It's trial and error mostly, paying attention to other people doing well or making mistakes, getting honest feedback from a skilled and trusted friend. Learning social skills is like learning to ride a bike, reading about it doesn't give you much of an advantage.
The younger you are the less it costs to make mistakes. I think a social job is a good way to learn because customers are way less forgiving than other people you randomly meet. You could volunteer for some social tasks too.
If your native hardware is somehow socially limited then you might benefit from reading a little bit more and you might have to develop workarounds to use what you've got to read people. It's difficult to learn from mistakes if you don't know you're making them.
One thing I've learned about the average human looking like a telepath is that most people are way too certain about their particular assumption when there are actually multiple possible ways to understand a situation. People generally aren't as great at reading each other as they think that are.
My native hardware is definitely limited - I'm autistic.
The standard quick-and-dirty method of predicting others seems to be "model them as slightly modified versions of you", but when other people's minds are more similar to each other than they are to you, the method works far better for them than it does for you.
My realtime modeling isn't that much worse than other people's, but other people can do a lot more with a couple of minutes and no distractions than I can.
Thanks a bunch for the suggestions!
This thread is for asking any questions that might seem obvious, tangential, silly or what-have-you. Don't be shy, everyone has holes in their knowledge, though the fewer and the smaller we can make them, the better.
Please be respectful of other people's admitting ignorance and don't mock them for it, as they're doing a noble thing.
To any future monthly posters of SQ threads, please remember to add the "stupid_questions" tag.