I’ve thought it over and decided not to write a post this week. It’s not that I have writer’s block. I am writing plenty in the dimensions of my profession that dominate my time (and actually pay me). And I have a back catalog of “bigger” pieces I might write later. But there is nothing I am compelled to write today. Which is what I want to write about.

We all share editorials and thinkpieces with each other, whizzing around social media and email, getting discussed over meals and beverages. This content is produced in mass and at breakneck speed. Some people are very good at it. Others less so. Some people, over time, rise to a level of recognition that they are offered plum spots at major outlets, lavished with salaries greater than I will certainly ever enjoy. Their names acquire significant fame, their opinions serving as the substrate for millions of conversations.

And then we savage them.

Sometimes we savage their works because they are signaling the wrong politics and identities. That’s just life. Sometimes we savage their writing because they’re rich and famous, which is annoying, but that’s just the tax a person pays for being eminent (see Swift, Jonathan). But often, more often than they would likely want to admit, we savage their writing as poor and ill-conceived because it is poor and ill-conceived.

Let’s be clear: these people are largely critical thinkers and phenomenonal writers. But they are also on a deadline. Opinions, unlike news, do not appear in our minds fully formed and their subsequent development does not always adhere to a regular schedule. My writings here are essentially an unpaid hobby. I haven’t missed many weeks this last 3 years, but I’ve missed a few. Some weeks I totally mail it in and just write a few paragraphs about a research paper I read that I thought was cool. The New York Times editorial page does not indulge such academic capriciousness.

Would I invest a lot more time in these posts if the NYT was paying me a hefty salary? Of course. I would have stockpiles of evergreen columns, folders of half-written ideas, a corkboard littered with post-its cataloging my every idea that might support a column. But even then I can’t help but suspect that I might occasionally find myself staring down a deadline with nothing I want to say, or with a drafted piece that I know isn’t very good.

And that’s why I think we get the so many big-name editorials that social media descends on like gleeful hyenas, merrily yanking and ripping until the every vacuous subject and failed predicate has yielded it’s final LOL. Why do columnists collectively produce so much dreck? Because they write too much. Scratch that. Because they publish too much.

Which is why I’m not writing a post this week.

(To be clear out of an abundance of caution, this observation does not pertain to those glorious blogs and substacks that mostly produce data-driven analysis and subject-matter deep dives, rather than bloated opinions designed to foment clicks. You are god’s perfect children, never change.)

The bombshell news in the tech world as of late Friday was that, in a sudden coup, the board of OpenAI fired CEO and tech entrepreneur Sam Altman, and demoted company cofounder and former president Greg Brockman. The exact grounds for their decision remain somewhat murky, but apparently Altman wanted to move faster with AI deployment and monetization than some board members were comfortable with.

The OpenAI organization burst on the scene in the past year with the release of advanced versions of ChatGPT. This “generative” AI technology can crank out computer code and human-like text articles and reports and images. Naturally, students have taken to employing ChatGPT to write their essays for them. And so, professors now use AI to detect whether their students’ essays were machine written or not.

Fellow blogger Joy Buchanan has addressed the rising problem of erroneous information (“hallucinations”) that can appear in AI generated material. There is a movement to slow down the development of AI, for fear it will lead to The End Of The World As We Know It (TEOTWAWKI).  (Interestingly, all of the business commentators I listened to today dismissed the alleged world-ending dangers of generative AI as largely deliberate hype on the part of AI developers, to create a buzz – which it has.)

Having hitched itself technically to OpenAI technology, and having poured something like $13 billion into funding OpenAI, giving it a 49% ownership stake in part of the business, Microsoft was obviously concerned about the effect of Altman’s dismissal on its own AI plans.  As it became clear that the board action would lead to substantial dysfunction at OpenAI, Microsoft carried out its own coup, by hiring Altman and Brockman to run a big in-house AI research initiative, and making it clear that anyone else who wanted to resign from open AI could have their old jobs back, under their old leaders, in Seattle.  And indeed, as of late Monday, nearly all of OpenAI’s employees had signed an open letter stating that unless the OpenAI board quits, they “may choose to resign from OpenAI and join the newly announced Microsoft subsidiary.”

