing the agentic boundaries outwards. If you are embedded in a culture that has broadly decided that it is impossible for ordinary human intelligence to prevent putative superintelligent entities from breaking out of their constraints, then cybersecurity seems secondary: alignment of AI goals with human goals is the only way to stay ahead.

I could keep going, but I hope that the point is clear. The Machine God metaphor, which is really just a catchy rephrasing of longstanding notions among a particular community of AI people, has broken out of containment. It is leading, more or less by necessity, to a radical simplification of debate. If you believe that the machine gods are coming, so that super-empowered goal directed superintelligence is right around the corner, then the only thing that matters is ensuring that this superintelligence’s goals are aligned with the ones that you yourself care about. If you stare into the sun, you only see its glare, missing out on all the other questions that may be important. Thinking about these questions involves adopting a more pluralistic understanding of where you might look for knowledge, and asking about the much broader variety of futures that face us than machine god: good or machine god: bad. That’s it.

1 As someone who does international security, sort of, I do not find Silicon Valley geopolitical speculation to be particularly well thought out or convincing. That’s not the topic of today’s post.

<https://www.programmablemutter.com/p/machine-god-metaphors-eat-your-brain> <https://www.programmablemutter.com>

Programmable MutterMachine god metaphors eat your brainHere are three propositions…Read more10 days ago · 26 likes · 9 comments · Henry FarrellProgrammable MutterTechnology and politics in an interdependent worldBy Henry Farrell

Brad DeLong here: Farrell seeks pluralism: multiple metaphors for a technology “that no-one, including its creators, seems to understand particularly well.” The “Digital God” metaphor is especially pernicious because it:

shrinks rather than expands your horizons of the possible, forcing you to focus on the single overwhelming question of what Machine Jesus is going to do as soon as he descends from that gleaming cloud. Is he going to usher in paradise on earth? Or is he actually a shoggoth in a Jesus-mask, condemning us to enslavement or destruction? Once you have accepted that Machine Jesus is in the process of descending, all questions besides determining his intentions for humanity seem irrelevant…

What do we have?:

• We have Farrell’s “MAMLMs are a socio-cultural technology”.

• We have Arvind Narayanan and Sayash Kapoor’s “The MAMLM boom sis a normal disruptive Schumpeterian GPT” of the kind we have seen every quarter-century since 1875, and saw before over 1775-1875 when we got the GPTs of steampower, steelmaking, and automatic textile machinery (and, I would say, the caravel and printing before that).

• We have mathematician Nestor Guillen, whom Farrell quotes at length making the argument I make that MAMLMs are well-viewed as front-end access nodes to the real ASI, which is not an Artificial SuperIntelligence of a machine but rather the Anthology SuperIntelligence of the collective time- and space-binding human mind and the knowledge it has produced and compiled over the past 5000 years:

Humans have always lived in a complex world and their survival has hinged on their ability to process information in amounts beyond what any individual can handle…. Herbert Simon… saw that one implication of bounded rationality is that humans have to (and do) develop systems to process information, as a group… states, stories, bureaucracies, markets, libraries, and more… cultural and social technologies… y which humans (not individually but as a group) navigate amounts of information so vast…. LLMs… are “wrappers” around an older cultural technology: the scientific literature. When you ask an LLM a question… you are effectively interacting with every person who has thought about and worked on those things — provided the data the LLM was trained on contained those contributions in one way or another…

That is, I must say, truly striking contrast to Scott Aaronson’s current view:

Scott Aaronson: The Age of Wonders and Terrors <https://scottaaronson.blog/?p=10062>: ‘It would be healthy for everyone to stop grinding their ideological axes, their sentiments about Dario Amodei or Sam Altman, for long enough simply to acknowledge that the wonders and terrors are here. They couldn’t be here more clearly if the sky had turned reddish-orange like in the Matrix movies…. For anyone who says AI doom sounds like an apocalyptic religion, that the rationalists/Singulatarians seem like a Bay Area cult, that Eliezer Yudkowsky gives off the vibes of a messianic prophet: yes, yes, and yes. But crucially, today you’re no longer being asked to believe in arguments and extrapolations, but only in the front-page news. Accepting the reality of the coming machine god after it’s solved Navier-Stokes and dozens of other longstanding open math problems (while dramatically ramping up in capability every month), is sort of like accepting Jesus after he’s returned to earth on the gleaming cloud.

It’s the epistemic bare minimum.

Yes, there’s still enormous uncertainty about what the rest of our lives will look like, but as far as I can tell, there’s no longer any real uncertainty that it’ll all mostly revolve around AI, and the extent to which we succeed or fail at directing its power toward human flourishing. By any accounting that doesn’t stack the deck, Eliezer Yudkowsky was right about what the greatest challenge facing civilization in our lifetimes was going to be, and you and I were wrong about it…

This reminds me of nothing so much as the story of Arkhidamos III Europontides of Sparta saying:

O Herakles! The might and prowess of man is undone!

upon watching a test-firing of a torsion dart-catapult built in Sicily and brought to Greece. Mechanized weaponry, he thought, had made the excellence of individual human courage and skill-at-arms obsolete.

Yes, the one-fifth of human work that is inside the blast radius of rapidly advancing information technology will spend the next decade and more grappling with “AI”, just has been the case over the past hundred and fifty years with whatever slice of the economy was then in the bullseye of the Schumpeterian process. But it is not a Digital God. It is not a power that needs to be “direct[ed]… toward human flourishing”. Thinking of it as such is not helpful.

OK: We have three metaphors. We need more. What are they?

Henry gestures at “cybernetics” a la Maxim Raginsky Thought and says “you’ll end up asking very different questions”:

“I need abstraction layers to monitor my abstraction layers” is:

• (a) the world’s s***tiest boasty bumper sticker,

• (b) a profound encapsulation of the modern condition, and

• (c) a diagnosis of the fundamental problems that the AI labs are making for themselves and not always capable of seeing, as best as one can tell from the outside.

Henry gestures at “cybersecurity” quoting Dan Davies:

Nearly all the examples of worrying agentic behaviour seem to have been seen only in one context – that of frontier research labs doing cybersecurity projects and screwing up their sandbox precautions…

But what else?

Subscribe now

Leave a comment

If reading this gets you Value Above Replacement, then become a free subscriber to this newsletter. And forward it! And if your VAR from this newsletter is in the three digits or more each year, please become a paid subscriber! I am trying to make you readers—and myself—smarter. Please tell me if I succeed, or how I fail…##crosspost-henry-farrell-machine-god-metaphors-eat-your-brain
##mamlms
##public-reason
##crosspost
#henry-farrell-machine-god-metaphors-eat-your-brain
#henry-farrell
#machine-god-metaphors-eat-your-brain
ai-abstraction-layer -metaphors
#real-asi-anthology-superintelligence
#cultural-social-technology
#normal-technology
#schumpeterian-general-purpose-technology
#nostradamus

My experiences with FaceBook products have always been that time spent getting up to speed on them is time wasted, and that time spent using them is time regretted. But if that is representative of anything, then there are hundreds of millions of addicted regretful regular users whose brains have been successfully hacked and hijacked and who then buy things based off of FaceBook/Instagram/WhatsApp ads. The market verdict is one of extraordinary success, with ads but only with ads:

All this is prelude to noting the sudden emergence of FaceBook Muse: an agentic AI system that has underlying economics I do not understand, but that Amazon fears enough to preëmptively block.

Share

From Sensor Tower via Paul Kedrosky:

Give a gift subscription

& Paul notes:

Paul Kedrosky: Meta’s Muse & Moat-less AI <https://paulkedrosky.com/metas-muse-and-moat-less-ai/>: ‘Meta's agentic AI app Muse entered the U.S. App Store’s top 20 on September 9 at No. 4. Nine days later, it was No. 1. Muse’s rise is a useful reminder that AI consumer products can gain share quickly, given buzz and distribution. Users can switch applications at negligible cost, while the underlying model intelligence is available to every serious competitor. The takeaway should be, in short, that there is, despite trillions in spending, minimal moat. The real assets are familiar ones: default placement, brand, habit, proprietary data, and an effective product loop. The AI itself is not a moat when competitors can access comparable models and cheaply reproduce the surrounding interface. Meta engineers concede as much in posts.

Muse may fall back out of the top 20 next month. Or it may get blocked everywhere, as has already happened at Amazon. That would make the point even more clearly: AI consumer software is beginning to look like a hit-driven market in which the model is rented, the interface is copied, and the users remain [—]mobile…

Share DeLong's Grasping Reality Weblog

Brad DeLong here: The key question is this: What is FaceBook’s monetization strategy for Muse? FaceBook has a monetization strategy for everything, and it—at least for me—makes committing to using their programs painful, and something that after the fact I wish I had not done, I think that the monetization strategy is that FaceBook watches what you do with Muse and uses it to better target its Facebook/Instagram/WhatsApp ads, plus extorting money from shopping websites that are willing to pay to have Muse direct traffic to them. It would surely be a bad bet for me to commit to Muse, just as it would be a bad bet for me to commit to Threads.

But that is just me. I am not representative here. Of anything. Paul thinks the lesson is that “consumer AI” is becoming a hit-driven market where the model is rented, the interface copied, users stay mobile. This smells much more like the streaming-wars media financial disaster than like a new platform-monopoly in progress. But—nobody knows anything here.

Subscribe now

Leave a comment

If reading this gets you Value Above Replacement, then become a free subscriber to this newsletter. And forward it! And if your VAR from this newsletter is in the three digits or more each year, please become a paid subscriber! I am trying to make you readers—and myself—smarter. Please tell me if I succeed, or how I fail…##facebooks-muse-on-the-apple-ios-app-store-chart-of-the-day
##mamlms
##chart-of-the-day
#facebooks-muse-on-the-apple-ios-app-store
#moat-less-ai
#agentic-ai
#move-fast-move-on
#next-big-thing
#paul-kedrosky

Barry Ritholtz has a personal solution to this problem of the firehose of worthwhile and important “content” that floods in not every day, but every hour. But his solution to the problem increases the burden imposed by the additional force he adds to an already overwhelmingly strong jet. Don’t get me wrong: I do not wish he would stop. I value it. Greatly. But it is a problem:

90% of everything is crap. So how do you find the other 10%? And how do you process it when even what is freshest and newest in that other 10% is 50x what you could possibly read?

Share

Barry Ritholtz has good advice on building a daily reading list that works: (1) curate viciously, (2) create a narrative so that what you read on any day reinforces itself, (3) be balanced, (4) but opinionated and judgmental, (5) while always looking for where one may have gotten the big story wrong:

CROSSPOST: BARRY RITHOLTZ: How I Build My Reading Lists

<https://ritholtz.com/2016/08/assemble-daily-reads-3-ez-steps/> <https://ritholtz.com>

August 12, 2016 2:30pm by Barry Ritholtz

How to Build the Perfect Reading List in 3 Steps
Enjoy your morning Ritholtz’s Reads? Here’s how they happen.
Bloomberg, August 12, 2016

The making of a good compilation tape is a very subtle art. Many do’s and don’ts. First of all you’re using someone else’s poetry to express how you feel. This is a delicate thing.

It is hard to do and takes ages longer than it might seem. You gotta kick off with a killer, to grab attention. Then you got to take it up a notch, but you don’t wanna blow your wad, so then you got to cool it off a notch. There are a lot of rules.

–High-Fidelity, Nick Hornsby

“Wherever do you find those wonderful reads each day?”

Every now and again, a variation of the above question comes in from readers. Since it’s a summer Friday — and really, it’s not like anything else is going on in the world — I shall give the people what they want.

“I’ve been assembling daily reads for more than 20 years, since I started as a trader way back in the mid-’90s. It eventually evolved to become a published daily reading list, seven days a week on my blog.”

Some of you may be wondering, “How difficult can pulling a dozen headlines off a feeder be?” Turns out, it is harder to do and takes longer than it might seem.

Here are the rules that guide my hand each day:

Curate viciously.

There is an embarrassment of riches in terms of online content. This creates an “abundance” challenge.

My lists are defined as much by what is included as excluded. What you choose to omit is crucial to making any list special. In the early days, I tried inclusivity, and the list got longer and longer. I gave up that pointless exercise years ago, when Google News made it wholly unnecessary.

I rarely use anything from the front pages of major newspapers or news sites, on the assumption that many readers have seen it already. Overly popular links on Twitter are also tossed. Most read, most popular, most e-mailed lists — nearly every site has a variation of this, and they occasionally point to something worthwhile AND relatively unknown. But for the most part, I steer clear of those.

In the wee hours of the morning, I open 50 or so tabs from a browser folder. Just about every major media outlet is included. Each afternoon, I do the same with a different 40 or so; these are more esoteric, special-interest and topic-specific. Alternating between a broad, wide scope and a narrow, deep one allows for a good mix.

Speed also counts. I follow Sturgeon’s revelation: 90 percent of everything is crap. Ruthless efficiency outweighs the occasional false positive.

Create a narrative.

I try to weave together a series of (apparently) unrelated items, which often have a commonality. It is a puzzle, finding that common thread.

Consider the primary topics that work their way through my links. Listing the subject matter in order of prevalence looks something like this:

1. Markets

2. Investing

3. Economics, Fed

4. Data analysis

5. Behavioral psychology, cognitive errors

6. Regulation, policy, taxes

7. Science, technology, math

8. Esoterica: unusual articles about current affairs

9. Surprising, humorous or amusing

10. Music, film, books, sports, arts, food and travel

All 10 categories are obviously not represented every day, but most make guest appearances throughout any given week. Knowing the readership of this site, markets, investing and economics have the greatest representation; there have been occasional days when that’s all it is.

The narrative comes from finding a common theme. It’s nuanced, but if you squint you might be able to see it.

Be balanced, but have an opinion.

It’s important to recognize that this reflects the views of one person — albeit one who is immersed in media as both a consumer and a producer. I have definite opinions about oh so many things. However, I am open to being convinced otherwise; that is why I use the “see also” or “but see” mentions to suggest there may be multiple opinions worth unpacking.

I am always surprised about accusations of bias. This week’s BuzzFeed link about Facebook selling guns was not, as several of you suggested, anti-gun. I enjoy shooting but believe in reasonable regulations, terrorist watchlists, etc. But the link was about Facebook, which seems to violate its own policies about nearly everything!

One last thing: Editors, not writers, create the headlines. A linked headline is often a “clickbaity” version of what the article is about. I have a novel idea: Please read the linked article before you accuse me of (fill in the blank).

