I thought it was interesting because it edged me a bit towards a bumpersticker-style thing I’ve been saying the last six or so months, which is that “verification is the new bottleneck”.

“Verification is the new bottleneck” is a sentence and type of rhetoric you see everywhere in which a writer posits that the demand for research output is downward sloping at the individual level. Before AI agents, the marginal cost of producing research was some amount that determined research quantity demanded. I’ve drawn below two pictures representing how I think of this shift in research output, which is self-explanatory and intuitive to economists, but may be helpful to walk through for non-economists.

The first graph is the old equilibrium determined by the marginal cost of producing research which I represent as a horizontal “constant marginal cost” for simplicity. At this old marginal cost of production, MC_b for “before AI”, we produce an amount of research up to the point where the marginal benefits are equal to the marginal cost, most likely fixed by time and skill. To do more would require more human capital and/or more time, and given the scarcity of both, the researcher would not do it.

But now the second graph. Now with AI agents, a few things happen. First, I and others argue that the production of research increases under AI agents because the marginal cost of producing research fell from MC*_b (before marginal cost) to the much lower MC^hat_AI which is at some arbitrarily small amount above 0 equal to epsilon.

I also put surpluses in this graphic though I am vague about just what “research surplus” means precisely, mostly because I’m not really clear myself. The old surplus was B, and it constituted I think the subjective net benefits of the research. But under the lower marginal cost, we gain not only more research (the elasticity of which depends on the dimensions of the demand curve in this case), but also additional surplus too. Specifically, the old work we would’ve done is done with fewer time inputs, and thus the old work now has net benefits of B+AI_1, which is the area under the demand curve up to Q*b. But as research output rose from Q*b to Q*_AI, there is additional surplus generated from the new work equal to AI*2.

The thing that I am unclear about is this, though. In principle, all of us have additional time from this shift that we now get to reallocate to other tasks in our lives. What that is or will be is shaped by heterogenous preferences.

But one thing that existed, at least in my life, was that the subjective quality of the old work has changed. By subjective quality I mean that the old work, Q*b, had two qualities inside that triangle B. The first quality, was that it was produced at all, which is revealed by Q*b. That it existed at all under the human mode of production means that it was created — a statement that admittedly does seem tautological, but I think it’s worth saying anyway. Q*b was created by humans because humans created it, which leads directly to the second point which is that the net benefits of B was that it was the (net) value of that output to the researcher consisting both of its existence (production) but also something a bit slippery which I think is just the innate understanding of the work itself that happens almost by accident. I think of it as a kind of Hobbs-like squatting by workers wherein the sheer act of working to turn the natural earth into something else created ownership. There was, in other words, almost like ownership to the output. Not so much ownership as in no one else could have it so much as a type of mental ownership. I made it, and therefore I knew it. Somehow the production of the thing and the understanding of the thing existed simultaneously, as you could not make cognitive output without understanding it too as the understanding of it was usually a prerequisite to doing it, and also the byproduct of making it at all.

But now, with AI agents producing, what even is B? Is it the same B as before or has even the new B changed to something else — like a B’? This is where the geometry of this starts to break down. I don’t think it is the same B as before personally because there is not there the same kind of Hobbs-like squatter rights over it since we were not the ones who made it. By which I mean the epistemological function of work itself is not present. The latent understanding of the work has gone away because understanding may not be produced indirectly as a byproduct when we are not doing it.

Now it can be, 100%. Lab managers understand by proxy the work when interrogating and working closely withe the coauthor and RA who made it, but often that is not the case. Primary authors will be held accountable, for instance, when their RAs make mistakes and they did not do the due diligence to be on top of it. That has happened before. We can all think of famous cases over the last 10+ years or so in which people were more or less deceived by duplicitous coauthors who had fabricated data outright. I think B was in other words highly contingent on trust with a coauthor, historically anyway, even with managers. They would more or less free ride on the trust of the coauthor or RA’s own subjective understanding of the work so that B could exist even in the minds of the non-producer simply because they trusted the producer.

