As the world around us shifts shape, it’s tempting to look at recent history, the US Constitution and the like for perspective. But instead, I keep thinking farther back, to when our institutions and assumptions were quite different, to the days of A Distant Mirror: The Calamitous 14th Century by Barbara Tuchman, an excellent recounting of the 14th century in France, at the beginning of a period of famine precipitated by bad weather and farm animal diseases, the Black Death epidemic, a schism in the Catholic Church, and purposeless wars, which together resulted in a population that shrank by roughly half. Now again we have more calamities and random events, a black-swan period collectively but each within the realm of overall probability. Yet the great pendulum of time appears not to be swinging back in the usual curve between two points, but off in some third dimension. What has changed since then…and how much, due to human nature, has not?

Back in the 14th century, nobles owned and ruled the land, and de facto owned the serfs who lived on that land. (Kings were in theory superior, but there was no way for them to keep tabs on what was going on in all the local fiefdoms. No spreadsheets to help, let alone surveillance cameras!) The noblemen learned how to fight - jousting in tournaments and launching battles “for real” - but had little sense of strategy; hence, a continuous stream of skirmishes but no lasting victory, resulting in the 100 Years’ War (1337-1453) and continuing turmoil without resolution. The clergy were trained formally, promoting the ideology of Christianity, but in practice they were just agents of another predatory institution, exacting entry fees to Heaven from the nobility. They also managed or mediated other relationships - including transactions such as marriages - among individuals and families.

As for the serfs… their role was to serve rather than exonerate the nobility, producing food and fuel as payment for the land they lived on, with little choice to do otherwise. Those serfs lived in drafty huts, with one another and with random farm animals as well. (Unlike slaves, who “belonged” to other people, serfs belonged to the land. If the land was transferred through marriage, inheritance or some such transaction, the serfs went with it. Now, in some sense, workers belong to companies, or suffer the daily challenges of gig or self-marketing work.)

Yet everyone was helpless against the Black Death and other calamities, which culled roughly half of the population and caused a great reset: Cheap labor became expensive - or at least, less cheap - and improvements in farming equipment made the workers themselves more productive, fostering a shift in the respect they garnered.

Fast forward to today, through the Renaissance (14th to 17th centuries), the Age of Exploration (15th to 17th centuries), the Scientific Revolution (16th to 17th centuries), the Agricultural Revolution (17th to 19th centuries), and finally the Industrial Revolution (18th to 20th centuries). Technologies and equipment including telescopes and reading glasses, the printing press, and ships and bicycles among others, brought new knowledge of the world at hand and far away.

As governments consolidated and democracy of sorts emerged, some of the nobility took over - mostly given their roles by (some of) the people rather than by God. Other nobles, especially the younger sons, shifted to owning factories, merchant ships and ultimately companies of all kinds, creating a whole new sector of private business with a professed ideology of free people and scalable free markets. Factories and machines grew in productivity, reducing the primacy of land and ultimately turning serfs into paid workers. The Industrial Revolution saw huge changes in society that included much more mobility - geographically, economically and even socially. In the Soviet Union and then Russia (and also China, which I know less well), the path diverged for a while, with corrupt Communism that was eventually replaced by equally corrupt capitalism and now a commingling of corrupt government and equally corrupt business enabled by and subservient to the state.

You might call one arc of the Industrial Revolution the “Trading Revolution,” best exemplified by the East India Company - perhaps the SpaceX of its (long) day, from 1600 to 1874. To quote Wikipedia, “At its peak, the company was the largest corporation in the world by various measures and had its own armed forces in the form of the company’s three presidency armies, totalling about 260,000 soldiers - twice the size of the British Army at certain times.“ Along with its competitors, it spread the use of or addiction to new delights from overseas: tobacco, opium, sugar, spices, tea and cotton, among others. Cotton in particular also fostered new industrial scale and exploitation of workers, with the use of slaves for picking it in the US South, and cotton mills, spinning wheels and the like in England. (My grandfather’s aunt and mother worked in the cotton mills of northern England starting from the ages of 4 and 8, respectively. More on my grandfather - who became “Sir George” - some other day.)

Along this same winding timeline and even as both governance and governments (and business) became more centralized, the coherence of the religious sector declined with the rise of Protestantism and many other sects, plus migration (emigration, immigration and invasion) in all directions enabled by boats, trains, and ultimately bicycles and cars, along with better maps. Still, the clergy kept more or less the same role, even though it now holds diminished power in a world where location no longer defines religion, and atheism, fragmented churches and religious sects flourish. Can Pope Leo help regain religion’s (not the church’s) stature?

Second verse, not quite same as the first

Somehow that crumbling set-up from 700 years ago resonates in the US today, not repeating but rhyming. Like the jousting nobles of old, our leaders perform more than they govern, offering ultra-processed content and personalities, the modern equivalent of bread and circuses. Abroad, instead of aid or investment, we now offer tournaments and wars without strategy. The zeitgeist favors performance and brands over delivery, money over value, identity over policy. Both businesses and politicians used to make promises of what they would do or deliver; people were buying results. Now people are buying or voting for products or leaders that will make them feel part of a tribe. Even as gambling and prediction markets try to turn sports and other aspects of everyday life into transactions and one more quest for money, the current excitement and (mostly) friendliness surrounding the FIFA World Cup show the need for and celebration of human tribes. We are meant to share those connections, superficial as they may sometimes be.

We watch as all around us as business and government intermingle in both directions, with leading noblemen-cum-businessmen taking on what look like government roles - from Elon Musk’s brief stint at DOGE, to a string of Wall Street executives “donated” by either party to the SEC and other regulatory agencies, to various others representing their own and perhaps Trump’s business interests in negotiations with other countries - neglecting the voters they supposedly serve. We are back to a world where people genuflect to nobles, not at altars.

Meanwhile, the US Supreme Court’s Citizens United ruling (2010) opened the way for those with money to spend it on helping politicians to win or keep office - and as of last month, the current Supreme Court gave them new license to spend it more directly on candidates (June 30). Politicians (especially those aspiring to or new in the job) spend more time raising money for their own campaigns/political survival than they do figuring out how to spend the public purse, ideally for the benefit of all.

Elsewhere, religion too often means identity and tribe more than belief. Preachers are valued for being “authentic” rather than for following the dictates of their church - let alone those of an actual religion. The clergy have lost some of their influence to the intellectual elites of the universities and to Substack writers, along with the proliferating think tanks that espouse conflicting views of the world and the proper role of all the other institutions. Those elites are selling access not to heaven as much as to Davos or the annual Allen & Co. Sun Valley Conference. (I won’t say where I was, but two weeks ago I overheard two men chatting; one mentioned he was heading to Sun Valley, while the second had other plans - unspecified. Did he have an even better offer…or did he hesitate to say that he had no offer at all?) Meeting the right funder or politician at one of these events can result in heaven on Earth.

For some people, money gets you political power. But for many rich people, money is often most important as a form of rank rather than of actual physical wealth or comfort, which is taken for granted. By contrast, for those who actually work for their money - for wages or a salary rather than a cut of various transactions - money is real and guides their everyday decisions and trade-offs, even as they are lured to spend it on ultra-processed temptations (Flamin’ Hot Cheetos or even fake shrimp for religious pseudo-observance, as well as AI- and marketing-infused content) of little human value.

How does AI change our (perception of) human nature?

But now, this fluid situation - reflecting long-term trends, the (re-)election of a single man (which may reflect some of those trends), the social separations and mistrust due to COVID, and the rise of social media - is likely to be upended yet again by the advent of AI and AI agents at scale. Human nature persists, but the rise of AI is changing our perception of our intelligence - which we may value too much - against our humanity and wisdom.

Just as new farming equipment and the replacement of oxen with horses made 14th-century serfs more productive and valuable, AI may do something similar: It replaces human intelligence in many aspects, but it also provides great capacity for those who learn how to use it. At the same time, AI cannot replace human love or kindness, or humans’ need for one another…and we need to recognize that more broadly. This seems to be happening: The widespread rejection of AI and data centers may carry more weight - and hold more truth - than the reactions of the (19th-century) Luddites.

This is our existential question: Not what will AI do to us, but how will we react to it? Will we understand how to value human existence and human attention in a world of seductive “artificial” offerings of all kinds? Can we understand the value of the artisanal? And can we feel that we play an important part in this future? What’s the clarion call that will wake us all up and shake us out of our artificial dreams?

As our current Pope Leo (2025- ) has observed, AI will disrupt and shape his papacy much as the Industrial Revolution did for that of Pope Leo XIV (1878-1903). Is AI a new kind of private land - to be owned and exploited by the new nobility? Or can it be a new open prairie, to be farmed by anyone willing to set up a homestead on an infinite plain? Will AI actually become a new sector, crossing and disempowering nation-states and business as some new and intangible yet powerful part of the world’s (or solar system’s) structure? Nation-states were defined by geography (along with their nobility). Businesses are defined by their ownership, which now crosses multiple geographies and jurisdictions. The new sector of AI agents may be even more elusive - and harder to hold accountable. As for us, are we defined by our online personas (enhanced with all kinds of AI tools) or by those we love and who love us - not because we were matched by an algorithm, but because we grew to love one another through continued sharing?

In one way or another, I’ve been pondering this for a long time. My favorite story about the need for clear vision in a world of murkiness comes from the time I was lucky enough to sit next to late Associate Justice of the Supreme Court Sandra Day O’Connor. I asked her about the difference between the way people vote and the way they sit on a jury. She was appropriately discreet and did not name any cases, but she said that while not all juries decided correctly (in her opinion), she had seen only two cases where the jury had not performed their duties with diligence and thoughtfulness. Voters tend to think they ought to vote but it’s more of a gesture than an action, whereas sitting on a “jury of your peers” sends the message that you are in charge of an outcome, one that may change several people’s lives. There is no intermediary, no crowd to join, little distance between you and the person whose future life (or death) hangs in the balance. The jurors take their job seriously and work to uncover the facts and understand the underlying law. In a world of performativeness vs. actual performance, we must treasure and realize our ability to fully understand truth and its consequences.

Author’s note: This spec for a video game is the byproduct of my four-month encounter with cancer of the esophagus, which ended successfully in April 2016. So I felt in perfect physical health during this whole ordeal – even though my mind knew I had cancer, growing slowly with each tick of the clock. I knew how lucky I was to have found it (regardless of how unlucky I was to have it in the first place), and I also was lucky to have connections, time and stamina to work for the best outcome, whatever that might be.

It’s a truism that the best way to bear torture is to distance yourself: Analyze your reactions; become an observer; imagine yourself playing a game. To some extent, that’s what I did in the face of my cancer diagnosis. I had to play the game consistently and patiently, but at the same time I had a full-time job to distract me between the steps.

Ultimately, the purpose of this imagined game is both to show how crazy the system is, and to share the experience with people unlucky enough to face it or powerful enough to change it.

It’s in the form of a spec rather than an actual game because no sane person would ever want to play it – nor would almost any sane person ever want to create it!

