This is less of a review and more of a reflection of how the issues raised in this book relate to my area of science. As a psychologist, it is refreshing to learn that psychology’s struggles with theories aren’t unique to psychology (and other social sciences), particularly when precise measurement is a challenge (which it almost always is when your subjects are human beings). As soon as you have difficulty measuring the phenomena that you’re interested in, good old human cognitive biases come into play, regardless of whether you’re a psychologist or physicist, regardless of your mathematical proficiency, regardless of the sophistication of your analytical models.
If particle physicists can be biased by subjective perceptions of mathematical beauty, imagine how much of a struggle it must be to deal with theorists (and outsiders) who feel that they can weigh in on your discussions simply because they have intuitive feelings about the constructs in question, which they may or may not justify with a couple of anecdotes. (Which is not to say that that can never be useful or valid in the process of refining one’s ideas, just that it leads to an inflation of self-assured interjections.)
Of particular interest (but only a brief point in the book) is the observation that this struggle apparently applies not only to theory-building, but also to statistical analyses of empirical observations. Hossenfelder provides examples of multiple published results (in particle physics) that were, in hindsight, probably the result of the type of researcher degrees of freedom that have been identified as a cause for many of the unreliable results (particularly false positive findings) in psychology and co. The fact that it seems to be a relatively new development for collaborations in physics, according to Hossenfelder, “… to settle on a method of analysis before even looking at the data (before the data is ‘unblinded’)” suggests that the current wave of open science measures that seeks to boost reproducibility and replicability in the social sciences isn’t too far behind fields that have typically been perceived as being more methodologically rigorous, and needs to be given time to grow. (Although it is important to stress that most of the core issues of these questionable research practices were known and written about by psychologists decades ago, so none of this is an attempt to absolve the field of its belated response to the blame that it is finally attempting to deal with on a more collective basis.)
There’s a fitting quote often attributed (in different forms) either to Richard Feynman or Murray Gell-Mann: “Imagine how much harder physics would be if electrons had emotions.” I currently don’t have access to any sources to verify whether either (or both, or neither) of them said or wrote that thought, but it sure rings true. As Sanjay Srivastava says, if we were to rethink the common continuum of sciences ranging from “hard” to “soft”, and instead differentiate between “hard” and “easy”, all of the extraneous variables that add complexity to the questions that psychology seeks to answer would probably make it one of the hardest sciences. Alongside theoretical particle physics.
Lost in math: How beauty leads physics astray (Sabine Hossenfelder)
Konrad Senf 