<https://ritholtz.com/2016/08/assemble-daily-reads-3-ez-steps/> <https://ritholtz.com>

Brad DeLong here: Barry Ritholtz’s guide to building a reading list is generous, disciplined, and quietly self-defeating — because the better his curation, the more it adds to the very flood it’s meant to tame. That’s the tension worth sitting with. His rules are sound. And, written in 2016, it reads as even truer and more important now.

There is no escape: abundance is the water we swim in now, and it rises every hour. Curation is thus an inevitable bet on judgment over completeness — you accept that you will miss almost everything, and you invest your scarce attention in a few people whose sense of the common thread you trust. Barry earns that trust. That is why I read him, and why I am passing him along to you: not as a way to read less, but as a way to read better while reading less than the impossible.

Whom do you let curate for you, and what are you willing to never see as the price?

Give a gift subscription

Subscribe now

Leave a comment

If reading this gets you Value Above Replacement, then become a free subscriber to this newsletter. And forward it! And if your VAR from this newsletter is in the three digits or more each year, please become a paid subscriber! I am trying to make you readers—and myself—smarter. Please tell me if I succeed, or how I fail…##reading-barry-ritholtz-how-i-build-my-reading-lists
##public-reason
##reading
#barry-ritholtz
#how-i-build-my-reading-lists
#daily-reads
#information-firehose
#sturgeons-law
#attention-economy
#balanced-but-opinionated

Claudia Sahm is in the business of trying to coax Fed Chair Kevin Warsh to behave better, that is to say: more like a Fed Chair and less like a Trump Lickspittle. Such pressure is useful, but I think we have to be clear-eyed about what is going on. What is going on is that Warsh has told different things to Trump and to the bond market, and so his principal concern now is to minimize the attack surfaces he leaves open to either. Hence, interpreting what he says as statements of policy or attempts to convey information is fundamentally misunderstanding Warsh’s intentions:

Claudia Sahm’s bottom line is this: A good policy decision was undercut by a communication problem: Warsh’s framework diverges sharply from the rest of the FOMC’s, and his refusal to articulate his own model turns routine transparency into destabilizing ambiguity. Warsh should fix this by doing the one thing his press conference conspicuously avoided: telling us what framework he actually uses, rather than reciting the standard tools—below-trend growth, the neutral rate, the SEP, forward guidance—that he rejects.

Concretely, she wants him to:

• state what role he sees for monetary policy if not macroeconomic stabilization;

• name the channel through which he thinks higher rates lower inflation

  • now that the labor-shortage buffer of 2023–24 is gone and

  • expectations are already anchored,

  • leaving demand as the obvious candidate he won’t claim;

• explain how he judges the appropriate level of the funds rate if the neutral-rate benchmark is off the table;

• articulate a stopping rule—which, as she points out, is just his own Jackson Hole hiking standard “with the sign flipped,” and requires no forward guidance to say aloud.

Share

Underlying all of it is a communication fix: use the press conference for its intended purpose—explaining and answering for the Committee’s decision straight to the public—rather than shortening it, barring follow-ups, and editorializing selectively on growth while staying silent on rates. Do that, and the confusion that drove market rates above what the committee intended goes away.

And she laments:

He continues to tell us which standard tools he doesn’t use. He has not said what he uses instead. As a critic of the Fed, that limited his impact; as Fed Chair, it is causing unnecessary confusion…

Give a gift subscription

But, Claudia, it is necessary confusion given that the point is to triangulate in order to limit the surfaces he leaves open to attack from either a disappointed Trump or a disappointed bond market. Don’t pretend he is playing chess when he is actually playing blind-man’s-bluff:

Stay-At-Home Macro (SAHM)My goal is to make economic policy and macroeconomics better.By Claudia SahmCROSSPOST: CLAUDIA SAHM: Questions for the [Fed] Chair [Kevin Warsh]

<https://stayathomemacro.substack.com/p/questions-for-the-chair> <https://stayathomemacro.substack.com>

Stay-At-Home Macro (SAHM)Questions for the ChairLast week, the Fed raised its policy rate by a quarter percentage point to a range of 3¾ to 4 percent, on a unanimous vote. That’s a notable shift: at the Fed’s meeting six weeks earlier, only three of the twelve voters had favored a hike. In its one sentence of explanation, the…Read more11 days ago · 57 likes · 22 comments · Claudia SahmA good meeting, and five questions the press conference left open.Claudia SahmSep 20, 2026Source: Federal Reserve, September 2026 Press Conference.

Last week, the Fed raised its policy rate by a quarter percentage point to a range of 3¾ to 4 percent, on a unanimous vote. That’s a notable shift: at the Fed’s meeting six weeks earlier, only three of the twelve voters had favored a hike. In its one sentence of explanation, the FOMC statement said that it “will support a timelier return to the Committee’s 2 percent [inflation] goal.” In the Summary of Economic Projections, officials centered on one more hike this year, with rates ending next year at that same level. Modest rate increases with a modest goal: that’s the outcome I argued for before the meeting. The Fed also ignored the political pressure not to hike before the midterms. I agree with the Fed’s decision, though I would never celebrate a rate increase.

A good meeting, but questions for the Chair linger.

It was a good meeting. As I said on NPR the next day, the Fed’s rate hikes are not a cure for inflation, but they will help keep a bad situation from getting worse. Real relief will require a resolution to the conflict in the Middle East and no new tariffs.

In some ways, that’s where this update on the Fed decision should end, but Warsh also held a press conference. He was clearer than in July, but his answers still raised questions, and today’s post uses them to frame mine. I want to understand Warsh’s approach. Monetary policy has shifted significantly over time, and there’s no reason to treat the status quo as perfect; he may see something the standard framework misses. Regardless, as best I can tell, his views diverge fundamentally from those of the rest of the committee, and that gap is what he owes us an explanation for.

It’s not an academic quibble. Markets read Warsh as hawkish, and pricing of future rate hikes rose as soon as he began speaking at 2:30 pm. The implied rate for the end of 2027 climbed about as much during the press conference as it had on the 2:00 pm release of the statement and projections, and the ten-year yield rose more while he talked than it did on the release. I doubt Warsh intended that outcome, and the committee almost certainly did not. I put it down to confusion, in markets and among Fed watchers, about how Warsh approaches monetary policy.

Share

Questions for Warsh.

After listening to Warsh’s presser, my questions for him range from foundational to operational. I suspect that once we have a better sense of his approach to monetary policy, the pressers will be more useful for markets and for the public.

What is the purpose of monetary policy?

Monetary policy is a tool of macroeconomic stabilization. It smooths out the bumps and wiggles in the economy around its trend: the Fed fights recessions and overheating alike, but it does not try to move the trend itself. Longer-run fundamentals belong to other policymakers.

Warsh spends more time on longer-run fundamentals than any Fed Chair I can recall. In his opening remarks, a prepared statement, he invoked growth several times:

The Fed has a role in sustaining the economic progress happening in America right now, and the rising opportunities that come with it. Those who are least well-off have the most to gain from a durable expansion, a solid labor market, and stable prices.

We at the Fed are unwavering in our vital and straightforward purpose: Full employment and price stability, and a thriving American economy that sets the standard for the world. [Emphasis added.]

These are important goals, but are they the Fed’s goals? Warsh himself was critical when past Fed Chairs strayed into topics like inequality. And talking up growth this much reads as hawkish, because when a Fed Chair dwells on how strong the economy is, markets hear a Chair who thinks it needs cooling. So, at a basic level, what role does Warsh see for monetary policy? If it is something other than macroeconomic stabilization, that is a sharp difference from the rest of the committee.