But put that aside, because now my point is that we have two new surpluses and that is AI1 and AI2. And I think in some way, these two new surpluses represent the additional gains subjectively that happens as a result of taking their hands off the wheel to some degree. It is gained subjective net benefits and it is gained subjective net benefits coming from manager-like trust that the work done was done well even if they did not do it, and even if they do not understand it.

I think for some, it is going to be the principal-agency problem, though. It could be the blind leading the blind. I trust the agent did the work well, but as the agent is not real, we do not actually know it was done well, and if our skills atrophy, we may not even possess the same comprehension of what it means to be done well.

Well, the question I have in my mind is that couldn’t we get there? I mean we have so much more time, don’t we? Couldn’t we just reallocate the time savings back to the verification task of that new and old material created and it become real surplus, something we both experience as positive net benefits and which is actual positive net benefits? That is, we experience it as valuable because we think it is correct, and it is actually socially valuable because it is correct. I think those might not be the same thing. I think they may be different.

See, the thing I keep thinking is that it is an assumption that the time savings we now have in our possession are even capable of being used to fill in those areas at all. We don’t know that it is. We don’t know, I don’t think, if the time savings we get from AI agents can be allocated to fill in that entire B+AI1+AI2 with subjective understanding. What if it’s not? What if the old way of understanding something was primarily an accidental consequence of the production process. Pollution is like that, for instance. Pollution is the byproduct of production. It’s an externality. What if understanding was always an externality — something which was created almost by accident when one set out to make some cognitive output using time inputs?

So I think I agree with this idea that it is not really the most accurate way to describe the situation we are now in to say that “verification is the new bottleneck”. I think it is more accurate to say “understanding is the new bottleneck” because it’s entirely possible that AI agents will be able to verify as well as produce, but that does not therefore mean we will understand it.

So I have been thinking a lot about this essay by Geoffrey Litt, as well as this idea Paul had put forth which is the unit of verification is the git diff operation. I now use the git diff operations like Paul suggested, but I am noticing that I am skimming and accepting them creating what Paul and others call cognitive debt — only it’s supposed to be that the cognitive debt is cleared by the sheer act of checking off those diffs. But I think I’m actually just silently accepting them without full understanding. That does not mean everyone else is so much as it means I am, and if you sense sometimes that you are doing the same thing over and over habitually, it is time to consider if perhaps you should try something different. Maybe not instead of the diffs, but in addition, where the goal is now two things, not one.

1. Zero Errors Remains the Constraint. I remain convinced that we must say to ourselves and one another that “zero errors is the constraint”, and not the objective function. Our goal should not be to minimize errors, because if you say that, then you will quietly come to accept the idea of “optimal errors”. And that cannot be what we move towards given the aggregate supply elasticities of research production with respect to prices is probably greater than 1. I suspect in the long run it is anyway, even if in the shortrun, due to fixed inputs, it is small.

2. Understanding is the constraint. Is understanding now the constraint or is it also the goal? I am thinking a threshold of understanding the work must become the constraint too. Paul usually quotes IBM to this point, at least indirectly: since machines cannot be held accountable for mistakes, they cannot be trusted to make decisions either. They either work to help us understand the work they’ve done, or we do not use them at all. I think it’s possible those are the two corner solutions.

3. Research output is the objective function. So then what is the objective function? It’s whatever we think is optimal in equilibrium but not perfected, and I think that could be the idea that we are trying to reach optimal research output, and optimal research output is the point where the social marginal benefits equal the social marginal cost, which requires verification and understanding.

So, what am I doing differently than git diffs. I am experimenting with a new skill called /quiz. It’s an extension of things I was already doing which were interviews used to extract my beliefs about which covariates to use and what the target parameter should be in situations where I could not really be sure, but I suspected I did know deep down enough to say something. But now what I’m working on is a little stranger of an interview concept.