Okay, your move!

Second Opinion is a rigorously paced VR game, with exciting visual effects including waiting rooms filled with people in wheel chairs, hospital corridors with dull walls lit by bright neon, clinical staff in white coats, a brilliant day that makes you glad to be alive seen through a filter of medical reports, and so on. You the player have just received an unexpected diagnosis of cancer of the esophagus. After a couple of false starts, you get to a hospital for tests with a doctor who counsels an esophagectomy – removing your esophagus and replacing it with pulled-up stomach tissue (leaving you with very little stomach). It’s a brutal but relatively safe procedure, with a short hospital stay and just a few months of recovery, they promise, though your quality of life will be permanently compromised.

Your mission is to manage your medical care for optimal results, based on your risk preferences, your ability to manipulate the system, and your persistence.

The Gameplay

You start with 1000 points, 10 points per step for this 100-step, almost seven-hour game. You need to make sure that you do not run out of them, because that will end the game prematurely. Otherwise, the game ends with your procedure, which automatically starts at 97 steps. Each step costs 10 points, but sometimes you get extra points back if the step goes well. Your final score is calculated at the final step, while you undergo the virtual procedure. More on scoring below.

You have to wait four minutes between steps in the game – long enough to be annoying, but not long enough to do much else while you keep playing the game. You can skip a step, but then you just lose the whole 10 points, and you may mess up the following steps too. Once you start the game, you can’t come back and resume it later, because time keeps passing, and you may run out of step/time points if you don’t keep playing each step at its appointed time. (For the best simulation, it should be played at work or during child care – because that’s how disease actually happens.)

However, though the steps keep time like a metronome, the actual play occurs in threads, weaving through the other threads, as you can see below.

Enough setup! The clock is ticking!

Initial Consultation

After your starting-point diagnosis in an independent clinic, you need to find a doctor and hospital to perform the surgery that will clearly be necessary. That requires several steps, at 10 points each. Calls to friends and your primary-care doctor for recommendations: 10 points each, times several friends. A fruitless half hour on the NY Presbyterian website attempting to make contact: 10 points! And finally, a hit! A friend puts you in touch with a doctor at Mercy Hospital (10 points) who actually returns your call (get 5 points back). That leads to a little appointment-setting dance (with voicemail) that costs 30 points, but now you have an appointment for early March.

Meanwhile, cue the second thread: Insurance!

Insurance

This thread continues for the duration of the game…and in fact persists for months after as random bills come in, some of them for amounts like $43.23, others for $12,582.49, offset by various insurance reimbursements such as $11,263.47 and $55.42, and others for $0.00, but meticulously linked to some particular service or other. (You can remove this linked-reporting feature, which is a browser plug-in loaded automatically by the game, by consulting a 12-year-old.) These bills take time to figure out, let alone pay (another step) – 10 to 20 points each.

You start by calling your insurance company to notify them of the diagnosis and to check that they will actually cover the procedure – resulting in a multi-step (four minutes apart) string of call-backs and voice mails, but all seems well.

Scheduling each step – a required medical consultation or test, for example – is a step in itself; so is calling the insurance company so they call the doctor, and calling the doctor’s office to ask her nurse to call the insurance company back. At each step, you usually get to choose from several options, including high-risk “obnoxious-intervention” measures (50 points each) that sometimes give you a positive response, but that might backfire – in a denied claim, for example – if people react badly. In the case of insurance, you can call an old friend who is now the medical director at the insurance company. (In my case, that turned out well.)

Initial Consultation – Part 2

Once you’ve set up the appointment and checked with insurance, the next step is simple: You get a treatment plan – in my case, the esophagectomy. That leads to more steps for scheduling, which takes some back-and-forth and also cannot be hastened by the user… Ultimately, the appointment is scheduled for early May…leaving time for a third thread.

Second Opinion

At this point, the game asks the user: Do you really want to lose your esophagus? For 50 points, you can ask Steve Jobs’ cancer doctor, whom you happen to know, for a recommendation of an alternative hospital and procedure. Your first reaction is that this would be rude and a little bit like cheating, but hey, this is a game where it’s important to win, so you take the bait. That leads to a bunch of separate threads: You have to go back to your original doctor to get a scan of the lump (2 steps), you have to pick up the CD-ROM (remember those?) in person (another step) and FedEx it to the doctor in San Francisco who will recommend a doctor in your hometown.

But this time you get lucky: With a connection from an illustrious physician, it takes only one round of play to get an initial appointment with another hospital – for two days later.

But then the complications return: The hospital wants to do its own scan and diagnosis before proceeding, and that hospital is not in your insurer’s network. Re-engage the insurance thread, which takes another five rounds of play – including the real-time cancellation of a PET scan ordered by the hospital but not covered by the insurance.

The second hospital, call it Redemption, wants to do a less radical “procedure” (i.e. surgery), requiring more skill and special instruments (a “laser thingy”), that removes only the cancerous tissue plus a small margin. It may need chemo/radiation follow-up, and recurrence is more likely. However, in 5 to 10 years, there may well be immunotherapy that makes recurrence a minor issue.

Meanwhile, this other option will actually cost the insurance company – and you – much less: On the order of $30,000 for an outpatient procedure, basically, plus half-yearly follow-ups, vs. over $100,000 for the esophagectomy alone. But your insurance company doesn’t have a relationship with Redemption. Can you successfully make the case to them? How many rounds of play will that take – including further calls to the voice mail of your friend, their medical director?

In short, do you want to risk all those points on a second opinion that may not work out? That’s up to you the player. And the game is already more than half over….

Strategic Play

One signal aspect of the second opinion thread is that you can run it partly in parallel with the rest of the game, though at some point you have to choose one path or another without really knowing the likely points attributable to either path. In fact, you are expected to learn how to juggle several different threads – insurance, doctor visits, transfer of records from one institution to another, and the vicissitudes of normal life (work and family) that proceed simultaneously.

How It Ends

The game proceeds along these simultaneous threads until step 97 – the procedure. This takes three steps – prep, anesthesia, and calculation of the final score. It doesn’t matter where you are on the other threads; these interrupt that flow. Prep is hard to express in VR; it mostly requires not eating or drinking.

And the anesthesia…It depends on the player’s attitude. I’ll always remember the long long minute during which I sank into sleep for my procedure, with my stepmother by my side. Whatever the outcome – tumor removed or rampant malignancy discovered, simple process or punctured tube – the four months of this horrible game would be over. It wasn’t quite like dying peacefully, but it was definitely a conscious, blissful passage from an unquiet life of uncertainty to some further stage where I could focus on either dealing with a new, shortened reality or resuming the old one with more appreciation.

The End

And at last, step 100: Your final score – which you receive when you wake up from the procedure — reflects the skill of your play multiplied by a random number between 0 and 10.

Right now, everyone - including the AI hyperscalers themselves - is talking about various forms of Universal Basic Income (funded by AI taxes of some kind) as a response to AI’s impact on the US workforce. Yet they are missing the real challenge posed by the widely feared loss of jobs: how humans can secure not just financial sustenance but also identity and meaning. If we do use it right, AI’s productivity can create enough economic value for all. But meanwhile, while we are considering how and whether Universal Basic Income might work, here is an approach that we could and should expand immediately, building on existing affordances: time-limited Universal Basic Service (UBS).

The adoption of UBS would work recursively, both creating, and helping us to recognize and support, the enormous value of human-to-human skills - most directly via better pay, but also with overall recognition by society. It should move us towards a world where the market rewards people for their human contributions, and universal basic insurance (including some safety nets for health care, housing, food, education, and child support) takes care of those who cannot fend for themselves or whom the market has failed. Through UBS, humans would learn what it means to be human, from other humans - skills and attitudes that AI cannot genuinely replace. Additionally, UBS would force people out of whatever bubble they happened to be born into, and give them the opportunity and challenge of collaborating with people they would otherwise never meet.

Right now, we are increasingly focused on efficiency, often supported by AI, and not enough on other valuable goals such as the slow learning and human connection you get from interacting with other people and the joy you get from making someone else happy. That’s what most parents offer to and receive from children, but it also applies to teachers, neighbors, mentors and the like. It’s what used to happen in PTAs, churches, synagogues and mosques, playgrounds, grocery stores, shopping centers, bowling leagues…all over the place. But now, instead of learning from one another in our communities, we are on our phones, scrolling social media, or working with efficient, automated systems that keep us disconnected from other humans. And when we do seek to mingle with others, we often look online to find people like ourselves, instead of peering outside our own bubble.

UBS would do the opposite: It would get participants out of their usual networks and geographies. In the model I imagine, everyone is invited (and as soon as possible required, I hope) to perform two years of service, sometime between the ages of 18 and 23 (the specific details, like age, matter less than the basic idea and principles). Whatever they do would be a full-time job, and should pay enough that they don’t need a side gig. The broad goal here is to invest in worker training, raise wages overall, and establish a higher valuation for human engagement in work; UBS should not be a source of cheap labor in disguise. Whatever the “output” of the service may be, it should also include the inputs of learning and teaching - in other words, human interaction, not paperwork or solo physical labor.

We can begin by scaling long-established programs such as Teach For America (TFA), HeadStart, AmeriCorps (including VISTA), the Peace Corps, Habitat for Humanity and the like alongside the UBS-like US military. Those outfits don’t just send people out into the world to serve - a one-way transaction. They offer supervisors who train the participants how to serve, and to communicate with and train community members (though this latter doesn’t apply to the military). Meanwhile, many go on to make careers in related fields - or even to work as trainers in the services that trained them. For example, says Yasmina Vinci, executive director of the National Head Start Association, which represents Head Start programs and the staff and parents of those programs: “When I was leading what is now Child Care Aware of America, we had a national AmeriCorps grant that placed members in communities across the country to help develop infant and after-school care programs. We tracked their retention rate and found that 64% of them remained in the early childhood field after their two years of service; many used their education stipends to continue their education and earn additional credentials.”

For context: Right now in the US, the overall number of people in these programs is small - about half a million - plus about 1.5 million in the military. (Note that the numbers were higher before COVID and DOGE, especially for Americorps.) There are a bit more than 4 million people in each single-year cohort between 18 and 23, so if each person’s period of service lasted 2 years, that would mean approximately 8 million people (minus those with some reason not to join) would be performing such service at any point in time (though it would take some time to ramp up). Assuming an investment of about $100,000 per year in training and paying the participants, that would cost about $800 billion per year to fundamentally improve the experience and human flourishing across our entire population. (You can imagine your own numbers for permanent UBI and for whom it would be provided, but it would certainly be a much bigger lift and focused only on financial security.)

All these programs teach their members how to serve: how to listen to others, how to model good behavior, how to resolve conflicts, and so much more. In short, members learn how to live. This experience would benefit both the “servers” and the people they serve, and adds value to the world. It would also contribute financially by reducing the current costs of so many deficiencies in people’s lives that lead to poor mental and physical health and ultimately to unemployment, homelessness and the like.

