Assorted links

1. Markets in everything, iPad edition.

2. Krugman further explains his views on Japan.

3. Derek Thompson’s skepticism about robots.  I say the correct view is not about “robots” putting people “out of work,” but rather how software and machine intelligence are restructuring relative wages, which in turn has subsequent implications for labor force participation rates.

4. How much healthier are the baby boomers?

5. Feynman’s restaurant problem.

6. How much consensus in the economics profession?


"I say the correct view is not about “robots” putting people “out of work,” but rather how software and machine intelligence are restructuring relative wages, which in turn has subsequent implications for labor force participation rates".

With all due respect, Tyler, is that not "he says tomaatoes, I say tomAtoes"?

The point in both cases, it seems to me, is that robots/technological progress is destroying jobs... as it always does and always did. However, the difference with the past post WWII is that these productivity gains are not shared. Not to self promote again but:

Last remark: How can you say that software & co is putting labor force participation rate under pressure and maintain a belief in TGS? I am genuinely curious as to the answer as I would have thought the two ideas to be quite irreconcilable.

If I purchase a Roomba and that saves me the trouble of vaccuming the floor, how am I not sharing in the productivity gain?
If a robot attends my hospital bedside and provides me with refreshments, how am I not sharing that gain?

Surely the improvements to quality and lower costs to consumers allow consumers to share in the gains from productivity, not just evil capitalists.

The issue is that a wide swath of the world's population isn't likely to ever be productive again. If producers are ~20% of the population, and the rest of the economy is cosumer surplus how does the 80% that isn't productive share in consumer surplus (government redistribution is an answer, but it tends to greatly increase negative externalities of the unproductive class).

Why are they unproductive? Because of robots?

What about comparative advantage? Even if robots are better at everything, the humans can do the things that humans are least bad at. Robot maids and waitresses aren't happening any time soon, and because of gains to productivity due to the robots, the consumer price of everything will fall, so a maid and or waitress can afford to buy more.

I don't see any economically rational reason why you would allow a large population of unused labor to sit idle. Furthermore, I don't see why they would *choose* to sit idle unless we're giving them transfer payments in the first place. In a state of nature, idle humans build themselves mud huts and farm and engage in trade, even if they can't afford to purchase a Lexus. Give the same person a welfare check and a subsidized apartment, and they'll stop being productive. Take away the welfare check and they'll find something to do, they'll plant community gardens and trade amoungst themselves, or starve. Unless we're for some reason forbidding them from growing their own food and building their own houses. That could be a problem.

From Office Space:

"Well, you don't need a million dollars to do nothing, man. Take a look at my cousin: he's broke, don't do shit."

We all have the option of doing nothing today. People choose to be productive even when transfer payments are available. My opinion is the goal should be the only people working are those that want to work. Do I want a waiter who would prefer to be on welfare? No thanks I'd prefer to get the food myself and wait for the arrival of robot waiter.

If you're giving them transfer payments, you're paying them either way.

So you'd rather pay them to stay home and do nothing than pay them to go get you a drink?

It may not be rational or economically efficient, but it's happened repeatedly throughout history. Just look at pretty much any slave society. You have the rich, the slaves, and a class of non-slave poor who eke out a meager existence, but who can't find work. For example, the mob of Rome or the hillbillies of the antebellum South.

If there is someone or something that can do everything you can do and do it cheaper than what it takes to keep you alive, you're going to have a problem.

It's not that they're unproductive, it's that they're not either as productive as their transfer payments, or as productive as their robotic competitors, nor smart enough to help automate something else. Technology is rapidly raising the bar for how smart one must be to add value to a firm above their wages (even at minimum wage/no wage--internship).

The television program "Gold Rush" is a vivid example that sometimes labor not only doesn't add to output but it can greatly reduce output. Increasing numbers of people are entering that category, and with better software it's become much easier to identify who those people are (thus the current hit to future employment prospects from even a short unemployed stint).

The real problems arise when the young men who aren't more productive than their transfer payments, and thus lacking other ways of gaining status (everyone in their neighborhood gets the same transfer payments) resort to violence to gain in status.

When jobs increasingly rely on creativity and innovation and montoring tools improve, employment has become a race to employ people like Jeff Dean:
CS is one of the easiest to id talent, thanks to signed code repositories, but it's slowly becoming easier in other disciplines as well. When the average employee works along with people like these, they not only don't increase productivity, they generally make the team less productive. As this ability to id super performers spreads to more professions, the bar to being productive rises rapidly.