My /quiz skill has two components. First, I am working on Claude Code creating a set of slides when invoked that teaches me something. And then I have Claude Code ask me multiple choice questions, one at a time, based on it. You know where I got the idea? The training videos that HR makes you do all the time at work. That’s how they do it — they give you something to read, oftentimes which includes a video, and then they give you a quiz. That’s where I’m going. I am going towards more or less “training video quizzes” to constantly create in me continuous understanding. On top of verification. On top of my detailed staged checklists.

We’ll see how this goes. I don’t know if I am over-engineering the production process using AI agents, or if it is optimal, but I am definitely moving towards something that is taping my hands to the wheels. I am tying myself to the mast to drift through the strait so that I do not go insane by the screaming of the harpies in the water, so to speak. There is simply too many things that can go wrong, and I for one do not get excited by the idea that a sniper will come along after me and find my analysis is just riddled with problems. The current environment in which problems in one another’s work is treated as scalps to be taken and pounds of flesh to be removed is far too antagonistic for my tastes. I am not interested in becoming helping someone get their 15 minutes of fame on a podcast as they discuss the problems with my work to be frank. But I also just want to be my best self. I want to continuously learn. Learning is why I got into this. I am, deep down, not really a competitive person so much as a romantic about learning, and if agents can help me learn in my creative work, I want them to sit in the cockpit with me. But if they are going to be auto piloting while I sleep, then my truest self would prefer I not do that as until the lights go out, I remain in love with creativity and creative tasks. I enjoy it, it motivates me, it fills my heart with lightness and joy, and my mental health desperately wants it.

So that’s all. I just wanted to share these thoughts. Now back to work trying to wrangle these agents into doing what I say even if they, in a very nice style, stumbling and mumbling as they do it.

Online dating is unique in comparison because unlike family, work and school-based dating platforms for matching people, online dating is the most stripped down version of a blind date we’ve ever had in American history. If you are referred, more or less, to someone by a friend, then positive association not to mention simply the ability to get information via your friend, and the fact your friend is unlikely to have a sketchy friend, let alone recommend them, all blind dates historically that happened via the friend, family, and school recommendation platforms would’ve actually had latent knowledge. But in the online era, an algorithm recommends and this algorithm comes from for-profit firms whose shareholders are looking for rising stock prices to justify the investment, otherwise they would be better off shifting their savings into tech firms, or real estate. Thus the supply-side or market structure side of the dating market from an informal friend/family/school/work equilibrium to an algorithmic platform equilibrium is likely to have been independently important as we try to understand family formation.

But there’s other interesting things going on. For one, did you know that 40yo “never married” marriage rates have risen to all-time highs? I don’t mean unmarried as that would include divorces. No, I mean never-married.

I had never seen that statistic before. The graphic is so pronounced. So now check out how the online dating experiences differ across these age (and sex) profiles.

I’m going to dig into a lot of statistics, but if you go here, you can find these too.

Let’s focus on my demographic — males in their 50s who are single. This first graphic is only related to the age range, but 5% of 50-59yo used online dating sites or apps in the last year compared to 15% of those and 18-49. The “ever used” is 23% for people in their 50s versus 43% for 15-49yo.

Which sites are the people in their 50s using? Overwhelmingly, they are using Match and eHarmony. Fifty-three percent of online daters in their 50s are using Match, whereas 28% are using eHarmony. Interestingly, though, of the 50 year old online daters, 19% are using Tinder.

You can see more statistics about the use of Tinder here. For one, it’s the most popular one, particularly among the youngest group. Its use falls a lot by age, as we saw above. The older people favor Match and eharmony, which I associate with more classic online dating which you might just call love-matching platforms. They were the early entrants. These are not used as much among young folks compared to Tinder and Bumble, which would likely mean that there’s segmentation in the dating markets which would also then affect who will and will not match at all. It goes without saying but is worth saying if the majority of people ages 18-49 use Tinder, Hinge and Bumble, but the majority people aged 50 and older use the classic love platforms like Match, OkCupid and eHarmony, then they are less likely to meet. This segmentation in dating markets across platforms is I think potentially part of the story of how online dating can be impacting marriage and family formation.

