Computer simulations are necessary for the philosophy of simulations, but are they necessary for the Higgs?
Margaret Morrison begins a chapter on the role of computer simulations in physics with a rather grand philosophical claim:
Hence, to say that the discovery of the Higgs was only possible using simulation is by no means to overstate the case.(Reconstructing Reality, 2015, p.288)
After a detailed case study on the Higgs discovery she concludes stating that “[…] the traditional philosophical categories are out of sync with contemporary practice.” (Reconstructing Reality, p.316). Apparently new categories and perhaps even a philosophy of simulation is needed to capture such practices better.
This is a well known stratagem employed by novelty hunting philosophers. They claim essential involvement of X to convince their readership that a philosophy of X is needed1.
In any case, such claims never go unanswered in the philosophical community and recently Krämer et al. (2024) argued that computer simulations were actually not necessary in any part of the process leading to the Higgs discovery. Specifically, so they allege, simulations were not necessary in any parts were Morrison claimed they were. This argument comes a decade too late to do serious damage to the field of philosophy of computer simulations (which has been on life support since the philosophy of AI regained its former strength anyhow). But there in there is something puzzling about such indispensability arguments. And that is perhaps representative of a whole class of them.
If you claim necessity or essential involvement (or the contrary) of any technology you better have some account of necessity at hand. Morrison really doesn’t, so we can only guess how she would have interpreted her modalities. Krämer et al. on the other hand appeal to a much discussed distinction between practical and theoretical necessity. A distinction that was cleverly introduced2 by Humphreys in “The philosophical novelty of computer simulation methods” (2009) to fend of his critics who were arguing that there is nothing philosophically novel about computer simulations (Frigg, Reiss 2009). I have long puzzled over this distinction because, while I have a good intuition about the theoretical necessities, I find practical ones extremely vague. Of theoretical necessity I like to think as relative to some formal system like (a) logic. Maybe one doesn’t need to be so strict and informal mathematics is okay too (or perhaps physics for physical necessity?). In any case there must be some semi formal steps involved in deriving a possibility or necessity result. I’m pretty sure that in Newtonian physics faster than light travel is theoretically possible, but in special relativity it is not. But practical necessity, especially practical possibilities of computations (which computer simulations are) I am not sure what to make of. Obviously, I can do all the steps a Turing machine can do. But wait! Humphreys wouldn’t like this, in his paper he scolds Stöckler for assuming just such a thing. Obviously, I can only trace the steps of the most simplistic calculations. Who in their right mind would execute the Turing version of a Finite-Element method anyhow? Just to prove a philosophical point… But the question of how to evaluate practical impossibility claims still stands. Relative to which practice should we interpret them? How much time are we given for our computations? A lifetime? A couple of lifetimes? The lifetimes of all humans currently living? The lifetimes of all humans who ever lived? Can future generations complete our calculations? Are idealizations like the asymptotics of computational complexity theory okay? Do we perhaps want something more realistic like average case complexity? What, in the end, are realistic constraints of practical necessity?
Like in the philosophy of mathematics there is an ultra finitist position, were only those calculations are practically possible which I can actually do (or should I rather say actually did? – how can I be sure I can do them before doing them?). But this is clearly an absurd requirement for practical possibility, even Humphreys could not be happy with it. Still, how much leverage do I get in practice? Can I compute the logarithm in practice, if the only thing I every do is using numpy? I mean, I’m pretty sure I could do the first few terms of the series by hand… But alas, ars longa, vita brevis.
To come back to the (un)necessary involvement of simulations in the Higgs discovery: Simulations have been used in its discovery, so much is clear. But as Krämer et al. argue that doesn’t make them practically necessary. If we believe their counterfactuals, the discovery would only have taken a couple of years longer without the use of simulations – a pessimistic estimate of 10 additional years seems plausible. But a decade is an eternity in science funding, especially if wars and pandemics soak up resources in contracting economies. Ah, the joys of evaluating counterfactuals! So maybe then, practically, the discovery was impossible without the use of simulations after all. Baseless speculations you might say – and I agree. But we only arrived at those speculations because we wanted to make a philosophical point.
Maybe we should take this as a hint that claiming or refuting necessary involvements of technologies in scientific practice is not a terribly interesting business, and perhaps the interesting point was that simulations were actually used in the discovery of the Higgs.
- The earliest example that comes to mind is James Moor’s use of the stratagem in a paper establishing the field of computer ethics. “The mark of a basic problem in computer ethics is one in which computer technology is essentially involved […].” (What is computer ethics?, 1985, p.267) ↩︎
- Actually the distinction between theoretical and practical possibilities can be traced back to his “Extending ourselves” (2004) ↩︎