Tuesday, July 08, 2014

Why Improving Things Is Hard

Someone who studies the effects of different diets comes up with evidence that consuming more salt causes high blood pressure and associated medical risks and concludes that people should eat less of it. Another researcher repeats the study looking at mortality from all causes and finds that people who eat more than the average amount of salt are no more likely to die early than people who do not.

Arguably you not only should not be surprised at the result, you should expect it. We have, after all, been "as if designed" by evolution for reproductive success. Dying, whether from high blood pressure or some other cause, makes it difficult to either help rear your existing children or produce more. Our bodies have built into them a sophisticated chemical factory for converting what we eat into what our body requires; reducing the amount of salt absorbed and excreting the excess should not be beyond its capabilities. If the body instead  absorbs all the salt consumed, that suggests that the disadvantages of higher blood pressure are on average balanced by other advantages.

That is one example of a more general point, suggested by my rather vague memories of what I have read about the effects of salt consumption and a comment someone made on an earlier post. For another, and this time entirely imaginary, example of the same point, suppose you discover that increasing the size of your car's tires improves its gas mileage. Before concluding that bigger tires are a good idea, it would be prudent to look at other consequences of the change—because you will probably discover that some of them are negative.

The logic is the same as in my previous example, although this time the design is by engineers rather than by evolution. People who design automobiles would like them to use less gas. If bigger wheels achieved that objective and had no disadvantages, someone in the past century would have discovered the fact. The current size of wheels is not an accident. It is the best solution engineers could come up with to the problem of optimal design.

The argument does not apply to everything; there is no reason to assume that either the climate or the population of the earth is optimal. But it does apply to anything that has already been optimized for some purpose, whether by human design or some natural mechanism. Any change from the present design that produces a benefit probably also produces a cost. That is why it is not already in the design.

Of course, even if something has been optimized, it may have been optimized for a purpose other than yours. Evolution designs organisms, including me, to do the best possible job of passing their genes down to later generations. That is its objective but not mine. Birth control is one of the ways in which humans subvert the objectives of the genes in order to better achieve their own objectives. 

For another example, consider trade barriers such as tariffs. There are good economic arguments to show that we would be better off if we went to complete free trade. That seems puzzling—if we would be, why don't we?

The answer is provided by public choice theory, the branch of economics that deals with the workings of the political market. A tariff makes the inhabitants of the country that imposes it worse off but the politicians who pass the tariff better off, since it benefits a concentrated interest group at the cost of dispersed interest groups. More concentrated interest groups are better able to pay politicians to do things for them.Trade policy is optimized, but for the wrong objective.

Another exception to the general rule occurs where optimization is slow and constraints have recently changed. Through most of the history of our species it made sense to get fat if you could in order to increase your odds of surviving the next famine. Now that famines are vanishingly rare, the same hardwired tastes produce less optimal behavior. Similarly, if gas prices have recently gone up a lot, existing car designs may weight fuel efficiency less heavily than  they now should.

Improving things is not always impossible. But it is often harder than it seems.

12 comments:

Darf Ferrara said...

Just to clarify the point you made in your penultimate paragraph, getting fat when you could (i.e. your brain telling your body to dump huge amounts of insulin into the blood as soon as food is available) is not simply a genetic trait, but is strongly coded for by how much food is available to the mother during the third trimester. The Dutch famine of 1944 is an example of how strong this affect is.

This strengthens your point that it is difficult with complicated designs, especially those with feedback mechanisms can be difficult to improve on.

Power Child said...

You say matter-of-factly that a tariff makes the inhabitants of the country that imposes it worse off. Is this really always true, especially keeping in mind your mode of thinking here about taking account for all other advantages and disadvantages?

Anyway, something I'd add to your statement about objects that are engineered is that they are never simply engineered, but rather they are engineered for a specific set of end users. Keeping the experience of those users in mind informs where the "optimal" point is.

David Friedman said...

Power Child

"Always" is a bit strong, but it requires rather special circumstances for a tariff to benefit the country that imposes it, and the pattern of what industries get tariffs does not fit that.

I would say that objects are engineered to maximize some objective. Part of that might be the set of users.

Brian Albrecht said...

This all goes back to one of the basic ideas in economics, which is too often forgotten. Everything has a cost and trade-offs are everywhere.

If tire A was truly better than tire B in every way, tire B would not exist. If they both exist in the real world, there must be something about tire B that we are missing. It is our ignorance.

Miko said...

Suggesting the Earth's climate as an exception seems odd given your first example. If we had just arrived on this planet from a spaceship, your suggestion would make sense, but evolution has been working for thousands of years to optimize all existing creatures for living in a climate similar to the current one, which gives us a strong reason to suspect that it is in fact near-optimal.

Shaddox said...

"Birth control is one of the ways in which humans subvert the objectives of the genes in order to better achieve their own objectives."

I wouldn't conclude this so hastily. Modern medical contraception probably hasn't been around for long enough to observe much about its effect on fitness, but it seems very plausible to me that the genes of people who use birth control could be more successful at surviving and reproducing than the genes of people who do not use birth control.

Samson Corwell said...

The answer is provided by public choice theory, the branch of economics that deals with the workings of the political market. A tariff makes the inhabitants of the country that imposes it worse off but the politicians who pass the tariff better off, since it benefits a concentrated interest group at the cost of dispersed interest groups. More concentrated interest groups are better able to pay politicians to do things for them. Trade policy is optimized, but for the wrong objective.

I think it's pretty obvious that tariffs are meant to improve certain domestic industries. This is not some profound insight. Some politicians (i.e., Alexander Hamilton) promote(d) tariffs as a means for improving the entire country, so I don't think this theory is entirely correct.

Joseph said...

If people who try for low-sodium diets are more likely to be anti-vaccine, that might explain the lack of effect on health.

mark wilson said...

This post reminds me of a quote attributed to former Univ. of Chicago economist Harry Johnson who said, "the world is optimal; we just have to figure out why."

David Friedman said...

Re your quote from Harry Johnson. My version of that is:

"This is the best of all possible worlds" could be the statement of an optimist or of a pessimist.

Anonymous said...

The "engineered for a purpose other than mine" is the exception that swallows the rule, I'm afraid. Using your car example, quite clearly cost constraints are a major factor in what goes in to a given car. They may be the "best" components to hit a price point and appeal to a generic model of the consumer of vehicles at that price point, but things can very clearly be made better for the end user's preferences in a huge number of ways, otherwise aftermarket stereos, hubcaps, painting, fuzzy dice, etc. etc. etc. wouldn't exist.

It is clearly also completely false in the digital. There are Microsoft Windows installers that enable/disable features based on the serial number used. DVD players that are easily improved to play disks from outside one's region, and pirated video games that can be played without an Internet connection, when the legal versions won't, and ebooks that can be moved between devices when publisher DRM is removed.

A lot of the DIY home market wouldn't exist if homes were sold to meet the consumer's desires rather than a median target and price point.

The list goes on for nearly everything bought and sold in non-custom goods.

Sometimes I think the world-modeling of economists is training for ignoring huge chunks of reality. Leaving with:

An Econ professor and a student were walking across campus. The student sees a $20 bill on the ground, and points and says, "Look! A $20 bill!"

The professor replies, "Nonsense. If there were, someone would have picked it up."

Unknown said...

The problem here is that high blood pressure is likely to be a mortality risk in an individual's post-reproductive years, which would not be selected against by evolution, since those that die would have already reared offspring.