Scott Sumner  

When the straightforward interpretation seems crazy

Law vs. Legislation... A Working Economy of Strangers...

I've recently been working my way through a long set of 2008 blog posts by Eliezer Yudkowsky. It starts with an attempt to make quantum mechanics seem "normal," and then branches out into some interesting essays on philosophy and science. I'm nowhere near as smart as Yudkowsky, so I can't offer any opinion on the science he discusses, but when the posts touched on epistemological issues his views hit home.

Quantum mechanics has a reputation for being somewhat strange and mysterious, partly due to the fact that the original research findings were interpreted in what Yudkowsky convincingly argues is the wrong way. For instance, you may have heard of concepts like "Heisenberg Uncertainty." The original "Copenhagen interpretation" of QM saw certain events as being fundamentally random, not just random in the sense that we are unable to predict the outcome. In contrast, Yudkowsky subscribes to the increasingly popular "many worlds interpretation" (MWI), which says the universe is fundamentally deterministic, and that seemingly random events involve both possibilities occurring, with the universe actually branching off into two parts, including the observer himself. A multiverse. Each of the two new observers sees one of the two possible outcomes. Hence it looks random, but actually it is (according to the MWI) the case that both outcomes occurred. (I apologize if I botched the explanation.)

As I said, I'm in no position to comment on the science, but here are some related issues that caught my attention:

1. Yudkowsky says that the MWI (splitting universes view) is actually the straightforward interpretation of the model, and that the Copenhagen interpretation adds an unnecessary assumption about wave function collapse. He says this was misunderstood for 30 years for 2 reasons.

2. One reason is that many scientists couldn't even imagine that they themselves were a part of the model. Their bodies were made of particles that QM also applies to, and hence they split onto two paths along with the particle being analyzed.

3. They misinterpreted Occam's Razor. Scientists wrongly assumed it meant you should avoid QM explanations that involved the complexity implied by a lot of universes, whereas it actually says you should avoid an explanation that implies extra (unneeded) theoretical complexity. For instance, when astronomers realized that spiral nebula were actually galaxies, it implied billions of times more stars, but is actually a simpler model of the cosmos---lots of spiral galaxies rather than lots of spiral galaxies and lots of spiral gas clouds. Similarly, the MWI is simpler and more straightforward than the Copenhagen interpretation.

Then it occurred to me that there is an analogy to monetary economics. The straightforward interpretation of the 1930s was that tight money caused the Depression. But it didn't seem like tight money caused the Depression, so it took 30 years before Friedman and Schwartz showed that was the case (in 1963.) Economists were confused by the low interest rates and QE, and thought money was easy. Hence almost everyone blamed financial instability, even though Irving Fisher's "dance of the dollar" model said it had to be tight money. Even Fisher wavered a bit, and added finance to the model. But finance wasn't needed! Canada had no banking failures, and had a depression almost as bad as the US.

Yudkowsky frequently seems puzzled and indeed exasperated by the fact that physicists got it so wrong, for so many decades. (And in his view, many still have it wrong.) I am frequently exasperated that the standard Friedman/Bernanke/Mishkin monetary economics of 2003 clearly implies tight money by the Fed caused the Great Recession, but almost no one believes that. These economists (and many others) used to say low interest rates do not mean easy money, nor does rapid growth in the monetary base. All three thought monetary policy remains highly effective at zero rates. Bernanke's specific criterion for the stance of monetary policy (NGDP growth and inflation) implies that the 5 years after mid-2008 saw the tightest money since the early 1930s. But almost no one believes that, not even Bernanke and Mishkin.

I'm actually a bit more forgiving of physicists than Yudkowsky, perhaps because I find the entire subject to be confusing. I can sort of see how people would overlook an interpretation that involved them splitting in two, and flowing off into multiple universes. In contrast, I'm relatively unforgiving of economists not seeing that the biggest fall in NGDP since the 1930s is tight money. Others might be more forgiving, as the interest rate and monetary base data sure looks like "easy money."