How will a higher federal funds rate lower inflation now?

Closely related is how Warsh thinks monetary policy brings inflation down. In the presser, Jennifer Schonberger of Yahoo Finance asked about the standard mechanism:

Now that you have hiked rates, do you need to push growth below potential, unintentionally pushing weakness on the job market to bring inflation down?

Warsh did not accept the premise that below-trend growth is part of the process:

First, we believe that the unemployment rate is basically running consistent with full employment. I don’t believe that we need to do harm to the labor markets to achieve our objective. I don’t believe that the two parts of our mandate, price stability and full employment, are working at cross purposes over the medium term. So, economic growth, that is ensuring continuous, sustainable, durable, economic growth, that’s the business we’re in, and the job we did today, the job we’ll continue to do, is to ensure price stability, which can mean that sustainable, durable, economic growth can go on for longer. The economy can be stronger, and as I mentioned before, the least well off can get the benefits of it.

There are times, such as 2023-24, when inflation falls with little cost to employment or growth. But that hiking cycle started with a labor market in shortage, nearly two job openings for every unemployed worker, so cooling it meant fewer vacancies rather than fewer jobs. The labor market is now in balance, and that buffer is gone. A channel that could deliver disinflation at low cost is credibility: if a hike convinces firms and workers that inflation will fall, the Fed has to take less demand out of the economy to get there. But market- and survey-based measures of longer-run inflation expectations have been stable, so there is little lost credibility to restore. That leaves demand as the channel.

So how does Warsh think higher rates reduce inflation, if not by restraining demand? Monetary policy operates through many channels, and it would help to know which ones he has in mind. The Fed Chair should be transparent about the potential costs of rate increases. Warsh’s framing last week was more upbeat than I associate with the start of a rate-hiking cycle.

What is the appropriate federal funds rate?

There are also operational questions. The Fed raised its policy rate to a range of 3¾ to 4 percent. How does Warsh know that is the right level? Steve Liesman of CNBC posed the question in terms of the neutral rate:

Previously most Fed officials have described the rate as modestly restrictive. And if you removed accommodation, could you give us your sense of where the Fed funds rate is relative to neutral? And some detail, if you wouldn’t mind, on your sense of is there a short-term neutral rate you’re aiming for and a longer-term neutral rate? Do you think in terms of those?

Warsh was clear that the neutral rate is not useful to him:

In a word, no. In a few words I’d say this, I’ve always been interested in a neutral rate as an academic matter. Back when I learned economics we used to think of it as the Wicksellian rate, the real equilibrium rate. It’s useful academically, it’s -- it’s a discussion to help us think about policy. Do I think it has any operational effect on decisions that we make today? No, I don’t.

The neutral rate has been a central concept in monetary policy for decades. Powell pushed back on false precision in the estimates, but it remained the benchmark for judging the level of the funds rate: above neutral, policy is restrictive; below, it is accommodative. What is odd is that Warsh framed the decision as “removing a dose of accommodation” and then distanced himself from the concept that defines accommodation. The rest of the FOMC has views on it; every participant but Warsh submitted a longer-run estimate, and the median is 3.25 percent, below the funds rate even before the hike. By the committee’s own numbers, policy was already above (longer-run) neutral. So if not the neutral rate, how does Warsh judge the level of the funds rate?

The word choice mattered. “Removing a dose of accommodation” was heard as saying the funds rate is still accommodative after the hike, with more doses to come. That is another reason the market priced in more hikes as he spoke.

How will you know when to stop the hikes?

Maybe Warsh doesn’t need a firm view of the appropriate level of the funds rate to know it needed to be higher. But now that the Fed has hiked, how will he judge whether to hike again, and when to stop?

The rest of the FOMC put its answer in the dot plot. The dots are not a commitment, but they sent a clear signal: a modest cycle of 50 to 75 basis points in total, far short of the 3 percentage points in the median hiking cycle. Warsh did not submit projections, and at the presser he distanced himself from the committee’s. Michael McKee of Bloomberg asked:

You said in Jackson Hole that you want to see inflation come down clearly and at sufficient speed, which is a standard without necessarily a measurable threshold... today you say today’s policy action will support a timelier return to the Committee’s 2 percent target, and yet, in the Summary of Economic Projections, the median pushes [reaching] the 2 percent target out to 2029... how can you square those two things?…

Warsh:

One easy way to square that, Mike, is those aren’t my forecasts. Those are the forecasts of my 18 colleagues, and I’ve tried to represent them dutifully to you. My business is to not give forward guidance... what I said in Jackson Hole in August is, we’re committed to a discipline, not to a decision. Today’s action starts to show we’re serious about this... But I’m ill-prepared to pre-judge those future actions…

No one is asking Warsh to pre-judge future actions. The SEP is a thought experiment, not a commitment: given what we know and where the economy is likely headed, what will policy need to do to meet the mandate? Warsh may not see value in the exercise. But he set a standard for hiking, confidence that inflation is moving to target clearly and at sufficient speed. The standard for stopping is the same sentence with the sign flipped, and he could say so without a word of forward guidance.

What is the purpose of the press conference?

Warsh has changed the press conference. Last week’s ran 30 minutes, the shortest in memorty outside of an emergency presser during the pandemic; Powell’s last five averaged about 50 minutes. Journalists were not allowed follow-up questions. And Warsh has not stuck to presenting the Committee’s views, as past Chairs did. When Victoria Guida of Politico asked whether he had learned anything about communication from the market’s fixation on the August CPI, he answered:

Markets over time will come to understand how this Fed makes its decisions, what’s relevant and not. And I wouldn’t want to editorialize that for them.

The press conference exists so the Chair can explain the Committee’s decision and answer for it. It is a hard venue for any Chair, and it is part of the job. Warsh does not want to give forward guidance on rates, which is his prerogative, but he is comfortable offering plenty on the prospects for American growth, so the editorializing is selective. It would help to know the principles behind his pressers: what he thinks they are for, and how he squares a shorter, narrower format with the Fed’s accountability. The Fed’s decisions affect millions of Americans. They should hear the reasons straight from the Chair.

Closing.

The Fed raised rates last week, and at least one more increase is likely. The start of a hiking cycle is always a serious moment, and this one is no exception: inflation is elevated, and disinflation is far from guaranteed. The Fed met the moment. A new Chair adds to the challenge, but that is surmountable. What we need is a better understanding of how Warsh approaches monetary policy. He continues to tell us which standard tools he doesn’t use. He has not said what he uses instead. As a critic of the Fed, that limited his impact; as Fed Chair, it is causing unnecessary confusion. Last week, that confusion showed up as higher market interest rates than the committee intended, and Americans pay for that.

<https://stayathomemacro.substack.com/p/questions-for-the-chair> <https://stayathomemacro.substack.com>

Stay-At-Home Macro (SAHM)Questions for the ChairLast week, the Fed raised its policy rate by a quarter percentage point to a range of 3¾ to 4 percent, on a unanimous vote. That’s a notable shift: at the Fed’s meeting six weeks earlier, only three of the twelve voters had favored a hike. In its one sentence of explanation, the…Read more11 days ago · 57 likes · 22 comments · Claudia SahmStay-At-Home Macro (SAHM)My goal is to make economic policy and macroeconomics better.By Claudia Sahm

Brad DeLong here: I understand that to stay court-acceptable in the finance-guru discourse in which she swims, Claudia Sahm has to pretend that Kevin Warsh is making communication mistakes rather than doing exactly what he intends to do.