The same argument has been made about international trade and is soundly rebutted by comparative advantage. Even if a robot is better than you at everything, there aren't an infinite number of robots. The robot labor is then devoted to producing things that the robots are best at, and the human labor is devoted to doing things that humans are least bad at compared to robots.

If this argument works for international trade, how can it not work for humans vs. robots?

"So you’d rather pay them to stay home and do nothing than pay them to go get you a drink?"

In many cases yes. I want to interact with people who are happy. An unhappy waiter makes the meal less fun. There is a tradeoff. I am not saying how much I am willing to pay for them to stay home. I think what we are paying in transfer right now is about right (in the US). I would be willing to pay a little more but not a lot more. My hope is as productivity increases we are able to allow more people to avoid doing jobs they don't want to do. Somehow our projections say we will need to pay more for the same % of people to stay home. This doesn't make sense to me. That increase is mainly based on increases in healthcare costs that I don't think will really happen.

Hi, Hazel

To your first comment, I would echo Bluto's reply. I need some kind of income to buy the Roomba. However low the costs, it's unlikely to be zero. The present-day situation, where a lot of people have insufficient incomes, with lots of insecurity as to their job safety and where debt is no longer able to hide the underlying condition of the middle class, speaks for itself.

NB: I've never used silly stuff like 'evil capitalists'. I love entrepreneurs in particular and the business world in general. I just strongly dislike rent extraction.

You also say: "I don’t see any economically rational reason why you would allow a large population of unused labor to sit idle".

I mean, hello? Where have you been lately? The rise of unemployment we've experienced is pretty damn obvious and I simply don't buy that people sit idle just because transfer payments allow them to do so. In Germany, people work for as low as 0.55 Euro an hour. Try bying any modern product, even one with prices coming down, with that kind of income.

Besides which, one of the best historical example we have of a world where the labour force has been displaced is Rome, as I mention in my blog post. The Roman elite did find it pretty dangerous to let all those nominally free people starve. They're much more likely to riot before dying. However, one thing they could not do is work at jobs where they were comparatively least inefficient compared to slaves.

Finally, you suggest people left on their own would do 'something'. Yes. I've seen that in emerging countries where no transfer payments or social safety nets really work well. The people start growing their own food and such. Guess what? They die early and they don't get to enjoy the productivity gains of the Roomba. Thus, we're back to square one. What's the point of having robots producing roombas if we're all forced back to mere subsistance farming?

However, one thing they could not do is work at jobs where they were comparatively least inefficient compared to slaves.

Really, why not? If there is unused labor, it can always be employed producing something. And if left unusued, it can create employment for itself by trading amoungst themselves.
One thing Rome also had was the grain barges from Egypt. That was their own version of transfer payments. And that's how Ceasar was able to rise to become emperor. You control Egypt, you control the grain shipments, you have the mob on your side.

I think our problem is that our regulatory code effectively makes it impossible for people at the lowest level to actually engage in any kind of productive self-employment. Building codes make it impossible for people to build their own homes. We just passed a law that forces crafters making home-made baby clothes to send away to get the buttons tested for lead. People get busted for growing some of the few high-value crops they could easily produce - marijuana. Or for having vegetable gardens in their front yard. Or operating a beauty parlor without a license. Or giving people rides for pay without a cabbie medallion. If you make it illegal for people to do anything to help themselves at the lowest level, yeah they're going to choose transfer payments.

Sorry, Hazel, I don't see the relevance in the Roman case of 'where' the grains were actually grown. Yes, Egypt was one of the key supplier. So what? It's like slaves + Egypt = kinda robots + China. It's not changing the issues.

As to those regulations you describe, I probably would support most of them. Buttons tested for lead = good. Some of the others would need to be discussed further. In general, it seems that in the USA, lots of local regulations are used to protect incumbents. That's bad. OTOH, small businesses' profit margins are already pretty thin, I think. They are not the ones achieving historical highs. That's the S&P500 type companies. To me, this suggests that more small cies would not be sustainable/a game changer.

Besides, let's imagine that, in world of robots taking all the work, the humans are then freed to build their houses by hand and cut their hair by hands/scissors...