For the rich or otherwise privileged, the basic dynamic of UBS is that you are obliged to share that privilege. You need to move somewhere else and live among people who don’t have that privilege - some of them co-workers and some of them the people you are serving. With luck, and smarts, you’ll learn something, whether it’s about the work itself, the community you work in or the day-to-day reality of people whose lives are different from your own. At worst, you will probably have a boss who is neither sycophantic nor servile. For the much larger number of less- or dis-advantaged, it’s a risk-free way to join a new community, learn on the job, and get training and the ability to operate in a different environment. With luck, you will also meet a peer group of models and mentors, as well as people who need your help. And you will learn the people skills you’ll need in a world that, with luck, will value and pay for them!

Each person’s experience will be unique, but I take some perspective from my own upbringing - which was out of the ordinary and privileged but hardly sheltered. My first real job (more lucrative than 25 cents an hour for babysitting at home) was at the Princeton Public Library, when I was 14. This was the place I had visited so often as a kid; now I was working there, serving patrons mostly just a bit younger than myself, under the watchful eyes of senior librarians who could actually check books in and out. I was restricted to shelving the books, helping kids to find them, and other forms of customer support. I also observed the parents, more closely than when I had been merely a visiting bookworm. There were the parents who insisted that their kids read Jane Austen or Hemingway, while the kids wanted Nancy Drew or Madeleine L’Engle. Some kids complied; others pushed back. And then there were the parents who supported their kids as they picked out history and science books. I loved this job: As a semi-outsider, I was observing how different parents could be from one another - and how differently their kids could react.

There was also my favorite book, which I probably recommended quite a few times: Danny Dunn and the Homework Machine (1958), about Danny and his friend Jay, who programmed the mainframe computer of Danny’s mother’s lodger, one Professor Bullfinch, to do their homework for them! Spoiler alert: They learned a lot more and worked much harder programming the computer than they ever would have actually doing the homework. (Some hint about using AI in there, eh?)

Though the library was outside the even more disconnected bubble of Princeton’s Institute for Advanced Study - where my father worked and the home of Einstein - it was still in Princeton. I had enjoyed a more radical experience the year before, when somehow I persuaded my parents to let me spend what would have been my freshman year of high school living with the family of a friend of my father’s, in a middle-class suburb just outside London. I shared a bedroom in the attic with one of the three girls (plus one boy) in the family, and attended the local girls’ grammar school. I learned so much living outside my own bubble - far beyond what you can find in textbooks; it wasn’t good or bad, poor or rich, just different. Instead of biking or riding the school bus, I took the Tube (subway) to school. The boy in the family had cerebral palsy and needed care and attention; the free-range living I took for granted at home wouldn’t work for him, and it meant all of us girls paid him special attention. And finally, there was British hardiness: First one down in the morning had to stoke the stove that warmed the kitchen but not much else. I returned home vowing that I would never ever again live in a place without central heating!

What I really learned is what I want everyone to experience in their own way, and which can also be done within the US just by living in a place with sufficiently different demographics or culture. Pretty much everyone thinks their life is “normal” (though not necessarily ideal). In fact, the things you take for granted are often strange and incomprehensible to others. It’s those basic, unquestioned assumptions - of course each of my parents has their own car - that underlie so much misunderstanding. How can you understand something that the other person - someone whose “normal” is taking public transportation or biking everywhere, or three months off in the summer - does not even notice, let alone know how to explain? That’s another fundamental part of this idea: You need to do it elsewhere.

Universal basic service should not be service-work by a few as a gift to some lucky recipients, but rather an activity that’s both an obligation for ourselves and a longer-term gift to and from society.

Of course there would be complications making that transition from a voluntary program to mandatory national service. This post is a prompt, not an answer; one person I’m addressing it to is Molly Kinder, who has written a lot about the impact of AI on jobs and who is now leaving the Brookings Institute think tank to start a do tank. How do you deal with the outliers in both directions? This program should apply to all: both those who have better alternatives than the jobs on offer via UBS (and may suffer from bone spurs), and those who can’t get a decent job and see UBS as a guaranteed stipend and training for a couple of years. And also to the disabled, some of whom may be the most able to benefit from a change of environment, an employer that wants to train them, and a system that wants them to succeed.

Brass tacks

The context in which UBS would operate is complex. The world’s biggest problems right now are intertwined: economic inequality, political polarization, environmental degradation, crumbling healthcare and education. One shared root cause is a divisive, individualistic culture that fosters broad mistrust of others and prevents us from collaborating to solve those problems. People think short-term and only about themselves, and don’t see or understand what they have in common with others. And no one trusts the inscrutable systems (run by inscrutable people, with or without AI) that control us all. UBS could address these dynamics on multiple fronts, helping people see past - and understand or even celebrate - their differences.

More than ever, the coming economy will need humans who know how to interact with other humans. Right now in the US, we have a surplus of young people without good jobs, and an overall deficit of workers in healthcare and education. These roles in particular benefit from people skills that young people have often failed to acquire during school, COVID and endless hours scrolling through social media. A few years of UBS can help them acquire those skills, along with experience and empathy for others unlike themselves.

The market has not worked very well in education and healthcare in particular because these are both areas where the workers do not typically get paid directly by the users, so there’s little direct “quality control.” Nor do they scale; they are not “efficient.” But they are key to human flourishing, even though the value created by these human-centric workers - healthier, happier and more productive humans, both the workers and the people they serve - takes a while to manifest and the long-run benefits are realized by society as a whole: wrong timing, different pockets. Fortunately, now we can use UBS to create and AI to track those long-term benefits. Any thoughtful person, let alone an economist, would argue that people are the country’s fundamental asset; keeping them healthy and educating them well pays off economically over the long run.

Once we have established the value of UBS with some compelling models, we should be able to fund them to spread broadly with general support, though it may be premature to figure out how to set it up in our current political environment. We can start with a non-universal form of Basic Service, which will help us knit our social fabric back together and teach humans how to teach and support other humans. If we can change the culture to recognize the value of human service both to the servers and to the recipients, the political path to spread it will emerge on its own. And then it should reinforce itself.

So, what to do first? As I mentioned above, we can expand the existing programs that right now are underfunded, under-resourced and under-promoted. They are good demonstrations of what works, but hardly critical-mass. Over time, I believe we should make participating in UBS a “normal” experience for all - like military service in Israel, where it is required of both men and women, and in many other countries - but for a wider variety of public service (firefighting, educational services and the like).

And now there’s some hope of serious funding, using those AI taxes and AI philanthropy that people are starting to talk about. This is exactly the way we should be applying those AI riches: investing them in our next generation of humans. Among other things, it might help AI’s public image.

Come on, Elon and Sam and Dario! How about putting a few of your many billions into Teach for America, AmeriCorps VISTA, the Peace Corps, Head Start, Boys and Girls’ Club, ExpandED Schools, Goodwill’s training programs, and a variety of post-incarceration or rehabilitation programs? Fund both the services, and the training of the workers in those services. Meanwhile, everyone, vote for more public spending to train and raise the pay of teachers and healthcare workers, including community health workers and doulas. That would address the affordability crisis - not by keeping prices unsustainably low, but by raising wages.

Is this all just an AI hallucination? Not at all. It is our collective opportunity and obligation to make it come true.

Disclosures: I’m on the board of (and donate to) ExpandED Schools, have donated to a local Boys and Girls Club, and spent two months living in Oujda, Morocco, with my first boyfriend, who was serving as a Peace Corps volunteer. My nonprofit Wellville.net worked with some excellent VISTA volunteers serving nonprofits in Lake County, CA. I also sit on the board of Charity Navigator, which is actively involved in helping donors find the best use for their funds - and their time!

The chart below shows some key investments from which I learned or earned, but it also reflects a fundamental underlying pattern: At least for me, learning and “earning” (i.e. getting a financial return) are inversely correlated. Instead of the usual, up-and-to-the right, increasing-goodness of a typical quadrant chart, the trajectory of the chart below is down-and-to-the-right. It shows that usually, the less money I made, the more I learned - often from fixing (or even just observing) mistakes. Both benefits are life-changing, of course: The money has allowed me to continue to invest in new companies or nonprofits, ideas and people, but I’m just as satisfied when my “return” is learning something new. Perhaps that balance is the real value.

Here are some of the stories behind the boxes, which are purposely imprecise.

LOWER RIGHT: I’ll start with the low-earning/high-“learning return” experiences. For example, I invested in 23andMe and joined the board in 2007 because I knew I would learn a lot about genetics - which at that point was a new and speculative field. I had already had my entire genome sequenced courtesy of the Personal Genome Project, but no one really explained to me what the results meant. 23andMe had the mission of reaching a much broader public with clear messaging, and ultimately building a robust dataset of both genetics and so-called real-world evidence (RWE) that could support medical research and find good targets for new or even old, re-purposable drugs. To do that honorably, we had to win regulatory approval for what the FDA considered medical advice - i.e. intelligible explanations of the data. We learned a lot when our general counsel left and the board did not pick up the slack and respond to the FDA’s complaints about various issues. After we finally made amends and started working closely with the FDA, I realized that however pointless some of its requirements, the agency (and regulation in general) was actually protecting us from less scrupulous competitors - at least in those days!

Meanwhile, in 2018 the company made a deal with GSK for drug development research. I left shortly after that in 2019, as the company upgraded its board and brought in more pharma expertise. But in retrospect that deal was a mistake: It was a joint venture in which each side would own half, but each side would also pay half, even though GSK - with revenues around $40B - was 200 times our size and had profits, while we had no corresponding earnings to support our side of the deal long-term. For 23andMe that deal was part of our future: Huge benefits to humanity from drug discovery using what would be the largest genetics-plus-RWE database in the world.

We missed the reality that for us, this was our mission; for our counterpart GSK, it was just one project among many. In 2023, as the biotech funding market collapsed and GSK faced stalling revenues, 23andMe couldn’t afford to continue its 50 percent contributions, and so it stopped contributing its share and would receive only royalties. Beware of misaligned motivations and resources! (This message might be useful for some companies contemplating partnerships with AI hyperscalers right now!)

But the underlying idea makes sense: Use the aggregated, unidentified data to discover genetic associations with diseases/conditions or responses to new drugs, and to understand how the biology works - or could be fixed with new or re-purposed drugs. That is proceeding; for example, the new 23andMe Research Institute recently published a paper in Nature identifying genetic predictors for GLP-1 weight loss efficacy and side effects for certain GLP-1 drugs.

From outside, I have followed the company’s trajectory into bankruptcy and beyond…into what is now a nonprofit acquired and run by 23andMe founder Anne Wojcicki that I hope can continue to carry on with its mission. One more clear learning: It’s really hard for a mission-driven company to survive in a profit-driven world. Thank you Anne for taking it forward!

