the fact that you still think of sex as a binary difference between xx and xy shows how far outdated your ideas about the science of sex differences are. even "biological sex" (not even gender) is a spectrum influenced by a variety of different factors
http://scopeblog.stanford.edu/2015/02/24/sex-biology-redefin...
"Light switches are in fact a binary difference, dimmer switches are a small portion of all switches."
"Life is in fact a binary difference, brain death patients are a tiny portion of all patients."
This seems analogous; we should design social structures (or allow social structures to emerge) that assume that individuals are male or female, but which handle the exceptional case of non-cisgendered individuals. Just like a software system that doesn't handle exceptions gracefully, a social system that lacks provisions for these is fundamentally flawed and can fail unexpectedly and potentially catastrophically.
Society also lacks "provisions" for violent criminals[0] but it still handles them and is yet to fail catastrophically.
[0] I'm not equating these two categories of people, merely illustrating that your example of comparing software exceptions to people is deeply flawed.
Metastable states[0] are a tiny portion of all logic values. They exist and are inherently unavoidable, but somehow massively complicated systems that fundamentally rely on the assumption that values are either 0 or 1 still work.
[0] https://en.wikipedia.org/wiki/Metastability_in_electronics