A family member recently experienced a medical mistake caused by Artificial Intelligence. More than a machine error, the problem stemmed from the conceptual landscape in which the tech was developed.

They saw a doctor of several decades’ interaction. Over time they’ve shared wide-ranging personal histories. While catching up, the physician mentioned an affliction that he, the doctor, had been dealing with for several months. To protect humans privacy, let’s call it a problem of the vagus nerve, which can affect both physical things like digestion and problems of the mind like anxiety. The conversation was recorded and then digitally transcribed by AI.

The patient’s after-visit summary, which arrived a few days later, stated that my relation had a severe and chronic medical condition. After an initial shock, they realized that the AI had ascribed the vagus-type issue to the patient, not the doctor. The physician had reviewed the summary but failed to catch this crucial detail. It would have gone into the patient’s personal health history if the patient hadn’t caught it. Perhaps no big deal in itself, except that in our healthcare system it could have been a tragedy. “Preexisting condition” can affect treatment decisions, or mean bankruptcy when insurers fail to pay.

Thinking about this encounter between two people and a robot eavesdropper raises lots of questions.

Primarily, what was the nature of the error? In Michael Crichton’s novel The Andromeda Strain, a scientist misses key data about a deadly extraterrestrial organism when a flashing red light causes an epileptic trance. In Aldus Huxley’s Brave New World, Lenina Crowne becomes distracted by erotic thoughts and neglects to inoculate an embryo against sleeping sickness. This leads to a tropical death 22 years later. In those fictional cases, the technology is perfect, and the humans are the problem. It’s odd to think that Huxley had more optimistic underlying assumptions about the reliability of tech than we do, but that is the case.

Isn’t the real world more complicated? You bet. In 2009, Air France 447 stalled and plunged into the Atlantic. Everyone died, from a confluence of bad data and information processing, combined with pilot error. Machines failed, while people lost situational awareness and communicated badly. At the Three Mile Island nuclear meltdown, poor information reached the control room and operators acted for hours on a false valve reading, even though there were other indicators that something was wrong. In 1988 the USS Vincennes shot down Iran Air 655 when the crew misread radar tracking data, something that might not have happened under less stressful conditions.

So is the problem that we tend to assume reliability, even excellence, from machines? To some extent. This is an “insufficient suspicion” argument, possibly an issue for my relative’s doctor. At the same time, automation tends to diminish skills and awareness, because we offload attention when we offload work.

Should the doctor be expected to commit additional anxious labor in vigilance over a supposed labor-saving device? That seems not just self-defeating, but impossible. We can scale up our physical and mental capabilities a thousandfold with automation, but inattention is an inevitable byproduct. And such an expectation would be deeply ironic, since the machine is supposed to relieve overwork and stress.

There are many other parts of life where tech has become widespread too fast to be well embedded, and accidents usually become the acceptable level of deviance. Think of the bad facial recognition AI which led to multiple arrests. The cops trusted the AI over their own procedures. They weren’t paying attention (and remember, the false arrests you hear about are only the errors that get caught.) There are lots of stories of people following driving directions to their ruin, rather than trusting what was literally in front of their eyes. Car crashes that begin when a driver is using cruise control happen at higher speeds and result in more fatalities than regular crashes.

Share

So why is there so little push to get rid of these products? Instead, we discuss how they can be made better and, after enough interactions, achieve the goal of flawlessness. The technology is always wonder-working, fast-improving, and perpetually leading to a better overall quality of life. Needless to say, this plays well in the profit-seeking environment where the world’s leading technologies are created, and where ideas that properly belong in sales and marketing often generalize into overall sensibilities about technology.

With the AI medical transcription, or for that matter with facial recognition and bad driving directions, this becomes the new problem: Is doubting our tools now part of the job? That’s not how a doctor relates to a syringe or a blood pressure cuff, since those are reliable and regulated products that work with exceptionally high similarity in each performance. AI errors, however many there are, are probably unique in each case. To catch them all, a doctor risks burnout on a whole new level, in paranoia about his own machines.

On the one hand, today’s tech designers need us to be paranoid; it’s part of their design. That is how products, particularly those with learning loops and over the air software upgrades, are improved. As I wrote elsewhere, AI is usually deployed like most other software has been over the past couple of decades, imperfect at launch and made better in the field. In this case, doctors and patients likely correct the mistakes, and the system updates. But we don’t really know, since tech companies are bad at sharing these details.

Two years ago a medical transcription company called Nabla used Open AI’s Whisper product, which reportedly had error rates as high as eight in 10 uses, and hallucination rates of about 1.4%. Nabla destroyed the original recordings, making audit checks and accountability a daunting task. This was ostensibly for “data safety reasons,” a consumer protection that also helps the company avoid liability issues. In fairness, a company called Abridge, which developed its language model in conjunction with Nvidia, was sued in April for processing information in unregulated sites without patient permission. There seems to be no good system for engineering both privacy and accountability.

Which takes us to the final question: What’s the environment in which this technology is created? Every technology has a larger economic and social context. At one extreme, you might have a situation like science under Stalin, where a researcher might go to the gulag for believing in the evidence of genetic theory. At the other end of the ideology spectrum, a market utterly free of oversight, in which drugs are launched and used with little testing and product safety, sorts out winners (survivors) from losers. Trust in any technology depends to a large extent on trust in the surrounding society.

For more than four decades we’ve had systems that increasingly favor the development of data at speed and with less regulation, which seems to be the way AI is entering even a doctor’s office. The patients given fabricated histories, or the doctor paranoid about his own tools, are like early users on social media who lost privacy in one or another Facebook experiments, or people who searched Google and got weird-looking results. The errors they encountered were steps in corporate learning.

We’re now part of the process of certification, the key elements of learning. Arguably this is a useful process, as long as the stakes aren’t too high, since it’s probably faster than going through a long procedure of standards-setting and regulation. On the other hand, “the hallucinating AI doctor will see you now” doesn’t sound like anyone’s Brave New World.

Share

For this process to work well, suppliers have to act in transparency and good faith, with an open willingness to say that their product is at best provisional, that one should expect mistakes, and that learning from failure will be open and disclosed. Instead, the products are more likely to be released from the market-based contexts in which they’re developed: User restrictions, trade-secret protection, mandatory arbitration, weak liability, deletion of source material, a failure to aggregate reported errors.

In short, development without discipline, and information control that prevents any of the supposedly high-value market transparency. Excellence is promised, but the execution is corrupted by self-interest.

I once nearly died of curiosity while swimming in the Pacific. In Guam there is a reef that falls away like a giant staircase, each step perhaps 20 feet below its predecessor. The change in light and temperature between each level creates a unique ecosystem of plants and creatures. I swam down as long as I could, fascinated by the multitudes of life and color at each level.

When I could go no further I looked to the distant surface, unaware until then of the distance I’d traveled. It was a terrifying return: I was desperate to inhale long before I burst through the surface. I reached air barely able to float myself to shore, where I lay for over an hour, careless of anything except recovery.

My blue lake dive was completely different. I was not investigating the lake’s ecosystem. I was enchanted by its singular undefinable blue, though Oregon’s Crater Lake, the deepest body of water in the Western Hemisphere, has a disorienting natural history for swimmers. Large pumice rocks float past, big as a basketball and light as a sponge. A hemlock tree has drifted there for 450 years, bobbing without becoming fully waterlogged. The water is icy even in high summer, but the swimmer leaves reluctantly, since the water’s clarity is captivating.

“Crater” is a misnomer. The lake, nearly 2,000 feet deep, lies inside the caldera of a onetime volcano, 12,000 feet high, that exploded in 5,700 B.C. Twelve cubic miles of magma blew out, and 4,500 vertical feet of volcanic cone collapsed into the sudden vacancy. Over the next 250 years the lake was formed entirely from sky, rain and snow melt. Even today the water is so pure that one may see objects 140 feet below the surface. Light continues its conversion to heat and fuel well beyond that, reds first and so down the spectrum, before fading to darkness. Light feeds ancient moss 600 feet down, the deepest photosynthesis on Earth, using one-thousandth of 1% of the brightness of the lake’s surface, short waves of residual blue light.

I witnessed my pure blue a few dozen yards from the beach, where the lake bed meets an underwater cliff. What was visible ground becomes endless unformed blue. It’s ever more concentrated the longer you stare down. Every swimmer I’ve watched there turns soon after the cross over the cliff, though they were in no more danger than they had been just a few feet back. The sight of blue darkness at the center of vision is simply too rich and troubling.

These were two Siren songs, then, Guam and Oregon. One sings of the numbered richness of the natural world. The other is entirely itself, and suggests a passage to something I can’t name. The ocean’s song almost killed me out of reckless fascination. The lake’s song touched me more deeply, but I too swam away. Even now at times, living instead of diving into the blue can feel cowardly.

Share

Blue has that hold on our minds. In “Bluets,” the poet Maggie Nelson writes of her love affair with this color, called “primal,” “the fingerprints of God,” with meaning beyond meaning. It is the color for a place, she quotes Ludwig Wittgenstein, where “explanations come to an end.” She associates it with her lost love affair, primal in its sexuality, that made her feel alive like nothing else, for all its “morbid heart.” In “On Being Blue,” William Gass associates blue with our interior life. He revels in its varied uses (blue laws, blue periods, having the blues, blue moons and bloods, and much blue sex besides) showing the reader the necessity and impossibility of truly depicting reality with language. And yet we must try as much as we can. We work in words and their limitations, even as we desire more.

Both writers notably take blue as a color of erotic life, our most direct experience of existence without language, and our closest intimation of the life force that will pass through us and continue when nature is done with us, to a place beyond speech.1 If sex is both pre- and post-language, this may be why most writing about sex is bad, and most of the decently-written portion of sex writing is about bad sex.

There are people in North Africa turn themselves blue intentionally by wearing indigo cloth; the Moslems who failed to convert them called them Tuareg, or “Without God.” They call themselves Imashaghen, or “Free People.” Western painters could not depict intense blue until precious lapis lazuli from Afghanistan, deep and pure, became available. Thereafter it became the color of the Virgin Mary’s mantle, the mandorla surrounding the risen Christ, and whatever else a rich patron wanted to show had inestimable value. While the Romans disdained blue, old Islam held it a color of dark magic, and China thought it signified torment. In today’s globalized and increasingly disenchanted world, blue is the most popular color by far, seen on our young jeans to our aging hair rinse.

The blue I saw in Crater Lake, unattached to anything, was blue in its infinite state, the living expression of Maggie Nelson’s thing beyond words, William Gass’ touchstone for the limits of language, the ineffable sacredness of the mandorla, an unbounded freedom I, we, intimate. As mortals, existing in only the tiniest crack in eternity, we yearn for more, the “more” of infinity, either physical or spiritual.

That was the blue that beckoned, and if it had killed me, likely that would have been incidental to the blue itself, as it was timeless.

* * * *

Infinity is today’s great seducer, the final dream of Reality, Unlimited. The scoreboards of our ultra-wealthy are superficially in the hundreds of billions, or now talked about as trillions with a straight face, but they might as well be “infinite money.”2 The unimaginable richness of every life on Earth is endlessly documented and stored, with more sensors on the way. We can’t store enough, we can’t stop counting and numbering things. The current fad for living forever, an idea no one can really grasp3 even though thousands seek it out, just hit a notable wall, but it continues.

But our closest romance with infinity today is with Artificial Intelligence. Large Language Models are constructed in ever-greater size, with the assumption that the only thing better than 10 trillion parameters is 20, no make that 30, and performance will continue to improve forever. Coursing through ever-larger data centers, eventually we’re supposed to have an AI “superintelligence,” which will far surpass us, supposedly the smartest creatures in known creation, and it (they?) will work on themselves to create new sensors and means of understanding, until perhaps the unified field of physics, long sought by our best minds, will be one more achievement on the way to infinite understanding.

Dreams of the infinite are paradoxically dreams of culmination. Which is to say they are dreams of apocalypse, a common fear and desire across thousands of human generations. What color would that take?

This week I was thinking about the beckoning blue I swam across, and our love of the fatal infinite, when I asked various popular AIs what color pure AI would be.

It would be blue, and it would not be enough.

ChatGPT

ChatGPT, along with Claude and Google Gemini, initially tried out the non-choice of “iridescent,” an attribute, like iridescent green, or a shimmering mixture. AKA a waffle, designed to please everyone — perhaps what you’d expect from software that scrapes everything, and gets fine-tuned to please its human and maybe sell some ads. Eventually ChatGPT came up with this blue.

“The cool rationality of blue and the imagination of warmer tones.”

Claude

Claude’s choice was more argument than mood. “Historically it’s the color of things that were rare and slightly suspect: royalty, mysticism, radiation warnings in old sci-fi,” the machine wrote. “Powerful, expensive to produce, and nobody’s entirely sure whether to revere it or be wary of it.” In other words, it came to its color by a language game. And failed: Reader, does this smack of royalty and mysticism?

“The last thing your eye registers before the spectrum keeps going without you.”

Google Gemini

Gemini got to this after offering me several choices of colors, and waffles like “let’s drop the fence-sitting: grey…the ultimate neutral,” followed by “red, an absolute refusal to be neutral.” It was like an encounter with the most embarrassing adolescent date, trying to figure out what to be so you would like them.

At last it produced this blue, a “synthetic” color without anything to latch onto. It came closest to the color I saw, which still isn’t very close.

“A synthetic color that feels like a physical void on the canvas, drawing your your eye in but giving you absolutely no texture or depth to latch onto.”

How could I have expected anything different? This is software made entirely from language, and it cannot intimate the primal and eternal world beyond language. True blue is everything before language and after. But language models have only words, and not the experiences we suggest with language. They sit apart from desire, since unlike us they are exclusively trapped in language, more able to depict the ecosystems of Guam than the unreachable color in Crater Lake.