How's that good? You've got a stone-age human society cohabiting with a 22nd century robotic one and that human society is going to have to redo everything done before, with the added risk that, as soon as they get a little indigenous wealth, it gets vacuumed by the robots-owners via trade (similar in some ways to trade between developing nations and developed ones when the developing nations' industries are not protected by infant industries trade regulations).

My point is: If we get to the point where robots could free us of work, I am not willing to go back to building my own mud hut and a hunter gatherer/early farmer lifestyle. I'd rather start a revolution.

If you want buttons tested for lead, as a consumer, you're free to volunarily purchase baby clothes that advertise they are tested for lead.

What you are talking about is making it ILLEGAL for someone to make clothing in their home, and sell it ot their next door neighbor, unless they get the buttons tested for lead. That kind of regulation essentially makes it illegal for people to be productive at all below a certain level of competence. it takes all the people below X level of productivity, and sets it to zero.

Now, I don't actually *expect* that everyone below a certain level will be reduced to hunter gatherer status. What will happen is that as Roombas get manufactured by robots, the price of a roomba (and everything else) will fall and make it possible for lower-productivity people to purchase a roomba. So the threshold of productivity you need to be at to participate in the all-robot society gets lower. There will always be some number of people who are completely incapable of reaching that level of productivity, but if you let the bar fall, more of them will be able to benefit from it. It will get to the point that you can purchase a Roomba with your deer-hide home-made mocassins that you made in your mud hut, which incidentally you will easily be able to replace with the pre-fab robot-made trailer. (besides the fact that deer-hide homemade mocassins will likely be a boutique product the that very rich will pay a lot for).

With all due respect, Tyler, is that not “he says tomaatoes, I say tomAtoes”?

No, it's more like, "he says tomatoes, Tyler Cowen says tomatoes as well, but in a way designed to communicate no insight while making him appear intelligent like he has some refined insight on the matter."

Now Krugman can have tempura on his face when this blows up, to match the egg on his face from the housing bubble.

Remind me not to go for dinner with Feynman or Gottlieb -- like many economists, they can do the math, but they can't provide a practical solution to even a simple problem. To wit, note that Gottlieb's solution means that you don't get to try your second dish until the third visit, your third until the 6th, your fourth until 9th, your fifth until the 14th, etc. That's just plain silly and non-useful even at a first glance. Then, notice also that the more visits you make, the more dishes you have to order -- for the first few visits, you eat but one dish, but at the fifteenth you eat six. Get lost! This is clearly not a good strategy. I think this is a simpler and better one: go out with your favorite friends, say three which is a good number. Order four different dishes randomly, then share them family style. Next time, go with three different friends, and repeat, but avoid the two worst dishes from the first visit. Then continue more visits using this strategy until you are either sick of this restaurant or have successfully ranked every dish on the menu. Here is then your choice: Gottlieb's mathematically optimized strategy, or my more socially optimized. Which one do you think will give you a better meal and a faster trip to your goal of figuring out the restaurant menu? Right, that's what I thought. Good eating!

"note that Gottlieb’s solution means that you don’t get to try your second dish until the third visit, your third until the 6th, your fourth until 9th"

No, M is the total number of meals you will have in that restaurant ever. You try a new meal every visit until you settle on your favorite after day D.

Pretty sure there's a typo and that M is supposed to be N, the number of dishes on the menu.

So if a restaurant has 10 dishes on the menu, you should stick with your favorite dish after trying ~3.5 or so.

To clarify, M is the total number of meals you will have but is constrained at a max value by the number of dishes on the menu. In theory, once you have tried all the dishes on the menu you don't need to try anymore to determine your favorite.

(Although the formula means you will never try all the dishes on a menu for a restaurant that serves more than one dish.)

No, there is clearly a mistake. It says: "Given N (dishes on the menu) and M <= N (meals to be eaten at the restaurant)". It doesn't make sense to say M <=N. Perhaps it should say D <= N ?

Oh, I finally get it. Apparently I forgot what the word "given" means.

you don’t need to try anymore to determine your favorite.

Except that at most restaurants the cooks change over time and dishes change over time and the number of dishes changes over time. (In most restaurants, there is no stagnation, more often a great decline.) This happened in several Thai places we used to go to: starting out the food was amazing, over time it got sweeter and much less amazing.


No, you have the wrong attitude to this problem: Feynman and Gotlieb are admittedly building a simple model to solve a complex problem, but the virtue of simplicity is that all their assumptions are stated up front to generate a determinate answer to the problem, or as it is often said: all models are wrong, some are useful. This particular model is, I submit, useful because it gives us a new way of thinking about a difficult problem

It's an interesting variation on the problem of optimal search. But I find it has much the flavor of the Games of Fair Division (such as how to slice a pie fairly) that mathematicians study; their notion of fairness often differs from economists'. In particular, economists are more likely to set up problems, and solutions, which involve negotiations, side payments, bidding, and heterogeneous tastes and outcomes whereas mathematicians (and physicists, in Feynman's case) are more likely to leave those out and use a model where the fair outcome is something along the line of equal shares for all.

In this restaurant example, Feynman ignores a couple of variables that would be top-most for an economist: what are the prices of the dishes? And what are the costs of search? The latter one might be profitably ignored, e.g. we might assume that you're going to be going to a restaurant anyway, and thus there's no additional cost associated with going to this specific restaurant. We could even try to claim that the model incorporates price by looking at satisfaction with a dish as combining the attributes of the dish with the price that you paid to eat it. But that sort of abstraction of the role of price is not something that would be most economists' first choice when modelling consumer behavior. I know that I'm checking out prices, as well as the nature of a dish, when I'm looking at a menu.

The real failure of Feynman's model is that it assumes the value of a dish increases linearly. In the real world, there's a power law in effect: The best dish on the menu is likely several times better than second best dish, which is likely several times better than the third best dish, etc. The utility of discovering the best dish on the menu may dwarf the negative utility of suffering through all previous experimentation, even if you have to try the entire rest of the menu first. You can't know ahead of time.

There is no credible alternative to exhaustive search. Sorry.

Nice to see at least some service being paid to the position that Japanese stagnation has been a myth.

Feynman's solution to the dessert problem, which admittedly he came up with as an undergrad, is much simpler.

2. It would be interesting to compare the GDP per working age of adult productivity growth of Japan to the US over the past couple of decades. Japan as a basket case has been exaggerated. Japanese protectionism in the 80s was exaggerated. Japanese youth suicide has been exaggerated.

'Almost 40 percent of the boomers are obese, compared with 29 percent a generation ago. '

Ah, the advantages of American suburbia.

Ah, the choices people make

Perhaps, but probably not in the purely cognitive sense of a choice to be obese as the happier outcome. That matters, because "people can choose" doesn't always mean choice in the slow, considered, rational sense. If people, good people, can choose badly ... we shouldn't have a political philosophy that pretends otherwise. We shouldn't prefer an imaginary populace.

In Germany, the fattest country in Europe, 67.1% of all men between 18 and 79 are considered overweight with a BMI of 25 or greater.

So glad America can provide false comfort to tubby Germans. Like Thornton Mellon said: "If you wanna look skinny, hang out with fat people."

It's kind of like how spendthrift Europeans provide Americans with false comfort about fiscal rectitude. What an alliance!

Boom goes the dynamite. Chance prior replies? I'd say 10% max.

Sure I'll reply. Though without a source, it is pretty hard to actually make a factual replay, so instead of just using a number, I'll use an observation based on two decades of experience. The most reliable comment Germans make after visiting the U.S. is just how large Americans are. Mainly because having a BMI greater than 25 is nowhere near the impressive feat of having a BMI greater than 30.

And when talking about comments - the ones that get deleted here tend to talk about things like domain names and their owners and internal GMU history.

Holy crap! Americans are even fatter than we are!


I notice myself reading comments less and less when I expect prior_approval to be around. E. Barandarian, when he was still around, was annoying, but at least I could imagine this elderly, gray-haired, gentleman magnificently and stubbornly typing his comment letter by letter, key by key. Did he wear a hat when commenting?

prior_approval is so predictable and just annoying, but because his comments are often longer than my screen, much harder to simply ignore.

+1 for this on the spot comment.

'because his comments are often longer than my screen'

Reading is a dying art in an age where screens are shrinking? - who says things are stagnating when we can just ignore them instead.

He'll probably make a "joke" about deleting his own comment.

