Greetings from Glasgow

I’m going to write random thoughts. But something I have been thinking about for a while is “regression to the mean”. Regression to the mean, which Galton first coined (though he said “mediocre” not mean) had to do with extreme values. If you draw an extreme value from some distribution, you are less likely to the second time.

I think mathematically this is related to the modern notion of probability though to be quite honest, I’m not sure. But the modern theory of probability emphasizes that the probability density function lays out a single unit of probability, almost like an stick of creamy butter, along some series of events, and it just is hard to consistently reach into the tails because so little of the probability’s mass is there in the first place.

You can pull two aces in poker, no doubt, and it happens all the time. But doing it again and again is not likely, and if you did, it’d most likely be because the dealer is cheating, not that you’re lucky. The pulling of extreme events is almost mechanically connected to the rareness of those events even being found and so it’s more likely you return to that place where more events can happen, which is wherever the probability’s mass is stacked up.

Well, sometimes in causal inference, you’ll hear about regression to the mean and event studies. Treatment effects and the outcome itself will revert to the mean. Some version of that anyway. But I just wanted to point out one thing. Technically the probability density function lays out a single unit of probability into a dense mass against the events in question, but it’s referring to a single variable. And the PDF for the untreated potential outcome, Y(0), and the PDF for the treated potential outcome, Y(1), need not be the same, and thus need not therefore have the same mean. The mean of Y(1) and the mean of Y(0) in fact are not the same if the treatment effect is non-zero.

This is why I bristle a little when people talk in terms of “the outcome” but do not specify which outcome. Is it the Y(0) outcome of the Y(1)? Because they aren’t the same variable and need not have the same distribution. None of the moments have to be the same.

That’s a random thought but I woke up thinking it. I’d need to find the old papers that said things about it but it’s been sticking in my head for a long time and just wanted to write down that simple thing before I forget it.

---

In the exact way I expect things to go when I travel, I got to my gate in Dublin on the way from Zurich to Glasgow, with a couple hours to kill. Actually two things happened, and I’ll start with the other one.

I left Zurich at 8am for Glasgow. I got through security, and was in the open area of the Zurich airport with two and a half hours to spare. I’m making great time! Why not get breakfast then, I said to myself, and so I did. I found a lovely and efficient restaurant to get my avocado toast and multiple cappuccinos. I did something on my phone that in hindsight was unnecessary and did not require my attention.

I took a long morning to relax, and with around 45 min to departure started walking to my gate. When I suddenly realized — I was leaving Zurich for another country. Which means security isn’t the only place we all heard; I still had border control and “the passport thing” I said to myself where I would stand in the most inefficient line and wait forever. So the reason I had gotten there early wasn’t security, I remembered, but also border. I got to it and sure enough — the world’s longest line, and it was now officially boarding time.

So I did that which I am loathe to do. First, I paced. I tried the automated passport kiosk thing three times, each time doing so because I absolutely did not want to do what I knew I had to do. I was going to have to beg the people at the front to let me break in line.

So I did. I went to the absolute front of the line, a line that snaked around like the small intestines, and explained that my American problems, of my own creation, were more important than theirs and would they consider letting me break. And then I did the other thing I didn’t want to do. I consistently made eye contact with someone as I said, requiring they do that which I also knew they had to do. I knew the trick of eye contact. It is very hard to deny a request from a stranger when you make eye contact, so if you don’t want to do the request, you simply don’t make eye contact. But if the person is homeless and has a genuine request, it is also very hard to not make eye contact forever, and once eye contact is made, it is very hard to then deny the reasonable request of the person which in this case was something undoubtedly the person had experienced before — the need for a favor from a stranger.

So we eventually did make eye contact because i politely kept asking and I was beside him, and he reluctantly let me in, and I made it through. Hopefully everyone else did too.

So that was the first thing. The second one was worse.

I got to the Dublin airport where I had another three hour wait. I went to my gate which was hidden in a small little room beside another gate. And at some point a staff member told us to move to the other nearby gate, so I did. And then I did what every well functioning human in 2026 does. I got on my phone and processed my emotional state with a chatbot, and did so with such intent focus that I did not notice they changed the gate back to its original, and I missed my flight by five minutes. I noticed at some point that I still hadn’t boarded, and the gate I was at didn’t have anyone standing. I turned around to an empty room and realized I’d messed up.

Me and another person had both not heard the announcement, and so then became on the same team, which helped me not feel too much like an extreme idiot. I regressed to my own mean, thanks to her also missing it, which was one of modest incompetence in other words but not extreme incompetence. We ran around, made it back to the checkout counter outside of the terminal, I got my suitcase back, and caught the last flight to Edinburgh. I got there at 8:30, waited no joke an entire hour for the suitcases to get unloaded, and then Ubered it to my hotel in Glasgow. I unpacked my suitcase, got ready for bed, splashed cold water on my face, and went to sleep.

So now I am in Glasgow in the Governor’s hotel which is the same hotel as last year. And I had a great time last year, so I’m excited by this today. I’m excited for a few reasons but one is that Glasgow is a great city. It’s full of Industrial Revolution history. And this is the Adam smith business school. Adam smith as you may recall lived and breathed at the university of Glasgow. He entered school here as a teenager (14yo!). He came back as a professor and taught logic and moral philosophy. He was the chair of moral philosophy for 13 years and said it was some of the happiest years of his life. I know the feeling. When you’re happy at your job, and it’s during plentiful intellectually productive years, you can’t help but infuse the job with that happiness. He wrote theory of moral sentiments while chair actually, which he wrote before wealth of nations, and if I’m not mistaken, moral sentiments was maybe his low key favorite of the two. He later became rector of the university.

So Glasgow and Adam smith are ride or die friends. And last year was smiths 200 year anniversary of the publication of the wealth of nations, which I did not get to attend but would’ve happily had the trip coincided. Smith, Ricardo and Marx interest me the most of the classical writers. Maybe one of these days I can come to town and just experience more of Smith’s presence here. But not this week; this week I am a teacher.

---

Here are my departing pictures of Zurich. I found the James Joyce bar. Did you know Joyce wrote a big chunk of Ulysses there? Given how complicated that book is, it would imply he therefore spent a lot of time there if he wrote a majority there. He is also buried there. I call this picture the portrait of the economist as an old man.

Zurich was beautiful and I had a wonderful time. I vibe coded my lectures which I thought would be simple given it was just migrating them over but in fact it messed up my tempo and the week was more fatiguing than I expected. So I am debating using my normal slides. But I think I will stick with it and keep building up the new material. That’s the summer goal anyway; to develop new material requires condensing the old material requires restructuring requires doing what I’m doing.

The new material is more covariates, more synthetic control, more continuous, more compositional change, more AI. So that’s the deal. And therefore I have to keep iterating.

---

And last but not least, the new book! I randomly was on Amazon last night and got a recommendation for a book whose name I recognized because it was my own.

I think Yale is still building the page because the cover art isn’t there yet and it’s not quite indexed to my name, but the pre-Order button is there, and so the book is now in a countdown. Amazon says August 25 is the release date so I think that means it is.

So, now me and Kyle are in a race to get the free version online. I say “me and Kyle”, but it’s mainly Kyle who will be creating the website. But that’s it! What a journey. Oh and apologies for this.