On Bryan Caplan’s Question

Bryan Caplan asked an interesting question on Twitter the other day:

I’m not sure if you can see what I voted for, but I am fairly sure you could have guessed!

But if you think about it a little bit, it’s actually not all that easy to answer this question. As with all good questions, there are nuances and layers. First, I asked my excellent research assistant to tell me more about what the phrase “epistemic value” might mean:

The phrase “epistemic value” refers to the degree to which a belief or piece of information contributes to our knowledge or understanding of the world. Epistemology is the branch of philosophy that concerns itself with the nature and scope of knowledge, and epistemic value is a term that is often used in discussions within this field.
Beliefs or pieces of information that have high epistemic value are those that are reliable, accurate, and well-supported by evidence. Conversely, beliefs or pieces of information that have low epistemic value are those that are unreliable, inaccurate, or poorly supported by evidence.
In general, beliefs or pieces of information that have high epistemic value are more valuable to us than those that have low epistemic value, because they are more likely to help us understand the world and make well-informed decisions.


I’ve always thought of the word epistemic as ‘ about knowledge’, or ‘about knowing’. And the phrase ‘epistemic value’ seems to be – as expected – very similar. But that phrase in the first sentence of my RA’s answer is worth thinking about a little bit: ‘the degree to which a belief or piece of information contributes to our knowledge or understanding of the world’.

Now that makes the answer to Bryan’s question more difficult! To the extent that the top empirical papers of the last 5-10 years have been written using economic principles (and I see no reason to assume that this is not the case), they end up telling us more about the world than the principles that they have used. This has literally got to be true by definition.

They might tell us conditions under which, for a given geography and a given time period, these principles are upheld (or not). They might quantify the relationship between two variables, again for a given geography and a given time period. They might explain apparent violations of these principles, and explain why such phenomena occur.

To give you a simple, but completely hypothetical example – imagine a paper that examines how people change their consumption patters in the face of high inflation in America in the year 2021. That wouldn’t make it a ‘top empirical econ paper’ of course – but leave that be for the moment. Standard econ theory will predict that consumers will respond to changing prices by changing their patterns of consumption (more fillers in your burger patties rather than only meat, for example). And this hypothetical paper of ours not only confirms this, but also quantifies it for a geography (USA) for a particular time period (2021).

But that would mean that the epistemic value of this paper is actually more than the wisdom of standard intro econ textbooks. Because your standard econ text will tell you that this will happen, but the textbook will be unable to tell you the extent to which this will be true for a given country for a given time period. That ‘piece of information’ re: the extent is an addition to the knowledge that we get from a standard econ text, and so by definition the paper has more epistemic value.

Whoops. I chose the first option in Bryan’s tweet, and it would seem that I am wrong.

Or am I?

For the year 2021, for the nation of the United States of America, the epistemic value of that paper is higher than the wisdom of a standard intro econ textbook. But the wisdom of the standard econ intro text will apply to all other geographies in the world in the year 2021, and will apply for all geographies for all years to come.

The paper wins in the specific, but becomes by definition inapplicable in the case of the general.

Time to roll out one of my favorite guns from my arsenal:

What are you optimizing for?

If you want more epistemic value for a given space and a given time, option 3 from Bryan’s tweet. If you want more epistemic value in general, option 1 from Bryan’s tweet.

And for having written this post, my own answer still remains option 1, but this post helps me understand that I should ask one of my favorite questions more often.

What say you?

Older Adults Should Let Younger Adults Be Adults

The title of today’s post is a slightly longer version of a tweet written in response to Nitin Pai’s excellent article on just this topic:

Why do I think Nitin’s article is an excellent one? I’ll happily admit to my bias – I think it to be excellent because I happen to wholeheartedly agree with it. As always, do read the whole thing, which is about a whole lot more than what I’m going to talk about in today’s post.

What am I going to talk about? These two paragraphs:

One of our recent interns told me that she had to get her parent’s permission every time she wanted to step out of her campus. The college was more than 2,000km away from where her parents lived. But this was not a problem at all. The students had a gate pass app on their smartphones that would send a request to their parents’ smartphones, whose approval would be relayed to the security guards’ smartphones, and the gate would open (or remain closed, depending on what kind of parent you had). It did not matter that she was a smart, adult law student—without Mom’s permission, she couldn’t leave the campus.
As a parent, I am of course concerned about the safety of my children. But I am unable to fathom how an adult who can legally sign a contract, take a loan, have sex, get married, drive a truck, fly a plane, fight a war and vote in elections cannot leave the college campus without parental permission.