No joke - this comment was also deleted -

'Nothing like equating the loss of job positions, possibly including tenured ones, with the start of mass persecution in academia.
Well, at least if one is an economics professor who has been well paid on the side over the years from a certain source.
So let’s quote the actual text –
‘First they came for the communists, and I didn’t speak out because I wasn’t a communist.
Then they came for the socialists, and I didn’t speak out because I wasn’t a socialist.
Then they came for the trade unionists, and I didn’t speak out because I wasn’t a trade unionist.
Then they came for me, and there was no one left to speak for me.’
And do note, this is the approved translation of the Martin-Niemöller-Foundation.…
Well, I’m sure that the Faculty Director of the Mercatus Center can be satisfied that none of the above listed people actually exist any longer in America. With the truly vague exception of trade unionists – though the trade unionists in Philly are unlikely to care about what happens to either owner of this blog.
And even my normal excuse that such abhorrent writing is ‘satire’ will not work in this context – Prof. Cowen has more than a passing knowledge of Germany, and its truly evil history, including the reality from which this quote arises.'

Not sure any footballs should be spiked until you give the % of Germans who are "obese" rather than merely "overweight".

Correct, our "obese" starts at BMI 30

Obesity is not caused by suburbia.

It is certainly a component. A desk job, and a long commute, contribute stress without exercise. Stress, and eating, without exercise, cause obesity.

See also the mean weight of NY's vs your average commuter city. The walkers win.

Obesity is more a problem among the poor, who are not as often out in the suburbs.

I believe that is a dated observation. The poor the first "20th century obese" perhaps.

Im not an economist, so forgive me please if this is a stupid or irrelevant question, but WRT Krugman and Japan, and, macroeconomics in general, it seems that every example that reinforces ones macroeconomic theory is held up as proof of that theory's correctness, and every counter example is explained away as being overpowered by some microeconomic force. I mean, i get that one cannot really experiment on a macro scale and come to some definitive conclusion, but how do economists even know that the theories they hold are true or relevant at all? IF demographics explains why Keynesianism doesnt seem to be working in Japan then how does Krugman know that there isnt some non-macro force or forces at work in places that he thinks of as Keynesian success stories as well? How does he know that there arent additional non-mcaro elements affecting Japan that further complicate his Keynesian dialog?

How does one prove or disprove Keynesian economic theory? Or Austrian economic theory? Or any macro theory? How do you guys know what you believe in isnt just total bullshit? Im seriously asking because i simply cant imagine a scenario where a macro guy says "this conclusively proves that my preferred macroeconomic theory is incorrect"

This is exactly what Krugman does with all of his analysis. If the data support his case prima facie then he just throws up a graph and case closed. When they don't support his case, he needs to delve down a layer. Or two. Or three. He peels back the onion layer by layer until he finds one that supports his case.

That's why, for example, sometimes austerity can be measured by "nominal federal government spending" and othertimes it's the "growth rate of inflation-adjusted federal spending per capita adjusted for demography excluding transfer payments but adding back in changes in state and local non-education spending."

Krugman may be guilty of that (most economics bloggers are), but that seems to be a part of debate in most subjects. It doesn't change the fact that circa 2008 he was able to explain with reasonable accuracy what the consequences of the financial crisis would be, and why fears of debt crisis and higher inflation were misplaced. Those predictions were based off his experience studying Japan in the '90s. For me that gives him more credibility on Japan than economists who have spent the past 5 years warning about inflation and "easy money".

Funny how his exact perscription, which never changes BTW, blew up spectacularly circa 2002.

Funny how you call his stance against deficit spending in 2002 the same as his stance today.

What you are saying is really just proof of my point. The fact that he was 'right' that time is held up as proof of his correctness. The fact that he has been wrong in others is explained away as complicated by other factors. Whos to say that he wasnt really wrong about the financial crisis and only appeared right due to complicating factors? If you are right when you are right, but when you are wrong its something else, then you are never wrong.

And you are right, this isnt just about Krugman, its about all macro. It all seems unfalsifiable to me. Under what circumstances would you say that a macroeconomic theory has been dis proven? How does that occur?

It's also worth pointing out that Krugman debated Sumner and claimed that BoJ was relatively impotent to ease further, but in the new column he seems to have shifted to the Sumnerian view that the BoJ is to be blamed for tightening every time inflation starts up which goes against his claims of monetary impotence thus requiring Keynesian stimulus. Sounds like a pure market monetarism explains better both Japan and the 2008 crisis.

I agree with you -- it seems nigh impossible to falsify some macroeconomic theories; at least there don't seem to be agreed upon standards for falsification. It seems like the problem is that relative to the complexity of the field we don't have nearly enough data. It allows almost any model to explain the data with the right calibration, so models get fragmented and end up working only in specific economic domains.

