Comment by zz3

My comment is too long, so I'm going to try to separate it into 2.

> Your comment about "drawing an arbitrary line" doesn't really fit with how biologists see this either, as it's not arbitrary at all but is based on understanding the mechanisms of reproductive function and development…

I talk to biologists all the time for work. I studied biology in college and I work in science and engineering. This is how we talk about data and models in science. What I'm seeing in your response is a fairly deep misunderstanding in how science works, and that's why it might appear to you like I'm not actually answering your question. I am not "proposing a new model" about sex being a spectrum.

Let's take a step back and look at how science works and hopefully we can address this misunderstanding. I'll start with the fundamentals.

We use words to communicate about the world, but they're an imperfect medium. As a quick example, I can call an ant hill a "mountain." What information I might be trying to convey depends on: (1) definition, maybe my definition of mountain is .5 cm, (2) context, maybe I'm speaking metaphorically, (3) relevance, maybe I'm speaking from the perspective of an ant. So key things in communication are (1) definitions, (2) context, (3) relevance. Whatever word I call the object doesn't change the object in any way. This is because all words are representations (or models) of the world, and all models are, by nature, false. They cannot possibly describe every aspect of reality. Words are only as useful as the information they convey. So how can we evaluate the information in words or models?

We evaluate models by treating them as black-box functions and comparing their output to reality. We’re trying to measure how useful or predictive the model is.

How do we do this in practice? We (1) propose a hypothesis, (2) decide which variables are relevant (3) decide on necessary and sufficient conditions or some kind of function. Then we run that function and compare the output of our model to what we measure in reality.

Let’s look at a binary model for human sex. Our hypothesis is that we can define a set of criteria or definitions such that the output is either 1 “female” or 0 “male”. The definition of binary means that it can be fully and completely described by 1 or 0, nothing in the middle. For example, TRUE or FALSE is binary.

Let’s hypothesize that human sex is determined by chromosomes, so therefore XX is female (1) and XY is male (0).

XXY exists (Klinefelter Syndrome). That breaks our model. We can update our criteria. [XX is female] (1) and [XY and XXY are male] (0).

XY + no SRY exists. That also breaks our model. We can update our criteria. [XX and XY + no SRY is female](1) [XY + SRY and XXY is male (0)].

Lots of intersex conditions exist. What does that mean for our model?

To skip forward, we still do not have a defined set of necessary and sufficient criteria where we can describe all outputs of human sexual development with 1 or 0. This means the assumption that the output is binary for our model is broken. Can we still sort everything into binary categories? Sure. Nothing is stopping you from labelling something, but we understand that we’re giving up information while we’re doing this. When we talk about binary models, what we’re referring to is an output of a function or model, not just applying the labels. We do this because we can obviously just apply whatever labels to anything, there’s nothing “scientific” about it. So what’s actually important in evaluating that model is the necessary and sufficient criteria we come up with: the actual function, model etc. This is what I mean by “binary system.” You can still obviously sort the data into binary categories, but the underlying data is fundamentally a spectrum.

This is essentially called a “proof by contradiction.” We had a set of necessary and sufficient conditions, a set of defined definitions (like binary), and we found counterexamples for each one.

Another perspective could be: if I see an individual, what are the chances I could correctly guess characteristics about them (chromosomes etc) based on their phenotype? For male and female, you might be able to fairly accurately guess certain characteristics about them, probably somewhere around 80%. Does that mean that it's impossible for us to sort every individual into two categories? No. But being CAPABLE of sorting or labeling them into two categories does NOT make something binary.

To be honest, the existence of intersex alone should be sufficient to tell you that male and female are not completely “binary” concepts. So we’re dealing with some kind of discrete or continuous “spectrum.” For convenience and simplicity, let’s say it’s male, intersex, female, so discrete but not binary.