But the rest of us do not have to do that! We do not have to pretend! We can say what is going on! Which is that, to improve the American economy, the single best thing Warsh could do is resign immediately! Sober probity is something a Fed Chair needs, and that someone like him who campaigned for the job inside the White House by doing a Trump Lickspittle cosplay act cannot regain.

Subscribe now

Leave a comment

If reading this gets you Value Above Replacement, then become a free subscriber to this newsletter. And forward it! And if your VAR from this newsletter is in the three digits or more each year, please become a paid subscriber! I am trying to make you readers—and myself—smarter. Please tell me if I succeed, or how I fail…##crosspost-claudia-sahm-questions-for-the-fed-chair-kevin-warsh
##macro-outlook
##chaos-monkey
##neofascism
##crosspost
#claudia-sahm-questions-for-the-fed-chair-kevin-warsh
#claudia-sahm
#questions-for-the-fed-chair-kevin-warsh
#kevin-warsh
#federal-reserve
#monetary-policy
#neutral-rate
#fed-communication

This week brought me news increasing my confidence with respect to one thing it is not: it is unlikely to be a hyperprofits engine for research labs with companies attached making money serving up calculations made in datacenters by simulated neural networks:

What is now called “AI”—what I insist on calling MAMLMs, Modern Advanced Machine-Learning Models; but perhaps CIP, Complex Information-Processing, would be better—these days is so many things:

Share

• First of all, it is wishful mnemonics: the opposite of the rectification of names, in that the very idea is to call it something that it is not, and all because Marvin Minsky was primarily a showman seeking a bigger budget.

• Most important, it is natural-language interfaces to information systems, especially, so far at least, databases of example code from successful software projects.

Give a gift subscription

But it is also:

• A potential disruptor and destroyer of platform monopolies’ existing megaprofits flows.

• Hence a spur to platform monopolists to spend fortunes to try to protect those existing platform-monopoly profits from this disruption and destruction.

• The next normal general-purpose technology driving the next generational doubling of human productive potential while leveling and rebuilding a fifth of the economy.

• And, on the other side, the favored post-crypto way to grift the gullible separating gullible, separating VCs and wannabes from their money.

• A bubble that knows it is a bubble—boosted by those confusing the long-run benefits to society from technology-driven investment from the actual profits to be earned by the investors.

• The Rapture of the Nerds: trying to build a Digital God as the latest and strangest of the many American religious-cult millennarian outbreaks since the First Great Awakening.

• Extraordinary plentiful tools for megasuperscale, hyperdimensional, flexible and extensible functional classification analyses at new levels.

• A better internet search-and-summarization engine that has not yet been polluted by SEO.

• A boilerplate, ritual, and slop text and image and video generator.

• Highly verbal software pets and coaches.

• Clever Hans at lightning scale and speed: trying thousands of things in the time it would take you to try one, and when and if it can be well and properly harnessed, tremendously productive.

• Quite possibly, the savior of authoritarian states that want to live East Germany’s STASI surveillance dream.

• Autocomplete on super-steroids: “that’s just kernel smoothing” vs. “you can do that with just kernel smoothing?!”; “that’s just a Markov chain” vs. “you can do that with just a Markov chain?!?!”

• A jagged frontier produced by a stochastic parrot that is being harnessed to build a bicycle for the mind.

Share DeLong's Grasping Reality Weblog

This week brought me news increasing my confidence with respect to one thing it is not: it is unlikely to be a hyperprofits engine for research labs with companies attached making money serving up calculations made in datacenters by simulated neural networks.

Today we have a review article for an $11,750 machine, the M5Ultra Mac Studio with 256GB of unified memory and a 2TB hard disk( cf.: the original 128K Macintosh’s price back in 1984 was $8000 of today-value dollars, with a Motorola 68000 processor, 128KB of RAM, and 400KB of floppy-disk storage per disk. 5 x 10^9 storage, 2 x 10^6 memory capacity, 5 x 10^5 memory bandwidth, 1 x 10^10 scalar and matrix addition and multiplication computrons, and with respect to the kinds of computations MAMLMs do “the honest ratio is division-by-zero: the 1984 chip couldn’t do it at all”):

Federico Viticci: M5 Ultra Mac Studio Review: The Dream Mac for Local AI Agents <https://www.macstories.net/stories/m5-ultra-mac-studio-review-the-dream-mac-for-local-ai-agents/>: ‘The M5 Ultra Mac Studio is a dream machine for local AI agents… powered by local models with great performance and no additional cloud costs…. The personal assistants I use the most are now entirely powered by a model running locally on a Mac Studio…. A team of agents… ran 24/7, for 99 days, to… transcribe my favorite WWDC sessions (using summarize plus LLM processing)… extract features of iOS and iPadOS 27… cross-reference features across different sources, and keep track… extract features and bugs from screenshots… work with the Notion API to organize…. Relying on the OpenAI or Anthropic APIs for this kind of always-on, persistent background task would be… cost-prohibitive, to say the least. So I pivoted to local AI…. The result is the iOS and iPadOS 27 review you can read on MacStories….

Having a model such as Qwen3.8-Flash-Next clear 100 tokens/second on short prompts and still write at 60 to 85 with 64K to 256K of context behind it… enables the kind of agentic back-and-forth between you and the model that feels great to use…. For my taste, 5-bit quantization hits the sweet spot on this version of the Ultra with a balance of intelligence, performance, and memory consumption…

Leave a comment

The natural local on-device comparison:

NVIDIA’s RTX 5090 is still faster than Apple’s M5 Ultra…. Prompt processing speeds are dictated by… one giant matrix multiplication, which is exactly the job NVIDIA’s Tensor Cores were built for…. The M5 Ultra read at ~1,700 tok/s; the 5090 delivered a staggering ~3,000…. Token generation… is bandwidth… so the 5090’s 1.79 TB/s against the M5 Ultra’s 1.2 TB/s gives it a steady ~25% lead…. [But] the 5090 doesn’t have… memory…. The moment I want to run anything exceeding 32 GB… the 5090 must offload model layers over PCIe to (much slower) system RAM, and that’s no way to live…

Get 75% off a group subscription

Cf.: the NVIDIA RTX 5090 launched eighteen months ago at a MSRP of $1999 costs $7,500 today, and that is for the graphics card alone:

The bottom line:

I have a simple, tangible result: the M5 Ultra lets me run local agents with incredible performance, with less time spent staring at a blank screen and everything happening on a single, compact, cool, and quiet machine on my desk. This would have seemed impossible a couple of years ago. But here we are…

Refer a friend

Running Federico’s agent team as claude-code-sonnet for three months at current API-prices would, I guess, have cost him at least twice the price of his machine.

Now, on the one hand, this kind of always-on, high-input-token workload and sustained agentic use is exactly what the hyperscalers are counting on to monetize the buildout.

And, the other hand, this workload means that $10K+ of Mac (or NVIDIA!) hardware pays for itself in a month. Profitable datacenter demand thus runs on frictions: people who need the technology’s Clever Hans trick—try a million paths, let the harness keep the winners with nine-nines reliability—but who can’t be bothered to hire a work-study IT tech to spin up what the open-weight community already gives away for free.