For some people, particularly the ones who fall in love with chatbots, that may be enough. Particularly if they’ve never seen the real thing, and then swum away, wondering forever about what they witnessed, and where they might be going.

Party, Boomer! Z! Millennial!

Cheer our semiquincentennial!

(You hipster Xers stay perennial

And sneer that it’s “sestercentennial.”)

*

Twenty decades back, plus five,

Our fathers fore did bring to life

Our freedom, our pursuit of bliss,

A path to wisdom, from excess.

*

America goes to eleven,

Going big is our hog heaven.

In fast food, finance, robot dreams,

We tally only in extremes.

*

Let’s start where we are truly Great —

That is, in terms of real estate.

We’ve Great Salt Lakes and fresh ones, tops,

Great Plains and Basins, Dismal Swamps.
*

We’ve Great Sand Dunes, Great Mountains Smoky,

Salt Dunes, Mammoth Caves…oh, okay,

That’s not “Great.” This esoterica

Adds a theme park: Great America.

Stride west blindfold, you’ll tumble in

Our greatest “Grand” - the Grand Canyon.

We’ve got Grand Coulees, Escalantes,

Mesas, Tetons, bet your aunties.

*

And let the devil have his due,

We’ve Devil’s Tower, his Golf Course too.

Devil’s Postpile, Punchbowl, Den,

Kitchen, Backbone, Canyon, Glen.

*

Great America has fried butter,

Corn dogs, root beer, fluffernutter.

We’ve snacks to go with all our moods,

Unique, extreme, and faster foods.

Share

Notwithstanding all surmises

You’ll never outcount our franchises.

There’s two-fourteen, says Wikipedia.

That’s trop for some, but we are greedier.

*

Ghost pepper death beef jerky, say,

With spray cheese and a chai latte.

Ranch dressing for your rattlesnake,

Pop Tarts, Funyuns, funnel cake.

*

It’s not our land without the mobs

Of people holding extreme jobs:

Mascots, bondsmen, muckety-mucks

And drivers of our monster trucks.

We’ve billboards full of litigators,

Quants and lobbyists, naked waiters,

A hot dog eating contest trainer,

A marketer who says “no brainer!”

*

Life coaches, televangelists,

Pool guards, giggers, and this list

Smiles on the kids content to slack

On slacklines, with a hacky sack.

*

Moonward fly in no-grav britches,

Fret if hot dogs are sandwiches,

Argue “both!” — and then eat each,

Extreme win-wins are what we preach.

*

But we are more than Team Gulp sodas,

Surfer dudes who chase tornadoes.

Our most extreme have found their answers

Within this land of second chances.

*

Consider lives like Robert Downey

Steve Jobs, Oprah, Clinton, Ali,

Honest Abe and Scott Fitzgerald —

With postlife acts unparalleled.

Share

And though my reader here may slump,

I count comebacks of Donald Trump.

We venerate a second chance

— It’s why we lead in immigrants.

*

Cheer less-sung folks like Ida Wells,

The Glen who gave us Taco Bells,

Norman Borlaug, Bayard Rustin,

Claudette Colvin, Cissy Houston.

*

We’d be remiss to leave unmentioned

Our epic story of invention.

There’s CRISPR, and the Ferris wheel,

The gadgets made by Ron Popeil,
*

Hamilton’s money, Jefferson’s desks,

Thomas Paine’s bridges, Madison’s checks,

Franklin’s bifocals and Washington’s mules,

Rush’s bloodletting -that Founder’s confused.

Today transistors, planes, vaccines,

Muzak, mass-product machines,

Robocabs don’t leave us waiting,

A-bombs, which we’re still debating.
*

We bent gravity, so what challenge remains?

Why, quantum computing connected to brains!

There’s umpteen more puzzles, I’m telling you, readers,

Like solving for death, and for squirrels in bird feeders.
*

There’s so much more that cuts our cloths,

From New Madrid to Yucca Moths,

Alleys (bowling and tornadoes),

Music, moonshine, lemonados.
*

We construct mountains from a molehill,

Our weed stores should be called “Da Tokeville.”

For Alex D., who knew us best,

Proclaimed our genius: restlessness.
*

It’s true: There’s little on this shore

We leave untouched by Mr. “More!”

We tinker, bust, and reinvent

-And will, unto the quincentent.

Share

What if, unlike after the past couple of catastrophes, the Powers That Be are unable to gin up a relatively quick recovery? Then we’ll return to the standard historical form: After a big crash, people stop believing in stocks and investing, and care less about companies generally.

One unintended, unremarked, consequence of that will be that Journalism will suffer far more than anything in tech. Business and financial news have been among the brighter spots in the 35-year downturn of Journalism, and advertising from these sectors is among the best in the industry.

What if there’s a bust, and then nothing interesting happens in the markets for 10 or 20 years? That would be one of the biggest cultural changes in over 40 years, ever since we began our national fascination with big companies, celebrity CEOs, and stock market riches.1 All, it should be noted, a likely consequence of the increase in business and financial advertising in that same period.

[Disclaimer: This is not investment advice.2 No one knows when a bubble will burst, or why these crowd manias suddenly reverse. Plenty of smart people have gone broke trying to time a market collapse.]

Let’s walk through this issue, with an eye to what happens if the market velocity slows.

Where we are, and the big new problem

The bigger a bubble stretches, the more it will splatter the rest of the economy when it bursts. Currently we have a world-historical big bubble. Evidence includes the $700 billion that just four companies will spend on data centers this year, with neither significant customer demand, nor a clear idea of what they’ll be able to charge customers for AI. And how did SpaceX, an unprofitable company associated with AI,3 come to have a market capitalization of $2.6 trillion after its IPO, for a single puzzling moment exceeding the value of Microsoft (2025 net income, $102 billion?) Or how come this year’s AI-related IPOs have a valuation of $4 trillion, more than five times the previous record?4

If all those (mostly make-believe) trillions vanish, it will be bad for investors. The real concern, though, is the effect on money and credit overall, owing to the less measurable, yet more significant, areas of emotion, psychology, and social culture. How people feel trumps everything else in markets.

The best example of this that I’ve lived through occurred in October 2008, when McDonald’s had trouble obtaining its normal overnight financing. It wasn’t as though people were going to stop buying hamburgers the next day; it was that in the early days of the financial-sector meltdown, no one trusted anything. Not even Ronald, or Mayor McCheese.

Soon after that we had big corporate bailouts. Central banks reduced interest rates so much that loans amounted to ‘free money,” and the big boys could borrow and invest with little worry about repaying interest. That strategy had also worked pretty well after the 2000 Internet bubble burst.

The goosing worked in 2000 and 2008 partly because inflation stayed under control despite all the free money. It may have been because China was suddenly selling things for cheap, or there was low corporate demand in a recession, or there was a general fear from workers that if they pushed for higher wages they’d lose their jobs.

By contrast, this time around America’s war of choice in Iran has jacked up prices for oil and fertilizer, among other things. And Europe’s massive heat wave is likely to put further pressure on global food prices, even if you haven’t read much about that yet.

If inflation persists, central banks won’t have as much room to reflate the market without ruinous inflationary consequences. That would mean that any recovery would have to happen without help from central banks.

Feelings, oh oh oh, feelings.

Most people recall the crashes of 20005 and 2008, with those relatively speedy “free money” recoveries. That may be a recency problem. The historical norm is a much slower and more spiritless recovery. Take 1968-1982, a 14-year gap when the market did more or less nothing, and it was neither interesting nor fun to work on Wall Street. Or 1929-1954, the 25 years it took for the Dow Jones Industrial Average to recover its pre-crash high. Or one I saw up close, Japan’s 35-year crawl from 1989 to 2024.

Pundits will offer all sorts of economic and financial reasons why some recoveries take longer than others, but eventually their talk turns to things like “animal spirits,” “exhaustion,” “capitulation,” or simply “fear and greed.” Emotions are at the heart of all mass investment.

In those long downturns, trust thins and stocks stall. The market has a few false recoveries, which turn off ordinary people. They put their money in their mattresses, or stuff it into old-fashioned savings accounts. Nobody invests.

Share

The experts will tell you that, as a society, we’re past all that. Ever since 1981, when people could put money from their paychecks directly into 401(k)s,6 there’s been a lot more participation in the market. In addition, computers have made it possible to model all sorts of things as financial instruments, one reason we got that subprime mortgage mess in 2008.7 And given the rise in online gambling, speculation certainly seems to be in the national bloodstream. For now.

When our bubble blows, expect to see all kinds of fancy financial moves to manage the catastrophe. Nice, but the key to recovery is still committing to invest for the long haul. If stocks stay down for long enough, and the game is just no damn fun, it could have quite an effect on the retirement plans of lots of boomers, as well as the inheritance plans of their kids. The mood about stocks could be sad and bored for a long time, if there’s no free money.

Which brings us to how this could take out Journalism.

The Industry Pillar, 1980-now

The Journalism business has been bad for decades. Ad pages peaked a few years before the commercial Internet arrived, and since then it’s been a pretty unrelenting bummer. Profit margins have gone from “twice the average of the Fortune 500” to “what is this word ‘profit’?” Overall employment has fallen by perhaps 80% since its peak, and one-third since the 2008 crash.

The work itself has leaned increasingly towards scoops and sensation over depth, since it’s now a more twitchy, mobile ads-based business. The biggest reputable success is probably The New York Times, my former employer, which some uncharitably allege does well because it’s got Wordle and recipes alongside news. Even so, since 1996 the company’s stock has risen only one-third as much as the S&P 500.

The one substantially positive story for Journalism has been the growth in business news coverage. When those 401(k)s became popular, there was an explosion of ads from stock brokerages, mutual funds, banks, savings plans, and other financial companies. And - brace yourself - a lot of what journalism covers has to do with where it can sell ads.8

The Society for Advancing Business Editing and Writing counted about 100 members in 1980, and today it has some 1700. Magazines like Money, Worth, Smart Money, and Kiplinger’s, plus related business titles like Entrepreneur, Fast Company, Business 2.0, Red Herring, and numerous newsletters and websites, owe their existence to financial ads. And those ads pay better than most. Local business publications blossomed. Many are now gone, but they lasted decades, much longer than regular newspapers, and they trained more than one generation of journalists.

All thanks to the national fixation on the stock market, business, and big money generally. Interests that also support cable business news, and for a long time stalwarts like Fortune, Bloomberg BusinessWeek, and Forbes, another place I used to work. BW was eventually sold to Bloomberg, which gets its main money from selling terminals. Forbes’ Brandvoice, an online PR service, brings in revenue.

Share

So what if, culturally, stocks and big money just turn into a bore? What if there’s coverage of a crash, and then years where nothing really happens in the markets? What if people stop caring about what happens day to day?

For an industry dependent on business ads, the news will be: Look out below. The rest of us will have to prepare for a new, and even tougher, fight for facts and truth.

The only likely substitute for the revenue from business news, the only format where people take passionate interest in new facts because it can make them money, is even more speculative than stocks: The aforementioned sports gambling and prediction markets. News that The New York Times was paying their star sports gossip $800,000 may indicate this is already a reality in contemporary journalism.

If so, the world can’t get boring fast enough.

It could only happen in a town like San Francisco, which has always punched above its weight when it comes to theatricality.

Today the main artifacts of the Internet, data and networked computing, have created Artificial Intelligence. San Francisco is still the center of the action, but an image capturing the moment would be a lone developer scrolling their smartphone behind the tinted windows of a robo taxi. They head to a private gathering where they’ll once again discuss whether they will create a new world, or simply destroy the existing one. High-end tickets to Burning Man1 now cost $3000, still less than a decent Ayahuasca retreat, but who’s got time for that?

As The Brains sang long ago, Money changes everything.

Share

That is the mood evoked by “City on the Edge,” Jonathan Weber’s memoir/chronicle of tech and politics in San Francisco over the past 30 years. I’m interviewing Weber at the 9th Avenue Green Apple Books on July 1. It’s free to any Curiosity Shop reader who wishes to attend.2 It’s an essential read for anyone who was there, or anyone who wants to know how we got here, in tech and society.

I don’t mean to say that San Francisco can rival New York in terms of blockbuster shows, though The Bay Area is decidedly outsized in terms of arts businesses and community support. It’s more that performance and display are a bigger part of the city’s life, not least its politics, than any other town.

The most famous political figure of early San Francisco was a madman who never held elected office, but declared himself Emperor of the United States, dressed ostentatiously, delivered a great many public proclamations, and issued a currency that was honored in several local establishments. He was a performer, and the city provided the stage, the audience, and the supporting characters. It’s like that a lot.

In 1964 LIFE Magazine declared San Francisco “the capital of gay America.” It’s been a place for joyful public displays of gay life, mass political anger after the shooting of Harvey Milk, collective mourning with the AIDS quilt, and eventually normalization with the first city-sanctioned gay marriages. In 1967 the city celebrated hippie life with the Summer of Love. Perennially San Francisco sees enormous protests against one or another U.S. policy, usually with characters in parodic costumes. If regular protests are your thing today, one website alone lists 45 in San Francisco between now and the end of July. For sports fans, the big athletic event, Bay to Breakers, is half footrace, half costume ball.

No wonder that Steve Jobs, who may have been a child of nearby Silicon Valley but who spiritually aligned himself with San Francisco’s culture of display, was such an exceptional showman. Apple Computer would never have gotten so far without his packaging. He gave his most famous (and very theatrical) keynotes at Macworld, held at San Francisco’s Moscone Center.

Besides tech, Weber’s history chronicles a particularly rich period in San Francisco political theater, the rise and collapse of what he calls “The City Family”: The political dominance of Willie Brown and his allies for 30 years starting in 1995.