Investors are still trying to figure out what all this means for Microsoft. A pessimistic take is that the corporation has to take a big write down on a $13 billion investment, if the OpenAI organization  (valued a month ago at $90 billion) loses its momentum. An optimistic take is that Microsoft may get the human capital crown jewels of this leading tech outfit for simply the cost of salaries (and signing bonuses), instead of shelling out to buy the enterprise as such. Also, having the technology all in-house would remove the vulnerability of Microsoft currently faces with having a key piece of its future in the hands of a separate organization. There is debate on how much the intellectual property held by OpenAI would inhibit Microsoft from forging ahead with its own version of ChatGPT.

According to Wikipedia:

Shares in Microsoft fell nearly three percent following the announcement.   According to CoinDesk, the value of Worldcoin, an iris biometric cryptocurrency co-founded by Altman, decreased twelve percent.   After hiring Altman, Microsoft’s stock price rose over two percent to an all-time high.  

According to The Information, Altman’s removal risks a share sale led by Thrive Capital valuing the company at US$86 billion.   A potential second tender offer for early-stage investors is also at risk.   Altman’s removal could benefit OpenAI’s competitors, such as Anthropic, Quora, Hugging Face, Meta Platforms, and Google.  The Economist wrote that the removal could slow down the artificial intelligence industry as a whole.  Google DeepMind received an increase in applicants, according to The Information. Several investors considered writing down their OpenAI investments to zero, impacting the company’s ability to raise capital. Over one hundred companies using OpenAI contacted competing startup Anthropic according to The Information; others reached out to Google Cloud, Cohere, and Microsoft Azure.

There is a slight possibility that the open AI board could take a big hit for the team, and bring back Altman and Brockman and then resign in order to keep the organization intact. If that happens, the deployment of generative AI would accelerate – – which might destroy the world.

THIS JUST IN: ALTMAN BACK IN CHARGE AT OPENAI

If there was a prize for “worst board decision of the year” it would have to go to the move late last week to fire Sam Altman. But just when you thought there was no more drama to be milked out of this scene, the news Wednesday is that the OpenAI board is out, and Altman is back in as CEO at OpenAI. Microsoft is presumably happy to have the organization intact, and it seems that those pesky timid souls who were trying to go slow on AI proliferation have been swept aside. TEOTWAWKI here we come…

Sometimes we think things have been incontrovertibly been proven, but we really only know them. Other times we think we know them but we really only think them. It’s always interesting when we something we think becomes something we know. We share those beliefs a little more often with a little more confidence. We start trying to tilt the balance of common knowledge one conversation at a time. I’d argue, however, that we would often be better served to wait until something is proven, as much as something can be proven. Or, at the very least, that our conversational was weight shifted far more when truly compelling evidence hits the scene.

I already believed that New York City’s infamous “Stop and Frisk” program was bad. That the bad outweighed the good. I was always careful to soften my language, to hedge my claims, however, because I always suspected there had to be some margins on which the program yielded some benefit to someone, somewhere, in some context. Criminal deterrence is real, after all.

I no longer feel any need to soften my language or claims one iota. There are research papers that change your priors. There are also ones that harden them into granite. Jonathan Tebes and Jeffrey Fagan have a new working paper they are presenting at conferences and quietly circulating that provides the single most compelling research effort into the effects of Stop and Frisk I have come across to date, one which makes the case that Stop and Frisk had now measurable effect at the margin to deter crime while, at the same came, causing significant harm to the young Black men walking the streets of New York City.

Please go through the slides, but let me summarize. Using a credible and clever event study design around the end of Stop and Frisk, Tebes and Fagan are able to identify the effect of stops on crime, finding an impressively precise null effect. They then look the effect of these stops of neighborhood schooling outcomes, specifically interruptions to instruction, persistent absences, suspensions, and graduation. The result, again, is very clear: Stop and Frisk was a disaster for high school age Black Men.

I’m just going to leave it there. Read the slides, read the paper when they release it, update your priors. And when someone tries to tell you at Thanksgiving dinner this year that New York City is going straight to hell because they ended Stop and Frisk, have the confidence to vigorously attempt to update their priors. Will it work? I’ve never met your family…but probably not.