Another great learning experience came from what I’ll call “Corporation X” (why be unkind?), a space startup that taught me something about space, but also about founders who refuse to let go. Through this company I got to visit a lot of interesting space facilities, try out some amazing flight simulators, and understand the intricacies of space travel from the ground up - even though I have not yet experienced them from LEO (low-earth orbit) down. However, once the founder’s ideas had gained traction, the company needed someone better equipped to run a company; brilliant engineers are often not the best managers. The company ended up going bankrupt, courtesy of that over-controlling founding CEO and a divided board.

And then there was Corp Y, where I learned how to address that problem. This one also had a tech CEO/founder who deserved veneration for founding the company, but didn’t have the management skills to run it. No one on the board - a bunch of techies - was willing to tell him so, and I rashly volunteered. It was really awkward. I didn’t know what to say, so I just told him the truth: “You’re amazing; you started this thing. We want you to be happy. But we think you’d be happier and more effective with someone else doing the boring part, so that you can continue to be smart and amazing.” After a few conversations like this he was smart enough to grasp the truth. We kept reassuring him that we loved him and loved the company. Thus I learned the value of telling the truth, even - or especially - when it’s hard. And it takes follow-up after a single step to remove a stuck-in-place founder and keep them on for the rest of their talents.

Then there was Corp Z - where a new CEO kept claiming he was the founder, and the rest of the board refused to challenge this un-truth. I did - and the CEO pre-emptively handed me a letter accepting my offer to resign. I signed it; the other directors were silent, and I had no interest in staying. A year or so later one of the directors approached me at an event and said he wished they had listened to me. Here I learned that there are some battles you cannot win as the only dissenter, and that it’s probably better to leave early than too late. I also learned to pay attention to the best way to spend - or invest - my time.

These “Corp” stories were not unique experiences, and if you think you know which companies they represent, rest assured that there are several possibilities for most of them. Through them, and quite a few more, I learned the basics of corporate governance and the necessary transition from founders, who make all the decisions and carry them out, to leaders who guide discussions, build consensus and delegate to a team who collectively know more than a single founder/CEO ever could. (And no, that team cannot be replaced by AI - especially not for the task of telling the CEO when they are wrong.)

CENTER BOX: The center box, with average learning and average returns, was a large part of my investing life - and of most people’s lives in many pursuits. If you spread the risk you can make a decent return, if you avoid the worst failures. But this was where I neither earned much nor learned much.

UPPER LEFT: These are the noteworthy financial winners that helped me offset the losses of my “learnings” and achieve or even beat the overall positive statistics of risk-spreading in venture capital (among 100+ bets overall).

Most relevant and interesting right now, I invested in SpaceX through an intermediary (like so many people, but much earlier than most). Though I can’t exactly figure out how much I own (let alone its valuation) because the intermediary thing is complicated, it was early enough that I’m pretty sure I’ll be a big financial winner - with luck a few million, though not trillions (LOL). Through this I got to meet Elon Musk a few times, but I had nothing to do with his success - however you define it! Likewise, I acquired early shares in Google and Facebook, through investments with funds such as Kleiner Perkins.

There were several more where I invested early directly with the founders as a small angel - Omada Health, Abridge.AI and Square - but they grew rapidly without any need for my help, so I neither contributed to their success nor learned a lot. But thank you, nonetheless!

LOWER LEFT: Here are the no-surprises negative outcomes: Invest a little, lose a little, but no real outliers. Fortunately, you cannot lose more than you invested - as long as you don’t try any clever tricks!

CENTER RIGHT: These companies helped me learn for free: a very small amount of money invested - mostly offset by director’s fees, options or office space - plus a lot of time. They also stand out because I was actively engaged with each for years (you do need to invest time as well as money in order to learn): WPP Group, Yandex and Meetup. I joined the board of WPP Group back in 1999, when it was basically just a few friends of the founder. At the first board meeting I showed up wearing my usual Silicon Valley uniform of blue jeans and T-shirt (though at least no hoodie); as I recall, the others all wore ties. This was my first serious board, and I learned a lot. I watched the company transform from a small, founder-mode concern to a larger, more bureaucratic outfit focused on the corporation rather than the whims of the leader. As the company grew, the board transformed as well: from that small group of friends into a formal affair, with committees, lawyers…and the usual tension. I was lucky to leave at the end of 2013, before things got really ugly with the awkward departure of de-facto founder Sir Martin Sorrell. One of the biggest things I learned is that term limits are a good thing - both for me and for Sir Martin. The world we started in had changed, and there were other ways and places for us to be useful.

I also learned a lot about the rest of the world from my perch at WPP. Suddenly I was on all kinds of invitation lists, and attended interesting events outside the tech world that I knew so well. We had six board meetings a year, typically a couple each in London and New York City, and one or two others in interesting places, with employees to meet and local dignitaries and customers to impress. Among those I remember best: Silicon Valley and Moscow (I felt like the host in both places!); Dubai; Cape Town, South Africa; Château Touffou (a castle in France near Poitiers); Cannes, France, for the annual June advertising week; Greece, Thailand and Indonesia for WPP’s ad-customer-centric Stream events; and a private tour of WPP customer De Beers’ main diamond mine in Botswana (where you get searched at the exit). Then there was the upstairs/downstairs experience at the London Olympics in 2012, which I later wrote about for LinkedIn. Best of all was that almost anywhere I went, I could find a local WPPer to introduce me to the local culture and give me a sense of what was going on locally. Over the years, I met country managers for Russia, Sweden, India, Russia and so many other places. While I didn’t make much money on the stock, the experiences and the education were priceless.

Then there are my Russian experiences, from 1989 to 2022. I got to know the country and especially its tech world intimately as an investor in and member of the board of Yandex (known as the Google of Russia). I also invested in a number of other Russian startups - solidly in the middle or lower-left quadrants - each with its own story reflecting Russia’s unique adaptation of mostly-corrupt communism into idealistic capitalism under a still-corrupt government. In the early 2020s, after spending some time contemplating Yandex’s transformation into a vehicle for Russian propaganda, I finally submitted my resignation at a meeting in Moscow the week before Russia invaded Ukraine, when the “special military operation” was still just rumors. (It took a couple of weeks for the resignation to be formalized with legal paperwork, but in retrospect I’m still proud of that date.) Meanwhile, the stock got frozen, and though it had appreciated, I had never sold a share.

Finally, there was Meetup, where I sat not only on the board but in the office for most of 2007 until 2019, about a year after it was acquired by WeWork. I saw up close - and once again - the challenges of a for-profit company devoted to serving a mission. For 23andMe the mission was (broadly) science; at Meetup it was community cohesion - and avoiding at least one board member’s wish to turn the company into a more lucrative dating service. From my seat on the board, I benefited from working alongside an amazing group of people dedicated not just to supporting communities of all kinds, but to training people on how to organize and sustain those communities. (That’s not something you get in the average high school, unfortunately.) At a certain point it needed further funding, so it sold itself to WeWork; sadly, it is now just one of many meeting-planner services.

LOWER RIGHT: This is the rest of the story of Yandex: The non-Russian part of Yandex got miraculously transformed into Nebius, based in Amsterdam. (Most such things in Russia happen in the passive voice, the result of unseen forces…) It resumed trading on the NASDAQ in October 2024 and is now one of the hottest AI stocks around. Someday I’ll go into the story in detail, but for now, just know I stayed on the board briefly before I resigned (due to Dutch governance requirements) - though I kept and still have all my Nebius (née Yandex) stock.

There’s one final story that has made all of this possible: Space Adventures, a small private company where I was a small investor. Its business was basically as a tour operator for trips on the Russian Soyuz spacecraft to the International Space Station. While NASA refused to let untrained people (except for a senator or two) fly into space, the Russian response was “Verrry deefficult….!!” followed by “How much?” For a few tens of millions, Space Adventures could arrange it. That was way beyond my financial capacity, but they kept inviting me to train as a backup…you never know! Finally, I decided to say yes, and for just a few million I got the chance to live and train as a backup in Star City for half a year, from September 2008 to March 2009. The lottery ticket of flying to the space station did not pay off, but I did get exactly what I had bargained for: training in space medicine and space plumbing, and a true immersion in Soviet Russia. Though things had changed dramatically since I first went to Russia in 1989, Star City was still a Soviet-style government facility; I needed a “propusk” - a government pass - to go in and out. As an American, I could not get a map; one of my worst days was when I could not find my classroom in one of the many nondescript buildings and arrived half an hour late. (Of course, the CIA knows the layout intimately, but that was not the point.) I would not volunteer to spend six months like that again, but I am so glad I did it once. Whenever I watch a space travel movie, it feels familiar enough that I have to remind myself I have not actually been up there. Some day - but not too soon! - I would definitely like to use my stock market winnings to retire on Mars.

UPPER RIGHT: But wait: One more thing! A totally different lottery ticket from Space Adventures actually did pay off. It’s really bad PR for someone to die on a space tour, so the Russians and Space Adventures together had every part of me inspected carefully, including a Russian-style endoscopy with valium only, plus a pleasantly anaesthetized confirmation courtesy of the Johnson Space Center. They found Barrett’s Esophagus, a precursor to cancer of the esophagus. No immediate danger, but after my return from Star City I followed medical advice and got checked regularly; in 2016, my US doctors found that I had stage 1 esophageal cancer. Stage 1 is easy to treat but is rarely discovered; without that space adventure, I probably would have died a few years ago.

That story reveals the most meaningful thing I learned from all these experiences - the 3rd-dimension Z axis of this chart, reflected but not visible: the importance of luck. In this case, I was lucky to earn a return of immeasurable value - some unknowable number of extra years of life. Indeed, I now focus on investing as spending on something or someone I believe in and believe I can learn from… but knowing that there may also be a lottery ticket attached. This explains the outlier from inverse correlation that you see in the upper right - where I paid a lot to learn a lot but got that return much later…through sheer luck.

I first heard about approval voting more than a year ago, from my friend Rob Sand. A former independent who registered as a Democrat in college to vote in a Democratic primary, Sand is now running for governor of Iowa and campaigning broadly to all comers. At town hall meetings he hosts throughout Iowa, he routinely asks those present to raise their hands in sequence if they are a Republican, Democrat, or independent - with applause for each. Then someone leads the crowd in singing the first verse of America the Beautiful. He wears a wonderful T-shirt emblazoned with

“Not bluer

Or redder,

But truer

And better.”

Read these words and smile…and then weep! We are so far from that ideal. How can we fix the country’s (and even the world’s) polarization? This challenge faces us while both US parties are engaged in an escalating re-districting war that will silence the voices of each state’s out-groups, instead of encouraging all voices to band together and flourish. This gerrymandering is structural change in the wrong direction.

What is the right direction? Sand’s campaign might offer a clue. He is appealing directly to all voters - not just to Democrats but also to Republican and non-partisan voters - with the message that we have a lot more that unites us than divides us. He says he’s loyal to Iowa rather than to the Democratic party, and part of his message to voters is about the need for structural change in elections: The current two-choice electoral system, he says, has led to a decade of one-party rule that has left Iowa families behind. In short, he’s running as if the rules of approval voting already applied.

