Skip to content

‘Twas beauty killed the particle physics beast

March 30, 2011

Some interesting things to contemplate (in the comments too!) if you are interested in physics and philosophy. This point about data mining is something I have been worrying about for a while:

One of the major problems with the supercollider, the most important (if not only) source for new empirical data of particle physcis today, is that it has largely become a data-mining exercise. In order to find any ‘evidence’ in it the scientists have to make non trivial statistical/mathematical and physical assumptions–and discard massive amount of ‘data’.)…So, here’s another exercise for philosophy of physics–what is the most appropriate methodology in the context of data-mining? In fact, that may be THE fundamental question facing economics and the philosophers that think seriously about it today.

The present capacity of automated experiments to produce gigabyte after gigabyte of data vastly outstrips the ability of anyone to analyze it except through hideously complex and black-box-like statistical calculations, which could be introducing all kinds of unknown biases. And as bad as this problem is for physics, it is even worse for economic data, medical studies, social statistics and other fields where ‘experiments’ are even more messy and uncontrollable.

I also worry about the tendency in theoretical physics to value ‘beauty’, ‘elegance’ and so on. I commonly hear seminar speakers motivated by a particularly nice mathematical form for their theory. Say for example you have some equations which are nearly perfectly ‘symmetric’ (in some technical physics sense) but are missing some little part: the automatic tendency is to look for the missing piece to complete the picture, as though the elegance of the theory corresponds to an underlying mathematical structure actually present in the real world, when it’s no more than a nice feature of some particular representation of reality.

King Kong/theory is confronted by harsh faceless experimental facts

A nice example is Maxwell’s equations connecting the electric and magnetic fields: they would be completely symmetric if magnetic monopoles existed to provide the magnetic version of electric charge. Not only that, but if magnetic monopoles do exist then they also (due to a nice result by Dirac) would explain the observational fact that all particles have charges that are a multiple of a basic unit of charge. Unfortunately there is no evidence for monopoles, but because of their theoretical prettiness the searches continue and tons of theories include them in one way or another such that they exist but are unobservable.

This seems like a potentially pathological way of doing things, where you notice something that would be nice to have, then require your theories to include it in a way that prevents you from seeing it directly. You could end up with theories that are conceptually and theoretically ‘beautiful’, ‘elegant’ etc., but almost by design don’t make measurable predictions (at least not measurable in anything but super long term).

"Sad, sad, terrible, gruesome news about my colleague Dr. Mbutu's scientific methods..."

I guess the question is, what are the root causes of this practice, which has (as the link states) been going strong for thirty-odd years now. Are the people working on all these theories just obsessed with mathematical elegance, rather than whether there is any realistic experimental test of their idea? Are they just making stuff up to get grants? Maybe it is some sort of survival of the fittest, system-generated thing? I’m imagining people who are convinced about the stylishness and promise of their theories, who are able to ‘string’ research boards along indefinitely by proposing things that are forever out of reach. Not out of trickery or because they want to be fashionable but because their real enthusiasm for theoretical perfection ‘sells well’ to like-minded people. Clearly this work is valuable in the same sense as abstract mathematical theorems, but the ultimate goal of physics in the past has always been to explain experiments with theories. There seems to have been a shift in motivation from explaining weird results to creating nice theories, and a corresponding explosion in theorizing that has no clear connection to any current or near-future experiment.

After spending a while doing my PhD in the neutrino theory world, which is kind of on the boundary between testable and untestable (my thesis was kind of on what might be realistically done to test some of the theoretical ideas) I feel like it’s becoming pointless to develop new theories that are so far from observations. The paradox is, these big untestable theories are (or at least began as) attempts to solve serious and undeniable observational problems such as dark matter, dark energy, inflation, quantum gravity, and so on. It feels like we shouldn’t need to go so far from experiments to create a theory that explains the data, but then that’s just my own bias of what theories ‘should’ ‘look like’…

If I’ve learned one thing from trying to explain ‘beautiful’ and ‘elegant’ theories like relativity and quantum mechanics to undergrads it is that physicists spend a long time struggling with the ugliness of the non-intuitive concepts and messy algebra before they are convinced that there is any sort of order or elegance to these things. And then we hide away a lot of the mess by inventing stylish notation. So much of what seems ‘natural’ and ‘intuitive’ to Physics PhDs probably only seems that way because we’ve become accustomed to it, like a husband and wife who at first grudgingly tolerate each other’s bad habits and at last wouldn’t have them any other way.

Advertisements
4 Comments leave one →
  1. March 30, 2011 8:15 am

    I have a three comments:

    1. Nice piece.

    2. I’ve always wondered about all this. Given that in a lot of ways mathematics is a human construct (at least according to some people), in any case it’s a pretty neat way of describing the world around us, shouldn’t it seem very surprising that real things do sometimes (often?) behave ‘nicely’ within our mathematical framework? Should we not be alarmed, or suspicious, rather than satisfied when we ‘find’ that this happens?

    3. Good work on fixing New APPS’ typo.

  2. Ned Gilmore permalink
    March 30, 2011 1:23 pm

    Amazing piece Mr Giles!

    I had to agree (even from a very weak physics standpoint) that the comment about ‘elegance’ reasonated a lot with me. Just from remembering 6th/7th form interactions with complex numbers and the associated clunkiness of ‘inventing’ a number that squared to -1 (“You can’t just make one up! It doesn’t work like that! This is cheating!”) I can agree that there is a very earnest, deep seated desire, even at the most basic levels, for the maths to ‘just look good’.

    Yet, you absolutely have to have one, if you want to be able to do things in physics.

  3. March 30, 2011 10:42 pm

    Clare:
    1. Thanks!

    2. Yeah it is suspicious…there’s a paper with a nice title about this: “The Unreasonable Effectiveness of Mathematics in the Natural Sciences. From my bit of a philosophy course that dealt with philosophy of science and some other reading I’ve done, it seems like pure positivism where everything proceeds via a well-defined scientific method and rigorous logic is not really a good description of how things work. And yet we can send robots to Mars based on these theories that ‘shouldn’t’ be so good. It’s suspicious but it works (in the overall scheme of science for the last few hundred years)!

    3. Yeah it looked pretty ugly…

  4. March 30, 2011 10:45 pm

    Ned:

    Yeah this tendency is even ‘worse’ in maths, but I think very few mathematicians are Platonists who believe they are investigating something ‘out there’ with their work. I think most mathematicians will gladly admit that what they are doing is investigating the consequences of certain axioms and rules of logic that are explicitly chosen rather than springing from an independent source. The difference in physics is that all these abstract things should in some way link up with ‘reality’ that is ‘out there’ and not fully arbitrary. (although it’s easy to get into arguments about realism too, and things like relativity and quantum mechanics even bolster the arguments against there being a single objective reality)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: