Hypotheses vs. Metaphors

What does intellectual progress look like?

The idea of “science” as the singular method for legitimizing knowledge is the current memetic winner, despite the obvious ways in which people use more than just “science” to know things.

Consider, this excerpt from the top conversation on letter.wiki: “Are The Methods Used By Science The Only Ways Of Knowing?”:

Which brings us to science. Some of the knowledge above isn’t seen as scientific knowledge because it’s not determined by scientists. But my definition of “the methods of science” is broad. I don’t mean institutionalized science of the sort I practiced, involving funded research, laboratories, and published papers. Instead, I see science as a set of tools, forged by experience, that help us find truth.

While there’s no one “scientific method”, those tools are familiar: observing reality, checking observations with other people, testing what your ideas predict, resisting your biases, and so on.  Observations generally segue into knowledge when repeatedly confirmed, adding more and more credibility to a hypothesis. Finally, empirical knowledge is always provisional, though of course some knowledge, like the formula for water being H2O, is unlikely to ever change. Still, nothing, including science, can tell us when we’ve attained the absolute and unchangeable truth.

The methods of science, then, simply involve applying the philosophy of naturalism, whose tenets, according to physicist Sean Carroll, are these: “There is only one world, the natural world; the world evolves according to unbroken patterns, the laws of nature; and the only reliable way of learning about the world is by observing it.”

This is not a process or a method, it is a description of an idea’s place in culture. This is the way that the word “science” has become a conceptual tool for understanding the “ground” of knowledge legitimization: if you can make the case that what you’re doing is science, then it can be said to produce real knowledge.

Interestingly, scientists themselves have gotten into a pickle of their own: in the current climate it is very difficult for anyone except an extremely famous and rhetorically clever scientist (i.e. a politician in a lab coat) to propose an idea. Instead scientists have to propose hypotheses about specific literal objects we currently have names for, or else they’re not being rigorous.

This is unfortunate, when one considers the fact that many of the ideas we consider really important in science were either vague, impossible to specify, or wrong in the given context when they were first proposed. Darwin’s On the Origin of Species does not refer to any experimental evidence (and the inquisitive reader is invited to consider whether experiments in evolutionary science are possible in the same way they are in physics), but rather refers to certain well-known (at the time) facts about the world and suggests an underlying idea that would allow us to thread them together in time. Mendel may have lied about his experiments (or not—I really couldn’t care less at this point), but surely thinking about Mendelian genetics really has nothing to do with that?

What we get from these scientists are metaphors—ways of thinking about a process that really exists as if it was a process one actually has the words to describe. It is not an analogy, because we treat these models as if they were the truth, until they are falsified and we make a new metaphor with which to see them. I use the word “metaphor” instead of “model” to emphasize that when we talk about the changing of species over time, it may be the case that we can’t even really say that species are discrete, but our perception of reality is extruded by the idea that a “species” evolves due to natural selection and that is the reality we are capable of witnessing in the abstract. For most people on most subjects, we cannot actually go and look at primary evidence, so these metaphors are literally the only way we are capable of having an acquaintance with the idea of a “species over time”. For investigative spirits who try to figure out more about “what’s really going on” we cannot keep all the evidence in mind at once, so we rely on a grab bag of metaphors to perceive a specific bit of evidence until we come up with an alternative, usually inspired by local observations that allow us to change the old metaphor or make a new analogy with something else we think we know about.

These metaphors are not hypotheses—they cannot be proven wrong in any meaningful sense because they float at too high a level of abstraction. Rather, they are a shared narrative that lead to mutually intelligible hypotheses that can then be tested. It is these shared narratives that underpin scientific communities, because without them we could not agree even on what evidence might mean. We would be stuck saying, “Sure, the light from that laser was reflected off of a massive object, but I don’t know what you mean precisely by ‘moon’.”

Currently, though, most scientists cannot propose new metaphors in scientific writing. So the famous ones do so in talks and blogs, and others find the closest thing they can that someone else wrote and attempt to tweak it into what they came up with. Of course, this disincentivizes spending too much time thinking about these conceptual tools in the first place. This is a very sad situation, but it is downright dangerous when it comes to inherently complex and fluid concepts like culture and language. We may have made GPT-3, but how do we plan to study it, if only Dawkins can coin the term “meme”? Imagine how many linguistic patterns that GPT-3 is picking up on that people are having trouble publishing about because their descriptions can’t be checked by another machine.

To move forward, we need to give-up on statistically verifiable hypotheses as the only means of knowledge legitimation—which means finding a way to make discourse where we can try to cooperate around what’s useful instead of “defending” our theses.