To describe approval voting, let’s start with some context, beyond the current, rightful fuss about re-districting (which is further eroding trust in the whole system). The current rules assume a good side and a bad side - usually defined by two powerful political parties - and ask where voters and candidates sit based on that single dimension.

But the real world of people’s voting preferences is closer to complex, unique-to-them decision matrices - like the table below, with the factors and the way the voter assesses them unique to each voter. Approval voting recognizes this complexity, and is structured to find the single candidate that the most voters - each with their own decision matrix - approve of. It’s like a Doodle poll for candidates (vs. time slots): “Which of these candidates (or time slots) would work for you? Pick all that meet your criteria - whether only one, or two, or even most of them.” The candidate “approved” by the most voters gets the role.

For an individual voter, one factor might trigger approval of a candidate automatically… or an automatic rejection. But the more candidates a voter approves of, the more likely they are to help get one of their approvable candidates elected.

The process is akin to ranked-choice voting, but much simpler and more likely to earn voter trust: Each approval vote for a candidate has equal weight, making things easier to explain and to implement, without multiple rounds or complex calculations for either voters or election officials. Your approval vote for a candidate carries weight; your failure to approve of others is silent, for better or worse. In short, the system rewards both candidates and voters who compromise, and ignores the rest.

Why approval voting?

All this process underlies a fundamental issue: Politics in general - and voting in particular - shouldn’t be just about who gets a seat. It should be about how they compete, how they earn favor, whom they pay attention to - both during the election and afterwards. It’s about returning to a focus on policies and services, not promotions and surfaces (another clever rhyme!). It’s about building a coalition, not destroying opposition. In short, it’s about culture and the rules public servants should operate by….

This is the spirit animating Sand. He is firmly focused on Iowa, and so he should be: underfunded public schools (partially due to $8K vouchers for private schools), rampant corruption, and Republican one-party rule over a Democratic minority and silenced independents. His approach seems to be working: Sand leads all candidates in a recent statewide poll (conducted by a Democratic outfit) and has a reasonable chance of coming out ahead in the general election as a de-facto independent candidate. Enough Republicans and independents - some privately and some publicly - have indicated their willingness to support him in the general election, in effect registering their disapproval of most of the Republican slate.

In response to my question about a broader future, he replies: “I want copycats! It’s not just about me and Iowa. Other states have to elect people like me [i.e. coalition builders] for themselves…Of course, right now that’s most likely in the middle of the country, not on the coasts”…

For now, approval voting can work at the state or local level; it’d be too complex (dealing with the electoral college, for example) to implement in federal elections right now. But my hope is that approval voting would lead to other, broader changes. First, the path to getting elected would not be to excite the most fanatic, extreme followers, but rather to reach out and meet the basic requirements of the largest number of voters. The overall pitches would be more positive, vs. bruising battles that push people to extreme positions and to demonize anyone who disagrees. Second, the approval process would incentivize candidates to appeal to - and ultimately serve - the greatest possible share of the population, rather than to stake out an extremist position on a single good-bad line.

The idea is not to use approval voting to find candidates for each party, but to use approval voting to make elections run for the people rather than by the parties, empowering the entire population rather than filtering candidates through the tight funnel of party approval - or through gerrymandered districts designed for the same purpose. (Note that even when term limits exist, they apply only to people; there are no term limits on the parties, which tend to focus more and more on their own self-perpetuation.)

At its best, approval voting makes people feel that their leaders value their opinions and voices, not just their votes or party affiliation. Sand’s town halls are a great model: They are not sales pitches; they are listening sessions.

But most important of all, approval voting surfaces leaders who aim not to rule but to serve the people. A role as governor, or president, or town crier, is not a prize or even an honor, but a duty to fulfill - an obligation.

Just as the obsession with money is ruining our businesses, leading to enshittification (H/T Cory Doctorow), so are the rewards and incentives of our electoral system contributing to the enshittification of our political system and the culture it produces. You could call approval voting an honorable business model for politics: address the largest TAM (Total Available Market) instead of a tiny, fanatic niche that gets you territory no one else wants. And focus on satisfaction and retention, not customer acquisition!

What are the next steps?

If implemented in a number of states, approval voting could improve political culture locally and prepare both candidates and voters for its expansion to federal elections. Consider the wisdom of U.S. Supreme Court Justice Louis Brandeis who said in a dissent in New State Ice Co. v. Liebmann in 1932: “...a single courageous State may, if its citizens choose, serve as a laboratory; and try novel social and economic experiments without risk to the rest of the country.”

What will produce broad approval of approval voting? Beyond Sand’s hope to bring it to Iowa, it has been tried in a few places, including Fargo, ND (2020), and St. Louis, MO (2024). Both of these experiments…well, they were learning experiences more than amazing successes - more or less shut down by the party leadership in North Dakota, and continuing in St. Louis - but only for primaries, not the final elections.

But generally, the idea makes sense - and now it has a powerful advocate. Many voters in the country yearn for a focus on policies over parties; indeed, both major parties are coming apart at the seams. Gallup reports that over the last 30 years the number of adults who identify as independent has soared, reaching a new high of 45 percent in 2025, up from the low 30s in 1990. Meanwhile, Democrats and Republicans are evenly matched at 27 percent.

In the current environment, any fix would be welcome; a larger number of people who feel they had a positive say in the outcome of an election has got to be a good thing, as would a system that helps good candidates prevail and avoid getting corrupted/mangled by it once they get elected.

Approval voting would help the system attract more people like Rob Sand and fewer like… well, the whole point is that I’m not going to name names. I’m encouraging more people to follow Sand’s lead. And yes, I have donated to him and his broader cause, even though I am a New Yorker.

Spoiler reveal - You may or may not have noticed, but I carefully wrote this whole post without using the word “win.” No, I don’t delude myself that we can totally eliminate it from politics, but I would still prefer to see it relegated to sports, gambling and TV contests. For more information on approval voting, see https://electionscience.org/ and https://ballotpedia.org/Approval_voting. For more on political trends, https://news.gallup.com/poll/700499/new-high-identify-political-independents.aspx. And for more on Rob Sand, https://robsand.com/issues/accountability-for-all/.

A critical and timely question: How should we regulate AI?

I’ll walk the talk and start with a disclosure…since the best and most honest way to gain trust is to be not just truthful, but transparent. So, in this post about one way to contribute to the regulation of AI, I disclose that I’m an interested party: not for money, but as an advocate and a named advisor with some learned experience. Of all the proposals and ideas out there - of which some large number ultimately must collaborate to succeed - one that is now in formation, and where I am best and uniquely equipped to be useful, is the Agent Community Foundation (ACF). It aims to help regulate the integrity of public-facing AI agents by inviting their owners to list them in a hoped-for new .agent registry - in the same way that websites are - under the purview of ICANN (the Internet Corporation of Assigned Names and Numbers). The owners of the AI agents registered would then be subject to certain requirements around transparency and liability.

Unlike websites, which are merely a portal to data and services (which could be agents, of course), the AI agents linked to by .agent would actually initiate actions, negotiate with people or other agents, and can change the world around them. This raises some questions: On whose behalf are they acting? Who guides their instructions and decision-making? AI agents in particular are difficult to control because their primary use is to work across and outside silos - and to negotiate with other parties rather than to follow rigid instructions set from above.

AIs cannot be “regulated” in the same way as humans, through moral and legal threats. Humans are finite - we fear death - and each of us is unique, whereas AIs can be duplicated and changed, and do not experience fear of death or moral jeopardy. You can’t just throw an AI into jail or even sentence it to death, although you can turn any single instance of it off. It won’t suffer - though of course you can train it to simulate suffering.

The best way to enforce accountability for an AI’s actions is to link it to people (often through the organizations they control) who can suffer and be held accountable. In the human world, we pay a lot of attention to verifying and tracking individual people - sometimes far more than is comfortable, and for the wrong reasons. Regardless, that practice underlies the principle of individual accountability. You - Jane Doe - are responsible for your actions. That principle is often messy in practice (e.g. Jane might come up with some exculpatory explanation in a court of law), but it’s the best we have come up with for humans.

AI agents likewise need regulation and certification, but not by the entities that depend on those AI agents for their profits. So, while I appreciate the calls for regulation from OpenAI, Anthropic, a16z and other AI players, we need a coalition of engaged, knowledgeable and unconflicted parties from outside the heavy hitters of the industry to set those rules and monitor compliance. Balazs Nemethi, a techy with startup experience in the worlds of blockchain, identity and KYC (“know your customer”), founded the Agent Community Foundation to create such a coalition. ACF’s mission statement says: “.agent should be to AI what .edu is to education: community-governed, vendor-neutral, not controlled by any single entity” with strict compliance requirements. Nemethi has the experience to lead such an effort, having served as the director of the Decentralized Identity Foundation, an initiative of the Linux Foundation, from 2020 to 2023.

ACF’s first big step, other than self-assembling its coalition, is to apply for the right to lead and operate a potential .agent domain name registry - along the same lines as the website registries for familiar top-level domains (TLDs) such as .com, .org, .cn, .info, and .ai. Those registries are legally controlled and assigned by ICANN, where I was founding chair from 1998 to 2000.

So, people have asked me, shouldn’t ICANN itself - international, purportedly bottom-up, etc. - be a model for AI governance? My reaction is complicated. Yes, ICANN provides useful infrastructure for registering domain names - inactive bits of text linked to websites - but in terms of regulation it is pretty laissez-faire. It leaves the details of registrants’ identity or activity to each registry. After stepping down as chair, I remained engaged as a board member and advocate for ICANN’s At-Large Advisory Council (ALAC). ALAC was supposed to represent the public interest, but had difficulty eliciting much engagement from the thousands and now millions of people who use domain names, as those names are hardly a key element of their business or personal lives. More recently, I joined the (successful) fight against ICANN’s attempt to sell the .org registry to a private-equity firm - another long and complex tale.

In a few cases - .gov and .edu, for example - the registrants of a particular TLD need to meet certain criteria, such as being a government office or educational institution. But ICANN does not impose such restrictions on most registries. For example, .org domain names signify nothing - they are available to anyone who will pay - though the original idea (and the marketing) suggest they are for nonprofits. More fundamentally, and laughably, some of the most profitable and interesting domain names are purely commercial efforts assigned by ICANN to various governments and then re-purposed for country-agnostic commercial purposes (though they fund the holding country). .ai, for example, is assigned to the island of Anguilla and provides a magical source of revenue for this small, British island protectorate; likewise, .tv is assigned to and generates revenue for the island nation of Tuvalu.

Nonetheless, ICANN’s infrastructure offers a scaffold for a well-managed registry to build the beginnings of a robust accountability and signaling mechanism for public-facing AI agents. This is where ACF hopes to step in. The domain name system itself is not especially political and is passive; it does not generate content or perform actions. And thus it’s a good host for effective regulation, managed not by ICANN directly but by a designated registry that could impose a variety of governance rules. In actual function, such a registry would be similar to the US SEC (Securities and Exchange Commission) - which ensures a level of accountability and transparency for public companies through external audits and the like - but for “public” agents. (The analogy to public companies vs. the shadowy world of private equity is clear.)