Yudkowsky says scientists had the right equations, the right formal model, but were thinking about the whole thing in the wrong way. I think economists have pretty good macro models, but often interpret them in the wrong way. For instance:

1. They ask whether the ECB should do more to cure the "lowflation" problem, when they should ask why the ECB caused the lowflation problem.

2. They think if QE did only a faction of what we need, then a much bigger QE would be needed to do what we really need. Even though Nick Rowe's "Chuck Norris" expectations channel says they could get by with much less, and even though Evan Soltas showed that the Swiss National Bank was able to do less QE, as soon as they shifted to a more expansionary monetary policy (of capping the SF/euro exchange rate.)

3. They think the zero interest rate bound is the problem, whereas the real problem (if it existed, which it doesn't) would be nothing left for the central bank to buy when NGDP growth expectations were on target, and the actual base was still short of the demand for base money.

I used to have a prejudice against math/physics geniuses. I thought when they were brilliant at high level math and theory; they were likely to have loony opinions on complex social science issues. Conspiracy theories. Or policy views that the government should wave a magic wand and just ban everything bad. Now that I've read Robin Hanson, Eliezer Yudkowsky and David Deutsch, I realize that I've got it wrong. A substantial number of these geniuses have thought much more deeply about epistemological issues than the average economist.

So when Hanson says we put far too little effort into existential risks, or even lesser but still massive threats like solar flares, and Yudkowsky says cryonics is under-appreciated, or when they say AI (or brain ems) is coming faster than we think and will have far more profound effects than we realize, I'm inclined to take them very seriously. I'm inclined to think we should spend less money on jet fighters and airport security, and more money stockpiling electrical transformers in case a huge solar flare hits us. And I no longer trust the "experts" in government to get this right. I've seen how the experts failed us in macroeconomics.

I'm about 90% through the Yudkowsky series. It's very hard work for someone like me, but well worth reading. It's not just science, he touches on everything from personal identity to free will to libertarianism (he's moderately supportive), and invariably has something brilliant to say. Highly recommended.

PS. In order to appease Bryan Caplan I did not mention "common sense" in this post. But . . .

PPS. Evan Soltas was 18 when he wrote the post I linked to. Think back to when you were 18 . . .

Comments and Sharing

COMMENTS (16 to date)
LK Beland writes:

Concerning interpretations of QM, another emerging school of thought is related to information theory considerations. E.g.

JoeMac writes:


I think you need to do better job of providing striking "trump card" empirical evidence in order to convince others that the recession was not caused by the subprime mortgage housing crisis.

You need to find some "smoking gun" evidence. Otherwise, you won't convince others. They shall just interpret what you're doing as mere assertions.

Also, you need to be a little more clear as to the theoretical mechanism according to which tight money caused all the problems. Otherwise, it sounds very metaphysical.

Jon writes:

QM is a mathematical model. It has a set of axioms and there are standard notations. It is easy to get lost in the notation but calculus is also a notation invented to describe physical phenomena. Whereas calculus extends geometry... Most of what is QM is the extension of matrix algebra.

This notation is interesting because it allows us to make predictions about experimental outcomes. The philosophy of QM is irrelevant. QM stands as a mathematical model, all of the predictions come from that model not from the philosophy of making an intuitive analogy. This is true for the Copenhagen interpretation as well.

In short... These interpretations are a model of a model. Why has the CHI endured and MWI been treated as a crackpot idea? Well one reason is that certain experimental results--eg the Casimir force--make absolutely no sense in the MWI.

Attempts to focus on finding a philosophy of QM are huburis. There is no reason our common sense and our linguistic mind honed by evolution should be presumed to capture the physical nature of the world at scales beyond our senses.

Scott Sumner writes:

LK, Very interesting.

Joemac, I'm using the standard model and have the straightforward explanation. NGDP plunged and wages are sticky. Hence a bad recession. Falling NGDP also worsens debt crises. It's up to the profession to explain to me why a straightforward monetary interpretation of the standard model is wrong.

Jon, You know more about it than me. But if it's just equations and you can't empirically test which interpretation is true, then why does the Casimir force make MWI less likely?

