Experiment vs. Theory: The Eternal Debate

Melissa at Confused at a Higher Level offers some thoughts on the relative status of experimental vs. theoretical science, spinning off a comprehensive discussion of the issues at Academic Jungle. I flagged this to comment on over the weekend, but then was too busy with SteelyKid and football to get to it. since I'm late to the party, I'll offer some slightly flippant arguments in favor of experiment or theory:

Argument 1: Experimentalists are better homeowners. At least in my world of low-energy experimental physics, many of the skills you are expected to have as an experimental physicist translate very directly to real-world problems. If you can wire up an electrical circuit for precision measurements, you ought to be competent enough to change a light switch or electrical socket. If you can change the oil in a Welch rotary vacuum pump, you should be able to change the oil in your car. At the very least, it gives you a better appreciation of what you're paying professionals to do for you.

Theorists, on the other hand, spend all their time playing with computers, and are thus less likely to be able to repair physical objects. The obvious counter-argument, though, is that high-energy experimental physicists spend at least as much time dinking around with computers as low-energy theorists do, so this may not be a general advantage of experiment over theory.

Argument 2: Theorists are better with computers. People who do theoretical science these days are usually very comfortable with computer code, which leaves them better prepared to deal with our modern computerized world than experimentalists, whose preferred mode of solution is likely to involve a big hammer.

The obvious counter-argument (aside from the high-energy experimentalists, who screw everything up) is that most physics theory involves esoteric things like FORTRAN code running on Unix machines, meaning that most theoretical physicists are just as befuddled by Windows as everybody else.

Argument 3: Experimentalists get better toys. Experimental physics apparatus looks like this:


That's something you can show off on a tour of the department and it's guaranteed to make an impression on prospective students and parents. Hot theoretical apparatus, on the other hand, looks like a stockroom at Best Buy. That could be a supercomputer, or it could be a rack full of discount desktops with some additional blinking LED's.

The counter-argument, of course, is that when experimental apparatus breaks, it's all expensive and custom-built stuff that takes ages to repair. When a theoretical apparatus breaks, you just order an identical replacement machine and have it overnighted to you.

Argument 4: Theorists have better boondoggles workshops. Theoretical scientists get to go to all sorts of summer schools and month-long workshops in California and Italy and so on. The nature of theoretical science is such that you can get a whole bunch of theorists together in some place within easy reach of a beach or a ski resort, and productive work can come from having them just hang out and talk to each other.

Experimentalists, on the other hand, are tied to their labs. Even if you're the kind of experimentalist who uses a billion-dollar facility somewhere other than your home institution, you're stuck in the lab when you go there. While a given user facility may be located someplace nice, when you're there, you're on the clock, with limited beam time or whatever. You don't get a lot of time to enjoy the ambience before you have to go home and analyze your data.

The weak counter-argument for this is that experimentalists are occasionally invited to these theoretical events, to serve as a reality check. This only works for a small and select group of experimentalists, though.

Argument 5: Experimentalists have to know a bit of theory, while theorists don't need to know anything practical. Most experimental physicists need to be able to hack a bit of theory. For one thing, you can't really begin to interpret your results without some understanding of the underlying theory. And it's often extremely useful to be able to bang together a quick-and-dirty theoretical model to see if your results have the right basic shape before you call in the real theorists to do the model without extreme simplifying assumptions.

Theorists, on the other hand.... Experimentalists are mildly amazed when a theorist manages to find two matching socks. Nobody expects a theorist to be able to debate the relative merits of round vs. square wire or solder two wires together, let alone align a Ti:Sapph laser. Thus, training in theory requires less breadth of expertise, and is the clear choice for students seeking an easier path to the Ph.D.

(In case the note at the beginning of the post and the general tone of the arguments wasn't clear enough, this is not entirely serious. While the arguments have some small basis in reality, they are exaggerated for humorous effect, and should not be taken as an accurate representation of my opinion of my theoretical colleagues.)

More like this

Wait, what are the merits of square vs. round wire? Please don't take my experimentalist id card away.

Everyone know that there is no difference between theory and practice.

-- Theorist.

By Alex Besogonov (not verified) on 13 Oct 2010 #permalink

I disagree somewhat with points 4 and 5.