Under Brown’s mayoralty City Hall got a makeover that dressed it up to look like a European capital. His handpicked successor, Gavin Newsom, allowed celebratory gay marriages in those glitzy halls, even while such marriages were still illegal in the U.S. Ed Lee was installed by the City Family; his role was to bring tech money into highly visible projects like the Twitter building to revive the rundown part of Market Street. London Breed, the last mayor in the Family’s story, came to ruin over homelessness and fentanyl, amid the urban vacancy of COVID that created horrifying theatrical backdrops for right-wing videos condemning the city’s dysfunction.

Weber was there for much of this theater, particularly the first part of the tale. He ran the editorial side of The Industry Standard, an overnight sensation of the dotcom years that set records for ad pages sold, but never really made money. For a magazine covering highly valued/soon bankrupt websites, that’s a formidable marriage of form and content.

The Standard had its own massive billboard (every tech company had a billboard, it seemed, often before they had a product.)3 Like many tech companies of the time, The Standard held wide-open, overcrowded parties. These were sponsored by the companies they covered, or by fusty older tech companies that wanted to look cool. Before everything went south, those parties, places to schmooze and be schmoozed, eventually became the product the magazine tried to monetize, taking “the rooftop” to other venues that could hold even more revelers.

When the dotcom bubble exploded The Standard went with it, monetized parties and all. To the extent the theater of tech continued, it was highly corporate. In 2003 I attended a Wallflowers concert sponsored by TMobile on Alcatraz. Even more memorably in 2012, Google co-founder Sergey Brin and a couple of skydiving Googlers had a virtual Google Glass “hangout”4 while leaping from an airship onto San Francisco’s Moscone Convention center. And around that time Oracle was throwing big parties out on Treasure Island. But these were staged by Oracle, the most straight-laced of all tech companies, for an audience of corporate buyers.

Most of us were more spectators than actors in this highly managed environment. If tech was still a performance, the subject was mostly revenue.

Attitudes among the groundlings hardened towards tech. There were protests against the (traffic improving, less polluting) Google buses, which seemed to the protesters like a mark of big money takeovers of the people’s city, all these newcomers with their high-salary jobs. Meanwhile, the City Family was hoping companies like Twitter5 could revive San Francisco’s troubled Mid-Market area. But this corporate display came up against more resentments, for example that the techies’ dog kennels cost more than the locals’ SRO hotel rooms (something I wrote about for The Times.)

Tech became the enemy, in the minds of many, and that feeling has continued to grow. In response, much of the tech leadership has withdrawn from its public profile. Tech leaders have retreated to exclusive private clubs and organizations (like Peter Thiel’s recently-exposed “Dialog” group.) The existence of those clubs stokes the paranoia of regular citizens.6 As I’ve written elsewhere, the new superclass lives separately from the rest of us, even as it governs. It’s fair to ask if they still know how to join a normal crowd, or enjoy the theater of a freewheeling public event with people unlike themselves.

Ironically the tools of tech that sprang out of a very social tech world, like smartphones, social networks, and chatbots, have instead isolated both the tech elite and the rest of us. For everyone, these tools and the tragic consequences of hyperpersonalized marketing, have made it harder to encounter and enjoy people different from ourselves.

Fatefully, the digital tech solutions to COVID, like cloud computing and personal video, probably saved millions of lives globally. They enabled work from home, and effected the first mass quarantine that didn’t destroy an economy. But there was a cost: People’s became less adept at normal socializing. It’s troubling that AI is now being built primarily by people in their early 20’s, the ones who were home alone under COVID exactly in their prime years for learning social interaction.

Share

It’s impossible to know if today’s private and siloed world will endure. Maybe tech culture will become more participatory again. Or maybe not: Competing with the “996” (working from 9 AM to 9 PM, six days a week) ethos of China doesn’t give people much time to develop their social muscles.

If the constant theatrical party is truly over, it will be a poignant reminder of a very familiar American story. It goes like this: People found what looked like a new Eden, rich with inspiring possibilities. It was pioneered mostly by the young and imaginative, with little money but lots of community and possibility. Then the money showed up, and there were new struggles. Some were corrupted, and others fell away. Instead of a dream, there is merely a progression of success, that thing William James termed “the American bitch-goddess.”

Lots of pioneer stories sound like that. Big money pushed out Prairie populism and defeated cooperatives. Big money discovered hippies could be a market for pop music, Levis and Pepsi. When the punks and artists moved into New York’s derelict Lower East Side, big money raised the real estate prices, something I saw firsthand. The only things to escape this trend, like public libraries and civil rights movements, tend to be harder to monetize.

Do what you love, they used to say, and the money will follow. They just didn’t say what that would mean.

Can’t guess the town I visited? Here’s another clue:

It may be the town’s most upright hotel.1

Las Vegas aches with obvious irony and overdone metaphors. Normal people drink cheap vodka at $300 a bottle, so they can feel rich. There’s neon everywhere, alongside shrieking slots, arctic-level air conditioning in 100-degree days, and little cards reminding us to conserve energy. People confidently place money on their highly rational roulette “strategy,” as the fateful ball bounces randomly on its wheel.2

I was there for the tech conferences.3 Brace yourself: Between morning donuts and late night beer, I heard about Artificial Intelligence. Specifically, about the rapid erosion of security and trust because AI makes it easier than ever to fool people. And about threats from malicious AI, which dictates that we must all assume we’re under attack, all of the time, even from people and things that until recently were normal parts of our environment.

Two aspects, one human and one machine, stood out in particular.

The temp from Pyongyang

I ran into my old friend and colleague Nicole Perlroth, who was a computer security reporter when we worked together at The New York Times. She wrote a great book about the market for cyber weapons, which include tips about vulnerabilities in business computer systems, and ways of spying on people through their own phones. She now invests in security companies and, working with the security company Rubrik, has a podcast about cyber threats. The first season, about Chinese spying, just won a big award.

Her new season is about the North Korean information technology specialists who work remotely for what may be thousands of U.S. companies. This has been going on for a decade, but it was supercharged by the COVID lock down, when companies ditched the office, and needed to put a lot of IT work online. The demand for AI coders has now increased opportunities for the North Koreans like never before. They use AI to produce more convincing fake identities, and chase more jobs.

Once hired, these attackers can tap into all kinds of passwords and secrets, plant malware, and otherwise set traps for future attacks. Meantime, their American wages help fund one of the world’s most hideous regimes.

One sad wrinkle inside this story is how the North Korean remote workers prey on clueless Americans, along with dupes in Ukraine, Argentina, and elsewhere. Since the attackers can’t tunnel directly into a U.S. company from Pyongyang4 without giving themselves away, they con people in the U.S. into hosting in their own homes the company laptops these workers have been issued. Then, the North Koreans tunnel into a seemingly innocent U.S. address, and the company can’t see where they really are. North Korea is exceptionally good at online social engineering, and duping folks who are looking for ways to make easy money. Last year a woman in Arizona was sentenced to 8.5 years for hosting one such laptop farm that touched 300 U.S. companies.

Now, thanks to AI faking, these guys (and they do seem to be almost entirely male) are flooding the market. Nicole appeared onstage with someone who figured out that a North Korean was applying for a job, so he notified the FBI, then sent the man a corporate laptop fully loaded with spyware. Even the delivery box had trackers. What they found was a cell of overworked IT fakers desperately looking for real work in the US, even while they were spying on their current employers. One guy applied for 27,000 jobs over a six-month period. Another actually landed 19 jobs over that period.

What can be done? Fortunately there are some useful tells that companies are starting to recognize. Faked resumes often indicate work experience that goes farther back than the age of the person who shows up for the online interview. Keystroke latency during an online test can show that the applicant is on the other side of the Pacific, and not in Florida.

Now companies are starting to close out interviews with suspected agents by asking them to say, “Kim Jong Un5 is a big ugly pig.” Since these humans are monitored in real time by the secret police, they can’t say that.

Just in case you think your job sucks: Try working in several IT jobs at once, while applying for 16,000 more, under the loving eyes of state security. You could almost feel for the enemy.

Share

The leopard at the watering hole

Elsewhere, much of the talk was about how Mythos, the powerful Large Language Model from Anthropic, has completely upended computer security. In late May Anthropic said that Mythos has found over 6,200 security vulnerabilities, across some of the most important software we use.

I heard a senior person from Anthropic security say that, possibly as soon as the end of this year, open source LLMs will have that kind of power, meaning bad actors will go on the hunt for new vulnerabilities. There’s now a race on to harden everything ahead of that, and maybe things will get patched in time, as happened with the Y2K threat.6 Still, even when the pros feel good about their system, they worry about flaws in some forgotten computer in, say a local utility or a water treatment plant. We’re all so immensely connected that such flaws could easily touch us.

Security threats used to be something that came to us from elsewhere. Now, as the North Koreans working in our payroll and password management departments show, the threats aren’t coming from outside, they are right at hand.

That is a big deal. Up to now we have kept away threats like hacks and viruses with firewalls and antivirus software. There were typically days, if not weeks, between an initial breach and a disaster, because the bad actors were manually mapping out a corporate system. Now it happens at what they call “machine speed,” meaning an AI might be in and done in less than a minute.

Game over? More likely, we’ll develop new strategies, new ways to work. We’ll assume everything is always under attack, and figure out how to work that problem. It’s likely that the new strategies will take cues from nature, where attack and defense have been refined for billions of years.

The industry has been doing this for more than three decades, exemplified by the seminal antivirus paper, “A biologically inspired immune system for computers.”7 Many elements in security have natural parallels: Firewalls are like the armor of turtles and armadillos; digital identity and authentication are like ants that can only enter their colonies if they emit the right pheromones, or else they get attacked.

If we used to have armor against threats from outside, now we’re like the animals at the watering hole, continuing to drink even as we watch the lion that drinks nearby. Such creatures have exceptionally fast awareness of behavior changes, and react quickly, often in ways that confuse a predator. They zigzag too fast for the attacker to focus, or sacrifice the least able of their group, much the way companies may sacrifice some nonessential data to gain an edge over an attacker.

Where tech itself is concerned, work and identity will look different, perhaps more than in any other sector. Smart people are working on new things all the time. The road to success may become as scary as it was during the worst nuclear days of the Cold War.8 Nicole tells me she’s never been more frightened, but she’s been telling me that for a decade. Historically when we experience shocks, we gain understanding, and learn to live in a changed landscape.

If that sounds like someone rationalizing their participation in a potential catastrophe - well, I have been staring at a man playing roulette. The whole time, he looked like he knew how to win.

Here is evidence from Vienna’s Klosterneuburg Monastery, which I visited on my break in Vienna. It currently has a delightful exhibition showing how we imagine unfamiliar spaces, from Jerusalem in the 14th century to today’s war in Ukraine. Klosterneuburg has been under continuous management for 800 years, enabling its good library.1

I especially liked the Cosmographia of 1574:

“Creatures of the Northern Seas,” you have my utmost respect, and not simply as a woodcut that launched 1,000 tattoos, first on the skin of sailors, and then hipsters.

The location of the monsters is notable. Long after much of the temperate world was known,2 the polar zones remained exotic. When India was part of Britain’s economy, and America was crossed by rail and telegraph, whole ships of polar explorers still vanished forever, their fates unknown. Over centuries, our farthest reaches were places of dreams and fear.

Examine some of these beasts close up, along with their descriptive notations:

“Whales as big as mountains, capable of capsizing large ships. The Icelanders call them devils.”

“Snakes, more than 200 or 300 feet long, capable of coiling themselves around ships.”

“One has terrible teeth, the other terrible horns and a fiery face. Its eyes are 16 or 20 feet in diameter, its head is square and its hindquarters small.”

Share

Let’s not forget deadly lobsters, and the beasts protecting us from them:

“The lobster is so strong that it can catch and strangle a human.”

“A ferocious animal resembling a rhinoceros. It has a pointed snout and back and feeds on lobsters. 12 feet long.”

Finally this guy who, while doubtless also fearsome, still seems like a rare good time.

“This bird is as big as a goose and has a pouch under its beak. When this is filled with water, it makes a cry like a donkey.”

I have omitted the snake birds that live under the snow, the bear/whale, and the animal that so loves to eat that it squeezes itself between trees to purge, then eat some more. You get the picture.

And tell me, doesn’t part of you also wish they were all true? Is not some part of you cheering on the rhino-type fish that defends us against man-killing lobsters? The thrill and terror is timeless. And it always sells.

Share

That’s more or less how I make sense of the SpaceX IPO, a scary document of exotic suppositions of a somewhat different kind.

According to its S-1, SpaceX’s mission, the core of a business at which many of the world’s rich people are about to throw $75 billion, is

to build the systems and technologies necessary to make life multiplanetary, to understand the true nature of the universe, and to extend the light of consciousness to the stars.

In case you don’t know what “making life multiplanetary” means, this government filing comes with a nice illustration of a death star-type orbiter around Mars. Which, for all we know, cries like a donkey when filled with water.

As for “understanding the true nature of the universe,” the SpaceX lawyers should have noted that theology has a historically dodgy ROI. “Extending the light of consciousness to the stars” sounds like a fun business model, but it should be noted that while Mars is far, the nearest (maybe)3 habitable planet outside our solar system, Proxima Centauri b, is about 76,000 times farther.

I won’t go into the particulars. Over at the Verge, Elizabeth Lopatto has written an excellent piece on the “financial nihilism” of this IPO. Even ignoring the money, here is a nice video on why, even with all that money they’ll raise, colonizing Mars almost certainly won’t happen.

But tell me, don’t you wish it was true? As it says at the top of the SpaceX website

SpaceX was founded under the belief that a future where humanity is out exploring the stars is fundamentally more exciting than one where we are not.

“Fundamentally more exciting.” Like far-off lobsters that gobble men. That, as they say, is entertainment, and it’s 76,001 times better than the facts.

In that sense, SpaceX is less a company and more like an ongoing media event, designed to be fundamentally more exciting than, say, exploring the now mapped oceans.4 We are, eternally, critters that look to have our imagination stirred, fundamentally excited by dreams and fear. Clearly it’s a much bigger deal than mere “rational” markets.