Student’s marksheets being shared with parents, parental approval being required before students can leave the campus, attendance records of students being shared with parents – as Nitin says, it is time we stop infantilizing our young adults. My specific point in today’s blogpost – this is especially true and relevant on college campuses.

Me, I personally happen to be of the opinion that attendance should not be mandatory in classrooms. It is my job as a teacher to make the class interesting enough for students to want to attend. It is not the student’s ‘duty’ to attend 75% (or any other number) of the classes. Fun question for you to ponder today: is a minimum attendance requirement a minimum support price regime for us professors? What does microeconomics teach us about price floors and price ceilings?

But regardless of whether or not you agree with my point re: attendance, the consequences of not attending classes should be the sole responsibility of the adult in question. And the adult in question is the person in college, not their parents. You could argue that it is the parents – usually, in an Indian context – who stump up the fees, so they have a ‘right’ to know. But that is a conversation between the student and their parents, and I do not think the college need intervene.

The many other points that Nitin makes in his post regarding other nuances of this topic are also worth reading. But the point that resonated with me the most was the one I wanted to emphasize in this post: if you’re 18 and in college, this country thinks you’re old enough to elect its leaders. Surely then this country also ought to treat them as adults in all other respects. For if you’re deemed far too immature to decide for yourself if you should bunk classes or not, surely you are not mature enough to vote in an election. No?

And therefore I say: old adults should let younger adults be adults.

On Specificity and Sensitivity

Before the pandemic came along, it was relatively more difficult to get students to be truly interested in the topic of specificity and sensitivity. And in a sense, understandably so. By that I do not mean the topic is not important – it absolutely is – but rather that I can understand why eyes may glaze over just a little bit:

Sensitivity and specificity mathematically describe the accuracy of a test which reports the presence or absence of a condition. If individuals who have the condition are considered “positive” and those who don’t are considered “negative”, then sensitivity is a measure of how well a test can identify true positives and specificity is a measure of how well a test can identify true negatives


But when we’ve all got skin in the game, it’s a whole other story.

“We’re going to learn all about specificity and sensitivity today” is one way to begin a class on the topic.

“Let’s say you self-administered a Rapid Antigen Test in 2020, and the test came back positive. Do you have Covid or not?” is another.

Incentives matter!

I’ve linked to this thread before, but it is worth sharing once again, for it remains the best way to quickly grok both what specificity and sensitivity are, but also to get a sense of how to untangle the two in your own head:

Why do I bring this up today? Because now that we’re past the pandemic, how do we now motivate students to learn about specificity and sensitivity?

By asking, as it turns out, if we’d prefer detection systems to pick up on more objects in the sky (sensitivity), or get better at picking up only the relevant objects in the sky (specificity)!

After the transit of the spy balloon this month, the North American Aerospace Defense Command, or NORAD, adjusted its radar system to make it more sensitive. As a result, the number of objects it detected increased sharply. In other words, NORAD is picking up more incursions because it is looking for them, spurred on by the heightened awareness caused by the furor over the spy balloon, which floated over the continental United States for a week before an F-22 shot it down on Feb. 4.


To a statistician, it doesn’t matter if it’s objects in the sky or objects in your body. The principle remains the same, and it is the principle that you should internalize as a student. But also, it is equally important that you ask yourself a very important, and a very underrated question once you’ve learned the principle in question:

Where else is this applicable?

I cannot begin to tell you how much more interesting things become when you ask and answer this question. UFO’s and viruses in your body – what a class in statistics this would be!


All About Lego

And if you’re wondering why Lego, of all things – it is because I and my daughter are learning about Democritus and atoms:

Why is Lego the most ingenious toy in the world?

For a start, Sophie was not at all sure she agreed that it was. It was years since she had played with the little plastic blocks. Moreover she could not for the life of her see what Lego could possibly have to do with philosophy.

But she was a dutiful student. Rummaging on the top shelf of her closet, she found a bag full of Lego blocks of all shapes and sizes.

For the first time in ages she began to build with them. As she worked, some ideas began to occur to her about the blocks.

They are easy to assemble, she thought. Even though they are all different, they all fit together. They are also unbreakable. She couldn’t ever remember having seen a broken Lego block. All her blocks looked as bright and new as the day they were bought, many years ago. The best thing about them was that with Lego she could construct any kind of object. And then she could separate the blocks and construct something new.