My field involves space flight hardware, so the dynamics of boondoggles a.k.a. workshops is different from other fields that don't. The experimentalists may be too busy working on the next instrument at certain times, but once the instrument is launched the experimentalists are just as free as the theorists to live it up in Hawaii or Aspen. And I suspect that even in fields where people are tied to labs, experimentalists who are sufficiently near the top of the food chain can go to these workshops and take the credit for the work their minions are doing back in the lab.

While it's true that mediocre theorists may be able to get away with not knowing practical stuff, IME the best theorists understand the limitations of data. Think Feynman, who immediately realized a conclusion was suspect because it relied on a single data point at the extreme of the range (he retells the story in Surely You're Joking, Mr. Feynman!, in the section "The Seven Percent Solution"). Or consider the nature of prediction: being able to tell the experimentalists that if their theory is right, then experiment X should yield result Y, and if it yields Z instead then the other guy is right.

On the issue of homeownership, I'm not so sure. A little knowledge is a dangerous thing, as I am occasionally reminded at my house (the previous owner was a civil engineering professor whose DIY projects have been an occasional source of amusement and/or head-scratching for my contractor). Being able to diagnose a problem is useful, but sometimes it's better to recognize that this repair job or renovation is beyond your skill and call in the experts.

By Eric Lund (not verified) on 13 Oct 2010 #permalink

The round wire vs. square wire thing is an old joke from NIST. The BEC groups there used to have a joint weekly meeting of both theoretical and experimental groups. They split into two meetings after a large fraction of one meeting was devoted to debating whether to wind the magnetic field coils out of copper tubing with a round cross-section, or a square cross-section. After that, the experimentalists still came to the theory meetings, but the theorists wouldn't go anywhere near the experimental meetings.

(Square wire is the clear winner, by the way.)

Good points overall, but I'm confused about one thing. Spending your day worrying about square versus round wires is an advantage or a disadvantage?

Either way, I think I'll stick with theory (point 4 is a winner, but it's not like I have a choice really).

As a theoretical biologist, I have a bit of a different perspective. This is especially true because of the divide in "real biology" (c.f. theoretical biology) between field biologists and bench biologists. Field biologists seem to have it pretty good... get to go on trips to exotic places and capture animals to understand some esoteric aspect of their demography or history. Of course, it's QUITE hard work and requires a lot of begging for funding among other things. Bench biologists have a different lifestyle (though I should mention that field biologists often become bench biologists when they get back from the field). They are probably more similar to experimental physicists, in that they will devise elaborate experimental procedures to test their ideas. These tests often fail and moreover I imagine that biological systems are even more uncooperative than physical apparata.

Theory in biology occupies an interesting position. In some sense it is derided and relegated to highly technical journals that have what seems to be the lowest standards of writing in the world. However, especially recently, theory has come back into fashion because data has been so much more forthcoming. This is especially relevant in population genetics (my area) because modern DNA sequencing methods allow field biologists to go out and collect a VAST amount of data, and they need to be able to say something about it.

Another interesting aspect of the theory/"experiment" divide in biology is that in my experience most biologists don't know very much math (especially more molecular-oriented types) and while they understand the qualitative behavior of many common models (e.g. the Wright-Fisher diffusion, coalescent theory, branching processes, Lotka-Volterra equations, etc.) they are utterly helpless when it comes to any quantitative analysis. On the other hand, my sense is that most experimental physicists could probably analyze the kinds of models I devise better than I could!

I have to disagree with a lot of what I read above. Maybe the theoretical physicists I know are an exception, but it seems like we know and do just as much as the experimentalists. When something goes wrong in their labs (meaning unexpected results), they come to us. I'm a theoretical physicist who has spent just as much time in a laboratory setting as my experimentalist counterparts! And let's not forget Richard Feynman, who was a renowned theoretical physicist whose work was later applied experimentally! In my opinion, the theoretical physicists are the groundbreakers. The experimentalists need us because while they are tinkering around, we generate the new ideas. And we need them because they check to make sure our ideas work practically.
And, I'm a theoretical physicist who changes my own oil, works on my own cars and motorcycles, and tinker with things around the house. And all of my theoretical physicist buddies are pretty much the same!

