> You're assuming people sample unifromly and at random from the population.

I'm not assuming anything. I literally explained that the only way it's possible is for someone to avoid the general population and only socialize in environments with extreme bias.

The more important point is that diagnosing autism is not something you can do by simply meeting people in social situations. It's something that takes training and experience by professionals, not an untrained person who sizes people up as they meet them in a social capacity.

Again, psychiatry is in its infancy. Many professionals use outdated models or stereotypes in practice. Living as an autistic individual can make it easier to clock other autistic people, because it's rare to meet someone who functions or thinks the same way you do and sticks out like a sore thumb. For example, "thinking in pictures" is not a universal autistic trait, but it's a pretty well known one.