Tuesday, October 21, 2008

Review: The Trouble with Physics

When I was an undergraduate in Computer Science, my instructors and I would occasionally have lunch and talk about Physics envy. You know, those Physics people could actually prove theorems and show that their experiments worked, while for us Computer Scientists to argue about Object-Oriented programming versus Functional programming just felt like kids discussing which flavor of ice-cream was better. While civil engineers could plug in equations and build bridges, the best software engineers build gossamer-like strands of code that support computing structures far better than any theory of Computer Science could give you, showing you that we don't have any idea how to teach you how to write code, any more than the English department knew how to teach you to be a great novelist.

The Trouble with Physics (Kindle Edition) might very well put an end to future young Computer Scientists having any more Physics envy, which might be a good thing. The first part of the book covers string theory, how it evolved, what problems it tries to solve, and what the major research areas are. I've done some previous reading on the topic, Peter Shor in his review of the book covers the technical part of the book quite well:

String theorists: We've got the Standard Model, and it works great,
but it doesn't include gravity, and it doesn't explain lots of
other stuff, like why all the elementary particles have the masses
they do. We need a new, broader theory.

Nature: Here's a great new theory I can sell you. It combines
quantum field theory and gravity, and there's only one adjustable
parameter in it, so all you have to do is find the right value
of that parameter, and the Standard Model will pop right out.

String theorists: We'll take it.

String theorists (some time later): Wait a minute, Nature,
our new theory won't fit into our driveway. String theory
has ten dimensions, and our driveway only has four.

Nature: I can sell you a Calabi-Yau manifold. These are
really neat gadgets, and they'll fold up string theory into
four dimensions, no problem.

String theorists: We'll take one of those as well, please.

Nature: Happy to help.

String theorists (some time later): Wait a minute, Nature,
there's too many different ways to fold our Calabi-Yao
manifold up. And it keeps trying to come unfolded. And
string theory is only compatible with a negative cosmological
constant, and we own a positive one.

Nature: No problem. Just let me tie this Calabi-Yao manifold
up with some strings and branes, and maybe a little duct tape,
and you'll be all set.

String theorists: But our beautiful new theory is so ugly now!

Nature: Ah! But the Anthropic Principle says that all the
best theories are ugly.

String theorists: It does?

Nature: It does. And once you make it the fashion to be ugly,
you'll ensure that other theories will never beat you in
beauty contests.

String theorists: Hooray! Hooray! Look at our beautiful new theory.

You get the idea. But Smolin's criticism of string theory goes even deeper than that. First of all, the underlying basic assumptions behind string theory are as-yet unproven. Yet the string theory community has not placed an emphasis on working on that critical foundation. Secondly, the theory makes no predictions that can be verified. That's because there are so many free variables in the theory that you can make whatever results you want come out of it. Worse of all, as a result of the dominance of string theory in theoretical physics departments, Physics (according to Smolin) has made no progress in the last twenty years.

How could this happen? And why do the rest of us (who don't get funded by NSF grants) care? At this point, most reviewers wander comfortably into the sunset, but I started getting intrigued, because Smolin's criticism here is a very deep one --- it's not just criticism of Physics and how it gets funded, but of the entire scientific process, and the kind of validation needed for a scientist to get support. This is a fascinating dialogue, and very relevant not just to Physics in general, but in all fields of academic study, but Smolin's criticisms are criticisms of the peer-reviewed promotion and grant system that subjects Physics to these academic fads that can stall progress for decades at a time.

The story goes like this: young scientists get their research agenda set by senior scientists --- that's because in graduate school, they are dependent on not just funding, but on the feedback of their senior advisers who write letters of recommendations to the hiring committees of the universities that will end up hiring (or not hire) them. Senior scientists need people to do work for them, so they tend to select technicians --- people who perfect the math and can do it quickly. In turn, when these technicians get hired, they in turn depend on the peer reviews written for them by senior scientists for their promotion-committees in order to get tenure. This leads to doing work that can be recognized quickly, but also work that is to a large extent less risky --- you're better off working on something that everyone else knows something about than thinking deep thoughts that might potentially revolutionize the field. As a result this leads to the entire field all working on one thing at a time (e.g., string theory), while other important avenues get neglected. It also leads to group-think, since those who reject the current status quo (e.g., folks who don't believe in string theory) find it hard to get hired as scientists, or to get tenure. In particular, Smolin singles out the one feature of the peer-promotion/hiring process that I also dislike, which is the forced ranking of "This scientist is better than that scientist" that's so frequently required --- it almost always leads to a bias towards the fast thinker who is a great technician, away from the deep thinker. (I always rebel against such simplified rankings of persons when I manage people)

The problem with this approach is that it does work quite well during times of normal science, when there's a promising theory and working out its implications are important. But when you need to re-think the underpinnings of Physics, it falls up short, because the kind of people most suited to that kind of work, usually tend not to be great technicians. Albert Einstein is the classic example --- he was not considered particularly bright, and basically could not get a job as a scientist. (He worked at the patent office in Switzerland while thinking up his most famous ideas) Smolin goes on to name several other scientists in the same mold, at least one of which literally quit the field of science for 10 years while he thought deeply and read a lot about the underlying problems for multiple decades before he was recognized as a great thinker on the topic --- and even then he still had a hard time getting hired!

One would think that university administrators, who are very competitive might recognize this problem and realize that unconventional thinkers are relatively cheap, high risk/high reward hires, and hire them, but of course, that's not how it works, since the hiring process at major universities is peer and committee based, as described above. The result is a stale-mate, where string theory might hold sway for several decades until one particularly fine thinker finally writes a ground-breaking paper that revolutionizes the field.

Smolin proposes several ways around this, mostly by making the hierarchy flatter, giving people more scope, and creating more opportunities for high-risk/high-payoff people. What's fascinating is that he thinks that businesses, like high tech companies and venture capitalists have the answer. Coming from the opposite direction, I know for sure that it does not, and in fact, one of the reasons I believe a recent, well-known large tech company was largely so successful mostly because it actively borrowed its hiring and promotion model from academia. Naturally, the jury is still out, and it may be that it takes several centuries for the peer-based hiring and promotions systems to calcify into a system that blocks progress for decades at a time.

Regardless, for providing such provoking food for thought and interesting reading, The Trouble with Physics (Kindle Edition). Even if you hated Physics in school (which I did --- and Lee Smolin did mention the stifling curriculum being one reason why we're getting fewer scientists today), you will find it a great read. highly recommended
Post a Comment