Thus my Visualization of the Cosmic All with respect to the future of MAMLMs is becoming clearer:

Will anyone other than NVIDIA ever show us real money from the AI datacenter buildout? Probably not.

Read more

Who actually made up the “consensus” forecast that the operating cash flow of the five biggest HyperScalers will triple from today’s annual $600 billion to $2 trillion by 2030, while their capital expenditures will grow by $500 billion less to give them $500 billion of free cash flow by 2030? And why? And how?

Apollo finds this time series in its incoming firehose feed flow, and reports. My reaction: WAIT!! WHAT?!?! WHERE DOES THIS COME FROM?!?! And the sudden upward bend in the net as the costs curve slows and the revenue curve accelerates is—interesting.

Share

The lines on this chart look authoritative. But they are a crowd forecast. And where has the crowd gotten its information from? What bottom-up detailed understandings of the technology and its deployment do the guys with the spreadsheets working for the sell-side actually have here?

The “HyperScalers” are: Google, FaceBook, Amazon, Microsoft, and Oracle.

Torsten Slok: Hyperscaler Credit Rests on One Consensus Assumption <https://www.apollo.com/wealth/insights-news/insights/daily-spark/hyperscaler-credit-rests-on-one-consensus-assumption>: ‘The credit story in hyperscalers rests on a single consensus assumption, that operating cash flow triples from $600 billion to $2 trillion, see chart below.

Give a gift subscription

Read more

Now modern advertising, public relations and corporate values statements typically carry no equivalent constraints. Contrast 1858’s Lincoln and Douglas, “divid[ing] time and address[ing] the same audiences” in seven three-hour debates, with modern corporate values statement radiating outward to “stakeholders” forever, and you can measure what we’ve lost:

Thomas Hobbes blamed the coming of the English Civil War (at least in part) on rhetorically skilled parliamentarians “with no arguments at all”. Gruen sees him as both right and wrong: right to seek an answer in fiduciary cognitive societal-scale institutions like science, but overwhelmingly wrong in his fever dream that, starting from geometry, we could maintain certainty in a logical chain of reasoning that would carry us to a science of justice. The real virtues of science are that it “reorders prestige to follow evidence, rather than the reverse” while making reasoning inspectable. Good institutions inject contestability, and grant standing to whoever can competently join the argument.

Thus our worries about the destructive rhetoric of spin and bullshit destroying public reason are not new, 350 years old. Or perhaps 2075 years old: recall how Quintilian reported that Cicero gloried in his having spread darkness over the jurors in Pro Cluentio. Or perhaps 2450 years old: do recall that Hobbes picked up on Thoukydides’s lamenting that during the Peloponnesian War

words had to change their ordinary meaning….. “Reckless audacity” came to be considered “the courage of a loyal ally”; “prudent hesitation”, “specious cowardice”; moderation was held to be a cloak for unmanliness…

Nicholas GruenWeekend readings and selected LinksCROSSPOST: NICHOLAS GRUEN: Hobbes on Spin & BullshitNicholas GruenHobbes on spin and bullshit What Hobbes understood about spin and bullshit…Read more12 days ago · 4 likes · Nicholas Gruen

<https://nicholasgruen.substack.com/p/hobbes-on-spin-and-bullshit> <https://nicholasgruen.substack.com>

The more things change, the more spin and bullshit pile up!Nicholas GruenSep 19, 2026

What Hobbes understood about spin and bullshit

Much of what is now said about public discourse — that manipulation and spin prevail over truth and candour, that pleasing abstractions displace description, and that profile confers standing more readily than expertise or disinterest — was said in much the same terms by the great 17th-century philosopher Thomas Hobbes about rhetoric.[1]

Hobbes’s account of what public speech is for could stand for our own time (even if his ultimate remedy was extreme — showing all the motivated impatience of the intellectual). The office of a counsellor is to make the consequences of a course of action manifest, so the person counselled may be “truly and evidently informed”; speech that works by stirring the passions is “repugnant to the Office of a Counsellour.”[2] Against that standard he listed rhetoric among the causes of a commonwealth’s dissolution,[3] and in Behemoth proposed that the Civil War was largely talked into being — by dissenting Presbyterian ministers, and by parliamentarians with great rhetorical skill and no arguments at all.[4]

For Hobbes, this was existential. Over a decade before the English Civil War, he’d translated Thucydides, dwelling on the way in which the tumult of civil war tore civilisation asunder, “the received value of names” being changed at will. Thucydides depicted a society losing its hold on its own vocabulary, and so on reality itself. A prime instrument of the transformation was described in rhetoric as paradiastole — the redescription of a vice under the name of a neighbouring virtue. Henry Peacham’s Garden of Eloquence (1577) gave the example of calling murder “a manly deede”, and Shakespeare illustrates it when Lady Macbeth goads her husband into murdering Duncan in his sleep: “When you durst do it, then you were a man”. We’re inured to this today from our political class. Kidnapping becomes ‘extraordinary rendition’, torture ‘enhanced interrogation’.

Seven such parallels between Hobbes’s concerns then and ours today are set out in the appendix.

However, two things have changed in the centuries since Hobbes wrote, not for the better. First, classical rhetoric was typically adversarial and, in its forensic form, retrospective: opposing advocates contested the appraisal of deeds already done.[5] Its practitioners therefore spoke in the expectation of contradiction: an advocate’s case had to survive another advocate’s attempt to break it. Modern advertising, public relations and corporate values statements typically carry no equivalent constraints. Though they may be deployed to pre-empt criticism, their puffery faces no equivalent adversary; their pleasing abstractions gesture to the future and float free of the inevitable frictions and dilemmas of reality: “Excellence in everything we do”.

The second difference concerns who is listening. Castiglione’s courtier was trained to inform his prince candidly on any matter he needed to understand.[6] The scheme assumed a prince who wanted to know, because he governed and stood to suffer for his own mistakes. He might still prefer flattery, but an honest counsellor would set his interest against his vanity.

By contrast, modern institutional speech is often different in both respects. No corporation hires anyone to argue that it doesn’t actually operate according to the pleasing values hanging in the foyer — that it lacks integrity, responsiveness or respectfulness in specific respects. The Catholic Church did something similar in having candidates for sainthood run the gauntlet of the devil’s advocate.[7] Hence the quiet coerciveness of so much institutional language: it is less an account of what is or was done than the peremptory assertion of a free-floating benignity which only the churlish would dispute. And its audience is diffuse — the honeyed, vapid words radiate outward to ‘stakeholders’ and forward in endless time, but never constrain what senior managers actually do.

Conclusion

Hobbes looked to science for a way out of all this.[8] I think he was right. And wrong.

What impressed him was geometry’s promise of certainty. For Hobbes, geometry starts from principles evident to anyone and arrives at equally secure conclusions. He set out to create a science of just and unjust, demonstrated, as he put it, from principles evident to the meanest capacity. His younger contemporary Descartes pursued a similar ambition: to take nothing for granted and build from indubitable foundations.

Even in natural science, if there are any indubitable foundations, they’re too thin, too emptily formal to take you very far. Hobbes’s task was harder still: not just to understand human motivation but to deduce the ideal institutions under which people should live.

But Hobbes was right to look to science for an answer to a problem that beleaguered his own time as it beleaguers ours: rhetorical tricks unmooring public institutions from reality and its inevitable dilemmas. Science reorders prestige to follow evidence, rather than the reverse. And it makes reasoning inspectable, so claims can be questioned on their merits by someone with no standing whatever.

