Thoughts on falsificationism

On multiple occasions, I’ve come tantalizingly close to comprehending falsificationism - Popper’s principle that scientific theories should be disprovable through empirical testing. In the Popperian sense, scientists try to falsify (disprove) hypotheses rather than prove them true. A theory that withstands attempts at falsification is considered good, but paradoxically, a theory that cannot be falsified at all is problematic - at least in empirical science.

That is for philosophers. It is very alluring for scientists to take the unfalsifiable turn, or the metaphysical turn, as it is sometimes called. Eminent scientists turning to philosophical topics is a recurring theme. I remember Karl Friston talking about this in a podcast, calling it the philosopause - a contraction of philosophy and menopause. I imagine that eminent scientists finally muster up the courage to speak on themes they have probably been thinking about for decades, but felt they couldn’t speak on because it’s not empirical science. In its worst manifestation, it may lead to what is known as Nobel disease (the link shows countless examples). In its best manifestation, scientists write amazing books where they finally think freely, and we get the condensed version of decades of unscientific thoughts. The reason why scientists cannot resist the desire to turn to unfalsifiable theories bears some resemblance to my own struggle with the topic of falsification. Perhaps falsificationism just doesn’t come naturally to humans, in the same way that expert-level physicists are unable to develop an intuitive understanding of Newtonian mechanics.

In trying to understand falsificationism more deeply, and perhaps develop some intuition around it, I created a flashcard about the contemporary response to falsificationism. What it says is that: (a) scientists seem to make truth claims all of the time, and (b) the hypotheses that scientists seek to falsify are found using theoretical frameworks that rest on information that they have often come to accept through uncritical assumption. Now, this is probably the tenth time seeing this flashcard, but only now have I begun to disagree with it. Theoretical frameworks can just as well be grounded on thoroughly researched and as of yet unfalsified hypotheses. At the same time, I realise that these critics may not be referring to the hypotheses themselves, but rather to the way they are purported to interact with each other within the framework. Can all implicit hypotheses underlying a theoretical framework be falsified? This question becomes even more intriguing when considering how scientific knowledge is actually generated in practice, as illustrated by this observation by Gwern Branwen’s when he appeared on the Dwarkesh Podcast:

I think another issue is that they shared the basic error that I was making about algorithms being more important than compute. This was in part due to a systematic falsification of the actual origins of ideas in the research literature. Papers don’t tell you where the ideas come from in a truthful manner. They just tell you a nice sounding story about how it was discovered. They don’t tell you how it’s actually being discovered. So even if you appreciate the role of trial and error and compute power in your own experiment as a researcher, you probably just think, oh, I got lucky that way. My experience is unrepresentative. Over in the next lab, there they do things by the power of thought and deep insight. ¹

It’s an interesting observation on the day-to-day reality of doing science. People generally just stumble upon truths through repetition and slow accumulation of related facts. It ties into the topic of this post in two ways: (1) it uses falsification in a peculiar and confusing way, and (2) it shows that science is humane. With falsification, he seems to mean misrepresentation or wrongful, rather than the scientific disproval of hypotheses. And on science, he is saying that the way everyone writes articles is in a way that uses some leeway in hiding the messy, trial-and-error nature of scientific discovery. Instead, papers present a polished narrative that obscures the actual process, potentially leading researchers to underestimate the importance of factors like computational power and overestimate the role of pure insight. This ‘falsification’ in scientific literature contributes to a skewed perception of how breakthroughs occur, perpetuating the myth of linear, insight-driven progress rather than acknowledging the often chaotic and resource-dependent reality of scientific advancement. Perhaps that is just another example of how our very human tendencies leak through into everything we do, from conducting science to understanding falsification. And perhaps, I’ve learnt a little better what falsificationism means.