← home

the master argument for trans inclusion

(or: induction with malintent)
~2887 words (~11 minute read)
01 04 2026


Words, words words. Hamlet, Hamlet


the argument

The statistical syllogism is an inductive form of argument which applies a general observation to a particular case to yield a probabilistic conclusion regarding the particular case. Here is an example:

  1. Most cats have four legs.
  2. Rosita is a cat.
  3. Rosita probably has four legs.

Or:

  1. (Approximately) 99% of American teenagers own a cellphone
  2. Ana is an American teenager.
  3. So, there is (approximately) a 99% chance that Ana owns a smartphone.

These sorts of arguments, while not confirming their conclusions with deductive certainty, nonetheless serve as evidence for their conclusions. In some cases, they serve as enough evidence to justify belief in the conclusion: in the first case, given that I know that Rosita is a cat, then I am justified in believing that she has four legs (given my background knowledge about cats). The justification granted by such an argument might be weaker than the justification offered by looking at Rosita's limbs myself, but it is justification nonetheless, and, combined with the truth of the belief, counts as knowledge.

That being established, here is a string of unrelated observations. Firstly, it seems like some sentences can be ruled as true or false by only looking at their structure. If we take a sentence like 'Mia is a dog and Mia is not a dog,' abstract away its content, and consider it only as a sentence of the form 'P and not P,' we are nonetheless able to judge it false. I need to know nothing more about the sentence than its form (I do not need to know who Mia is, what dogs are, etc.) in order to know that it is false. We can make these sorts of judgements categorically, in fact, and judge that any sentence of the form 'P and not P' is false. The same can be said in the affirmative case: every sentence of the forms 'P is true if P,' 'S is F or not F,' and 'If P is true, then P could be true,' and so on, are all true. It seems, further, that these categorical truths about sentence-forms can be applied to particular instances of these forms to yield judgements on their truth. A sentence like 'Mia is a dog or is not a dog' can be ruled true, again, without any knowledge of its content, if we only attend to the fact that it is an instance of a necessarily true form: 'S is F or not F.'

Consider the following form: '(adjective) (plural noun) are (plural noun).' Instances of this form include 'Nebelung cats are cats,' 'little dogs are dogs,' and 'friendly slavs are slavs.' While this form may not be strong enough to guarantee the truth of its instances (like in the examples considered before), it is nonetheless the case that most of its instances are true. A smart gambler would bet on the truth of a sentence P given that his only knowledge regarding P was that it is an instance of this form.

But 'trans women are women' and 'trans men are men' are of this form. So, probably, trans women are women, and, probably, trans men are men. Organizing these observations in the form of a statistical syllogism yields something like our toy arguments from earlier:

  1. Most sentences of the form '(adjective) (plural noun) are (plural noun)' are true.
  2. 'Trans women are women' is of that form.
  3. So, 'trans women are women' is probably true.

And, optionally, the truth predicate can be convention-T'ed away:

  1. 'Trans women are women' is probably true if and only if trans women are probably women.
  2. Trans women are probably women.

Call this the master argument. This argument, prima facie, at least, does very well. The statistical syllogism is a cogent form of inference, and both of the premises are very plausibly true. The sentence in question is clearly of the relevant form: 'trans' is an adjective, 'women' is a plural noun, and both of these are followed by 'are' and then the same plural noun, 'women.' That most sentences of the form are true seems straightforwardly obvious, but for reasons of completeness, I've collected a sample. This footnote contains 100 sentences of the relevant form. Every sentence in the footnote is true. So, inductively: every sentence of the form is likely true. If you'd like to generate your own sample (to strengthen the inductive case or to verify I've not cherrypicked the examples), you can use the same generator as me, linked here.


objections


counterexamples

There seem to be counterexamples to the first premise. Former teachers are not teachers, fake astronauts are not astronauts, and fool's gold is not gold. These counterexamples are of no relevance to the argument, however: the first premise is a claim about most instances of the form, not a claim about all. The existence of American teenagers without smartphones does nothing to undermine the fact that 99% of American teenagers, regardless, own smartphones.

There may be a stronger way of presenting these sorts of counterexamples, though. Could one not pose an argument that fool's gold is gold, making the same rhetorical moves as I have? Something along these lines:

  1. Most sentences of the form '(adjective) (mass noun) is (mass noun).'
  2. 'Fool's gold is gold' is of that form.
  3. So, 'fool's gold is gold' is probably true.
  4. 'Fool's gold is gold' is probably true if and only if fools gold is probably gold.
  5. Fool's gold is probably gold.

This argument has a false conclusion, but is of the same form as my argument for trans inclusion, and relies on (practically) identical premises. How can this presentation of the objection from counterexample be resisted?