1

As well as the world’s most imposing gift shop:

2

Check out this image from 1482, concerning the types of humans living at the edges of the world:

(also-if I ever vanish, seek me in this library)

This sort of conjecture of exploration - one-footed men, six-armed savages, headless and dog-headed humans - was long since disproved by the time the North Sea monsters were illustrated. The imagination must go somewhere.

Ink-Stained Mensch

I arrived here from London, where I hid out from some truly epic heat by visiting The National Gallery. There I caught this familiar gent:

He is Desiderius Erasmus, the groundbreaking 16th-century humanist/editor/aphorist/thinker/satirist. A self-made guy, Erasmus is one of those thinkers, like Voltaire or Aquinas, whose way of seeing influences future centuries. Even while alive, he spurred the Renaissance in Northern Europe, contributed to what became the Reformation (though he had issues with the people involved with that), and changed how we think about books and facts.1

He has over 400 publications to his name, but his real fame and influence is due to his networking; Erasmus was known to write up to 40 letters a day, often witty and profound, and counted among his correspondents many of the influential people of his time.

Which made this detail in the Holbein painting jump out:

Check out those cracked and inky fingernails, and the ink stains our eminent sage has left on the fine tooled leather covering that book. The writing on the book’s edges praises Erasmus’ labors with the pen, comparing them to the labors of Hercules.2 The fingernails and the ink stains are further evidence. It’s an extraordinary touch, since at that time such a book cost the equivalent of several months’ pay for a working person, and Erasmus has ruined the binding with his inky fingers.

Erasmus sent this painting, which became one of his most famous likenesses, to the Archbishop of Canterbury, a patron of the scholar. No mere laborer, the Archbishop still doubtless knew the value of a fine binding. I hope he enjoyed the painter’s lightly barbed humor.

I am married to a fine bookbinder so the desecrating stain leapt off the canvas. Strangely, there’s nothing in the National Gallery’s descriptions of the work that touches on this detail. I’ve written the NG about the stain, and will let you know if I hear back. Meantime all reader speculations welcome.

Share

Foundational Truth

Last Sunday’s piece was about the fears that AI will make us humans lose our humanity. On Monday the Pope published his take on the same subject. It was a little more preachy, but what do you expect? 3

There’s a big payoff in paragraph 99, though, concerning whether AI is intelligent:

99. It is not possible to provide a single, comprehensive definition of AI. What can be stated, however, is that we must avoid the misconception of equating this type of “intelligence” with that of human beings. These systems merely imitate certain functions of human intelligence. In doing so, they often surpass human intelligence in speed and computational capacity, offering tangible benefits across many fields. Yet this power remains entirely tied to data processing. So-called artificial intelligences do not undergo experiences, do not possess a body, do not feel joy or pain, do not mature through relationships and do not know from within what love, work, friendship or responsibility mean. Nor do they have a moral conscience, since they do not judge good and evil, grasp the ultimate meaning of situations, or bear responsibility for consequences. They may imitate language, behavior and analytical skills, or even simulate empathy and understanding, but they do not understand what they produce, for they lack the affective, relational and spiritual perspective through which human beings grow in wisdom. Even when these tools are described as capable of “learning,” their way of doing so is different from that of a human person. It is not the experience of those who allow themselves to be shaped by life and grow over time through choices, mistakes, forgiveness and fidelity. Rather, it is a form of statistical adaptation based on data and feedback, which can be very effective, but does not imply inner growth.

His eminence has a lot more on the ball there than most of our vaunted Tech Leaders.

Curiously, we both referred to the Tower of Babel. I wrote that it was an example of a mythic loss of humanity, a belief that our striving to glorify ourselves instead changed us into something eternally fractious. And yet, here we are, always worried that something will make us less human.

For the Pope, AI is potentially another God-defying tower:

conceived without reference to God, supported by a uniformity that eliminated diversity and that chose homogenization over communion. When a city is built on pride and the claim to self-sufficiency, communication breaks down, languages are confused and people no longer understand each other. The result is not unity, but dispersion. Babel thus reveals the limits of any effort that, however grandiose, arises from self-affirmation, sacrifices human dignity for efficiency and aspires to reach heaven without God’s blessing.

I thought those were quite different readings, to different ends.4 Then on my first day in Vienna I happened to see Bruegel’s famous painting of the Tower of Babel.

Perhaps like you, I’ve seen this image so many times that I thought I knew it well. In my case at least the real thing makes me stop and look at all the details. Which in the case of Bruegel are most of the story.

One of the things that makes his work so wonderful, and as it turned out, so visionary for the coming democratic centuries, was how much he depicted the reality of ordinary life even amid epic grandeur.5 Take a look at the foreground, where the king is being worshiped:

Four workers are on their knees to him, but most of the crew are just doing their stone-cutting jobs. There’s even someone descending the hill bearing one more heavy slab for the giant tower, circular and sheathed in marble, like the Colosseum in Rome. While that ancient structure was a ruin in Bruegel’s time, this tower is still going up, with everyone working hard at its construction.

Crucially, the technology is also state of the art:

Those are just the kind of machines used for building the great cathedrals of Europe. Human-powered cranes6 like the one on the ledge also lifted cargo on and off boats in the growing world of international trade.

But look behind the king: The base of the Tower is already flooding, and bits are falling away even as the work proceeds. They’ve built the tower on sand, and apparently at some kind of unsustainable angle on the docks side.

Everyone is hard at work anyway. In modern terms, it’s as if they’ve laboring under a ridiculous contract, and though they know their project is a disaster, what the hell, they’re getting paid. We’ve all seen jobs like this, whether it’s a cockeyed government deal (pick your favorite military folly) or a fundamentally insane business proposition (looking at you, Quibi.)

Bruegel is also great at taking a slightly comic direction to make a deeper point, the same one Leo nailed in his encyclical: You can build big, with state of the art technology, but if you don’t look at the larger context, the landscape and the foundation, you’ll crash in ruin.

When the cost is in stone, marble, or digital quick bites, it’s just another catastrophe. When it’s in human souls, the loss is much worse.

Share

Gotta run. Those dancing ponies don’t wait.

Historical myths such as the Tower of Babel or the gift of fire Prometheus brought to mankind illustrated people’s sense of their decline from a previous idyllic state to their present woe.1 Plato worried that writing would diminish our ability to remember, and thereby diminish us. Goethe drew on the historical figure Faust, an alchemist and proto-scientist, to create a literary figure who loses his soul in his quest for ultimate knowledge. In different ways Karl Marx, Sigmund Freud, and Charlie Chaplin riffed on how modern mechanistic society changed and diminished human beings.

And yet, here we are, human enough in spite of it all. Yet also different from the pre-Socratics, medieval alchemists, and grumpy 19th century Germans. Different, and yet still human. Which might lead one to respond to the “Tech will dehumanize us!” concerns with, “of course it will - it’ll make us into another version of humans, which isn’t the same as who we are today.”

Change has always been a problem. Over time our responses have changed, from the “don’t go there!” of Babel and Plato to the “here’s how to adjust to it” of Freud and Marx.

Today we’ve largely capitulated to the reality of change, and we discuss adjustment to AI in terms of aligning our interests, of co-evolving for mutual benefit.

At the symposium one man said humans and machines were hybridizing. People are now like centaurs; their humanity is sustained by machine intelligence. AI, meantime, is like a minotaur; its machine essence is increasingly taking on human characteristics. Soon the human centaur and the AI minotaur must integrate. A woman spoke of designing AI systems after biological systems of cooperation, aligning the needs of humans and machines.

Both evoked the creation of mitochondria 1.8 billion years ago, when an early cell engulfed a bacterium. Mitochondria created a more capable entity, eventually resulting in our rich and complex multicellular world.

It’s dramatic to align the present moment with the most consequential development in the history of life. While it’s a compelling metaphor, though, what does “co-evolution” mean? In cooperation, both sides have needs they seek to satisfy. But what are the needs of a bunch of software? Does a Customer Relationship Management software application “need” to file all the current hot prospects? Why and how will AI evolve needs?

For all the formidable brainpower going at this human/machine problem, I suspect we too often fall into the trap of taking the conversational interface of AI as evidence of its evolving intelligence. It talks to us, so it must think. Recently even the eminent Richard Dawkins, author of The Selfish Gene and much besides, fell into just such an error.

Initially it seems forgivable. The computer interfaces we’ve evolved over the past few decades have become increasingly captivating. Why else would people scroll on their screens for hours? No blame there.

Share

AI has been deliberately engineered over time to increase human engagement. That’s why almost every answer you get from it tells you how brilliant you are, or why the chatbot relies on the language of support and continuation. Flattery keeps you talking to it. This design comes out of a technical culture that largely sustained itself and grew with advertising, in which more time spent online was always better.2

We do not have conversations with AI in a human sense, with authentic emotion, challenge, a feeling of mutual enthusiasm or conflict that moves us. Genuine human interaction includes things that are difficult, that often challenge our personal vulnerabilities, and that ultimately make us more empathetic. Which is to say, more human, and more willing to be with other humans.

On the other hand, the social and mechanical tweaks that have made the interface so seductive, to a point where people would rather scroll on an increasingly AI-driven phone, or engage with a chatbot instead of another person, are now proving more dehumanizing, and possibly more of an existential threat than anything that could come from an autonomous foreign intelligence.

The Financial Times recently ran an article on the shocking global drop in fertility rates (article here, my visit to the childless world here.) It included a single chart that should bring us all up short: a graphic of what happened to the fertility rate in a diverse range of countries3 after smartphones became popular:

There is no co-evolution here. Just fewer humans, talking endlessly to themselves.

-Larry McMurtry, quoted in David Streitfeld’s Western Star

Novelists are liars. It’s nothing personal, just the job.1 They make up people and plots in the interests of exploring higher truths, like “here, then, is the human condition,” “this is the tragic psychology of desire, reverberating down the generations” or, more prosaically, “people love a really good yarn.” Sometimes they lie to reinforce cultural touchstones, like “the cowboy was a self-sufficient loner who tamed the wilderness, thereby wiping out the need for him. It’s noble and tragic.” There’s good money in that kind of reassurance, and considerably less in debunking popular illusions.

Share

The best writers are frequently professional betrayers, too, ransacking their personal history for material, revealing family secrets, even rewriting their own life stories to make them more interesting. Many others among us do these things too, but for smaller audiences (sometimes an audience of one, residing between our ears.) Novelists go big, though, and if artful enough in their betrayal they are forgiven by everyone, except their families.

I’ve been thinking about these eternal truths about lying and betrayal in new ways since reading my friend David Streitfeld’s literary biography of Larry McMurtry, Western Star.

McMurtry and Streitfeld were friends for many years before Streitfeld broached the idea of a biography. McMurtry put him off for many more years before equivocally okaying the idea. It’s fair to assume that by then McMurtry knew that Streitfeld, who I can attest is passionate about capturing facts and exposing hypocrisies, would explode many of McMurtry’s abundant personal fictions, even as he captured McMurtry’s long project of exploding the most cherished myths about Texas and the frontier. Which even from the jump shows McMurtry’s interesting allegiance to exploding personal myths, too.

To what end is tragically hard to say, for this book also depicts a time when quality fiction steered the culture, and writers competed fiercely to nail exactly what it meant to be alive in their times, debunking myths and steering people to a greater awareness of their own lives. This meant not simply reading Kesey, Mailer, Bellow, Roth, Oates, McMurtry, Lessing, Morrison, and many others, but internalizing what they had to tell us all, and following discussions of their work in the popular press.

Not long ago some 400 daily newspapers carried book reviews. That figure has fallen by 90%. All the major papers had separate sections devoted to books and ideas; the founder of Amazon.com recently closed the Books section at his Washington Post, leaving The New York Times the only high-profile paper with a notable standalone Books section.

Very quickly, life on the frontier of Literature has come to seem almost as remote as life on the frontier of the Old West. That type of (may the muses forgive me) thought leadership role for Literature is long gone. It surrendered first to movies. Now all manner of digital media has created a culture of hyper personalization that threatens any attempt at cultural consensus. Memoirs abound, meaning that betrayal of family history chugs along, but for the most part these are personal stories. We don’t look for them to explain our shared psychic realities.

I’ve written about feeling ghosted by the culture, the hazards of hyper personalization, even the new personal soundtracks in our heads. I have much sympathy for Streitfeld’s elegaic description of this bygone era.2 Even so, the most interesting thing is Streitfeld’s sustained passion for naming places in that halcyon literary past where McMurtry and others don’t have their stories straight, where they have their facts wrong, where they outright lie. As I said, McMurtry knew what he was getting when he let Streitfeld in, which is to say some part of him must have wanted his own myths demolished.

Share

Streitfeld debunks false memories the most when writing about the 1950s and 1960s, when books mattered most. McMurtry says his childhood home was bookless, and no one read to him. His father remembers otherwise. McMurtry can’t really decide if he grew up on a place called Idiot Ridge. His first literary agent was not, as McMurtry wrote, a once rich person from a castle on the Danube. He misstated her age by a decade. He says he didn’t hang out with the LSD crowd of his friend Ken Kesey, but others say he did.3 Another friend spins stories about himself, Streitfeld writes, “even less tethered to reality than Larry’s.”

Once film takes over the culture, Streitfeld becomes more forgiving of myth-making. During a long section on the filming of McMurtry’s The Last Picture Show, a fictionalized take on his hometown of Archer City, Texas, there are no fact checks or corrections. Perhaps all of those Hollywood people tell their stories straight, let the chips fall where they may…oh wait, they don’t. They are from an industry best described by the title of producer Lynda Obst’s memoir, “‘Hello,’ He Lied.” It is customary in Hollywood to bullshit even more than you are aware of, since it’s a professional tic. Maybe it doesn’t matter in the same way that lying did when the world was different, and the fictional truth-tellers ruled.