Last week, ICANN formally announced an application window for new TLDs - the first since 2012 (!). In this widely anticipated round, ICANN is offering a newly enhanced “Community Priority Evaluation” (CPE) process for supporting community-based organizations in bids for certain domain names, including .agent. Its new call for registries prioritizes CPE applicants (vs. regular for-profit applicants) by allowing them to avoid the normal bidding process for standard, mostly commercial registries. ACF hopes to win its role as .agent registry under CPE, since it lacks the financial resources to win in a standard auction. So far, no one else is known to be applying for this TLD under CPE, though both AI tools provider Vercel and crypto outfit Unstoppable Domains and likely others may join the auction if one happens.

Meanwhile, if ACF’s application does win approval, it will be allowed (and then committed) to set and enforce rules around the use of .agent domain names. This won’t solve all the potential dangers of agentic AI - nor will any single regulatory effort. But it will offer the opportunity (if done right) to set up a visible, operating exemplar for one important aspect of the many facets of AI regulation: a system for honorable, public-facing AI agents to be registered and authenticated, with a strong regime of “Know Your Agent,” as carried over from KYC in the worlds of banking, crypto and other trust-dependent and risk-rich markets. Its remit is a combination of provenance - What’s inside something and where does it come from? - and accountability - Who’s accountable for the performance and integrity of what is being sold and used?

The suffix .agent, for agents sold mostly to individuals and small businesses, may not serve a large part of the overall AI/agent market financially, but it’s likely to be the most visible consumer- and public-facing TLD for agents. Larger, B2B customers will likely have their own security and direct contracts with the owners/sellers of the AI agents and software they use; there will be more direct accountability and larger revenues, and more lawyers and legal bots involved. By contrast, the .agent agents will of necessity need to be fairly locked-down, and that’s a good thing. The whole idea is to limit their ability to do harm….

The CPE applications are due August 12, and with luck ACF’s (along with others for other TLDs) will be decided - and granted! - sometime in 2027. If ACF does not succeed and the right is instead auctioned off, the winner would probably be one of, or a coalition of, the interested AI players including the ones mentioned above. They, says Nemethi, “would likely then control the naming layer for every AI agent on the internet: discovery, trust signals, registration policies, pricing. Any organization building in the agentic space would operate under that entity’s terms” which might be quite loose, or even favor some particular slice of the market.

ACF has already gained support from a large number of AI players, such as Hugging Face, Replit, Alibaba Cloud, Oracle, Datadog and Netlify, as well as luminaries such as Tim O’Reilly, the founder and CEO of O’Reilly Media and founder and co-director of the AI Disclosures Project. But (to no one’s surprise), no one from OpenAI or the other big-name players has yet signed on. At least some of them are considering supporting ACF but are still consulting with their lawyers (and investors too, no doubt).

ACF cannot and will not raise, say, $60 billion from “interested parties”; instead, it wants to grow bottom-up and genuinely represent the interests of AI agents’ builders and users rather than of the underlying developers. It wants to put in enough regulatory friction to represent and support the honest players - those who ultimately will benefit from a trusted, transparent AI ecosystem that rejects the bad guys.

Personally, I have signed on to declare support for ACF’s application and to serve as an advisor. I see my role as urging serious governance, whether to ACF or whatever other leaders will listen. Beyond a variety of obvious measures, I advocate that a .agent name should require some kind of liability insurance for any accountable organization or person registering an agent. Yes, that makes things more expensive and adds friction - good friction. I’d trust the judgment of an at-risk insurance company over that of a marketing firm. Otherwise .agent would be just another TLD among an exponentially growing crowd of unvetted…what? people? scammers? rogue bots created by vibe-coders in fits of “creativity”? lonely agents programmed to act as companions?

This is the kind of targeted approach to AI regulation needed in so many situations - as opposed to grandiose, all-encompassing laws and requirements that are likely to be rigid and out of date even before they are enacted.

So here’s the ask: ACF is preparing its application for ICANN - and looking for both financial support (the application process alone costs at least $500,000, including fees and lawyers and the like) and endorsement by everyone from would-be builders and users of .agent agents to validating bodies and would-be regulators. What we need is your help to form and guide such an active community of builders and users, not of “interested” vendors and related businesses.

You can join the Foundation as a member if you are interested in helping to set up the good friction and rules you want for yourself and others to observe. We invite you to join and help set up working groups addressing issues such as agent identity (protocols), agent trust criteria and standards, community (membership and outreach), and of course grants and support. Every new member helps with our application to ICANN, and your active involvement will help the mission itself. So please visit agentcommunity.org/why-join and sign up now; it “takes <1 minute,” promises Nemethi. But do commit some time, and temper idealism with skepticism. Defining and ensuring integrity is a complicated business; there are a lot of shady ships in the sea, and one bad actor, enabled with AI and automation, can bring down a lot of worthy efforts.

As for me, I look forward to learning a lot and helping to scratch a small, positive dent in the universe.

We - the AI collective and its more aware users - spend a lot of time wondering about LLMs. How much do they think like humans? Can they actually start reasoning, like us? Are they like that little girl (me) in the photo above, ready to start moving from parrotting big words to understanding their meaning?

But perhaps we’re asking these questions in the wrong direction…. as I’ll explain in a moment. First, let’s explore the “context window” - the shared understanding that makes a brief conversation, or a prompt and response, meaningful to both sides - for these questions.

Yes, LLMs are indeed improving rapidly, because AI developers are adding reasoning “adjuncts,” helpfully defined by Google AI mode (using Gemini) as something added or connected to a larger, more important thing, but not an essential part of it. Rabbit hole alert: That definition prevails because the people who (formerly) created dictionaries care a lot about grammar, and this definition is grammar-centric: about words or phrases that add a little information - such as the phrase “such as” followed by an example - but don’t change the overall structure of a sentence. But in the case of LLMs, the definition of “adjunct” changes; it often adds the more valuable part. The LLM is merely the front end to a complex but fundamental situation or analysis for which the adjunct provides the necessary accuracy and nuance.

When an LLM encounters certain sequences or collections of words, it might call out to an adjunct to handle a math problem, explore some logic, observe a specific date range, check on safety rules.... The adjunct adds specific knowledge or rules that enable the LLM to incorporate complex notions such as how gravity interacts with the density of an object, how a sleeping person might react differently from one who’s awake, how things might change over time. Indeed, the whole point of the rabbit hole above was to show the value of digging below the surface in order to understand the context behind a “simple” answer.

Meanwhile, LLMs are now running (for higher prices) in ever larger “context windows” (also known as “memory adjuncts”), which enable them to collect and hold in memory the details of underlying specific interactions with specific users (including other AI agents) or data. Those windows are the amount of context an LLM can collect and hold in memory as it interacts with you, or another agent, or anything else. In the beginning - just a year or two ago - a typical context window was just a couple of prompts and queries: the equivalent of a few pages of text. Now it’s increasing to 1500 pages or more - the size of a standard term sheet with exhibits, but still just a small part of an investor data room holding financials, projections, risk factors and the like.

That’s what’s happening with AI. What’s more important, I believe, is how this relates to humans - except that we are changing in the opposite direction over time. Increasingly, we are passing more of the burden of exploring various adjuncts and context windows over to our computers. We are also missing the value of such exploration. I learned more checking my facts for this Substack than I did writing it (H/T my wonderful, inquisitive editor, Christina Koukkos!). We humans are no longer taking the time to experience and understand the context for what we should know – by reading long documents, comparing alternatives, or “reading” the people in our lives by having deep conversations. A quick text is good enough.

Ironically, we are getting better at thinking like the ancient, adjunctless LLMs of just a few years ago. Have you ever spent an hour or two listening to people opine on a topic you know nothing about? If you don’t spend the whole time scrolling on your phone - you wouldn’t ever do that, would you? - you’ll probably walk away able to offer a surface-level comment or even venture a few opinions. “Yeah, the artistry was amazing”! “I liked the way the moderator put that guy from Stanford back in his place.” “GDP growth was really disappointing!” “But the Dow’s at 50,000!” “You have to keep thinking about the Iranian people, not just the weapons.” But what do you actually know about the Iranian people - their history, their complicated politics, their points of view? What part of the GDP grew fastest? Was it useful things, like more housing, or was it increased spending on remedial healthcare, private prisons and prediction markets? What superficial summaries did you collect but never quite explore with your brain’s equivalent of adjuncts - the stuff you actually had to study at school or through some other training?

Whether scrolling through TikTok or listening to pundits punditing - or perhaps both simultaneously - we pick up the vibe and the words, but not the meaning. Kids in school are interacting with their phones, not with other humans. Those working from home are scrolling on their phones as they attend (not participate in) Zoom meetings, and scanning the Fireflies summaries later. TLDR and “clipping” (thanks for the clip, Semafor!) are the terms of the day. We are thinking like LLMs - fluidly, easily, but without understanding the underlying complexity. Even as our computers understand us better, we are beginning to understand ourselves less. The context windows between ourselves and our friends and colleagues - developed over time, through direct interactions and working through problems together - are shrinking.

The people sounding the alarm the loudest are teachers, who see their students failing to understand the people and the situations around themselves. In so many ways, they are failing to learn what it is to be human, to be conflicted, to want to ace the test but also spend the time to get more likes on TikTok. It’s more than understanding how things work out; it’s looking under the hood to understand why. What’s the underlying motivation of the people I’m talking to? Are they trying to manipulate me? Or, as when I was in Russia, what were people not saying? As AI legend David Waltz once said, “Words are not in themselves carriers of meaning, but merely pointers to shared understanding” - in short, a context world - not just a window.

Indeed, one reason I can make this Substack relatively short is that you already know this stuff; you just may not be paying attention. And by not paying attention here - and there, and everywhere - you may ultimately lose the capacity to do so. So, formally, I’m suggesting that you PAY ATTENTION. Don’t get stuck inside your own opaque “context window.” Don’t just “consume” ultra-processed information; process it yourself. Just as food slop corrodes your metabolism and often results in diabetes, so does ultra-processed information corrode your mental metabolism and give you “information diabetes.” Don’t outsource your understanding of the world to AI summaries and 30-second videos; experience the complexity and see the outliers that don’t fit the quick summaries and the norms that AI reverts to. You just might assemble a new truth that AI summaries would dismiss or disregard; most scientific discoveries (the roundness of the Earth, relativity, evolution and the like) were once heresies.

In the end, we are rightly interested in not “missing the forest for the trees.” But to truly know the forest, you need to wander among the trees, smelling the leaves and tripping over an exposed root or two.

You probably noticed the two amazing victories against predatory business practices last week, when two juries in Los Angeles and New Mexico ruled against YouTube/Google and Facebook/Meta for harming and endangering children by design. Let’s acknowledge the irreversible devastation already wrought, bask in victory for a moment … and then figure out how to move forward, away from filing lawsuits and toward preventing the damage in the first place. Spoiler alert: The answer can be found in the most boring of market sectors: insurance. But first, let’s briefly explore what happened.

