> But people on the autism spectrum make up such a small population of humans and software engineers (we are over-represented compared to other industries, but in my experience we are still very much a minority overall) that it cannot be nearly enough to account for the sex-imbalance numbers we see today.

Imagine a scale that goes from autism to whatever the opposite of autism is, with neurotypical as the middle. The hypothesis is that men are more likely to be on the autism side of dead center, even for people who are still in the neurotypical region. A higher percentage of men being diagnosed with autism is evidence of the hypothesis being true.

> So I guess I'm still firmly in the "Its a pipeline problem" camp.

They're kind of orthogonal. It clearly is the pipeline problem -- you can't hire people who don't get the relevant qualifications. But it's possible that biological differences contribute to the pipeline problem.

In reality anything is affected by everything. So it makes sense to work against the pipeline problem regardless, because some of it is caused by non-biological factors that could be addressed. But some of it may also be caused by biology, which means that even if some progress is attainable, expecting the end result to be a 50/50 split may be unreasonable.