When you compare online daters who are 50+ to the younger 18-49yo, there does appear to be fairly similar stated reasons. The bulk are trying to find a long-term partner or spouse, though it is in its point estimate larger for the older folks than the younger ones. But the younger folks are much moreso using the platforms for casual dating and casual sex than the old people. It’s interesting — the older people are looking for partners and friends at higher rates than the younger folks.

Younger men and women tend to experience the online dating phenomena negatively at around the same rates, whereas older men interestingly actually experience it more positively than anyone, and considerably more positively than women in the same cohort (60% versus 43%). This was the first time I found differences by gender.

But check this out. Men are more likely to pay for the services, even though technically they don’t have to. The kind of positively priced services on dating apps are subscriptions, roses, boosts, and whatever else. These are ways to expand your search as well as get more notice on the other side of the market. My conjecture has been that beautiful people, and probably beautiful women in particular, wake up every day to an app filled somewhat with dating-spam. If they take even a week to not go back to the app, I imagine the most attractive have a lot of what is like “unread emails”, only it’s roses, likes and other forms of communication. And so my theory has been that the congestion alone, and dating fatigue just from that dimension of online dating, will place people at the top or bottom depending on things like the algorithm and payment. Hinge uses, or claims to use, a version of the Gale-Shapley algorithm for recommending people to one another, but I would be willing to bet that payment will push you up higher in that stack of unread emails, giving you a better chance of being noticed. This is not a great measure, but there’s some evidence for it. See here how the majority of women feel overwhelmed by the communications by the amount of messages they receive. And even though there is a huge amount of people excited by the people they have seen, there is even more disappointment.

Still, notice how bit the gap gets between women and men once you get into your 60s. And look how the gap between male payments grows compared to young people, too. The ratio of male 50-64yo to male 18-49yo who pay is 1.34, but the ratio of male 65+ to male 18-49yo who pay is 1.5.

You do see growing ratios for women too though. The 50-64 ratio of females to 18-49yo is just over 1.5.

Interestingly, among those who have paid, they are more likely to say it was a positive experience than those who didn’t pay. Not saying it is causal, but it might be, and in some ways you’d expect so insofar as you are paying for a better experience presumably. But again, the ratio among men to women is one of the main stories — men experience online dating more positively than women at a ratio of around 1.9.

What about “unwanted communications”? Who experiences that the most? Well, men experience it the least, which is not surprising. But there’s a lot more heterogeneity than I knew about to be honest. When you break it down by sex-age, you see that women 18-49yo experience it the most, whereas men in the same age bracket experience it the least (tied with 65+yo men). At least having received unwanted sexually explicit messages or images that they did not firs request.

But within women, the disparities are the largest. A whopping 56% of women ages 18-49yo have received such, and falls precipitously. Women aged 50-64 actually see it at a rate that no male category ever experiences.

But you also see repeated communications from someone after you’ve requested it to be over to continue far more often for women 18-49yo than any age group of men. But also women 50-64yo say the same thing. And then there’s other things — threats of physical harm are the highest for women 18-49yo, And been called an offensive name too.

But going back to the 40yo “never married” statistic. Look at all the heterogeneity in “single” for Americans across race, sex, age, sexual orientation and education. The marriage disparities are largest for those with the least education, which I guess I didn’t know. It differs a lot by sexual orientation. But it’s super pronounced within race too with 47% of all blacks reporting being single versus 27% of whites. And there are differences by age as well.

But then what are the stated reasons for single people choosing to not date at all, as there’s a decent amount of exiting of the dating markets altogether. One of the main stated reasons is just that non-daters who are also single actually like being single. That’s the most common reason. The other is that they have more important priorities. The third is that they are too busy. And that’s the top three reasons that Pew found, in that order. Feeling like they’d be successful finding someone is actually a much lower reason. And feeling too old — that’s pretty far down the list. Even Covid is way down the list, and this was a survey from 2022, which was pretty deep into the pandemic.

