In the promo for his piece: Post-empirical science is an oxymoron, and it is dangerous | Aeon Essays , Jim Baggott is identified just as a “popular-science author”, but in fact he’s also a real scientist with research awards and applied experience. And this shows in the quality of his essay which objects to much of the sensationalist mumbo-jumbo that passes for popular science writing these days. But there are other real scientists among those he objects to and I think there is plenty of room for people to work on (and write about) speculative ideas that do not yet have testable consequences.
In his article, Baggott refers to an earlier article by Massimo Pigliucci entitled Must Science be Testable? which carried the subhead “String wars among physicists have highlighted just how much science needs philosophy – and not just the amateur version” (to which I responded with a comment at the time).
For me, YES, science must be testable, and No science does not “need” philosophy – especially the “professional” version.
That’s not a statement about how the world “is”, but about how I choose to define the word “science”, and I suspect that most of my scientific colleagues feel similarly (in general terms though not perhaps in all the details).
More completely, I think of science as the game (a word I shall not try to define!) of making the most complete and accurate predictions from the most compact set of assumptions. It differs from some games in that the “scoring” system is not precisely defined and is largely a matter of personal taste (kind of like judged aesthetic events like gymnastics and diving in the Olympics).
String Theory is a part of that game because it is an attempt (still in progress) to interpolate between General Relativity and Quantum Field Theory in a way that reduces the number of bits needed in order to describe the assumptions while not failing to at least match the accuracy of any of the separate theories’ predictions. One can argue as to whether it is likely to succeed, but any claim to have properly constructed it will be testable in several ways – not all experimental. One is just whether it is logically and mathematically consistent. Another is whether it has limiting cases which match GR & QFT. But then it will almost surely make predictions in situations where the other two break down, and so it will also probably be possible to design experiments which detect whether or not those predictions are correct.
A “theory” which I would not consider real science is the “many worlds” interpretation of quantum mechanics. It is designed to exactly mimic the predictions of standard QM and requires exactly the same mathematical tools but just dignifies some of them with the label “real” which sadly does not seem to have any real meaning; however I can see how the interpretation might earn some points if we added a criterion of aesthetic appeal or conceptual accessibility to the scoring table.
There are also some similar-sounding “multiverse” theories that go beyond the many worlds idea and may possibly one day make predictions re observable effects from “nearby” “branes” and such; and the “mirrorverse” idea that Leah Broussard is proposing to explain some observed anomalies in neutron decay time measurements seems to be one of these. As yet it may be even further from being properly constructed than string theory, but I think Baggott is overstating the case against it.
Baggot and others like Sabine Hossenfelder can argue that certain lines of investigation are a poor allocation of scientific resources, but that is very far from making it appropriate to identify them as “pseudo-science”.