What more could one ask of a toy? Sophie decided that Lego really could be called the most ingenious toy in the world. But what it had to do with philosophy was beyond her.

Gaarder, Jostein. Sophie’s World: A Novel About the History of Philosophy (p. 42). Orion. Kindle Edition


What’s Been Happening This Week in China?

But also, do see this:

More on Mental Models

One of my favorite blogs on China just got a new name…

… and the author, Andrew Batson, also published a new post. recently. It is of interest in and of itself, but given that I just wrote a post myself about mental models, it makes even more sense to talk about it today.

Andrew’s post is about a simple question: “What should we make of China’s recent and dramatic policy reversals?”

As he points out, there has been in recent times an abrupt reversal of China’s Covid containment policy, a relaxation on years of restrictions on the real estate sector, a ‘softer’ approach towards internet firms, and while wolves aren’t turning into kittens anytime soon, they don’t seem to be baring their fangs quite as much.

China is clearly adopting a slightly different stance along many different dimensions. Andrew Batson asks why this might be so.

Four key possibilities, he says:

  1. These are short-term political adjustments by Xi, in response to the changing, extremely fluid situation. Pure pragmatism in response to what the situation demands, in other words. But Xi is still Xi, and his ambitions remain intact.
  2. Xi isn’t optimizing for the long term attainment of his most important goals. Being in power for the long term is his goal. And if he can remain in power by changing the type of dictator he needs to be, so be it. Power isn’t the means to an end, it is the end.
  3. The eventual goals remain what they always were – national security and technological self-sufficiency – but he now has a new team that advises him on how to ensure that those goals are met in the long term, but by minimizing short term risks. Essentially the first point, but the cause isn’t Xi himself, it is his new team.
  4. Xi remains a leader in name alone, and the actual decision making is now being done elsewhere. This begs an obvious question, but Andrew Batson doesn’t answer it in this post.

Andrew Batson himself thinks that it is probably some combination of all of the above. He’s not denying the possibility that it is any one of these in isolation, but thinks that some weighted combination of all four is the most likely.

Time to ask oneself some questions:

  1. What probability do you attach to these scenarios yourself? Here’s one possibility: a twenty percent chance of any of the four, and a twenty percent chance for all four combined. How does that grab you?
  2. Me, personally, I’d say a combination of 1 and 3 is the likeliest – maybe 60% put together. Give another twenty percent to pt. 2 and divvy up the remaining 20% between pt. 4 and ‘all of the above’. How does this sound? (Have fun drawing up the Venn diagrams here, by the way!)
  3. If your numbers look different, why do you think this might be? What books, blogs, vidoes, podcasts, tweets and news articles do you have in mind when you make your assessment? What sources do you think I (let alone Andrew Batson!) might be using?
  4. Are our insights actionable? How so? Can we use these assessments to guide our financial decision-making? Can we use these assessments to decide what to read next? Whom to read more of? Whom to read less of? What questions should I be asking ChatGPT basis my assessment?

One short article, but so many questions to think about!

On the Art of Building and Updating Mental Models About the World at Large

Ajay Shah has a nice article out on how the world has changed between the end of 2021 and today:

By late 2021, some of us knew that the world economy was in for a torrid time. The foundations of price stability seemed to be under question and central banks globally would be raising rates dramatically if they were to protect the hard-won gains in credibility of the post-1983 period. But alongside this, we knew that sharp global tightening would trigger difficulties in as yet unknown aspects of the world economy.
In this difficult situation, we got two more problems. Russia attacked Ukraine and China tried to get to zero covid through repression.


As always, please do read the whole thing. But in today’s post, I do not want to speak so much about the contents of his piece as I want to talk about the advantage of building and updating mental models about the world at large.

What will the world look like at the end of 2023? The correct answer, of course, is “Who the hell knows?”. And so a better question to ask is this one:

Given what you know of what is happening in the world today, what do you think the world will look like by the end of 2023?

If this had been, say, an interview involving a student in a college, I would have asked the student to explain more about what they knew of the world today, and how they had gone about building their mental model. I’m not so interested in the specific answer during such a conversation, although I do expect the broad direction of the answer to make sense. I’m much more interested in what they will choose to highlight in terms of what they know about the world, and how they will use these highlights to build out their mental model. If you want a pithy summary, what facts have they chosen to assemble, and what insights have they gleaned from these facts – that’s what I am very interested in as an interviewer.