After “The Last Picture Show” was a hit, McMurtry is headed to becoming largely “Lonesome Dove, Inc.”, a highly-paid storyteller who becomes wealthy off his long novel and4 miniseries about a couple of trail drivers past their prime, out for the validation of a last roundup. As McMurtry accedes to the sentiment and payoffs of Hollywood and television, Streitfeld interrogates people’s stories less frequently and finds them wanting only a handful of times, compared to the scores of “gotchas” during the literature-dominant earlier time.

His closest fact-checking the later years is for McMurtry’s work on “Brokeback Mountain,” the writer’s own last roundup, when he worked the art as he used to (and scored the big bucks and the important awards, whose ceremonies he is pleased to attend.) Even there, however, Streitfeld has a new tone of forgiveness. His final fact check, concerning the mistakes on a sign in Archer City honoring McMurtry, ends with, “it’s a start.”

Like many people who live by their wits, McMurtry presents interesting paradoxes. While he aimed for greatness on the order of a Dickens or a Fitzgerald, both of whom wrote themselves out of obscure circumstances to cosmopolitan triumph, McMurtry perversely reveled in his supposed marginality, and the obscurity of the Texas Panhandle. In fact, he was captivated by the land, or more precisely the land’s unique effect; his characters are entirely the product of their landscape and their psychology is resolutely place-specific. People in his books are consumed by memory, whether it’s a Vegas showgirl dwelling on the times when her body was best, or somebody recalling the candy bar left to bake in the Texas heat, or any number of frontier types lost in a time when they had more purpose. Yet everyone moves forward, seeking a new place to bring their memories.

On a personal level, McMurtry fled his hometown, a cultural wasteland, only to return there and build a bookstore with 600,000 volumes. Inflicting that much culture on your culturally arid roots must go down as some of the most performative capitalism ever attempted. The locals didn’t care one way or another, but he persisted. A person of recondite tastes, it’s doubtful he saw himself as nothing but the product of his pinched landscape and his grudging frontier ancestors, but he avidly chose not to escape, too. He even arranged the lie that he died in Texas.

Yet for all that, he agreed to have his own myth fact-checked, almost certainly knowing how much wouldn’t stand up. Much like the myths of Texas itself, which he tirelessly tried to explode, but ultimately profited from better than any other writer. And which, ultimately, he chose not to escape. There were in the end four books each based on “The Last Picture Show” and “Lonesome Dove,” fiction based on the landscapes of his personal and family past.

A paradox joins two things that contradict each other, yet their combination nevertheless reveals a truth. You could say the same of a novelist, that professional liar, signing up a biographer, his friend, to ferret out his lies, so people will know him for real as the fabulist he was.

4

To me, tedious and laughable. Sorry America, but that snake scene is straight out of Toho Studios.

Not that there’s anything wrong with that.

Johnny Yuma was a rebel

He roamed through the west

Did Johnny Yuma, the rebel

He wandered alone.

I’m not interested in Johnny Yuma, a short-lived show with these lyrics in its theme song. I only wanted coffee. And yet, came a familiar imaginary audio:

He got fightin’ mad, this rebel lad

He packed no star as he wandered far

Where the only law was a hook and a draw

The rebel, Johnny Yuma.

Johnny Cash recorded the theme song to the show, and it was on some cowboy record I played enough that it still shows up in my head every now and then. It is among my oldest1 extant earworms.

I have many more ancient songs and advertising jingles in my mind, and they pop up periodically in no discernible order. I might be discussing computer network infrastructure with a client, and in the background “Build Me Up, Buttercup” is rocking away. Reminiscing with my wife about our children’s earliest days, Ace of Bass’ “The Sign,” Pharrell’s “Happy,” or maybe even “Little Willy,” by The Sweet, will intrude. Not a year goes by without an internal revival of any number of horrifying jingles for Winston cigarettes.

I saw the sign
And it opened up my eyes, I saw the sign

(Oh crap. I really wish I hadn’t mentioned that one. Now I’ll be thinking of “The Sign” while I’m trying to write. Unless I’m really unlucky and start hear “Summer of ‘69,” or worse, “New Slang.” It’s easier to shake off a tattoo than to unhear that one.)

I don’t particularly like any of these songs and ads; they are just a small part of my preposterous mental inventory. You’ve got them too. I once heard an eminent novelist of my generation confess during a talk that his prime thought just then was an earworm of Carl Douglas’ “Kung Fu Fighting.” I know a successful VC who can joyfully break into a bygone jingle praising a supermarket’s beets.

Earworms are music you don’t try to remember, or maybe even want in your head, that persist because of overexposure. The word “earworm” first appeared in English in 1978. It comes from the German Ohrwurm, which means “earworm.” They hatched this word in the nineteen-twenties, at the dawn of recorded and broadcast media.

It riffs on the superstition that earwigs burrow through human ear canals to lay eggs that hatch and consume brains. While not true, it’s an excellent analogy for an aural infection via radio, jukeboxes, the neighbor’s record player, shopping music, or any other ambient onslaught. The Germans figured out the anxiety of the technology early on, once again displaying a skill that extends to figuring out the anxiety of almost everything.

Share

Yet now this ear beast seems endangered. My inventory is aging fast, and I bet yours is too. Instead of earworms, I believe future generations are cultivating a new kind of unplanned companion.

While recovering from “Johnny Yuma,” I realized that I might silently hear 2011’s “Somebody I Used to Know,” or even “Shut Up And Dance,” from 2016. But I have basically had no unwelcome audio visitors dating from the years since then. What happened?

“You aged out, ya spavined dimbrain” is one possible reason. Arguable, but I checked with a couple of younger adults. They report a similar lack of earworm recency.

I think consumer tech is making the earworm obsolete. In 2014 the first wireless earbuds were crowdfunded. In September 2016 Apple co-opted that idea with the Airpod. Music, already becoming personalized with streaming services, rapidly became a more internal experience. It was no longer what commercial interests sprayed in every direction, the way they had

No, you didn’t have to stooooooooop so low

Have your friends collect your records

and then change your number

I guess that I don’t need that tho

Now you’re just somebody that I used to know.

(Man, that one’s savage.)

Check for yourself: Here are the billboard hits from 2014 and 2017. You probably recall a lot of them, voluntarily or not, even if you don’t like them. Some will even hatch in your mind like a ravenous nymph and start chewing.

Now try it with the songs of 2020 and 2024. Anybody unwillingly got “Blinding Lights” in the brain? I thought not. Today you are more likely to discover songs accidentally from Tik Tok, or some other source where the algorithms aim to please your taste. You probably don’t get slammed with “Johnny Yuma” moments by the ones you never liked. Which is the maddening characteristic of the earworm.

good, like a cigarette should.

(Damn! Just…stop!)

This is an interesting example of one technology, recorded music and broadcast, creating a mental state, and a successor technology, wireless earbuds and personalization, extinguishing it. But there’s more going on, since technology amplifies pre-existing parts of ourselves.

The earworm is technology leveraging our mental bias towards easy memorization. This goes back to the furthest reaches of the oral tradition, and the “music” in the metrics of The Illiad. More recently, Mark Twain wrote a story about the madness induced by a stupid rhyme. Evelyn Waugh wrote about being the last Oxbridge generation where the lads got drunk and recited memorized poetry by the armload, before records and radios and radio ads showed up and made the practice of memorization even easier.

Given enough time, people may treasure their earworms as markers of the past, even if at one time they were an annoyance. I once attended a Christmas dinner where the older generation sang the radio advertising jingles of New York hotels, 40 years after these were in the ether. Let me know if you want to hear my version of

plenty peanut stuff in the middle

that the camel praises in an ad for Gaucho candy bars.

The recession of the earworm is only part of the story. We’re at the start of a new and unknown tech-based relationship with sound. People now wear their earbuds for up to 10 hours a day, talking on the phone, listening to podcasts and music, or canceling out the world’s noise (which is its own kind of audio.) Given what I’ve learned of tech and business, if this turns out to be bad for hearing it’s more likely that earbuds will include augmentation for those turning deaf, than it is we’ll abandon the shiny product. The things are here to stay.2

Share

Behaviors are already changing to accommodate the new reality. When I see someone purposefully wearing earbuds I know that I have to catch their attention before speaking to them, a turn from how casual conversation worked just a couple of years ago. Sometimes I even weigh whether what I’ve got to say is better than “The Rest Is History” podcast.3 As it happens, a story in The Wall Street Journal today describes the new office irritation, earbudded people audibly whispering to their AI while at the desk.

This is just the start of a greater transformation. Last month researchers at The Paul Allen School showed off what they called Vuebuds, a small camera in an earbud connected to an AI. It can identify objects on command. Increasingly we will see more chatbots implanted in earbuds, so wearers can have a conversation with a make-believe intelligence whenever they want, or even review with it things that happened earlier in the day.

We lose the earworm, and maybe gain more of what I call “the eyemoth,” or random visual snippets that recur in the eye of memory. In the movie “Citizen Kane” the aging Mr. Bernstein describes the phenomenon:

“One day back in 1896 I was crossing over to Jersey on the ferry, and as we pulled out there was another ferry pulling in. And on it there was a girl waiting to get off. A white dress she had on, and she was carrying a white parasol. I only saw her for one second. She didn’t see me at all. But I’ll bet a month hasn’t gone by since that I haven’t thought of that girl.”

I already have many eyemoths of my own: a red ribbon on a late night bus in Seattle, decades ago, still pops into my head. So does Captain Hook’s crocodile tormentor hiding among the seacoast rocks of the wallpaper in my childhood bedroom. A stilt dancer’s back flip in Singapore, but little else of his performance. A train conductor’s hat on a padded seat in Pennsylvania. A cliff face in the Sierras. They probably mean about as much as an earworm, but they stick in my visual memory, more welcome than their aural cousins.

AI buds might enable eyemoths on command. Or perhaps they’ll make eyemoths vanish like Waugh’s reciters, as life becomes even more of a movie, in entirely new ways.

A bad thing? A good one? Really, just a thing. It’s odd to prefer one consciousness over another, particularly if one is preferred only because it’s familiar. History shows we still remain human despite all the figurative and literal noise.

Still, I’ll miss earworms. Except for “The Summer of ‘69,” which I’m resigned to hearing on my deathbed.

1

This banger, a milestone in the history of technophilia. Phones are cool! Red records are cool! Everything is cool! And yes, I can sing it.

The man was Iceland’s Víkingur Ólafsson, performing a solo concert as the 2025-26 Artist In Residence at Berkeley’s Cal Performances. He played what he calls “Opus 109,” a project which consists of a Bach prelude, a Beethoven sonata, a Bach partita, a Schubert sonata, and the final Beethoven sonata (the titular opus.) He played it without a break between movements, or the standard applause for individual pieces. This new work illuminates a creative interaction spanning more than a century. He plays it entirely from memory.

I should state here that I am a musical idiot, and know little about the history of music or the technical attributes of composition. Yet I can witness beauty and admire excellence. While many of the nuances of Opus 109 were as lost on me as they may be on you,1 anyone would be astonished by a performer flawlessly executing his extended meditation on influence and relationships, inspiration and response, and recapitulating the past in order to create something new. All played in the key of E, making it easy to hear as a single journey, with the silence between each note equally critical to the outcome.

Afterward Ólafsson spoke about the recent death of Michael Tilson Thomas, musical director of the San Francisco Symphony and a giant in the world of classical music. “He did so much for the musical community,” said Ólafsson, “we don’t know what it means yet.”

That seeming aside was for me another revelation of how he saw a single human influence on the world - long and deep, with an effect not fully realized simply by the ending of a life. Like “Opus 109” itself, his comment was a statement about how our works exist both in the moment, but also in the context of our vanished past and our invisible, potential future. Living in an age of omni-directional surveillance, where everything is supposedly stored for review by our AI prediction machines, this sensibility borders on rebellion.

Share

Those words landed with me perhaps because earlier that day I’d read2 about something called Historical Language Models. Also called Vintage Language Models, or VLMs, these ingenious creations use only text from specific historical periods. Thus, the idea runs, you can query the model and get an answer that reflects the knowledge of that particular time, uncontaminated by later findings.

On my third try with it I found a pretty bad error,3 so clearly VLMs have a way to go. Improved, though, they will be interesting tools for investigation and inquiry. Particularly if optical character recognition improves in cost and capability, and more non-English sources enter their corpus.

Still, I doubt there’s anything remotely accurate in the inventors’ claim that “while we don’t have time machines yet, we can simulate this experience” of talking to someone in the past via a VLM. Or in Benjamin Breen’s intriguing idea that this is a hypothetical first step in creating a collective consciousness of past humanity.

Breen moves the question away from the technology itself and back to authentic human consciousness as a thing in time, which Ólafsson explored for us in past and present, and which his aside about Michael Tilson Thomas extended to the future.

Plainly, if there is something like a collective consciousness within any past society, it’s unlikely to exist solely within captured and analyzed text. In 1900 about 20% of the globe was literate, and in 1800 the percentage was about half that.4 Go further back, and it’s tiny. Whatever was printed was largely for people with the wealth and leisure to read. Their interests may not equal people who struggled for bread and told folktales.

Share

Adding more foreign-language texts will make for a richer corpus, but will also introduce new questions: How does the contents of, say, an Armenian newspaper in 1890 affect the consciousness of, say, a Japanese businessman of that time? The answer starts with, “Which businessman?” While there may be a collective consciousness, it still exists within individuals.