I mention the other day the popularity of an informal whisper network called “Are We Dating the Same Guy?”. Now that is not the same as experiencing unwanted actions from the other side, as nothing about experiencing unwanted actions from the other side necessarily means you are in danger from them. But given the “stranger blind date” equilibrium I posited, I think it is indicative of the problems participants in online dating face since they are matching with truly unknown types — which I think is a major new phenomena. Well, not surprisingly, a lot of people feel that online dating is actually unsafe. It’s actually the majority opinion in the point estimate. But it’s a far more common claim made by women, and despite older people experiencing it the least, interestingly it’s far more common a position that grows with age — even though the incidence of unwanted actions experienced falls with age. So that’s an interesting paradox.

Lots of interesting facts about online dating. Lots to learn. Unfortunately, not as much data out there that I wish, but this is still great.

After 16 episodes of working with Claude Code to study the effect of abortion clinic closures on marriage rates using a natural experiment, so to speak, in Texas called "House Bill 2" and continuous difference-in-differences with a treatment of "change in distance", Caitlin and I start our descent. Caitlin shares her own thoughts about returning a poisson estimator, and we continue to wrestle with the implications of differences in econometric estimators and interpretation, as well as what we are willing to live with. Plus a lot of discussion about the implications of AI for our lives as researchers manages to creep in its pretty little head!

I wrote at length a fe…

Read more

Read more

Read more

Read more

Read more

I will struggle to call it the Remix as a result of my affection for the Mixtape brandi…

Read more

Read more

Read more

Read more

Read more

Caitlin Myers and myself have been studying the effect of House Bill 2 in Texas, which closed half the state's abortion clinics, on marriage rates using "continuous difference-in-differences" estimation for 14 episodes straight.

But behind every good diff-in-diff estimate lurks a parallel trends assumption, and so we asked Hannah to look into one confounder -- oil! fracking! energy! Perhaps things related to technologies that enhanced the extraction of oil from the ground, like fracking technology, could have alone shifted marriage rates by bringing men into communities but only for work, temporary residence, not permanence.

We talk about a literature where men in transit can alone reshape the marriage markets, which if those events happened disproportionately in our treatment counties, but also our control counties, then parallel trends mechanically *might* be violated. Because remember -- parallel trends is a) about things differentially impacting the treatment and control units on *outcome trends* and b) therefore can impact the treatment group *or* the control group. Hannah comes back with research she did for us on it.

If I run diff-in-diff without covariates, and find a number, but then run it with covariates and find a different number, this is bad and diff-in-diff is invalid.

It’s a gut check, vibe thing for sure. Of course, though, there is such a thing as “co…

Read more

On Substack

New on Substack: Subscriber Perks, saved Notes, scheduled Chats, MCP, redesigned podcast player, and more

Whether you’re spending your Substack summer tinkering with your existing newsletter or podcast or launching your first one, we’re shipping features that make it easier to publish, customize, and distribute your work, and engage with your audience. We look forward to hearing what’s most useful and what you want more of. Here’s the latest…

Read more

a month ago · 1502 likes

If you want to connect Claude Code to your substack, here’s the instructions to do so. But let me lay out what I’ve learned, and how I’m going to learn from what I learned going forward. I had Fable 5 do the analysis for this con…

Read more

Greetings folks. I am back from Europe. I left on May 9th (not the 11th as I previously said) and got back on the 4th of July to Boston. That’s 56 days in Europe across I think I counted half a dozen countries and approximately 10 cities. Quite a romp. I made new friends, saw new places, visited old ones, including Amelia in Sa…

Read more

Read more

1. University of Zurich for a …

Read more

This week is a hoot! Caitlin and I finally get some estimates of the different target parameters using continuous treatment diff-in-diff! We discuss in detail how the estimator works and then go through our analysis. Watch to the end to see our reaction when we finally see some output in a beautiful deck!