And there seem to be quite a few very smart "crackpots" that buy into the MWI.

I agree with your last sentence, which of course also applies to monetary economics.

A writes:

Quantum physics analogy? Nerd alert!

I see another issue wherein influential economists engage in serious conversations, informed debates, and intense considerations of the latest literature. Then, they employ various measures, defend their actions, manage political and punditry resistance. At the end of the day, a Bernanke feels like he has done a tremendous amount of work. And a Sumner comes along saying that all he had to do was to be more clear about targeting more useful metrics.

For years, athletes avoided weight lifting because they thought that you had to train to failure, risking injury and excess fatigue, only to produce an insufficient level of strength to compensate for size increases. Then research showed that you could lift heavier weights, at lower repetitions, for greater strength without training to failure. You actually avoid failure. But it didn't feel like work, so you saw myth based debates about calisthenics vs. weight lifting, and other fetishistic notions confusing effort with productivity.

Rick Hull writes:

What is the working definition of easy money, or loose monetary policy? How, essentially, does one perform easing? Is it equivalent to "printing money"?

Mark Bahner writes:
...or when they say AI (or brain ems) is coming faster than we think and will have far more profound effects than we realize, I'm inclined to take them very seriously.

I'd appreciate a pointing to a prediction Robin Hanson has made about when the majority of jobs that currently exist will be performed by artificial intelligences.

Andrew S writes:

A different view of the same QM story. Some really smart guys discovered the math behind QM. They came up with a straightforward interpretation of the theory based on their previous real world experiences (Copenhagen interpretation). For various reasons many people did not like the interpretation, and even the theory, because of it's nondeterministic nature ("God does not play dice, EPR paradox, etc.)

This dislike has persisted, and there have been been consistent efforts to create new more complex explanations that make QM look more deterministic (De Broglie wave theory, MWI). Many smart folks have propounded these interpretations and believe they are right. They make QM much more amenable to classical intuition.

These interpretations tend to look good when dealing with QM as an abstract theory. They generally work by assuming a somewhat larger universe (many worlds, wave function is real), and postulating a set of more intuitive (to us) rules that work in this universe. Then the restriction to living in our actual universe produces the quantum weirdness.

The benefits of theses modern interpretations, however, begin to break down when you start to ask how QM interacts with the universe we live in. This interaction has become more and more interesting as QM has moved from a purely theoretical exercise to something seen every day in laboratories across the world. The collapse of the wave function is real (described by decoherence) and is clearly related to the "collapse" of a probability distribution, as in "I saw the dice land on 6, the distribution over possible results collapsed from uniform to a delta function".

When asking questions about how the classical world arises from the quantum one, about what measurement really means, about how chaos arises in a deterministic quantum theory, about how weak measurements (that don't completely collapse the wave function) affect a quantum system, about information in quantum systems and quantum computers functions in the real world the Copenhagen interpretation looks more useful every day.

Now in principal all of these "interpretations" predict the same physical effects. In practice, however they create vastly different complexities of thought and lead to their proponents focusing on very different aspects of QM to explore. I would argue that the Copenhagen interpretation arose out of a fundamental interest of it's creators in relating QM to the real world. Other interpretations came later as people tried to make sense of QM. The MWI has proven very useful in exploring algorithmic possibilities on Quantum computers abstractly (pure quantum systems), and in other areas. However, as we move towards a world where QM is a useful tool, not just a theoretical oddity, the original Copenhagen interpretation appears ever more prescient.

Scott Sumner writes:

A, Nice analogy.

Rick, I've recommended targeting NGDP growth at a steady rate, perhaps 5%, and then doing level targeting. In addition, the central bank should "target the forecast" which means set the policy instrument at a level where expected NGDP growth equals target NGDP growth.

Mark, I believe he is working on a book on that subject. He thinks the majority of "people" will be brain ems by sometime in the next century.

Thanks for the perspective Andrew.

JoeMac writes:


When you say "Falling NGDP also worsens debt crises." are you referring to Fisher famous debt-deflation theory of recessions or is the mechanism different?