I don't think it should be. I grant fully that this argument is evidence that fool's gold is gold. But not all evidence warrants immediate assent for its hypothesis: in the presence of countervailing evidence, we can be rational in denying views which nonetheless see some evidential support. One could, for example, look at a clock displaying a time of 12:05, but read a time of 12:10 on their (more reliable) phone. While the clock's reading serves as evidence that it is 12:05 (it is typically reliable), there is stronger evidence to the contrary of the clock's reading which outweighs its evidential support for the proposition that it is 12:05. In the same vein, there is evidence to the contrary of the above parody which outweighs the parody's justificatory force. Specifically, this evidence is the non-subsectivity of 'fool's' in the phrase 'fool's gold'. Non-subsective adjectives are those whose applicability allows one to conclude that the noun term does not really apply to the subject: the inference from 'rosita is a fake cat' to 'rosita is not a cat,' for example, is valid. Subsective and intersective adjectives, on the other hand, allow one to conclude that the noun term really does apply: the inference from 'mia is a good dog' to 'mia is a dog' is valid. In the case of 'fool's gold,' the 'fool's' is of this second sort of adjective, and to infer that something is not gold from the fact that it is truthfully called 'fool's' gold' is valid. The statement 'fool's gold is gold', then, is self-undermining: calling the gold 'fool's gold' is, itself, admitting that the 'gold' is not gold. In light of this, it can be conceded that the above parody is evidence of its conclusion, but that we are nonetheless rational in rejecting its conclusion in virtue of the stronger countervailing evidenced just outlined.

shadowboxing

The transphobe could make the same moves that I did in the last paragraph of the prior section. In order to resist the conclusion of the argument, they can grant that the argument is evidence, but take it that there is some countervailing evidence that rebuts the conclusion. A possible available option for such countervailing evidence could be extrapolated from my own argument, too: the transphobe might say that 'trans' is non-subsective and means something like 'fake,' just like the 'fool's' in 'fool's' gold. What evidence is there for this semantic thesis, then? Not much, it seems. A cursory glance at a few dictionary entries yields poor prospects for this transphobic rebuttal: Merriam-Webster, Cambridge, and Oxford do not offer definitions of 'trans' along these lines, and both Merriam-Webster and Cambridge offer trans-affirming definitions of the phrase 'trans woman,' while Oxford's is neutral. I can see no other sources of impartial support for this thesis.

Rather than trying to argue that 'trans' means 'fake,' a weaker objection might be forwarded by the transphobe: the semantic content of 'trans woman,' at the very least, requires that the person in question be assigned male at birth. But people who were assigned male at birth, the argument might go, cannot be women. This only pushes the question back a step, though: why suppose that AMAB people cannot be women? The master argument, detailed earlier, seems to provide evidence that a categorical claim like 'all AMAB people are not women' is false by means of counterexample. Just as we would deny a definition of 'U.S. teenager' which makes it such that, analytically, U.S. teenagers cannot own phones (despite strong evidence in favor of such people), we can deny that trans women analytically cannot be women because we have strong evidence (namely, the master argument) that they are. So it is on the onus of the transphobe to propose an argument for their semantic thesis which outweighs the evidential support for the existence of counterexamples to such a thesis provided by the master argument. With another cursory appeal to dictionaries, though, we run into the same evidence we saw earlier: most English-language dictionaries define 'trans woman' by using the term 'woman.' Cambridge's definition of 'woman' (reading 'a woman who was considered to be male at birth') analytically entails that trans women are women, and Merriam Webster's (reading 'a transgender woman; a woman who was identified as male at birth') does so as well.

There are, of course, more issues to be considered in the second, weaker reply. There are far more cases (plausibly ones stronger than the cursory appeals listed here) for antitrans semantic theses. These cases, however, are to be tabled for present purposes and dealt with in a separate post. Before moving on to other objections, I want to make clear a more obscure upshot of the preceding: the master argument, at the very least, places the transphobe on the dialectical backfoot. That is, because it offers significant evidence for the validity of trans identities, the default presumption, in the absence of any contrary evidence, is trans inclusion. It becomes the transphobe's job to offer a substantive and powerful against trans inclusion, one which is able to sufficiently countervail the justification against this semantic thesis provided by the master argument.

other parodies

Rather than operating at the meta-level, the transphobe could generate a parody statistical syllogism which operates directly at the object-level. This way, they can construct a similarly underhanded inductive argument that isn't prone to my prior complaints. Such an argument could go along these lines:

  1. Most people without XX chromosomes are women.
  2. Trans women do not have XX chromosomes.
  3. So, probably, trans women are not women.

To make clear the issue at play here, consider a very similar argument:

  1. Most people without uteruses are not women.
  2. Women with MRKH do not have uteruses.
  3. So, probably, women with MRKH are not women.

The problem, in both cases, is symmetrical: these arguments are ignoring relevant information and updating on only one proposition, rather than the conjunction of this proposition alongside the relevant information. This leads the arguments to haphazardly include a large population of people (i.e., cis men, who could not possibly fall within the categories of women under discussion) within the sample we're considering when questioning the possibility that trans or intersex women are women. When we consider the other relevant facts, however, (viz. the transness or intersexuality of the relevant groups), we realize that they could not be cis men, and the population used in generating the statistical information in the first premise of either argument shifts. 'Most people without XX chromosomes who are trans are not women' is false (or, at least, is not a dialectically effective premise in an argument for the conclusion that trans women are not women), and 'most people without uteruses who have MRKH are not women' is false too.

Consider a more straightforward example: I know that the ground is wet. This is good evidence that it is raining. But I also hear a sprinkler going off. The sound of the sprinkler explains away the alternative hypothesis of rain, and, were I to update my beliefs based only on the wetness of the ground (rather than on the conjunction of the wetness of the ground and the sound of the sprinkler), then I would significantly overestimate the chance of rain. It is not obvious that the same sort of shadowboxing from earlier is possible here. Classes of relevant information which might impact the sample used in generating the master argument's statistical premise (like the semantic content of the words in question) have already been considered and discarded.

conclusion

The argument speaks for itself.

ฅ^•ﻌ•^ฅ home