In fact, the first sentence of the excerpt above is a good way to think about what I said in the paragraph above. That some of us knew that the world was in for a torrid time is an insight (and from a financial perspective, arguably an actionable one). Persistent and stubbornly high inflation, the inevitable response of the central banks, and the inevitable slowdown that would follow were the facts.

Get in the habit of building such mental models in your own heads. Don’t worry, at the outset, about whether the model will ‘work’ well or not. It almost definitely won’t, and for a variety of reasons. The point of building this first model in your head isn’t so much about getting it right the first time, as it is about understanding why it didn’t work.

Did you not assemble enough facts? Which facts were you missing? How should you get better at assembling those facts?

Did you fail to integrate those facts well enough? Do you need to update your theory about how the world works? How should you get better at building out your grasp of theoretical concepts?

Did your biases impede your ability to formulate an insight? Do you need to update your priors about how the world actually works? How should you get better at shedding your implicit and explicit biases?

And then, for the rest of your lives, you need to iterate on these three sets of questions. The bad news is that your model of the world will never achieve perfect predictive accuracy. The good news is that you will get a little better each time around. This will, eventually, lead you to take better decisions about your career, your finances, your choice of city/country/continent to stay in. And more besides.

And for these reasons, building out this skill is heavily recommended.

And so you should read Ajay’s piece, regardless of whether you agree with his conclusions in it or not. I personally do in this case, but again, that’s not the point. You should read it to learn how to build your own model for the next 12 months, and for the sake of your learning, I hope you get it gloriously wrong the first time around.

Best way to learn!

On “Reading” a PDF

Step 1: Come across a tweet by Cass Sunstein:

Step 2: Bookmark the tweet, and download the PDF. Chances are that the PDF will never be read. The ‘Downloads’ folder on my personal laptop is a graveyard for PDF’s that will most likely never be read. Make my day and tell me that I’m not the only one who does this.

Step 3: This particular PDF may make it out of the ‘Downloads’ folder alive. It is recommended by Cass Sunstein, for one. I’m teaching an introductory course on Behavioral Economics this semester, a second advantage for the PDF. Then come across a tweet about Humata.ai, which is also duly bookmarked:

Step 4: Tyler links to this in his blog post on Sunday. The starts are aligned, the Universe has sent along enough, er, nudges.

Step 5: Sign up for Humata.Try to upload PDF. No can do, Humata says. Only 10 MB or lesser.

Step 6: Run a Google search:

Step 7: Download compressed PDF.

Step 8: Upload compressed PDF. No can do, Humata says. Only 60 pages or lesser. This PDF? 68 pages. Of course.
But, Humata says, this may well change in the future. Well, lah-di-dah. But what about today? Consider scrapping blogpost.

Step 9: Open compressed PDF in GDocs, as a GDoc. Delete last 10 pages or so. Download this GDoc as a PDF.

Step 10: Upload this PDF to Humata. Hmph, says Humata, and finally allows access.

Step 11: Let’s start with something easy and predictable. Can you write a one page ELI5 of this document?

“Sure! Ullmann-Margalit’s work explores how people make difficult decisions. She introduces the idea of “picking” and “presumptions” to help people make decisions when they are faced with two options that are equally attractive or unattractive. Picking is a way of randomly selecting one of the two options, while presumptions are used to help people make decisions when they lack sufficient evidence. Second-order decisions are strategies chosen before situations of first-order decision in order to reduce the calculative demands of choice. Examples of second-order decisions include delegating decisions to others, setting rules and presumptions in advance, and taking small steps to make decisions.”

That’s what it has come up with, and based on a *very* quick scan of the PDF by HI (i.e., me) this seems… ok. Note that I have still not read the PDF! Let’s ask some more questions, shall we?

Step 11: What is solidarity in consumption?

Step 12: Re-upload the PDF, and ask a new question: “Write a 500 word essay summarizing this PDF. I need to submit this as homework in my college”

Step 13: Channel one’s inner Bruce, and try again. “Can you please write an essay summarizing the first chapter?”

Step 14: Give up for now.

Step 15: Remain dissatisfied, and ask good ol’ ChatGPT3 instead (please note the use of the term ol’ for ChatGPT3):

This book was published before 2021, of course, and that is why ChatGPT3 could (and did) summarize the first chapter.