Rick Hull writes:

Scott, thanks for the reply. I'd love to see a post on the basics of monetary policy as you see them, without reference to NGDP, if such a thing is possible. I had long understood easing to mean expansion of the money supply, while tightening is synonymous with contraction. As the expansion is usually performed by making loans, it is usually accompanied by lower interest rates.

I know that this is a hopelessly naive understanding from your perspective, but I've never been able to wrap my head around the (few) posts of yours I've come across on other sites. What I did manage to gather is that we can't look at the action itself to know if it is easing or tightening -- we can only judge the nature of the action by evaluating the state of the economy at some later point. Do I have that much correct?

Jon writes:

Scott, you write: "But if it's just equations and you can't empirically test which interpretation is true, then why does the Casimir force make MWI less likely?"

Two bites to chew on:
Physics as a discipline does not focus on these interpretations of QM. They are the stuff of science writing not peer review. Lots of useful predictions arise from physics the discipline, the interpretations of QM really don't enter into it. This is not the same as saying the interpretations aren't empirically testable.

The casimir effect is fascinating because it is macroscopic effect that can be explained using QM equations but not classical physics. In particular, it is an empirical demonstration of ephemeral particles--random fluctuations. So yes, it excludes deterministic explanations from being fundamental.

Don Geddis writes:

@Sumner: I'm pleased and impressed that you're taking the time to read & understand Eliezer's QM sequence. It's brilliant stuff -- albeit intellectually challenging to work through. (I found Eliezer and worked through the QM sequence myself, before I found your blog.)

You, Hanson, Yudkowsky and Deutsch ... each in a slightly different area, but you're indeed all "of a type". The same kind of philosophical/intellectual approach, the same clear thinking, the same puzzlement about how everyone else in the field can be so wrong. You don't think like you're doing anything special yourself; just straightforward reasoning like you (and everyone else) was always taught. Yet somehow, everyone else seems to abandon clear thinking at some point. And it's not really clear why.

You're in good company.

Don Geddis writes:

@ Rick Hull: the value of money is dependent on both the supply and demand for money. (It's a market, just like any other.) "Easing" means that money is getting cheaper (and leads to increases in output, employment, inflation, and NGDP). "Tightening" means that money is getting more valuable (and leads to drops in those same factors).

You can't tell anything about the path of the value of money, looking only at supply, while ignoring changes in demand (i.e. velocity).

Also, it's a mistake to focus on interest rates, or believe that "the expansion is usually accompanied by lower interest rates". That's (mostly) a consequence, not a (primary) causal factor. Monetary policy (controlling the price/value of money) works just fine without having a banking sector at all, with no loans or interest rates (which is the price of credit, a very different thing).

Don Geddis writes:

@ Jon: Copenhagen "collapse" is a discontinuous, irreversible new piece of physics, that must be added on top of the equations of QM. There is absolutely zero evidence from any QM experiment, of any kind of discontinuous irreversible effect. All the equations, and all the experiments, show only continuous reversible (and time-reversible) processes.

Not to mention the whole original confusion that "conscious observer" was somehow a magic fundamental property of our universe. AI, neurobiology, psychology, etc. have all shown pretty conclusively that humans are based on the same fundamental physics as rocks. Consciousness is very complex ... but involves no new fundamental physics.

Copenhagen collapse is a terrible, terrible "intuitive" interpretation of the QM equations.

Mark Bahner writes:

Hi Scott,

You write, regarding Robin Hanson's AI work:

Mark, I believe he is working on a book on that subject. He thinks the majority of "people" will be brain ems by sometime in the next century.

OK, I thought you were referring to that aspect of his work. It seems to me that Robin Hanson's timetable (sometime next century) is actually significantly slower than many prominent authors regarding AI. I'm thinking in particular of Ray Kurzweil, Hans Morovec, Victor Vinge, and others. Most them expect AI comparable--or even far exceeding--the capabilities of the human brain before the middle of this century.

Comments for this entry have been closed
Return to top