Well, it's not so much that he can't write code (he can) but more that he markets his hacking skills to an extent greater than what his open source software indicates.
He compares himself to Stallman and Torvalds without having anything nearly as impressive as an OS kernel, C compiler or debugger.
I will say though, that he's good at introspecting on code and articulating points about programs better than most (hence the high quality of the book)
> He compares himself to Stallman and Torvalds without having anything nearly as impressive as an OS kernel, C compiler or debugger.
Does he compare his hacking abilities to Stallman and Torvalds?
My impression is that he considers himself a tribal elder, and compares himself to Stallman and Torvalds in that respect. But that's not quite the same.
I'm pretty sure he has worked on compilers and OS kernels. They just weren't as popular as GCC or Linux. I don't think judging one's programming skill based on the popularity of one's projects is rather fair.
People who can't build excellent software have written many good books.
Writing a good tech book requires deep understanding of a subject.
Writing good software takes discipline, creativity, and (interestingly) not necessarily a particularly deep analytical understanding of underlying concepts. People have written truly excellent software on poor CS foundations.
Edit: Yes, it was written by a guy named Carl Harris and ESR took it over when Harris no longer wished to maintain it: http://en.wikipedia.org/wiki/Fetchmail
If ESR can write a really good software development book, why do some people thinks ESR can't code a damn?