By Theoretical ph… (not verified) on 13 Oct 2010 #permalink

As with every question about academia, the answer is "it depends on what field you're in" -- and "physics" is way too broad to specify the field. Most of the experimentalists I know spend much more time with computer programming than I do, because they're working in several-thousand-person particle physics collaborations, and while they might have had some role in building part of a detector, now the machine is running and experiment is all about data analysis. They also go to summer schools in nice places, and seem to have more conferences that revolve around skiing than I do.

Theoretical Physicist at 1:07 ... OMG, you sound like a string theorist, although I guess you are probably having a go at our host, lol! I am a theory person who spent most of my youth working as an experimental physicist. I find that even though I was raised on practical skills, a dedication to Theory does eventually lead to a sock problem. And real theorists know bugger all about code. As an experimentalist I used FORTRAN etc, but a real theorist just does maths.

Experimentalists have to know a bit of theory, while theorists don't need to know anything practical.

Hahaha...that makes my night! I can now stop browsing blogs looking for that final fulfilling tidbit of humor/info before retiring for the evening...

I'll deal with the damn socks in the morning...

In quantum optics theory we pretty much always care what people can measure, so we do need to know at least the state of the art in experiments, if not the cross-sectional form of the charged fermion transport tubes. However I fully admit to not having a clue about electrical circuit design more than can be done with Kirchoff's laws (what ARE those triangle things?).

By The lane ranger (not verified) on 13 Oct 2010 #permalink

As a theoretical physicist who has worked with microbiologists and ecologists, I can relate to what Josh says. We do indeed have a better grasp of the models than the ecologists and we have a hard time explaining them how to appropriately modelize. On the other hand, I think microbiologists are kind of laughing our simplistic models away as not sufficiently realistic or interesting. They want to go to serious biological stuff immediately like resilience and function-diversity relationship, while we are still grappling with a very basic model for diversity.

By Raskolnikov (not verified) on 13 Oct 2010 #permalink

Thereotical and Experimental scientists, both are dependent on each other. There are limitations to both the approaches. Like, if you go only with the theory, you may over-idealize the system and end-up with practically impossible solutions. On the other hand, if you go with the experimental approach, you need to have atleast basic knowledge of the theory in order to design your experiments and to predict the result. Unless you have an Idea about what to expect, you may not be able to draw any conclusions from the results. Experimental and therotical approaches, both are two sides of the same coin.

if you arrange theoretical and experimental alphabetically, experimental comes first.

cogitate on that theorists!

Nice ventilators on that supercomputer.. Like superman on crutches.

By rijkswaanvijand (not verified) on 16 Oct 2010 #permalink

Why are you still using an oil sealed pump??

I wish stupid Blogger would cross-link stuff, I just happened to come across the post. This was a fun read!

I will just proudly say that I run esoteric home-made FORTRAN codes on my Windows cluster! To the disgust of many of my UNIX-purist theorist brethren. :)

I work really really closely with experimentalists, and have picked up a lot of details (materials processing, device design and fabrication...) which definitely make for way better models and overall better physics. Dirty, nonideal systems are fun! Plus people ask me about experimental details when I give talks anyway. Not sure if my experimental colleagues have picked up any details on the theoretical aspect of the work -- every time I throw any equations out there, they start looking very drowsy. ;)

"Argument 2: Theorists are better with computers."

That would ignore experimentalists writing complicated programs to control their experiments (LabView anyone?) and to take data, and to reduce data, and to fit the data to models.

Though any given experimental lab usually needs or has only one or two such programmers at a time.

According to Yogi Berra, "In theory, theory and practice are the same, but in practice they are different."

By BrucePeterson (not verified) on 18 Oct 2010 #permalink

high-energy experimental physicists spend at least as much time dinking around with computers as low-energy theorists do,

I swear I thought it said they spend as much time DRINKING around the computers as low energy-theorists do. :)))

By AppliedPhysicsProf (not verified) on 18 Oct 2010 #permalink

Herpetology version, very short:

Experimentalists have it better because they can say "Today I got to work with a majestic 14-foot long King Cobra!"

Theoreticians have it better because they never have to say "Speaking of which, is there any hospital with large anti-venom stocks within a 10-minute drive?"