However much old text we log, it will not be coherent. Our civilization is now passionate about counting and storing everything, but what we retrieve from the past is haphazard. Just last year we learned something about Shakespeare’s relationship with his wife, thanks to a 1608 letter that survived because it was used to make the binding of a later book.5

The 14th century poet Francois Villon’s wrote, “Where are the snows of yesteryear?” to describe our vanished living past. All the LLMs in the world will not change the reality of that loss. We can guess something about what’s in other people’s minds here in the present, often with success, and with even more sympathy we can approach a past consciousness, but only with great humility. Our approach will not be dramatically improved with a language model.

Share

FWIW his third encore was recent. And equally haunting.

Like Ólafsson, we can think about influence, and how a creation in a later time both reflects the influence of the past upon us, and changes something of the character of that past work. It’s given a new meaning by the later sympathy. And in addition, human creation is an unknown gift to the future. Time is made out of the infinitely present moment, infinitely receding, as it creates the unseen future.

Even with the most powerful technology the present moment is irredeemable. That is, many would argue, the wellspring of beauty.

Then there are the physical dimensions of memory and thinking. Ólafsson developed the idea for Opus 109 after performing Bach’s Goldberg Variations over 100 times in just a couple of years. That repetition, as much as the sound, led him to hear their influence on Beethoven, a composer that he says his mother taught him must be played with a feeling in your gut. To this mix of sound and feeling, add that the pianist possesses synaesthesia, or the mingling of senses, so that he describes music in certain colors, the way you’d say a blade of grass is green. The Schubert piece was performed as a reflection on how Schubert may have been influenced by the preceding Beethoven sonata. More boundary-crossing influence.

Thinking of the highest sort can be physical as well as mental, emotional as much as reasoned, adrift in time and focused on the endless Now. LLMs are great repositories and sorters and relationship-makers, but they are human-made tools, like a monkey wrench, but with a pretense of greater meaning. They are never in time like Vikingur, or you.

Unintended Consequence of the Week: On Mt. Kenya, sheep farts now smell like roses.

According to the Financial Times, Kenya’s flower-growing industry is in a crisis brought on by the U.S.-Iran conflict. Kenya is the world’s fourth-largest flower growing economy, and its prime export destinations were Tehran, Dubai, Riyadh, and Doha. With fewer Mideast hotel guests and celebrations, however, and less-reliable air cargo services because of our global jet fuel crisis, its flower sales are down 50%. Since you can’t store roses, they are fed to the sheep. A strange footnote to the ongoing global impoverishment caused by this conflict.

Share

3

See the screenshot below.

In fact, Stevens was born in Reading, PA. That’s possibly an acceptable error, since in 1930 he was barely known. The real anachronism, in this model that supposedly stops at 1930, is that his first volume of poetry, "Harmonium," was published in 1923. He did not publish his next volume, "Ideas of Order," until 1936. Somehow this “time machine” is traveling into the future, too.

On Thursday I dropped my collected works into an AI and was told that they add up to 199,000 words, a suspiciously round number.1 Add a few lists of whale arcana, and this would amount to the length of Moby Dick. Should some readers now wonder if following me has been a good use of their time, let me say - Ahab dies, taking almost everybody on the Pequod with him. No idea what happens to the whale.2 Now you’ve got that covered, too.

Friends who write actual books assure me that this work is nothing like the daunting anguish they experience, bleeding away at the keyboard to further The Human Project. Fair enough; I view this weekly offering more like letters to friends and like-minded strangers. To the extent they reach you that way and, like a good letter, sometimes stir your mind or heart, I’m happy in the work.

Every 50 entries, I note a few personal favorites from the recent group. Here are links to lookbacks 50 and 100. As usual, the essays fall into the categories of Technology, Culture, and Personal History. As the Shop’s overall theme is “the search for meaning,” those categories are proxies for, respectively, ways we shape the world, ways we shape each other, and ways we come to terms with the people and things that shape us.3 It’s odd that understanding the self, supposedly the only thing anyone can truly know, is the most elusive. But, as they say, here we are.

Here is my list for essays 101-149:

Personal History

1. Passage Of My Mother, In parts One, Two, and Three. A story about primal grief in rising China, ending with a visit from an accidental Bhoddisatva.

2. Dad Testifies. My father describes to Congress a crisis of meaning in American life. Those concerns collapsed in decades of material excess, until that con ended badly. Dad was right.

3. Nino Shows His Painting. Reconnecting with an exuberant childhood hero, and discovering his dark and hateful side.

4. The Education Of Peter Kline. A co-written tale of Britain’s class system, set at the mad end of hippie India. We shall not see its like again.

5. Confessions of A Tooth Fairy. On being up late, creating fictional memories to be recalled decades later.

Share

Culture

1. Our 14% In Their Perilous Court. A personal favorite. Knowledge workers are the viziers, knights and scribes to our modern kings, and we feed on fear. The longest piece on The Shop, one of the most read.

2. New Tales Of Old Tools, In parts One, Two, and Three. Humans made the same tools and art for millennia. Then we evolved perpetual change. Why? Forget cultural Darwinism; our brains learned how to be dissatisfied.

3. Two Ghosts And A Chance. A halting goodbye to the literary culture I was raised in, which became over the last couple of decades as irrelevant as a dead civilization’s god.

4. Human Moments In Hell. Brutal truths on the American frontier, from the threadwork on a straitjacket in Oregon.

5. Hamlet, Post Mortem. The prince dies, and in 70 final lines his story of personal torment is distorted for PR and political spin. The eternal tragedy.

Share

Technology

1. Horseless Carriages, Digital Paint, AI. The design problem of naming and explaining new inventions. Plus, why the main AI logo is called “the butthole.”

2. Two Mirrors Of The World. How the real threat from AI is the perfection of hyper-personalization, and its role in the death of civic institutions.

3. A Dark Data Center Crisis. What a financial hiccup would mean for expensive, fast-aging semiconductors. The most-read story of the past 50.

4. Some Unintended Consequences Of AI. All computer architectures change work and society in ways impossible to foresee. This time it’s all about trust.

5. Kale Salad With Ash. An exuberant raconteur foresees extraordinary monsters emerging from the price-collapse democratization of AI and biotech.

Looking over this list, I’m pleased with the range, grateful for all the feedback I’ve received, and astonished at the momentous times we live in, as we gain and lose many structures of our world. To the extent I’ve been thought-provoking, entertaining, or otherwise useful to my readers, I’ve done my job.

My weekly letters to the world provide a small part of my income as a professional writer.4 After the Substack and Stripe fees, the take home isn’t dreadful, but it’s nothing anyone could live on. But the effect of this Substack on my work as a whole is enormous. Here I get to explore larger ideas which animate the particular work I do with my clients. Conversely, I bring to these letters ideas that come from working with people in tech and seeing what they are building. If the current wisdom is to own a niche, this kind of wide-ranging exploration is mine.

Letters to friends, of course, should have no niche.

We can’t help ourselves. We see the new thing, it looks useful, we’re almost always going to employ it, grow it, take it to the limit. Even if its use will utterly remake our world, including all kinds of unforeseen consequences. Maybe we do this for the sake of the unforeseen consequences. Nature has made us creatures who crave novelty like no other species. Even when the search for novelty is against our interest, as many of us can attest to, over 52 weekends a year.

This is one of several reasons why we’re going to keep on developing Artificial Intelligence, even while reasonable people think the current speed and scope of AI development could frustrate, enslave, and/or kill us all. They call for a big pause, until we have a better sense of what we’re getting into, or a global pro-human policy that keeps humans in charge, diffuses power, doesn’t fool or enfeeble people, and makes companies accountable.

With no snark, I note that successfully meeting the needs of understanding what we’re getting into, elevating humanity, or holding companies to account has little space in the historical record. That said, godspeed, pro-human pausers.

Share

As we consider their challenge, though, it’s useful to look at a few examples of people successfully abandoning a burgeoning new technology. Why was it done? How did it work? And we should also speculate about the possible consequences of that refusal. Choosing to do nothing is, after all, its own type of action

Boats, guns, and llamas

The best example of closing off a new technology is the Ming Dynasty’s destruction of its deep-water navy. Around 1430 the eunuch admiral Zheng He began the first of seven “treasure ship” voyages to establish trade with Southeast Asia, India, and East Africa. Following a power struggle between the Ming court’s eunuch faction and its Confucian scholar faction, China burned those ships, destroyed records of their construction, and made oceangoing voyages illegal.

Zheng’s largest ships were perhaps 300 feet long, and there were at least 60 vessels in his fleet. Compare that with the craft of Portuguese explorer Vasco da Gama’s 1497 trip to India, which slapped into high gear Europe’s Age of Discovery. Da Gama’s longest ship was about 85 feet, and his “fleet” was four ships, two of which made it home.

Imagine how different today’s world would look if the Europeans had encountered an established, sophisticated trade network using much better ships. The precursor to Western Imperialism would be a very different matter. Today, ironically, the West worries about controlling the new technology, AI, while China seems hellbent on its development.

A century after da Gama, under the Tokugawa Shogunate Japan mostly closed itself to the West. The rulers kept up with some developments in Western learning, but adopted little of it. In particular, they sharply curtailed the manufacture and possession of firearms. Not coincidentally, in 1616 the ruling class codified the Bushido samurai ethos, elevating swordsmanship as a spiritual practice.

Japan remained closed to outsiders for more than two centuries. When Commodore Matthew Perry forced the country’s opening in 1853, his vessels could fire explosive shells from far offshore. The samurai must have looked woefully at their few 16th century-style matchlock armaments.

Maybe if they’d kept current, Japan could have met the American incursion with its own violence against foreign shipping in nearby waters, much like Iran is now doing. On the other hand, the sudden flood of Japanese aesthetics into the West after a long hiatus played a critical role in the development of Impressionism.

These are two examples of blocking technological development to preserve the status quo. Confucians in Ming China saw that foreign trade imported new ideas and new power, in opposition to their fundamentally conservative Confucian outlook. Japanese elites elevated the violence of swordsmanship, which takes years of practice. Guns were a weapon any hick farmer could pick up and use to kill a samurai.

European powers tried to do the same thing. For example the ribbon-making loom was suppressed in Germany, and the stocking-making machine and the sawmill were curbed in England, all to preserve the livelihood of specialized workers. Geography did not work in their favor, though, the way it did for Japan and China; in Europe the tech and the goods it produced simply took hold nearby.

Share

Sometimes a revolutionary technology is beyond the purposes or imagination of the ruling powers. The Inca had the wheel, but didn’t use it for much besides children’s toys. Agriculture and transportation were geared to their vertical society in the Andes, with the uncooperative llama as the significant beast of burden.

Perhaps the lesson is we are really slaves to our geography.

More tech, more problems

Or at least, we were. Now big developments in technology are global. Before the Internet came on the scene, the great bans and discontinuations in modern times centered mostly on nuclear power. The world has fought just one nuclear war, and once another side had the weapons, governments started looking at ways to control and limit the technology (sadly, elimination has never had a chance.)

Russia does appear to have a “Dead Hand” doomsday machine, but we’ve also had test-ban treaties. As best we can tell, no one has built a hydrogen bomb salted with cobalt, which could render places uninhabitable for centuries. If nukes eventually reach places like North Korea, well, the slower the better.1

Looking at another existentially threatening technology, the 1975 Asilomar Conference on Recombinant DNA delivered voluntary guidelines on ethical approaches to working with genetic clones. Those guidelines probably didn’t slow down the development of genetically-modified organisms, though, or of state-controlled bioweapons labs,2 but even developing a framework is a good start. It chimes with what people are now trying to do with AI.

The big difference between Asilomar and AI is that Asilomar happened well before the technology was broadly out of the lab and effective. AI is already in schools, corporations, and much of the technology you’ll encounter today. That makes its control exceptionally difficult.

One “out of control” context is an AI job apocalypse, which would represent a threat to the social order beyond anything caused by the 2008 financial collapse, or other economic depressions. Those caused financial losses, or higher unemployment; if AI takes out massive numbers of jobs it will curb what amounts to identity itself, since for the most part we are what we do.

Confucians were similarly worried about their means of identity fading away under widespread foreign influence. And to this day, people and groups in Japan navel gaze about whether the country has lots its Bushido ethos. The amount of individual empowerment with sophisticated AI is both a boon and a threat, in some ways akin to the Tokugawa strangulation of firearms.

Banning things that don’t exist: Just say now

Today we’re sometimes like the early industrial crowd trying to ban a particular technology application, so that “no sawmills” is like “no AI on the battlefield,” or “AI in journalism, only under the strictest and most controlled circumstances.” It’s probably going to pop up there in some form, and the ban will then be adjusted.

Sometimes we’re like Asilomar, talking about curbs on something that’s possible, but isn’t quite here yet. But the the AI pause/ban situation is different in that, unlike genetic work, AI is remarkably easy to develop and share, so it’s harder to control. In that sense, AI is closer to stocking looms than genetic engineering.

Additionally, there’s already a lot of money at stake in keeping things going, testing all the limits. That makes AI development a bit like leaded gasoline, carbon-emitting energy production, or smoking: People knew these things were hazards for decades before they were curbed, but there was a great deal of money in pretending otherwise.

Share

Perhaps the unique dimension of AI in the annals of turning off a technology is this: When we talk about banning some types of AI, like conscious and malevolent machines that act against our interest (by, you know, killing everybody) we are talking about banning technologies that not only show no sign of existing, but that we still don’t really understand how to build.

A total folly that is utterly worth our time. A truly unprecedented event, or a mordant comment on studying history?

IF YOU HAVE READ THIS FAR: My thanks, and please spend a little time at “The American Bystander,” America’s last great humor magazine, edited by the estimable Michael Gerber. I’ve written for them, and also support them with money. You should think about a donation, the stuff is great.

About forty years after the printing press became widespread, the abbot of a monastery argued that if his monks were freed of tedious hours and months of writing out manuscripts they’d have lots more free time to be corrupted by Satan. This complaint appeared in a printed book, which is a good reminder that, like it or not, we tend to work within the dominant technology.

