I had been intending to write something here on the blog about this essay by George Ellis, so when I was contacted by someone at Inference about writing a letter in response, I did so for publication there. It has now appeared in their latest issue, with the title Theorists Without a Theory.
The topic is one I’ve addressed here all too often, but the main point I was trying to make is perhaps a new one. When I was writing here about the controversy over inflation one thing that struck me was that the pro-inflation side was responding to arguments that their theory didn’t solve the problem it was supposed to by in effect saying “the real theory is much more complicated” (see the paragraph beginning “Besides our disagreement…” on page 3 here). One way of seeing part of what is going on here is that most of what gets advertised as “theories” of inflation are actually more appropriately described as toy models. They involve a single inflaton field with a simple potential and unknown couplings to matter, intended as a toy model for the real theory (which will have lots of fields, complicated potentials and specified couplings to matter). An aspect of the controversy is one side pointing out that this theory doesn’t solve problems it is supposed to solve, with the other side arguing that it’s just a toy model.
People sometimes note that there’s a terminological problem with “string theory”, in that the public is often told that “theories” are solidly tested parts of science, which is not true in this case. The actual usage among physicists is different though, with “theory” often used to mean a specific mathematical model or set of models, with no implicit claim of a successful experimental test. A lot of the problem with the usage “string theory” is that no one knows what the actual theory is: it’s a conjecture that a theory with certain specific limits exists. The main point I was trying to make in this piece is that to a large degree the arguments over the scientific status of string theory (and of its supposed landscape and multiverse) revolving around its lack of testability are moot, since the underlying problem is something different: that there is no real theory to argue about. String theorists often try and evade this problem by a terminological shift: string theory is not a “theory”, it’s a “framework”. “Framework” is a much more ill-defined term than the already ill-defined “theory”. A theorist who says “I have a framework, not a theory” is actually saying nothing more than that they are a theorist without a theory.