Instead of a vote by a broad public for a TV show, the juries each comprised 12 thoughtful adults committed to listening and learning in order to discover the truth. They read documents and listened to testimony from experts and company executives and whistleblowers, and probably considered their own experiences. They weren’t legal scholars grounded in section 230 of the Communications Decency Act, which limits the liability of online platforms (i.e. passive websites) for content posted by users, and that disregards the role that underlying algorithms and affordances might play. They were normal people, with a sense of ethics.

Certainly, one of the most satisfying things about the recent verdicts was the use of appropriate analogies to describe what was going on. The companies tried to present their case as freedom of speech and consumer choice/autonomy vs. over-regulation and restriction of children’s freedom. Hard to argue with, right?

Maybe. But once you see behind the curtain, perhaps it’s more like cigarettes, opioids and - MAHA alert! - ultra-processed, metabolism-damaging food. Just as sugars (including artificial sweeteners), ultra-processed foods and additives mess with your metabolism and physical health, so does ultra-processed content mess with your mental metabolism, reducing your ability to process content while at the same time luring you to “consume” (not to interact with) more and more of it. Call that condition “information diabetes.” And now, we face ultra-processed AI companions and advisors - usually luring you into something that’s bad for you, and driven by a business model that wants you to be addicted rather than challenged. (They call it “customer retention.”)

Those analogies and metaphors won’t go away, precisely because they are so apt. Social media offerings and now chatbots and AI agents are not (free) speech; they are services and tools, run by algorithms and with personalization, akin to drugs or medical devices that mess with your mind.

So, we won a few lawsuits. Where do we go now? Is there a way to fight in advance, instead of waiting until the damage happens, when it’s too late to do anything other than sue to collect money and forced apologies? The judgments awarded were $6 million (LA) and $375 million (New Mexico) for what could - should - be just the first of many such trials - or settlements. But who would take money in exchange for the life of their child? That question is almost too horrifying to ask. Yet every day all sorts of companies are prioritizing money over the lives and well-being of children and adults around the world.

Let’s not even try to quantify the value of human lives … but now we can quantify the likely damages if you lose a lawsuit. The suits against Google and Meta follow well-worn paths established with Big Tobacco, and more recently with opioids and the Sacklers. When you bring in liability, suddenly the whole set-up changes. The Sacklers and their company Purdue Pharma faced their first lawsuit (for deceptive marketing) back in 2001; many lawsuits followed. In 2019, the company filed for bankruptcy to evade the liabilities. Late last year, the lawsuit was finally settled, for about $7 billion, to be paid off over the next 15 years. (Note: In that case, the Sackler family owned the company; in these new cases, the liability would be held by officers and directors of Meta et al.)

But here’s one big idea - a different analogy - that might help things move faster this time around: fire insurance, which doesn’t wait for the damage to be done.

Take what happened with fire protection of buildings in the early 20^th century: Fire insurance companies and their inspectors led the way, ultimately joined by the government, in establishing building codes and formal regulations requiring building owners to buy fire insurance. The message was clear: If you have the capacity to harm, you have the obligation to avert the harm, or be ready to fix or pay for that harm. This led to a huge and sustainable business focused on safety-as-a-service (well before such terminology existed): Insurance companies diligently inspected properties to ensure that the building codes were being observed, that oily rags were disposed of, that emergency exits were accessible and opened from inside…and to notice things early rather than wait for an infraction. Following their own financial incentives, they went beyond formal fire codes in making their insurance projections and premium-pricing decisions.

In short, the best insurers of all kinds - fire, home, cybersecurity, general liability, etc. - reduce risks by inspecting the premises, digging deep into records and pricing coverage justifiably high…or simply refusing coverage. The question is not just how you fill in the forms, but how you actually behave. What are the tells, whether dirty rags or termites…or disaffected employees leaving at high rates (with non-disparagement agreements)?

In short, what we need is not a huge and quickly-out-of-date regulatory regime, but a few simple rules that mostly mimic what already exists. The industry itself should ask Congress to pass a law requiring social media/chatbot/AI agent companies (and their accountable senior officials and board members) to purchase liability insurance in order to offer their products to the public, even - or especially - when they are free. (That’s in the industry’s best interests overall, because it will keep out, or at least deter, the bad actors who might win without guardrails and who will tarnish the whole industry.)

Then the SEC should require any public social media or AI company to disclose their insurance payments - which private company investors will be savvy enough to ask about to protect themselves. Ironically, collecting that information is a good task for AI operated by the insurers: inspecting not just fire doors and oily rags, but searching through the kind of internal documents and operational tells that eventually got the Sacklers and Meta and many more on the hook. They can assess all that “sentiment analysis” valuing profits over customer lives, or discover an internal conversation (true story revealed in court!) where one Meta researcher wrote “IG [Instagram] is a drug,” and a colleague responded “LOL, I mean, all social media. We’re basically pushers.” That’s a better indication of risk than corporate mission statements.

So, if you know any insurance people, get them involved - or send them my way, please! They have a huge role to play here. The goal is not to win a lot of lawsuits, but to assess the risks that lead to them. Use the need for insurance not to spread the risk, but to prevent it. Good insurance companies are known by the risks they turn down, by the unworthy customers they refuse.

Long ago, on a summer Sunday, I was walking to my pool - the Downtown Athletic Club - and randomly took a path on the walkways beneath the towering World Trade Center.

A month later, I was in Bulgaria, at a conference about reconstructing the governance of Eastern Europe. The participants were wise people from the West - politicians, lawyers, activists and journalists - plus their counterparts from the East - would-be politicians, lawyers, activists and some existing journalists - eager to learn and to build a new world of transparent, freely elected governments; sensible laws; and fair courts and judges.

After hanging out a bit with the participants - what an amazing group of noble, altruistic people to be part of! - I went to a small, windowless office to check my email. My business partner in New York City had sent me a brief note: “We are all okay!” Weird, but whatever! There was more important stuff to read. Someone popped their head in to tell me that there had been an airplane crash in New York. Sad, I thought. But whatever! Then I checked the news: The World Trade Center I had walked through a month earlier was on fire and pouring out smoke.

All afternoon, both in that little office and back in the crowd, people came by to console me, to express concern and hope. It was about the best place in the world I could imagine being in at that moment: With people who had lived through dictatorships, corruption and censorship, and hoped to build a better world. People who believed in the power of the Internet and freedom to unite us all. People who were actively working for that better world - running for office, writing laws, working to uncover and document the past so it couldn’t happen again.

The next day one of the women attending and I started to talk. With a little encouragement, she said what others would not: “We are really so sorry for you guys. You mean so well; you have such good principles. But really, you have no clue what it’s like for the rest of the world. You’re like a rich kid living in an elegant mansion on the hill. You’re nice and generous. You invite us to your birthday parties. Your butler treats us with respect.

“But you have no curiosity about our lives. We live in a different world. We’ve had the Irish troubles; we had World War II and the Iron Curtain. Our countries were bombed and some of us were sent to prison camps. And of course many of us…were lost. We suffered under dictators like Stalin, Tito, Mussolini. So we wish you well, but we wish you just understood our reality!”

After the conference, I kept to my original itinerary, and stayed in Europe. There were long security lines at the airports, and the flights back were all canceled anyway, but everything else was in order. As for me, I felt odd. I was just glad I wasn’t in New York. Fortunately no one I knew was missing, and I wasn’t needed…the street where I lived, 12th Street, was within the residents-only restricted area authorities had established to keep crowds away from the wreckage of the Twin Towers; I would need some kind of pass to get home.

When I did come home, I returned to devastation. For the next few weeks, you needed a pass to cross the barricades blocking everything below 14th Street. I found a replacement for the Downtown Athletic Club at a smaller place nearby, and for six months I swam there, just blocks from the former WTC. The whole area smelled like a damp fireplace - which we now know probably caused lung cancer for many of the rescue workers and for others less directly involved except by location. Every morning, when I arrived at the pool for my daily swim (before the place fully opened up), I saw volunteer workers serving coffee to emergency workers in the lobby. One thing was clear: What you could destroy in minutes, you can fix only in years.

This all came flashing back to me when Iran started the war. (Just kidding!) I thought more about the wisdom that woman in Bulgaria had shared. I imagined the US as a rich, teen-age kid, welcoming, but oblivious to the lives of others outside his family’s mansion. The butler politely treats the visiting kids as if they belonged.

The gate in front of the mansion on the hill is high, with a guard house nearby. But things are starting to crumble. The foundation of his mansion is rotting, succumbing to mold and neglect. You can now see the beginnings of collapse. The mansion is crumbling and falling apart… The lights are on in some windows above, but there are weeds below. The foundation is rotting at the seams, succumbing to mold at the edges. To the side, you can see a stream of sewage from under the wall running down the hill, with a few broken syringes, some fancy torn clothing, a discarded Meta Quest 3 headset.

And it’s the kid’s fault. He lives alone with his staff. Somehow his family, who built the mansion, are gone.

From the outside, the kid is no longer seen as oblivious, but at times vicious and predatory, and at other times like a kid throwing his own toys at the wall. Once respected at least by some, he is still feared, even as his words are disregarded…the same as AI slop, but coming from a human. Perhaps that kid, obsessed with AI, thinks it’s easy to reprogram things, press a button, and start over. But society - along with buildings and other infrastructure - isn’t programmed. It has to grow, like any complex, living thing.

When will we wake up from this nightmare? And when we do, what can we do when we’re awake? Will the damage be beyond repair? Or can we learn enough to grow back better?

And what if the relationship in question is not between you and another person, but you and a thing - food or money or AI? It would help us all to heed this relationship truth in our interactions with things. Indeed, I’m writing a whole book about how to be human in the age of AI. While many people are writing books about AI from all sides, I’m more concerned with people and their relationship with AI….which may require some counseling.

The trick of having a good relationship with AI is understanding it - and its limitations. It is to avoid being seduced by it or misled by it…but also to avoid blaming it. For a healthy relationship, you need to understand what business models or other motivations lie behind the company and people that created it. And then you need to figure out how to get it to do what you want.

Before we get to AI, let’s explore couples counseling between people and “things” in other arenas. I like to half-joke that I have invested in several couples-counseling companies - one about money and the other about food. These apps help you recognize that food and money don’t need you or shame you; they are just inert, passive objects to consume or trade. But you react to them!

The first is TheBeans.io, as in “counting beans.” Its mission is not so much to help you manage your money, as to prompt you to consider how you think about your money. Do you see everything as a trade-off, or do you focus more on value for money? Do you buy things you don’t really need, when what you actually need is an expression of love from your partner or gratitude from your boss? For example, why do I blithely spend $300 on a hotel room but think twice about a time-saving Waymo ride for $20? Partly, it reflects my relationship with money through life-long training. I started paying for hotels long ago, so I don’t much notice spending that money. But spending money on a private car hits different somehow; I don’t drive and have always defaulted to subways and buses and lifts from friends whenever I can. A couples counselor might suggest that I value the time lost in transit too little!