Two decades later, a monk in Rome became fed up with how much the Pope was spending on artistic glories. He made good use of the new technology to print up a lot of tracts about this and other churchy ills, swamping the Catholic power with his abundant content and setting in motion the Protestant Reformation. This theological revolution killed millions of people. It also transformed our sense of the individual in the world, which over the years brought about the scientific method and democracy. Unintended consequences are a natural product of time.

Today we’re making guesses about the unintended consequences of another technology, the computer systems that power Artificial Intelligence. To do so, let’s set aside unhelpful metaphysics, such as questions about what Satan will do next, or whether scraping the Internet and searching for patterns (what the current version of AI does) will result in consciousness. There’s zero evidence that machines are conscious, even if the head of Anthropic keeps pushing his tempting but unhelpful “we don’t really know” angle. Conscious or not (again, they’re not) it doesn’t seem to affect what most people do with these computers.1

We can, however, focus on how AI will change both the computer industry and the way we’ll all think and work.

Share

There are useful guideposts for figuring this out in the patterns of previous computing transitions. The trillion-dollar spend on new AI data centers is just the latest and biggest part of a larger development of the so-called “architecture” of computing, which is to say all the chips, networks, servers, mainframes, terminals and everything else that goes into a computing system.

If you’re over 20, the repeated architectural changes in computing have already upended your life. Here’s how, in four over-simplified2 steps:

1. Until the Eighties IBM’s mainframe computers dominated computer tech. Then came an architectural shift to computer servers, networked to personal computers (so-called “clients” of the more powerful servers.) The architecture moved from a centralized model of mainframes to a distribution of computing power. Computing spread like crazy.

One unintended consequence was that networks, which had existed inside mainframes as a way of passing data and instructions from one place to another, became a huge industry on their own. Cisco got big, and IBM never saw it coming.

The other effect, the social one, was the utter decimation of middle management, thanks to electronic messaging (aka email.) It got the mid-90s management theory name of “Re-engineering the Corporation,” but really it amounted to one company department talking to another, without a human middleman conversant in the languages and cultures of both places. Top management got rich by firing thousands of mid-level lifers, and after some mutual incomprehension departments communicated directly.

2. Then the Client/Server model was re-architected into the Internet. Thanks largely to the world wide web, public networks exploded, and the browser became a client that could go anywhere. The widespread use of open source software became a central part of computing architecture. More decentralization, more people and money in the industry.

One consequence of using unbranded commodity servers and the open source Linux operating system was the decimation of server companies like Sun Microsystems, which for many years had been the big winners in Internet architecture. Google and others realized they could ship not-yet-perfect software and fix problems as they emerged. Internet as a computing platform! Move fast and break things!

On the work front, the Internet became a much bigger way to sell and ship products than anyone expected. Today I can have a new transmission for my Subaru delivered to my house tomorrow, using the low-carbon route. One unforeseen giant leap for the culture of instant gratification.

3. Smartphones were a huge technical challenge, but they have proven so addictive that it has been worth it for both businesses and consumers. A big unintended consequence was the dominance of just two companies, Apple and Google, neither of whom had any previous experience in mobile phones. Nokia and Motorola never saw it coming.

The big two ran the operating systems, the app stores drawing millions of developers, and the data collection. All that online data created today’s unexpected social consequences of a global surveillance society, in everything from targeted advertising and home doorbells, to control of the public, worldwide. The “compute from anywhere” promise of mobile architecture turned out to be “work and be seen from anywhere.”

Share

4. Cloud computing, the last big architectural transition, is why we have AI today. Architecturally it was a revolution in scale, with million-server data centers that use software-based infrastructure, microservices that break big jobs into more manageable bits, and failure-tolerant systems beyond anything Google had deployed earlier. Couple those things with all that computing power and with all of the data people posted to the Internet and created on their phones, and commercial AI became plausible.

The unintended technology consequences are still being worked out, but they include centralized outages that can knock out dozens of businesses at once, and terrifying security threats. On the plus side, cloud computing made work from home and video meetings possible, saving millions of lives in the first tech-era global pandemic.

Broadly speaking, tech’s unintended consequences break down into technical biases to speed and economy that incumbents don’t see coming, and social changes that often imitate what the new architecture is producing. The Jesuit thinker John Culkin was getting at this when he wrote that “we shape our tools and thereafter our tools shape us.” I doubt he was thinking that the workplace would imitate the computing architecture, but that development makes his observation even more profound than he knew.

Look at the workplace over the lives of many readers and their parents. First, rows of desks and stacks of paper bearing the results of mainframe punch cards, with lots of managers behind closed doors. Then cubicles and email, with the middle managers gone and everybody else picking up the slack. Next, producing and consuming endless amounts of information on web sites, with attention suddenly the scarce commodity, and the start of working late into the night from a connected home. Mobile computing made everywhere the workplace.

Now we are disaggregated piece-part workers in the cloud, with gig work, influencing, side hustles, and distributed teams on video calls. Work is often a series of microservices. For many people the contemporary world is similarly fragmented, delivering inordinate power and subservience, put together haphazardly, confusing us with the number of calls on our attention, moving fast and breaking things.

Into all this comes the AI architecture. On the tech side, right now everyone is buying Nvidia chips, to a point where that company’s CEO recently posted a model of AI computing that began with electric utilities and went on to earth moving equipment, before spending a little time on the actual computer parts of the architecture. Everything was in service of the mighty chip. With a few tweaks, one could make a similar model of the McDonald’s hamburger world, starting with growing hay for cows before including electric utilities, earth movers, and computers, all in service of the patty.

No offense to Nvidia, but IBM, Sun, and Nokia all sounded like this in their day. Something cheaper that we don’t see coming may be headed our way.

Certainly, there are already a lot of open source projects aimed at getting more out of the computing, networking, and data storage demands of the chips, probably by depositing a lot more computing intelligence at each step of the process. It will be interesting to see whether some big change happens in the chip world before that trillion-dollar data center bet, which will take years to complete, is fulfilled.

As always, the more interesting unintended consequence to consider is what the new architecture will do to the way we work, live, and think.

One big initial change of AI computing is the interface. People talk to the chatbot, and it talks back. It feels more like conversation than programming, and very quickly even smart people start giving the machine a personality.

That may ease off once the “it’s conscious!” hysteria goes away. But when you couple the conversational confusion with the reality that AI is essentially statistics-based (what’s called “probabilistic”) computing, of a depth even its designers can’t quite replicate, you start to see a bigger problem. Probabilistic computation is why you can talk with AI, why those conversations sometimes go off the rails, and why AI hallucinates. Both technically and socially, there’s an emerging crisis of trust here, abetted by things like AI’s deepfake images, which create mistrust in the real world.

Share

When Martin Luther set in train the Protestant Reformation, there had already been a century of rising individualism. Merchant wealth made it possible for rich traders to strike medallions of themselves. Painters like Da Vinci and Michelangelo were collected not just for their technical excellence, but so people could boast of having that name on their walls. Writers as diverse as Machiavelli and Castiglione3 talked about the world as a kind of theater in which the individual had to figure out and act their part.

Seen against that context, our postmodern suspicions about truth and our pervasive culture of irony seem like feeders to a colossal crisis of trust brought on by AI.

Which, almost certainly, will be met with a tech fix to the architecture of machines and society.

I am The Tooth Fairy.

Under that shimmering mantle, I have managed a telepathic parrot, a multitude of ice cream-fueled unicorns, and a dance band staffed by artistic snow monkeys. I have traveled in a globe-spanning zeppelin command center, and built an enchanted castle, the resting place for billions of children’s teeth, not far from the storied headquarters of Mr. Santa Claus. He works but one night a year; I never ceased, owing to my steady and abundant global customer base. Unlike humans, most of the critters I managed were obliging and required minimal oversight. Real estate, though, that’s a hassle even when it’s make-believe. Ask anyone possessing magical wealth. Or take my word for it, in this and all other matters.

Critically, The Tooth Fairy also had a marker pen of sparkly golden ink, which for several years I wielded for my audience of two.

I was not, like some, a Tooth Fairy manqué, casually swapping out a lost tooth for the usual quarter and calling it a night. Not that there’s anything wrong with that: The kid gets up in the morning two bits to the good, everybody is happy. Fine, if fleeting. When my initial son lost his first tooth, though, I yearned for a Tooth Fairy backstory, and some understanding of the consequences for the magical sprite’s peculiar nocturnal fetching.

I began with a simple thank you note, informing him that his lower front tooth was now part of a Castle of Eternal Smiles, a live-in grin of joy, made from the teeth of children the world over. Specifically, it capped a glittering rampart.

Admittedly, a sprawling construction made entirely of teeth sounds rather primal, even a touch grisly, to adult ears. For the most part though, kids possessing gift money are too gleefully ADD to dwell long on such things. Besides, they’ve heard worse - read your Brothers Grimm.

The letter was a big hit the next day, particularly for its loopy golden signature I’d written with a party pen that happened to be next to the keyboard. I signed it as a southpaw to throw off any youthful sleuthing.1 The boy ran around our bed waving his document, his younger sibling in tow with a vision that he too would someday enter this larger pantheon of magical beings.

What I had not counted on was the obvious, that between them there were 39 teeth to go. This presented a significant call on details and plot over the next few years.

With the next tooth, the Tooth Fairy revealed the existence of her magic parrot, who locates every tooth under every child’s pillow every night. [Note to the data-driven: our planet enjoys about 950 million people between the ages of five and 12, who lose on average 2.5 teeth a year, which works out to about one every 146 days; this amounts to 6.5 million teeth a night, or 4,519 teeth a minute. You’re goddamn right that parrot is magical.] His name was Midnight Pinkie, perhaps because he works all night, on a globe where it’s always night someplace. And yet he finds time to keep TTF fully apprised of what’s happening in the lives of two of her favorite kids.

Share

It then developed that Midnight Pinkie was assisted by a unicorn named Sparkle Bob. One spots, one transports. Then, that Sparkle Bob liked ice cream, and could they leave some out next time? The tooth after that revelation yielded further evidence in support of the myth, for the next proffered ice cream bowl was left empty by their bunk beds. Mirabile dictu!

Another unicorn, Sparkle Keith, showed up to help with the younger boy’s teeth. Another one, Sparkle Norman, worked in the greater Himalaya/Mongolia/Siberia region. The Tooth Fairy also maintained an Enamel Archive, where the history of each tooth and what it had chewed was recorded. There were the musically-inclined snow monkeys. There were skipping turtles, in charge of an Easter Egg hunt at the North Pole. There were many doings with Santa (the firstborn running into our bedroom yelling “The Tooth Fairy knows Santa!” is a cherished morning newsflash.) There was a tour in the command zeppelin to the South Seas. Iceberg races with Royal Ventana, the penguin queen of Antarctica. Snowmen they made with her, and took back to the North Pole. They heard of dances at the Cloud King’s house. They wrote back to her, asking for a ride on Sparkles Bob and Keith.

At the time I was, in my day job as a financial journalist, reading a lot of quarterly reports and shareholder letters. Which may explain why sometimes the Tooth Fairy would begin an update with “Big doings around the kingdom these days…”

As some of you have guessed, the Tooth Fairy drinks. For some reason, children’s teeth seem to come out unexpectedly during dinner. Well I remember draining a second glass of wine as I smiled at my son’s newly gap-toothed grin, and prepared for a second writing shift for the day. Mrs. Hardy would look at me lovingly across the table, mouthing something like, “your move, buster.”2

As everyone slept I was back in our cottage, raveling my shaggy dog mythology from the keyboard, sometimes abetted by a slow third glass, while raccoons padded on the roof above. It was great fun, and always a relief to make my deadline, painstakingly signing my letter in southpaw gold.

In some ways it was some of the most distinctive writing I’ve done. All effective writing has an audience in mind. It answers the question, “who is reading this, and where do I want to take them?” This is true even when the audience is oneself and the destination is “Hopefully, illumination.”

Here and elsewhere, though, one’s children are a special category, if only for the way we know them. Everyone else we know, everyone, comes to us with their history, and so much of any close relationship consists of us trying to explain ourselves to each other. Holding my firstborn, however, I knew the only connection utterly unsullied by time, and therefore possessing a unique purity. Not to mention, all the cleaning, and toting, and sometimes myth-making, thereafter.

Late into the night as TTF, I wrote for them with that connection in mind, and often in anticipation of how overjoyed they would be at the new knowledge. When I wrote for print newspapers, I’d sometimes see people on the subway smiling, even laughing, at something I’d written. That feeling was close, the way a lightning bug is close to lightning itself.

Today I might have AI do the whole thing for me, issuing a few simple prompts and seconds later possessing a far more coherent mythology I could deal out over those years.

Reader, thank Goodness I’m not looking back on how I did the whole thing with AI. Nothing like the same audience relationship.

As I write this, I have another confession. Until now I never revealed my identity to my sons. They simply ran out of teeth, and The Tooth Fairy sailed into the horizon in her airship, accompanied by all her magic friends. One of those children will soon clerk on the Federal Court, while the other manages a team of a dozen AI engineers. Most likely they have moved on to other delusions, more suited to the world of adults. As must we all.

And if by chance they have not - Boys, I made this whole thing up.

The kingdom ended a $4.7 billion contract for construction of an enormous freshwater lake. It was the crucial centerpiece of Trojena, a projected desert ski resort that was also going to contain a golf course, hotels, restaurants and (brace yourself) shopping. Compounding the tragedy, Trojena was going to host the 2029 Asian Winter Games; now they will be held in Almaty, Kazakhstan, where there are already mountains, and snow falls in the winter.

Against the loss of lives and real property in the Mideast these days, the probable end of Trojena is a minor tragedy. What makes it worthy of note is that Trojena itself was a subset of NEOM, a truly spectacular folly on which the kingdom has already spent some $50 billion, with little to show. And, by extension, the collapse signifies a generally sad forecast for the next couple decades in the Mideast. Their loss won’t be good for the rest of us, either.