So when I say something like "Krugman is right on Japan", it would probably be better to say "Krugman's predictions about macroeconomic behavior in a liquidity trap without NGDP targeting by the central bank have been accurate relative to many other economists". As a result I trust his worldview more in this particular situation. I don't necessarily agree with him that central banks are impotent to solve the problem on their own -- I still view that as a largely untested hypothesis. With some luck we may see it tested soon in Japan.

People makes a arguments for how the droids are taking our jobs and then proceed to substitute breathlessness for argument in their "optimism".

This reminds me somewhat of the immigration debate and jobs. At the critical juncture the debate becomes about feelings.

In both cases, the idea is to be invited to give talks to people who think they are immune to losing their jobs so that the speaker can salve their collective consciousness. However, they needn't worry. They too will probably lose their jobs. They won't even be able to compete with the sex droids in the "new service economy".

So what's the _real_ answer?

Here's something he could be talked about but isn't because it is depressing to think about how hard it would be to implement given the powers that be:

A citizen's dividend paid equally to all adults that is financed by collecting the economic rent of civilization.

Since the powers that be have increasingly turned to rent-seeking, they will, of course, resent the very idea that "their" rents should belong to everyone. And they're in a position to make sure that doesn't change, aren't they? As a result, things _will_become dystopian.

Thanks. I've been trying to come up with some way to come up with a logical way of thinking about this issue other than this pessimistic argument, but I've not been able to summon the required creativity.

"Here’s something he could be talked about but isn’t"

Except it's talked about all the time.

1. I thought it would be "There is no great stagnation" ;-)

Well that was the title the last time it was posted here...gotta keep it fresh. I had to search for "potty" to find the old post, see the first link. (Not entirely surprised that a Krugman-related post came up with this word too.) Now that this thing is on Amazon Prime, which it was not the first time posted, I am not sure I can resist. My two year old son loves his apps on my iPad but he does not like his potty time much. Still this would have to be one of my stupidest parent purchases (in a while).

"the Bank of Japan has always pulled back on monetary policy when the economy looks better, instead of doing what it should, which is to keep the pedal to the metal until the inflation rate is solidly into positive territory."

Of course. The problem with Japan is that there are no True Scotsmen there!

(5) Isn't Feynman's restaurant problem just an example of the optimal stopping problem?

It's similar, but not the same. I thought it would be the , too, but the answer is different because you can return to your prior best choice.

#5: yeah, but how much did he tip?

I hope robots put us all out of work. The future should be just like The Jetsons.

Available on your iPad6, Fall 2023.

From #5: The average total ratings in a sequence of meals that includes n "new" dishes and b "best so far" dishes can be no higher than the average total ratings in the sequence having all n "new" dishes followed by all b "best so far" dishes.

Interesting and clearly true. So it's always best to start with a flurry of exploration before settling in on a favorite. However, I think the maximization function should include some utility value for diversity itself, rather than just maximizing the average rating.

#4 is a great example of yellow journalism. For example our parents were not considered hypertensive until they were over 100+ their age.

3) I buy broadly the "robot problem." One way to tell it might be that mechanization on the farm freed workers for the factory. Automation in the factory freed workers for ...

We need a sink with high labor requirements initially, and productivity gains for the future. We don't have one. "We freed factory workers for Walmart/Starbucks" is not such a nice stage of civilization.


Possibly, though I think kickstarter is pulling down the big bucks.

Though we might both be thinking of the top few percent successes at those places.

#3 to me it is all about the deltas. That is robots eliminate jobs faster than adjustments can make new jobs that you get involuntary unemployment. There are plenty of things that need doing that humans can do better than robots. Replacing the minimum wage with a wage subsidy might help in times were robots eliminate jobs fast.

Are the people who are skeptical about robots parents who have raised a kid for 18+ years?

I'm not the best way to submit links to you two, but this price action on legos is up your alley.

Feynman's restaurant solution leaves out discounting. As M (number of expected meals) grows, he would say I should try more and more dishes. But that's not what I do because, in addition to wanting great meals on average, I'm interested in having a great meal *today*.

The function to be maximized should not be the average meal, but the net present value of all meals, discounted at an appropriately risk-weighted rate. I haven't done the math formally, but I know that when I'm at the table, looking at the menu I am strongly inclined to order what I know will be good.

Comments for this post are closed