“Give it to me—the real news”
“So I will”
“Well, Dadamashay, let me see what skill you have. Tell me the big new news of these days, making it ever so small.”
“Listen”[1]
When, I was a graduate student, somewhat after the time of the Vikings in long boats, my thesis supervisor, Prof. Bhaduri[2], took me with him when he went on sabbatical to Copenhagen, a Mecca for nuclear physics at that time. When we were leaving there, his officemate gave him a small Mickey Mouse figurine so he would know what kind of physics to work on. Well another man might have been angry, And another man might have been hurt, But another man never would have[3] stressed during his seminar that he was using a Mickey Mouse model. A yes, Mickey Mouse science, the simple model or calculation that brings out salient features that are all too often lost or obscured in the complete calculation.
We all know what big science is: the big detectors at the Large Hadron Collider (CMS has a 12,500 ton steel yoke) or the Super-Kamiokande (50,000 tons of water). That is big science. Even theoretical physics does big science: the massive calculations of lattice quantum chromodynamics (QCD) or the nuclear shell model. Now, there have been attacks on big science, either the LHC or lattice QCD, as being inherently evil because they are so big. Would you believe, even books written on the topic? I strongly disagree with that view. Large science is an essential part of science. Big is needed to answer the questions we want answers to. However, there is more to science than that. We need the little to complement the big, the simple to complement the complex. As a post-doc, I was returning from a somewhat annoying conference with Gerry Brown[4] (b. 1926), one of leading nuclear physicists of that generation, when he turned to me in exasperation and said that people did not realize how many hours of computer time went into his simple estimates. There is an interesting concept: using computer time to justify simple estimates, simple complementing the complex. The purpose of computing is insight, not numbers[5] and the simple Mickey Mouse models are essential in generating that insight—even when they are justified by complex calculations.
The simple models are useful in a number of ways. First, they are useful in checking the results of complex computer calculations. I have learnt through bitter experience never to believe the result of a computer calculation until I have “understood” them (and not always then). That is, until using some simple model or estimates, either explicitly or implicitly, I can reproduce the main trends of the results. In trying to do that, I have frequently found errors. Never trust a number you do not understand.
Second, we want to understand what aspects of the model are important in reproducing the results and which are coincidental. Scientific models are designed to predict future observations, but which aspects of the model are crucial to that endeavour. It is through the use of simple models that we can most easily explore the dependencies of the results on the assumptions. We calculate some nuclear cross-section. Is that bump significant? What, if anything, does the location of the bump tell us? What about the turn up near threshold? Is that an artifact? We want to know more than merely if the calculation fits the data. It is here that the simple models come in. They give us the insight into how the models can be improved and what assumptions are not necessary and can be eliminated.
Finally, and most importantly, it is the simple models that allow us, as people, to understand the results. It is not just for the layman that we need the simple models, but for the expert as well. A prime example would be the non-relativistic quark model. Its success calculating the properties of the excited states of the proton was touted as proof of the quark model but all it tested was the symmetries built into the calculations. The simple approximations to the non-relativistic quark model revealed it pretentions. But as a Mickey Mouse model, the non-relativistic quark model gave us insight into QCD that would have been difficult if not impossible to obtain otherwise.
I suppose one could hook up the computers directly to the experiments and have them generate models, test the models against new observations and then modify the experimental apparatus without any human intervention. However, I am not sure that would be science. Science is ultimately a human activity and the models we produce are products of the human mind. It is not enough that the computer knows the answer. We want to have some feeling for the results, to understand them. Without the simple models, Mickey Mouse science, that would not be possible: the big news made ever so small.
To receive a notice of future posts follow me on Twitter: @musquod.
Tags: Philosophy of science