For those unfamiliar with NEOM, the 2017 brainchild of Saudi leader Mohammed bin Salman, the modernization project’s original centerpiece was originally conceived to include The Line, a 170-kilometer long vertical city, just 200 meters wide but 500 meters tall, sheathed in moving mirrors and housing 9 million people. Add to that and Trojena the industrial robot-based manufacturing city of Oxagon, to be built out into the ocean.

The Line’s promotional video is itself a thing of bubbly beauty. Promising an end to “dysfunctional and polluted cities that ignore nature,” it was to be “organized in three dimensions” with flying drones, “equitable views,” and a transit system that would make it possible to cross the megacity in just 20 minutes. There was to be a hidden marina, home to luxury cruise ships, which would sail under an arch bearing an inverted 30-story building that hung like a glamorous pendant. That last bit was scrapped after engineers pointed out that the force of Earth’s rotation would turn the pendant into a pendulum, which would break off and crash into the harbor.1

NEOM-related construction went on for some three years, before The Line was significantly scaled back to a couple of kilometers late last year. It was then expected to house some people and maybe a bunch of data centers. That was in January.

At the end of February, though, America and Israel attacked Iran. At this point Iran has no military forces to speak of, but has become a new, and possibly lasting, threat to most of the Gulf Economies. Since NEOM is well within drone range of both the Iranians and their allies the Houthis, condos there might be a tough sell to most foreigners.

Even if the war ends in a couple of weeks, as hoped, the new status quo is likely to lead to all kinds of recalculations in the region. Dubai, with its luxury condos of $50 million and up, will see its rich foreigners price into their purchases the risk of being a drone target. Qatar, which very quickly had become a producer of fertilizer, particularly for farmers in poor countries, has seen its exports go to zero. It was also one of several Gulf States giving billions in foreign aid2 to poorer countries, a relatively recent phenomenon that may now suffer from a cash crunch and new defense spending needs. Rory Stewart’s projection of a 10-15 year economic setback in the Gulf seems likely.

If the threat from Iran were to disappear,3 the country’s previous status as the single shadowy bogeyman that Arab countries could agree on would probably disappear with it. In that scenario these countries would rejig their security strategies, and in the absence of Iran they could well go back to antagonizing each other.

All of which makes NEOM, The Line, and all the rest of that ambitious project a relic of a bygone time, even before it’s built. Great follies are normally a beautiful thing, something we can all point to later to show how crazy the dreaming became. In this case, however, we will miss the impossible future that people once foresaw.

Earlier this month I was at the Montgomery Summit, an investor conference focused on advanced technology.2 Brace yourself: There was a lot of AI talk. Two companies in particular stood out to me for the way they are using AI to play human-type games, but on a whole new level of complexity.

One of the companies, Xyme, is focused on transforming chemistry by locating and programming new enzymes, molecules that speed up the catalytic events transforming one material into another.

We do that now, when we turn crude oil into things like gasoline, fertilizer, plastics, pesticides, and a bunch of other things we’re currently reminded3 are critical to modern life. Nature plays the same transformation game with all kinds of enzymes, using far less energy and with far less waste. Positing and creating new enzymes has been extremely difficult and time consuming, but Xyme says it can discover new materials with a combination of synthetic chemical data and AI attuned to quantum physics.

Buzzword compliant, maybe, but the week after we met they announced that they have designed a novel enzyme. You can read about it here. The potential is to produce the biofuel ethanol for half the price of gasoline.

This development shouldn’t come as too much of a shock. Eight years ago Alphabet’s DeepMind released AlphaFold, which predicts protein structures. Its creators won the 2024 Nobel Prize in Chemistry. Making previously nonexistent enzymes is a far tougher proposition, but the basic use of AI in understanding and manipulating the physical world is now an established reality.

Share

It’s illuminating in this context to think of science, as many scientists do, as a game of seeking understanding (I tend to think of writing that way, and “play” it in part to understand myself and the world better.) AlphaFold, and now Xyme, are programmed to play a game of discovery.

Before protein folding, DeepMind was renowned for AlphaGo, a machine that beat the world’s champion Go player. Go is considered exponentially more complex than chess. AlphaGo got so good by learning the initial rules of the game, then playing millions of games against variations of itself. In competition it made moves that were highly unlikely to be made by a human, but which were still effective.4 You could say that Xyme aims to play the chemistry game in a whole new, nonhuman, way.

The other company, Lila Biosciences, first came into the public eye a year ago with a promise to create AI that plays the ultimate game in science, the scientific method itself. This means that on its own the machine will, with enough experience, not simply point to hopeful stuff to check out, but conceive of new experiments, carry them out, and apply the new learning from the results to make broader and deeper observations about reality than we have achieved. It’s the human practice of observe-hypothesize-design-experiment-conclude, but umpteen times faster.

Today the company says it is running all sorts of experiments in diverse subjects, for example materials science and biology, around the clock, much the way AlphaGo played millions of games of Go against itself in order to achieve a new proficiency. Lila has played through over 600,000 variations of MRNA molecules, a company executive said. It has worked on advanced magnets without using rare earth minerals, and it has split water into hydrogen without the use of precious metals.

To be sure, all of these results need to be checked independently, and the economics of any new discoveries have to make sense within the marketplace. Yet if this takeover of the scientific method itself bears out, it would have big implications for how industry works and how academia operates. It would also, for some game-playing scientists at least, be similar to the shattering loss that Go champion Lee Sedol felt watching AlphaGo play at a whole new level.

As I wrote elsewhere, our civilization’s two engines of growth have been the market and the method. Uniquely among civilizations, we’ve set aside teleology, or any final fate of things, for a triumphant world of middles, where prices continuously change without a final price, happiness is pursued rather than achieved, leaders transition without ascending to heaven, theories prevail only until they’re improved. Since we didn’t have to work our ideas backward from our particular sense of God, we have manipulated the world better than anyone.5

But outsourcing the scientific method itself to an algorithm, one so enmeshed with methodologies like Deep Learning that we can’t really understand what’s going on - that might represent the final end of our method. We ourselves won’t even think up what’s next.

On the one hand, this could be the final demonstration of what AI researcher Peter Norvig pointed out in 2008, that decades spent on theories of how language works mattered less for computer translation than a year of really big computers with lots of data, just bashing through the problem.6 On the other, we would increasingly be consuming results without looking for new things ourselves through the scientific method.

And if history is any guide, we will then find something else to look for .

A decade of my life was enmeshed with Japan, first as a traveling salesman and then as a foreign correspondent. I left in 1994, when Japan’s median age was 38 years old. Today it is about 50, and aging fast. On three other visits since repatriating I worked with tech companies, so I saw many younger people. This time I got the bigger picture.

I’d heard the weird stats and chilling stories. Sales of adult diapers have outstripped sales of infant diapers in Japan for a decade. It has more pets than kids.1 In the countryside, emptying out for years, houses sell for $2,000.2 Rural villages have dummies in the playground to remind themselves of children. The number of vending machines is off 23% from its peak - these pricey outlets for busy people have lost share to cheap markets, where old people dawdle and bargain hunt.

Lived experience is always more intense than reported data. Tokyo’s population is larger than when I left, because the countryside is emptying out. But it is strangely quiet. It’s much more vertical, thanks to three decades of government-led construction to jump start a flagging economy. But there isn’t much street life, compared to when people popped in and out of two- and three-story dwellings. They don’t blast music and other noise in the streets as much. Once-cacophonous pachinko parlors, absolute onslaughts of noise, light and burning tobacco, where crowds of individuals hid from work or family to smoke and gamble on vertical pinball, are now scarce and empty.

Empty playgrounds. Vans converted to hold five wheelchairs. Schoolkids playing soccer alone, or on underpopulated teams. Places famous for being crowded, like the crosswalk in Shibuya Square, have smaller crowds.3 The “passenger pushers” who ensured that rush hour subways were loaded at over 200% capacity are now rare sightings (it’s amazing what you can miss.) When you see a baby you feel like telling someone.

This was a visit to one very unique place, and not all positive,4 but each sight was also a vision of the developed world’s future, coming at us fast.

Share

The world fertility rate is now 2.2 births per woman, barely above the population replacement rate of 2.1 births, and falling. Japan’s birth rate of 1.2 births per woman is tied for seventh place, along with Chile, Italy, the United Arab Emirates, and Thailand. This group of countries is diverse enough to warn anybody off glib cultural reasons to account for Japan’s declining population. It’s strange that people treat this country’s dwindling population as unique, like cherry blossom viewing or kabuki theater.

What does make Japan distinctive is how long depopulation has gone on there. They stopped having lots of kids 50 years ago. The ensuing problem was foreseen earlier, too: In the economic boom of the early nineties we wrote stories about a government plan to deploy the country’s astonishing (and, it turned out, illusory) wealth to buy Northeast Australia, which they would convert into a giant old folks’ home. Japan is now the major country5 with both the highest median age and the greatest percent of population over 65. More than the estimable habits of eating fish and taking the stairs, Japan has lots of old people because there haven’t been that many kids here for a very long time.

“Japan just gets there first,” says my friend Peter Tasker, over an enormous shared donburi one night in Shibuya. Peter is a longtime resident and excellent observer of Japan. “While everyone else thought China was going to be this great partner, Japan was getting bullied by China. Now lots of countries have China leaning on them. Japan was first with a catastrophic natural disaster that destroyed essential infrastructure, with the tsunami in 2011. While Germany went nuts and shut down the nuclear power plants and went back to coal and gas, Japan upgraded its system and got people on board. Now Japan is building for an old people world ahead of everybody. Lots of things that are going to happen elsewhere have already happened here.” Peter has lived in Japan for most of five decades. He plans to stay on partly because it’s so well retooled for older people.

Share

Of course, Japan’s reality is not the global one. The populations of many countries are still growing fast. And it’s good to remember that demography is destiny.

Among the 40 countries with the highest birthrates, ranging from 3.6 to 6.1 babies per woman, all but three are in Africa.6 Many of these are small countries, but Nigeria, with 4.5 kids per woman, has 223 million people, the world’s sixth-largest population.7 Outside Africa, Pakistan, with the world’s fifth-largest population, also has a fertility rate of 3.6 kids.

What these countries have in common isn’t their location. It’s that they’re poor. And lots more than rich people, poor people need jobs.

Does this mean rich countries are about to see migrations by the billion, a scale unstoppable for even the most repressive immigration police? Possibly, but these migrants would be headed into developed economies that won’t be growing that fast, owing to aging populations.8 And, if you believe the projections, an increasing amount of physical work will be done by robots. Japan had about 358,000 migrants in 2024, but it already has 450,000 industrial robots, a high number thanks to its early concerns about an aging population. Stories about domestic robots for the elderly are now legion. Each new robot nurse is one fewer home healthcare worker.

In other words, on top of the hostility this enormous wave of migrants would face, there may not be that much for them to do when they get to the rich world. Barring climate-related catastrophes that make their home countries utterly unlivable (in which case we’re all screwed, since climate doesn’t happen in just one place) there won’t be that many reasons to leave the languages and cultures they know.

Share

Just before I went to Japan I was at a financial conference in Santa Monica, California. The final speaker, Michael Milken,9 put up some demographic slides that told much the same story - more adult than kid diapers in Japan, more dog strollers than baby carriages in South Korea, China’s unsolvable aging crisis, the burgeoning populations in Africa.

Milken then made an excellent point to the people in the room, who represented hundreds of billions of dollars in available funds.

“The world can’t take the migration of a billion people,” he said. “We are going to have to create jobs in Africa. Give people opportunities where they live.”

He was exactly right. I hope the aging money guys in that room weren’t hard of hearing.

Meet Young Bacchus, the bronze head of a Roman statue from the first half of the first century AD. He currently resides in the Getty Villa in Pacific Palisades, Calif. I had the good fortune to hang out with the lad/god this week while moderating some panels at the March Capital investor conference in Santa Monica. Of which more later.

One of the things I like about this statue is the way it seems to get drunk as the viewer changes position. In the photograph above his wide eyes take in the viewer, his mouth is slightly open in wonder, and the garland rests lightly on his locks. The photo was taken while I viewed the bronze directly, bending my knees to do so. Then, I looked at him again while standing up, which put me about 3 inches above his eye line.

Whoa there, son. Looking a little muddled. His eyes have a vacancy and the lips are slack. There’s too much stuff in his hair.

This Bacchus looks a few too many cups into his evening. Somebody better call the lad a chariot before things get out of hand.1

Share

He’s not the only Bacchus who drinks as you change your point of view. In the Bargello Museum of Florence there’s another, said to be Michelangelo’s second professional sculpture, made when he was about 22 years old.2

Michelangelo’s Bacchus is mostly intact, and you can see how the artist plays with the god’s posture. Approached from the left side, he’s stable, cup straight on as it heads to his mouth:

Then you circle around him, and from the back there’s a decidedly lopsided gait, and the little satyr that at first was frolicking at Baccus’ foursquare leg now seems to have both legs on the ground, supporting the boss a bit:

Then you continue around to the front, where Bacchus is definitely wobbly. He looks through the handle of his now-tilted cup to drunkenly meet the viewer’s eye. Nice!

Not bad for a 22 year old no one had ever heard of.

The following year he did this:

It had a couple of angels too, but they’ve flown off.

Word is that Michelangelo, a 23-year-old nobody, overheard someone attribute his Pietà to another sculptor, a famous craftsman. That night Michelangelo returned and carved his name on Mary’s sash. He never had to sign anything again.

1

I’d be remiss not to mention the wonderful geometry of the Getty’s cases, which do such a nice job here of framing the Bacchus with thin white lines. Check out this winged lion, a Phoenician image cast by Iberian artisans in 500 BC:

It’s bigger for being so well-encased.