But you gotta try, right? It’s your duty. And then make yourself a drink or take an extra slice of pie knowing that you earned it.

First, read this story from the NYT about the biggest producer of content on OnlyFans. TLDR; a couple have a compound in Florida and a full stable of employees (writers, editors, accountants, cooks, etc.) all being coordinated around the gigabytes of data generated by the supply and demand of their sexy content. If they were selling in a less stigmatized market, whis would be taught as a case study at business schools.

What I found interesting is how it simultaneously validated and assuaged all my fears about the opportunity to emotionally manipulate large numbers of people by using highly granular data. Don’t get me wrong, that’s arguably the story of every information-deficient marketing campaign ever, but I’m not talking about coarse, subliminal manipulation (“Look at this fully self-actualized person drive a car that signals their worth to strangers and their father”). I mean direct, interpersonal maninipulation through the fabrication of intimate parasocial bonds. The ability to allow customers to create their own, bespoke, false narrative in which they have a relationship with a beautiful stranger. At scale.

It’s that list bit that matters. What this couple have deconstructed is a formula for producing intimate parasocial relationships worth thousands of dollars to customer at scale.

The exploitation of fabricated relationships for income is the story behind the worlds oldest profession, not to mention most scams, for a very long time. The ability to produce them at any real scale, however, has been far more elusive. When someone pulls it off they’ve usually created a cult, whether it’s a new religion or a political cult of personality, and it’s worth taking note of. So have these Onlyfans creators laid out the blue print for future politicians, social entrepreneurs, and general power seekers? Are we at the beginning of an industrial revolution for social movements?

Actually, I kind of think they do have a blue print, but it’s going to be a minute before it crosses the chasm to other sectors because most fields that rely on parasocial relationships to grow don’t have the luxury of immediate profitability that sex work does. You might start your social movement with the ambition of analyzing every bit of data so your stable of employees in your Smithian pin factory of communications and content can rapidly grow your follower base, but you’re not going to have any money to pay them. What people tend to forget about sex work industries is that they generate revenue from minute one (that’s exactly what lures people into making what have historically been less than optimal long term personal decisions). By comparison, religious and political aspirants are a bunch of broke boys.

Religion and politics look like they have a lot of money until you consider size of the customer base (most people) and the sectors they influence (nearly all of them).  $14 billion was spent on federal election campaigns in the United States in 2020, the most ever. That sounds like a lot until you realize that a) it’s 2 and 4 year cycles and b) the federal government spends $6 trillion per year. By comparison, $5.5 billion poured through just OnlyFans, just last year. (Do I even need to convince you that new religious and social movements are notoriously short on cash?)

The story of each stage of the internet is the same thing over and over: a group of people couldn’t benefit from scale before but now they can. Social minorities looking to date couldn’t find each other before, now they can. People buying and selling pez dispensors can’t find each other, now they can. People with extreme beliefs were socially ostracized now they can find and reaffirm each other. People selling content to niche audiences used to have to find their customers through large media companies now they can do it directly. Bernie Sanders and Donald Trump both had disproportionate impact on American politics in part because they leverage the internet to disintermediate their ostenisble political parties. That’s the internet bringing scale to parts of the American electorate previously too distant from the median voter.

Power and ambition be damned, however, aspirant leaders are still not going to be able to build what two people selling naughty pictures in Florida were able to do because most people don’t want to pay for politics. Just ask every newspaper in the country struggling to stay afloat. We’re entering a new age of scale in the fake relationships being sold to us, but it will only be for the kinds of relationships we actually want. That doesn’t mean those will be emotionally nutritious relationships, but choice will remain intact. Portfolios of relationships for a lot of people are going to change, but I suspect its going to look less like Evita Peron and Jim Jones, and a lot more like Taylor Swift and Frito Lay.

Marketing will become more granular, more personal, more intrusive, and more effective. If this fills you with anxiety, I hope you can take some solace knowing its mostly going to happen for the stuff you are willing to pay for, like love, family, and your sense of self-worth. Data-enabled relationship fabrication will grow in market share as artificial intelligence crowds out the classically information driven side of marketing. The uncanny of valley of cringe is a customer relations disaster, a trap whose lines are invisible and always moving. For AI to learn where the boundaries lie is to move them. Which means this decidedly human labor market will grow all the faster. And a blue print for selling naughty content from a Florida couple will find its way to selling you damn near everything else.

