What is Science Anyway?

A partial discussion on the nature of science.

This article is a full reply to a discussion with D. L. Shultz’s about his article on soft vs hard science (not fully safe for work).

So interesting! Science is indeed science, but I’m wondering whether everything that is scientific is necessarily science? I can take a scientific approach to understanding what I’ve written in my journal (how many words, which words, dates written, content analyses, etc.), but does this mean I’m doing science? I currently am leaning toward no, but am also loudly proclaiming that knowledge need neither be scientific nor based in the scientific method to be valid knowledge.

Before getting into science, let’s talk about mathematics. People often think that mathematics is all about numbers, counting, etc. But really, mathematics is the study of formal logic. It is the process of taking a collection of axioms and definitions, and seeing what we can produce with them, when we apply a system of logic.

Taking a number of axioms, including the ones necessary to formulate the natural number system and probability theory, and using the standard logical framework that we tend to use in mathematics, which requires that a proposition is either true or false, and never both, and combining it with the axiom that our most recent empirical observations are true, we’re able to construct mathematical models, which can potentially be falsified, by our empirical observations. That’s science.

Facts in Science

Evolution is fact is something I hear a lot. I understand why people might think that it is indeed fact, after all, it’s been tested over and over and over again, and all these tests seem to confirm evolutionary theory. But science is not about confirmation. It’s about falsification of theory, and if we give it a little bit of leeway, prioritization of theory. The reason why is because of the nature of science that I mentioned above. In science, we construct mathematical models, which can make predictions about observations we might make. Then we can use statistics to attempt to falsify the theory, in a form of statistical proof by contradiction.

But proof by contradiction does not allow us to determine that our assumption is correct. That’s because proof by contradiction starts off with the assumption that our position is true, and then we find a contradiction and so find that the position is actually false! That’s how hypothesis testing works as well. It starts by assuming that our hypothesis is true, and we work from there (The Basics of Hypothesis Tests).

And so?

So what does it mean to “do science?” It really just means constructing these mathematical models and testing them. But I would suggest a slightly more strict definition. Does testing a theory for the millionth time, just to learn how it’s done, in a classroom setting, mean that you’re doing science? Maybe? But if you’re developing new theory, or conducting a new test for existing theory, and are publishing your results in a scholarly format, whether in an academic paper, or preprint server, etc, for public scrutiny, then you’re doing science.

What about knowledge that isn’t scientific? Well, as I said, science is really just an extension of mathematics. So mathematics itself is very important. Our theories are, after all, just mathematical models. Therefore mathematical knowledge is absolutely important, and yet it isn’t necessarily scientific.


I wonder what strong objectivity would look like in the “hard” sciences, and how scientists account for themselves in their research.

This section will be brief. Science is objective, if we’re looking at the actual science, because it is simply an extension of mathematics. Statistical analysis is not subjective. If we have a properly developed theory, we know which kinds of statistical tests to use, and those tests will always give the same results for the same theory and data.

But theory selection is itself biased and there is always a degree of subjectivity in whether people accept findings and in whether they’re properly aggregating experimental outcomes together. In the modern era, we have new techniques, such as systematic review and meta-analysis, which objectively look through the entire body of experimental outcomes.

But there are those who still rely on subjective analysis. That’s why I do not accept scientific consensus: the consensus between scientists. Inherent human subjectivity can easily sway consensus, as it did in the case of Newton and Hooke. But people live and die, while the science remains. The continued accumulation of theory and evidence moves us forward over time, and the biases of individuals, and even groups, are lost to time, making the overall process objective.


A core maxim of mine is “Human subjects are not objects,” which means that humans are based in subjectivity, not objectivity, and that therefore any attempt to claim objective knowledge about human subjects requires objectification of the subjective. It is necessary and important that science eliminate subjective influence from inquiries about objects (e.g., atoms, chemicals, cells, etc.), but if the object of study are human subjects and/or subjective experiences (e.g., stress, emotion, belief, etc.), then the scientific method can only be used if the subjective is objectified, which is necessary but not sufficient for unethical treatment of humans…

I think part of this issue can be “resolved” by looking at science, and then looking at everything that is added to science. Science must be objective, as Shultz says. As I mentioned earlier, science generally boils down to very systematic and formulaic models. And really, science itself is just an extension of mathematics to the empirical world: we construct a model, compare it to observational data, and make a statement about probability from it (statistical proof by contradiction).

But then scientists do something else. We need to understand science in our own subjective way. So we take all our experimental data and results, and we construct that narrative. We take our radiometric dating, our results from smashing particles together, our comparisons to highly rigorous mathematical models like general relativity and quantum mechanics, and we construct a narrative.

In the grand scale, our narrative that we have constructed, and wrapped around our theories and data is as follows: roughly 13.8 billion years ago there was a massive explosion of energy, which quickly expanded. Over many years the results of this explosion cooled, and the energy coalesced into electrons, quarks, etc. Then eventually those coalesced into atoms, and then molecules, stars and planets…. well you get the idea.

Commenting on the original point of Shultz’ article, I suppose you could say that the “soft sciences” have a lot more narrative and more layers of narrative, but at the end of the day, the science itself is still just science.

And there you have it. A lot more could be said on science, and I have said a lot more in my paper on reforming science .But I think this article covers enough to understand the basics of science, and also the narrative which is attached to it. The mathematical aspect of science keeps science objective. Statistical analysis is an objective way of measuring how data relates to science. Meanwhile, the narrative allows us to attach our own subjective understanding onto the theory and evidence.



Originally published on Spiritual Anthropologist.