That does not mean that public policy should be value-free in the way often associated with science, still less that scientists should rule, though scientific consensus should presumptively enjoy high standing. Successful scientific institutions do not abolish rhetoric or hierarchy; they introduce contestability at every step of scientific reasoning and accord standing to whoever can join the contest competently. Yet so much of our own governing code — values statements, for instance — comes down from on high, and it’s not really for challenge.

By contrast, Toyota’s commitment to getting it ‘right first time’ at each step of the production line is enforced from the bottom. If there’s an error on the line that the process worker can’t solve, they pull the andon cord for more assistance, and if the problem can’t be solved as the production line moves, it is stopped. In other words, teams are given a live veto power to protect the agreed-upon standard.

My proposed Evaluator General seeks to build something more elaborate in government on the same principles so that evidence and front-line learning have standing against managerial imperatives. Each turns a professed value into an institutional power.

Such arrangements do not require agreement about ends. They need to be built into institutions rather than left to individuals whose efforts will mostly go unrewarded or be punished.

We have barely begun to take that seriously in public life.

Appendix: Seven rhetorical discontents, then and now1. Speech engineered to move rather than to inform

The complaint today: Edward Bernays opens Propaganda (1928) by describing his own trade as the “conscious and intelligent manipulation of the organized habits and opinions of the masses,” carried on by an unseen mechanism he was content to call an invisible government. Drawing on his uncle Sigmund Freud, he argued that the appeal should be made not to the rational part of the mind but to the unconscious — a man may buy a car not to travel in but because it stands for something he would be embarrassed to name.

Hobbes’s complaint: Hobbes defines the “gift of eloquence” as the ability to make good and evil, expedient and inexpedient, honourable and dishonourable “appear to be greater or less than they are in fact,” and to make injustice appear as justice, in whatever way suits the speaker’s purpose.[9] He is echoing the rhetoricians’ own vocabulary of amplification and extenuation back at them: their claim was that the powerful speaker knows how to magnify what favours his cause and diminish what tells against it, quelling or inflaming an audience at will.[10]

2. Indifference to truth, as distinct from lying

The complaint today: Harry Frankfurt’s On Bullshit isolates a pathology distinct from lying. The liar is still tethered to the truth — in order to conceal it. The bullshitter, in Frankfurt’s phrase, is indifferent to how things really are.

Hobbes’s complaint: The rhetoricians had promised that their art would unite oratio atque ratio; that is, speech with reason. Yet Hobbes argues that what we get is oratio praeter rationem — speech running alongside reason without touching it.[11]

3. Pleasing words displacing accurate ones

The complaint today: George Orwell’s “Politics and the English Language” (1946) argues that political speech consists largely of euphemism and question-begging because it is mostly the defence of the indefensible: villages burnt and inhabitants driven out become pacification. Political language, he concludes, is designed to make lies sound truthful, murder respectable, and give “an appearance of solidity to pure wind.” Today kidnapping is ‘extraordinary rendition’, torture ‘enhanced interrogation’, and to lie is to ‘misspeak’.

Hobbes’s complaint: The figure had a name: paradiastole, the redescription of a vice under the name of a neighbouring virtue. Every virtue lies between two vices, so every virtue has a bordering vice it can be slid into. By the end of the sixteenth century, on Skinner’s account, ‘paradiastole’ was widely denounced.[12] In Hobbes’s time, Shakespeare had Lady Macbeth redescribe murder as manliness; Henry Peacham’s rhetoric textbook had already listed murder as “a manly deede”.

4. The frame as a non sequitur

The complaint today: “Lifting productivity is crucial today with governments keen to reduce budget deficits while satisfying voters’ rising expectations.” Something that should be adopted for its own sake is presented as answering some special strategic priority. I dubbed this ‘strategisation’ and it’s everywhere. Strategisation supplies the appearance of a reason for action but not its substance.

Hobbes’s complaint: Rhetorical inventio proceeded by running a matter through the loci communes — pre-set topical headings, described by the theorists themselves as seats or storehouses from which stock maxims could be hunted out and applied. As with strategisation, the link is rhetorical, not substantive. Hobbes argues that accommodating one’s arguments to received opinion cannot in principle yield knowledge.[13] Hobbes’s concerns also surface much earlier, when he served as secretary to Francis Bacon, whose 1620 Novum Organum carries the same analysis in its charge that leaders of science order existing customs and what is already known rather than devising methods of invention.[14]

5. Standing conferred by fluency rather than by being right

The complaint today: Jay Rosen calls the working ideology of the political press “savviness”: a claim to authority based on practical, unsentimental realism — knowing how politics “really works” while treating substantive positions as ideological or naïve. The savvy journalist does not claim to have the better analysis but to stand closer to the ‘real action’. Reading a David Speers column at the height of the lockdown debate, I documented the phenomenon line by line. The advice of a Chief Medical Officer whose judgement had already proven faulty was relayed uncritically, though New Zealand’s better-performing officials advised the opposite course. That didn’t register. Meanwhile the concerns of a highly credible public-health expert were described as “armchair” expertise.

Hobbes’s complaint: The classical and Renaissance rhetoricians equated the good orator with the good citizen — Quintilian’s vir civilis, the man fit to govern cities by his counsel. In Tudor England mastery of the studia humanitatis helped legitimise the gentry’s claim on government. Eloquence could therefore qualify a man to be heard on any subject, while mastery of a particular field could disqualify him. The schoolmaster Roger Ascham discouraged mathematics on the ground that men wholly bent on such sciences prove “unfit to live with others, & how unapte to serve in the world.” To paraphrase Michael Gove (admittedly unfairly and out of context), he’d “had enough of experts.”

6. The dissenter renamed rather than answered

The complaint today: Irving Janis’s account of groupthink explains how anyone arguing against group consensus meets direct pressure. Rather than being answered, they have their motives or temperament questioned.

Hobbes’s complaint: William Cavendish’s 1611 A Discourse Against Flatterie offers a catalogue of ways to dismiss a person by redescribing his virtues as faults. Honesty becomes fastidious eccentricity; gravity, dullness; repentance, superstitious melancholy. The objection is never answered; the objector is simply redescribed as the sort of person whose objection need not be taken seriously. Hobbes, then Cavendish’s tutor, later revived one of these examples in Leviathan.[15]

7. The right consideration loses for want of a sponsor

The complaint today: Mike Bracken, writing on his Whitehall experience trying to embed digital capabilities into government, describes endless policy cycles of carefully controlled documents circulating between private offices and departments, in which — his own gloss — “Rarely, if ever, does user need get a look-in.” With users unrepresented by anyone with standing in the system, their needs have no practical way of counting.

Hobbes’s complaint: Hobbes’s own 1660s account of why his science of justice made no difference: it had been demonstrated from principles evident to the meanest capacity and was nevertheless “cover’d and kept under here by a cloud of adversaries,” since most people learn their politics from the pulpit. He concedes that the democratical gentlemen who beat him had no arguments at all, only enormous rhetorical skill — and concludes that reason has little chance of being heard against interest, and that the most pernicious doctrine will beat the clearest proof if pressed with sufficient force.[16]

[1] This essay grew out of my reading of Quentin Skinner’s account of Hobbes’s fascination with and horror of rhetoric in 17th-century England. Skinner argues that Hobbes, after an early humanist phase, turned against rhetoric but then decided that, if he couldn’t beat the rhetoricians, he should join them, and made his masterwork, Leviathan, rhetorical with a vengeance.

