NPR’s 13.7 blog has a new post entitled “The Most Dangerous Ideas in Science.” In it, the author discusses two of the more controversial ideas in modern physics: string theory and the multiverse. The debate on whether or not these theories are useful endeavors is presented as a mostly philosophical disagreement between those who believe that theories that could be true even if we can’t get any evidence either way are interesting and those who believe that theory should be much more closely tied to experiment. The potential danger seems to be that theoretical physics could end up in a place where it is completely divorced from even any attempts at experimental verification/falsification.
The post is actually a little difficult to comment on because it contains so little information outside quotes and paraphrases from others. I think both sides – at least, as presented in the post – have some decent points. Requiring theory hew too strongly to experimental data might prevent a lot of possible advances in theory. Predicting new things with little to no evidence is one of the most important things for theorists to do. That helps experimentalists know what interesting things might lurk in their data. Additionally, even intentionally unrealistic theories could yield new insights into real world physics.
At the same, I do agree that it would be dangerous for theory to become wholly detached from empirical science. Ellis and Silk’s main point, which I don’t think is explained particularly well in the NPR post, appears to be that the idea that we should accept the truth of something like string theory without proper evidence is dangerous and not that string theory is itself dangerous. That is something that I don’t have a problem with. Accepting an idea due to ignorance of better alternatives or incredulity at the possibility of better alternatives is certainly logically unsound. However, I wonder if Ellis and Silk go too far in their criticisms. The things they criticize (string theory, the multiverse, and also many worlds) seem to me to really be frameworks within which potentially testable models can be built or (in the case of many worlds) are explicitly philosophical musings on how to interpret empirical theories. A framework could be difficult or even impossible to falsify, but any given model that fits within the framework likely can (at least in principle) be falsified. Ellis and Silk even admit that particular types of string theory yield empirical predictions and that some multiverse theories are also testable.
Even if we never find a viable way to experimentally test string theory, a string theory consistent with the Standard Model would still be a breakthrough because it would provide an alternative model to the quantum field theory framework in which the Standard Model was built. A viable string theory that describes real physics would be very elegant, but that wouldn’t make it true. It wouldn’t make it false either. As it is, we only know that the Standard Model largely accurately describes subatomic physics, but we have probably no idea how many alternative theories also accurately describe the same data. Choosing between effectively identical models isn’t scientific, but that doesn’t mean that theories that replicate our understanding of the universe but also include some unfalsifiable assumptions aren’t useful or interesting. Why not let theorists come up with as many ways to end up with something that looks like our universe as they can? If these ideas are fundamentally flawed, we should show that that is so. Otherwise, we should at least accept such theories as possibilities (not as the truth) and consider if and how they can affect our universe. Honestly, I’m not sure if many of the people involved would disagree with that.