The other is EatMyWay.com, about you and your food. It helps you examine your feelings and habits around food, often based on early training: Do you automatically finish your vegetables first and save the steak as a reward? Do you eat so fast you can’t actually taste the food? Does certain food call to you? Does it shame you for wanting it (the way your mother used to)? As for me, I know I sleep better if I don’t eat dinner, but the value of time with friends (or business contacts) at dinner often outweighs the missed sleep. Relationships often require trade-offs - but ideally, I can make them thoughtfully.

Then there’s a very different example that shows that couples-counseling techniques can be used - usefully and carefully - to create a fake relationship where none existed. In this example, it is done in a way to make it explicit that it’s not a real relationship, even though some might find it creepy. As related by Stewart Brand in his recent book “Maintenance: Of Everything,” the US Army was trying to solve the problem that soldiers in the Vietnam War weren’t taking sufficient care of their poorly designed M16 rifles; those rifles were routinely getting gummed up with dirt, backfiring and otherwise harming the soldiers themselves instead of the enemy. So, the idea was to create a comic book that nudged the soldiers to treat their guns like lovers. A bit icky, but transparent and bold enough to be funny rather than manipulative. And it probably leaned in effectively to what was on soldiers’ minds anyway.

As Brand recounts: “It was a jokey 32-page comic book drawn by the revered comic artist Will Eisner. Its first three pages, titled ‘How To Strip Your Baby,’ illustrated the nineteen steps it took to disassemble the M16. The text began with flirty language: ‘You want to know her inside out, every contour and curve, every need and whim, what makes her tick’ ” …and outlines how to manage a relationship with that pesky, breakdown-prone device.

What we need now is a class of couples counselors mediating between us and AI; understanding their counterparts in these human-AI partnerships is vital both for individuals and for society as a whole. Such an AI counselor should start by helping you see clearly what it itself is: Not a companion or a participant with an agenda, but rather a metabot that looks out for you vis à vis other bots and AI affordances. How can we develop and sustainably fund such a genuinely useful and disinterested tool? A good therapist doesn’t automatically support the client’s worldview, unlike so many sycophantic, seductive chatbots. A good therapist starts by asking questions “as an observer,” not suggesting answers. And they focus on you and your needs, not on the “other” that you do not control. As in “real” life, you need to know who is friend and who is foe.

Your counselor should be a trustworthy AI tool that is genuinely on your side in a world riddled with tools and bots that claim to help you. It should remind you not to automatically believe what another AI or AI-based business tells you. It could help you ask important questions and tell you awkward truths about your interactions with each of the various AIs you deal with: Is the AI a tool you control? Or is it working for some person or company trying to evoke some behavior on your side: buying something you can’t afford on credit, watching an ad, renewing a subscription, among other possibilities? Such AI bots may be trying to manage your behavior to keep you paying; They are not predators; they are parasites. The couples counselor AI would remind you to avoid succumbing to the sycophancy of a bot that offers you short-term satisfaction by always taking your side - and lures you into long-term addiction to behavior that you will regret.

The whole point is for the person to manage their own behavior and reactions rather than to engage in yet another fraught relationship. There is nothing “there” on the other side. With people, yes, it’s about another person. But with AI - it’s only a tool. Who is using it? You…or someone else.

Somewhat surprisingly, the awkward reality is that having an AI bot (not a human) ask awkward questions may often be the best response to our current situation of stressed-out people and scarce human therapists. As Peterson Health Technology Institute concludes in a discussion of this issue: “Patients are counting on us to lead with guidelines that promote access and better clinical outcomes. Expanding access [through AI] without appropriate safeguards isn’t supporting innovation - it’s irresponsible. At the same time, failure to act isn’t caution - it’s a choice that serves no one.” Yes indeed…as long as these AI therapy bots are built to purpose, trained on vetted therapy techniques, and not generic bots like Claude or ChatGPT!

And FWIW, please consider this Substack post as coming from a bot (though I am a person!). In the end, I’m not sure whether this is a political/social problem created by lack of trust, or a mental-health problem created by the loss of human interaction, or both. Consider me your first counselor: All I have done is ask the questions. It’s up to you to find the answers!

I have been thinking a lot about friction recently, in this age of optimization and efficiency. I first learned about the challenges of optimization when I worked with Federal Express as a securities analyst back in the late 70s and early 80s, when the company was just getting started. They were on a mission to be optimized, but learned the hard way that the more efficient you are, the tougher it is to respond when anything messes up the carefully optimized setup. In other words, when friction intervenes, you need alternate paths and equipment to route around it.

What I did not yet understand was the value of friction itself. It can give you time to reflect, to appreciate the unexpected, or to connect with people you might never have met. In this age of AI, people seem to be rushing through everything, “consuming” summaries of books or articles they never read or contemplate, asking an app to choose their wardrobe, and relying on an AI agent to plan their vacation travels. Why not just tell the agent to go itself and send you a summary? Once the robots get agile enough, should I use one to swim for me every morning? Just kidding…or maybe not. I have seen ads for bots that can answer friends’ questions online, saving kids “the awkward interactions” of communicating for themselves.

So a couple of weeks ago, this observation by Anand Ghiridaradas, as a guest on Ezra Klein’s podcast, caught my attention (about 20 minutes in):

“[The Epstein story] is revealing how these elites make decisions about trust — that I think are really different from the way folks at home go through the world and make decisions. I think you make character judgments about people, about how honest they have been and therefore will be.

These billionaires, these superelites, these superlawyers are working on a whole different kind of system. Their system has to do, as you say, with how loaded with connections you are in this network, how high your stock is on a given day in this network.

What Epstein figured out was how to game this. He figured out the vulnerability of this entire network, which is that these people are actually not that serious about character. In fact, character may be a liability for some of them, may be an unnecessary source of friction.”

In short, friction is not inherently “bad” or “good.” Sometimes it should be optimized away, like interacting with an airline booking system. Other times it can be useful. Or even delightful. In order to really see the forest, you must walk among the trees, trip over exposed roots, and smell both the fresh leaves overhead and the fallen leaves rotting underfoot.

Efficiency may get you where you were going, but perhaps the journey itself is the point. Last week was busy for me; my Out-of-Office response was headlined JFK-SFO-DCA-MCO-LGA - a very efficient way of describing a five-city marathon! Fortunately, I had left New York before the big storm on Sunday night; see a screen shot from my favorite Schadenfreude site, above. But a different “exception” caught up with me.

The diciest leg of my trip was San Francisco to Washington, DC, to attend a mere two-hour soirée at a noisy restaurant. This evening gathering was just a pre-meeting for the main event, on philanthropy, the next day…and was the only part I could join in person. I had to be in person at another event in Orlando the following day. Even a relatively short flight delay – just a small bit of friction – would have rendered the whole excursion pointless.

My flight was scheduled to arrive in DC at 4:30 for the 6 pm event. Miraculously, I got in early and had a great evening! I was glad I had taken the risk, and it was worth getting up at 5 am the next morning to make my 7 am flight to Orlando for a healthcare event. My efficiency bet had paid off!

The reason I took the risky flight was to argue – in person! – for the importance of human connection – i.e. friction – in philanthropy. True charity is not about allocating your dollars wisely (although that certainly helps). It’s about human empathy and connection, being present. You can’t outsource your own attention.

There’s no way I felt I could make that argument effectively from a distance. Done right, charity serves both the recipient and the donor. It’s not - or should not be - a question of the size of the letters on the hospital wall or which table you get at the gala. It should work in a way that gives you satisfaction from the giving, not from the transaction. When’s the last time you gave some friend or relative some object you liked, because you could see that they would actually love it far more than you did? At the DC soiree, by talking and eating with people dedicated to helping others manage their giving, I was modeling this idea: connection over efficiency.

The next morning I awoke to a text that my DCA-to-Orlando flight was delayed by 3 hours. Of course, being very efficient and a good planner (LOL), I was able to use my “swimming insurance” - an Equinox membership - to take my daily swim just a short Uber ride from my hotel. I had a wonderful swim and then walked to the nearby Foggy Bottom Metro stop to head to the airport, lugging my huge suitcase plus my backpack and a cloth bag holding my wet bathing suit and other treasures I’d need on board or too valuable to check - contact lens case, hairbrush, a couple of expensive wearable trackers, and the like.

I had just missed a train, but I had time to spare (it was only around 8 am and my flight was scheduled to board a bit before 10). A sign showed the next train due in six minutes. I sat down on a bench to check my email, with barely a moment before the next train came in. I quickly gathered my stuff, including the heavy suitcase, stuffed with books from an AI event in Silicon Valley I had attended over the weekend, and entered a crowded train. At the next stop the train emptied and I found a seat with space to put my backpack and cloth bag down - and realized I had only the suitcase and the backpack. Could I have left it behind? There was no other place it could be. As we entered the next stop, I realized the chances of retracing my steps and finding my bag were low - but better than if I did not even try!

I scrambled off, lugged the suitcase to the other side of the tracks (via escalators, thank goodness), and headed back to the Foggy Bottom station. I wasn’t quite sure where I had been sitting, so I walked from one end to the other - and then a policewoman waved at me. She seemed to be smiling. “We have your bag,” she said as I approached. “Come with me.” I followed her up to the control center (or whatever they call it) and she smiled as she handed me my bag. “Follow me,” she said. “I’ll take you back to the platform.” I didn’t want to invade her privacy (though the modern part of me wanted to take her picture), so instead I asked her how long she had been doing this job. She told me she used to work as a corrections officer, but the work was punishing…. Finally, as she dropped me off where I had started, I asked if I could hug her. Of course she said yes, and as I hugged her I felt warmth and relief and love for this woman I barely knew.

Of all the amazing things that happened during that week of meetings with important people, thought-provoking talks and beautiful musical performances, this is the one that I’ll remember most fondly: inefficient, full of friction, and immensely human. It was not an intellectual discovery, but something much deeper. I hope never to make the same stupid mistake again; it might not turn out so well a second time. But I am so glad that I did make it this time. I’ll never forget the warmth of that hug with a random human being doing her job of being human to someone in distress.

PS - If you happen to recognize this amazing woman, please show her this story. I would love to meet her again. But I also know that there are millions of public servants out there just doing their jobs with dignity and respect - and they deserve respect and gratitude back from those they serve.

My prompt to Gemini: “please draw a Venn diagram of three motives for personal philanthropy: tax optimization, mission, status (your name somewhere).”

I guess it translated “your name somewhere” into John Smith, as opposed to the notion of performative philanthropy. I know how to fix my errors. :) Please stand by; I’ll get it perfect before I post.

And context for readers: I’m working on a sample chapter for my book Term Limits; it’s on - guess! - term limits of nonprofits. H/T Vu Le https://www.linkedin.com/in/vu-le-b395a69/