I enjoy Graham's essays and find many of them quite insightful, but they tend to be rather light on empirical evidence, which makes it hard to say how much of the insight is real.
Several points:
For many of the things he writes about, we can take his clout and background as evidence that his insight is 'real'. It doesn't have to explained via careful science to be true. I think the fact that Paul Graham is saying something, for certain subjects, makes it highly likely to be very-mostly-correct. I'll happily believe it to a high degree unless I have reason not to. I believe the evidence that they're mostly-correct is that he wrote them, and the evidence that things he writes are mostly-correct is that they have been so in the past
This isn't really a full post, but merely a note of potential interest. Paul Graham (who runs Hacker News) has several very interesting and thought-provoking essays located on his personal website. To me they fit very well with the style of thinking employed and advocated by many people on LW and I'd advise that nearly anyone interested in LW check out his work.
I especially recommend Keep Your Identity Small, What You Can't Say, and What You'll Wish You'd Known, but nearly every essay up there is interesting to me in some way. Many of them are directly relevant to issues of rationality, while others are only indirectly related, but either way I found them worth my time.