I was struck by the similarity between Hobbes’s concerns about the befuddlement of public thought by rhetoric and my own about the saturation of contemporary discussion of public life by spin and bullshit. The essay began with my asking Claude to document the first version of the appendix to this essay, from which I drafted the essay.

The Hobbes material throughout is drawn principally from Quentin Skinner, Reason and Rhetoric in the Philosophy of Hobbes (Cambridge University Press, 1996), with page references given in the notes.

[2] Hobbes, Leviathan, ed. Richard Tuck (Cambridge University Press, 1991), ch. 25, pp. 179–80, quoted and discussed in Quentin Skinner, Reason and Rhetoric in the Philosophy of Hobbes (Cambridge University Press, 1996), pp. 345–46.

[3] Skinner, Reason and Rhetoric, pp. 284–85, on Hobbes’s stigmatising of the art of rhetoric as a cause of the dissolution of commonwealths, in The Elements of Law, part II, ch. 8 and De Cive XII.

[4] Hobbes, Behemoth, or the Long Parliament, quoted from the St John’s College Oxford manuscript in Skinner, Reason and Rhetoric, pp. 432–34.

[5] See Skinner, Reason and Rhetoric, pp. 41–42, on judicial rhetoric’s concern with past actions.

[6] Castiglione, Il Cortegiano, in Hoby’s translation of 1561, on the courtier’s duty to inform his prince frankly of the truth of any matter he needs to understand; see Skinner, Reason and Rhetoric, pp. 66–87, on the ideal and its English reception.

[7] Indeed, perhaps it’s a sign of the times that the Catholic adversarial office was wound back in 1983, and canonisations grew a great deal more frequent: John Paul II canonised close to five hundred people, whereas all his twentieth-century predecessors together had canonised ninety-eight.

[8] Skinner argues that having completely repudiated rhetoric in The Elements of Law and De Cive, Hobbes changed course in Leviathan, trying to beat the rhetoricians at their own game by incorporating their persuasive arts into his civil science. See Skinner, Reason and Rhetoric, pp. 327–425.

[9] Hobbes, De Cive X.xi, quoted and translated in Skinner, Reason and Rhetoric, p. 270.

[10] Hobbes, De Cive XII.xii, quoted in Skinner, Reason and Rhetoric, p. 270; and The Elements of Law, part II, ch. 8, quoted at pp. 269–70, on amplification and extenuation.

[11] Skinner, Reason and Rhetoric, pp. 283–84, summarising Hobbes’s reversal of the rhetoricians’ formula; the underlying text is De Cive XII.xii.

[12] Humfrey Braham, The Institution of a Gentleman (1555), borrowing Sallust’s complaint that we have lost the true names of things; Stefano Guazzo, La civil conversazione (1574), on an age in which courtesy is read as flattery; Ben Jonson, Catiline (1611), chorus to Act IV. Skinner, Reason and Rhetoric, pp. 167–68 and 178, notes that both forms of the corruption were taken to be gaining ground; his summary judgement is at p. 179.

[13] Hobbes, The Elements of Law, part II, ch. 8, quoted in Skinner, Reason and Rhetoric, pp. 269–70. On loci communes as storehouses of pre-existing argument, and Hobbes’s assault on the doctrine, see Skinner, Reason and Rhetoric, pp. 113–17 and 262–63.

[14] Francis Bacon, Novum Organum (1620), preface, discussed in Skinner, Reason and Rhetoric, p. 263. On Hobbes’s service to Bacon, Skinner, pp. 225, 236–37.

[15] William Cavendish, A Discourse Against Flatterie (London: Walter Burre, 1611), pp. 41–42, quoted in Skinner, Reason and Rhetoric, p. 169. Cavendish deleted the passage when revising for Horae Subsecivae (1620); Hobbes, his tutor and secretary from 1608 and possibly the source of the example, restored it in Leviathan, ch. 4, forty years later (pp. 169–70, 340–41). The catalogue of the Detractor is in Cavendish’s essay “Of detraction” (1620), quoted at p. 170.

[16] Hobbes, Behemoth, quoted in Skinner, Reason and Rhetoric, pp. 432–34.

<https://nicholasgruen.substack.com/p/hobbes-on-spin-and-bullshit> <https://nicholasgruen.substack.com>

Nicholas GruenHobbes on spin and bullshit What Hobbes understood about spin and bullshit…Read more12 days ago · 4 likes · Nicholas GruenNicholas GruenWeekend readings and selected Links

Brad DeLong here: Nicholas Gruen says that we are worse off than we were in Thomas Hobbes’s day. In many ways, we are. Back then rhetoric was adversarial and often focused on the past as much s the future. Opposing advocates contested the appraisal of deeds already done, so anyone making a case spoke expecting to be contradicted. On July 24, 1858, Abraham Lincoln wrote to Stephen A. Douglas asking:

Abraham Lincoln: <https://papersofabrahamlincoln.org/documents/D201056b>: ‘Will it be agreeable to you to make an arrangement for you and myself to divide time, and address the same audiences the present canvass? Mr. Judd, who will hand you this, is authorized to receive your answer; and, if agreeable to you, to enter into the terms of such arrangement…

And Douglas agreed. Within a week <https://papersofabrahamlincoln.org/documents/D201067> they had set up seven debates between the man who would be chosen as U.S. Senator from Illinois if the Democrats were to emerge from the fall election with the Illinois-Senate majority and the man who would be chosen if the Republicans were to emerge. The opener was to speak for 60 minutes, followed by a 90 minute rebuttal and then a 30 minute surrebuttal. Douglas and Lincoln were to switch openings debate by debate. These bear no resemblance to the modern “debates” that candidates’ handlers setup,. (In large part, I think, because handlers have little confidence in their candidates and think the risk of having something that is not a joint press conference with little or no follow-up is too great.)

Thus we have this: contemporary anxieties about “spin” are 350-year-old news. And we have this: the distinctive modern pathology is unopposed rhetoric: values statements and PR that pre-empt criticism as they are designed to never face an advocate on the other side.

Now do consider modern science’s own prestige hierarchies, replication crises, and capture by funding interests. Those do tell us that “prestige follows evidence” is aspirational. And it is not true that modern institutional speech faces “no adversary”: we have journalism, litigation, regulators, social media pile-ons, whistleblowers, and o on. Evidence and “front-line learning” do not point to determinate answers that merely lack standing. Disagreement is, often, genuinely about values or truly contested facts. It is only a fuzzy middle for which framing certain things as “facts” and devaluing others can move opinion needles.

But I found this very much worth reading this morning.

Subscribe now

Leave a comment

If reading this gets you Value Above Replacement, then become a free subscriber to this newsletter. And forward it! And if your VAR from this newsletter is in the three digits or more each year, please become a paid subscriber! I am trying to make you readers—and myself—smarter. Please tell me if I succeed, or how I fail…##‎⁠crosspost-nicholas-gruen-hobbes-on-spin-bullshit
##public-reason⁠
##‎⁠crosspost
##nicholas-gruen-hobbes-on-spin-bullshit
#nicholas-gruen
#hobbes-on-spin-bullshit⁠
Based on the post’s content, here are 20 hashtags:
#thomas hobbes
#spin-and-bullshit-and-rhetoric
#rhetoric
#lincoln-douglas-debates
#contestability