Housing remains the most expensive monthly outlay for most Americans. There are signs of things getting better, but the fact remains that for those living within the first or second ring of suburbs surrounding a given city, space is at a premium. For this holiday season, give the gift of not taking up any more of that precious, precious space.

Don’t give them instant pots or juicers for their already full kitchen counter. Don’t give them clothes to go in their overflowing closets. Don’t give them knick knacks, tschotckes, or decorative thingamajigs that will rapidly migrate from shelves to bins to (shudder) storage units. Don’t get them stuff.

I’m not going to say get them “an experience” because we’ve all become a little weary of that cliche. What I suggest is getting them a luxury that might fall at the margins of their budget. Get them a massage. A facial. A stretching session (that’s a real thing). If they line wine or whiskey, get them a bottle they’ve never tried. The stuff they already like is easy, but my expectation is that is probably already in their budget. Gift them the risk of trying something they might not like.

Books are acceptable because there is an entire ecosystem that exists that take books from one home to the next once it has been read. Get them a subscription to Amazon Music or a download code for a new game on their PlayStation. Get them tickets to a concert or play. Get them a two hour cleaning service.

Babysit their kids for a Saturday. Tell them their hair looks nice and really sell it that you mean it. Leave them alone for a couple days so they can recooperate from a long week. Just no more stuff please.

I had the opportunity to present a new paper about theft to the faculty and students at two law schools last week. The questions and comments were interesting throughout, but I noticed a pattern in several of the questions from students: were we attributing too much rationality and sophistication to criminals?

Citing Becker (1968) as a useful exercise in applying economic parsimony to understanding how punishment and enforcement deter crime is one thing, but I think it’s simplicity sometimes undercuts a really important intuition that I hold to strongly: crime is boring. More specifically, a lot of crime (not all, of course) is a product of a banal calculus that arrives at the conclusion that my expected life (probabilistically) is better if I take this illegal action. These crimes seem irrational is because of two behavioral errors, not on the part of criminal, but on the part of the observer.

The first mistake is failing to realize you are observing the conclusion ex post, after the outcome has been revealed, and your ability to observe it is almost exclusively because the action failed i.e. they got caught. You know they got arrested for shoplifting an item that won’t significantly change the quality of their life. This feels like a mistake, but what you can’t observe is how many times they or someone else has taken the same action without negative consequences. If 1 in 10,000 thefts worth $500 are caught, then that’s probably an optimal choice for a lot of people.

The second mistake is implicitly assuming the same level of constraints that apply to your life apply to the criminal actor. We all know the question asking whether it is a sin to steal bread to feed a starving family, but that same logic applies in broader and less severe circumstances when considering the ex ante rationality of a choice. The cost-benefit analysis facing a potential thief is far different if they already carry the stigma of a criminal record. If their labor market opportunities are limited. If rent they have insufficient funds to cover is due is three days.

I find it interesting that people who disavow the salience of IQ and the people that place IQ at the center of their core model of humanity both seem to consistently underappreciate the sophistication of most human problem solvers. I’m not saying we all get the math right. Quite to the contrary, not only do we make constant errors in judgement, but the duress of operating under difficult constraints likely makes optimal decision-making all the more difficult. But those errors are relatively modest relative to the humans who are, at a baseline, tremendously sophisticated. They want and need resources and they can conceive of myriad manners in which to acquire them. Applying a lesser sophistication to people who steal from a CVS and sell to a middle-man who turns it over at a street corner or on Amazon puts any policy design at a disadvantage from the start.

If you want to divert people out of illegal markets and into legal labor, you’re better served treating them as people who made the optimal decision. Your ambition should not be to change their decision-making, but to change what the optimal decision is.

President Biden winning the Democratic nomination is currently priced at 72 cents on PredictIt, implying a roughly 72% chance of winning the nomination. Not the general 2024 election- where he is priced at a mere 43 cents- but the Democratic nomination.