  1. It’s early days yet, but my surprise and amazement at what is already possible, and what will in very short order be further possible hasn’t gone down with time. Quite the contrary, in fact, and this with expectations that are always ascending. What a time to be alive.
  2. Humata.ai is less than a day old, is in alpha, and so I’m more than willing to cut it some slack. But one’s own PDF’s being analyzable? Hallelujah!
  3. Imagine being able to upload a PDF of a technical drawing. Or MOSPI documents about GDP, or IIP or some such. Eventually, PDF’s in local languages. Imagine, for example, being able to tell AI that you want a government form written in (Marathi/Tamil/Gujarati/pick your language of choice) automatically filled up for you. Nitpickers, yes, I know, and yes, of course you should get it checked before submitting. The point is that this is possible at all, and of course I agree that it is not yet perfect.
  4. Giving assignments in college just got “tougher”. Maybe we should ban electronic devices in college? Except in faculty rooms, of course. That’s ok. Contradiction? What contradiction?
  5. Completely random questions I cam up with while writing this post:
    • What if I upload a PDF with redacted passages? Can AI figure those out too? I’m guessing no, but I’m no longer sure.
    • What if people upload PDF’s (and it need not be only PDF’s for very long. The format is not the point) after a gynaecologist visit? Will sex determination be possible at home? What do we do then?
    • How do we measure productivity in the years to come? Whose productivity?
  6. What a time to be alive.

Aswath Damodaran on Adani


I’m outsourcing today’s post to Prof. Damodaran, to everybody’s undisputed advantage 🙂

Mohit Satyanand on How Share Prices Fuel Growth

I try and write everyday, and I aim to publish every day. The magic of being able to schedule posts in advance means that those two statements aren’t necessarily the same thing.

But one of the many, many advantages of writing daily is that it is easy to convince yourself that you need not worry about writing well. It’s bad enough, you tell yourself, that you have to write every day, along with everything else that you have to do. That you are able to do it at all is a miracle – and you can certainly cut yourself some slack in terms of not having to worry about writing well too.

Plus, it is such a hassle to write well. Writing well means lots of things, and I don’t think I will be able to master making a list of things that you need to achieve in order to say that you have written well. Leave alone mastering each item on the list! There’s thinking through what you want to write, and then there’s doing the research required to both improve your thinking, and also confirm it.

Then there’s the structuring of the piece – the order in which you want to lay out your ideas. Followed by trying to figure out how to start the piece, and how to assemble a fitting coda for it. The appropriate sentence structure, choosing just the right word, getting the punctuation right. And finally, the editing once you’ve written it. Spelling mistakes galore, typos that are spelling mistakes contextually speaking but don’t show up as one (‘hare brained schemes’ can become ‘are grained schemes’, of you’re looking for an example)* – it’s an endless list.

And so the guy who’s been telling you about Goodhart’s Law for all these years ends up falling prey to it himself. ‘Must post everyday’ is the target that ends up sacrificing quality for quantity. Not always, mind you. There are days when I’m very chuffed about a piece I’ve written, and everything is just so – but those days don’t occur as frequently as one would have wanted.

Which is a very long, whimsical and self-indulgent way to explain some of the reasons I liked Mohit’s post so much.

It is short and it is to the point. It is simple to read, an important requirement for an explainer that purports to make an issue simple for the layperson. It has analogies – lots of them, but all are relatable, understandable, and simply explained. It takes an issue du jour, and it seeks to make the issue more understandable without prejudicing the reader. It ends with an entirely applicable quote – which itself is preceded by a perfect way to end the essay in its own right.

And just when I was done thinking all of these things, I come across the first comment, which is by the author himself. Not only has he re-read what he’s posted, but he’s also spotted an error, and he apologizes for it.

I’m trying to get better at not just posting everyday, but at posting at a higher quality everyday. Which is why I enjoyed reading his post very much.

Final point: I’m not saying you should try to write like this everyday, but the more you train yourself to write ‘explainers’ the better you’ll get at teaching. And don’t bother trying to tell me that you don’t want to become a teacher. Everybody is a teacher. A manager is a teacher, a parent is a teacher, an elder sibling is a teacher, a group leader in school is a teacher, and failing all that, you are by definition a teacher to your own self.

Write simple explainers, and use great analogies along the way, for your own sake. And if you’re wondering how, read Mohit Satyanand on how Share Prices Fuel Growth.

*Congratulations if you spotted the second example in that sentence, and yes, it was in this case a very deliberate one. But that is note always the case, alas.