To me it seems crazily underpriced to put the odds of an incumbent president being renominated by his own party at only 72%. Yes, his approval ratings are underwater, and yes he’s old, but the base rates here very much work the other way. No incumbent president has lost a vote to be renominated since Chester Arthur in 1884. It think its extremely unlikely Biden would run for nomination and lose; it makes more sense that he would choose not to run, like LBJ in 1968, but I see no indications of that.

I think Biden will only fail to be renominated if he dies or experiences a major decline in his health by the convention next August. This is certainly possible for an 80 year old but the odds of it are well below the 28% implied by PredictIt. A recent WSJ article lays out the details:

a nonsmoking male with Biden’s birthday, in good health, would be expected to live nine more years after next year’s Election Day, while for one with Trump’s birthday, it would be 11 years.

WSJ focuses on his chance of finishing a second term and doesn’t give an estimate for just making it to renomination, but my own look at actuarial tables shows that the average 80 year old has only a 6.5% probability of dying within a year. The chance of dying or getting a disabling health condition in a year is of course higher than that, but the convention is actually less than a year away in August, and the primaries will be done by June. Plus the WSJ article gives several reasons to think Biden is in better health than the average 80 year old:

First, the median includes people who drink alcohol. Regular drinking of two or more drinks, three or more times a week, shortens life expectancy by about seven years. Both Trump and Biden are teetotalers, in addition to being nonsmokers.

“Those are two of the biggest killers right there,” said Bradley Willcox, a professor and research director at the Department of Geriatric Medicine at the University of Hawaii. “When you eliminate excessive alcohol intake and smoking, one thing you’re left with is genetics.”

Here, Trump and Biden picked their parents well. Trump’s mother lived to 88 and his father to 93, though late in life he developed Alzheimer’s disease. Biden’s mother died at 92—living long enough to see her son become the sixth-oldest vice president. Joe Biden Sr. died at 86. That is even more impressive than it sounds: When those four individuals were born, life expectancy was around 50. 

Biden and Trump are each highly educated at a time when the life-expectancy gap between the educated and uneducated has been growing. They are wealthy, also a strong predictor of longer life. They receive excellent healthcare.

Add it all up and I think Biden has over a 90% chance of being renominated, so being able to bet on him at 72 cents seems like a great deal (even if it means tying up money that could now earn 5% interest elsewhere). PredictIt has betting limits and high withdrawal fees, but other prediction markets are in the same ballpark; Polymarket currently has Biden at 75c.

For similar reasons Trump is may also be underpriced to win the nomination, currently at 68 cents on PredictIt. He’s not an incumbent the same way, but he’s enough of one that I don’t think any of his electoral opponents can beat him for the nomination; he’d have to beat himself by dying or withdrawing (very unlikely), or be beaten by the legal system (he’ll continue to have trouble but I don’t think it will be enough to get him disqualified or in prison by the June convention).

It’s boring and its not my preference, but I think we are headed for a rematch of 2020. On the bright side, 80 isn’t what it used to be:

Disclaimer: Not investment advice. I did put my money where my mouth is here, and so am now talking my book

The Republicans can’t seem to elect a speaker of the house. This seems like a classic collective action problem, where the prisoner’s dilemma is a holding everyone back as they all try to individually free ride, hold out, defect, etc. But I think there is something deeper happening here that is actually quite simple.

Republicans are struggling to take a collective action because too many of them have no interest in producing any public good at all. As Mitt Romney observed, Jim Jordan has never authored legislation nor even sponsored a bill that passed. Humans have spent thousands of years devising clever mechanisms for solving collective action problems in pursuit of public goods, but the crux has always been a public good that could be pointed towards as a point of coordination and motivation for pushing through uncertainty and personal risk. If you are congressperson who has no interest in changing or defending the body of law, then the goal of electing a speaker within your party has no interest to you. In fact, failure to elect a speaker is actually preferred if that failure is generating attention that builds awareness of you and your personal brand.

Democrats should heed the warning of what they are observing as well. A small minority of pure attention seekers can shut down an entire party at the choke points both purposefully and accidentally placed within our constitutional republic. Collective action is everywhere and always fragile, not least of all when some participants have no interest in the public goods that might be produced.