The points in your question that I'd agree with are that more books on specific topics are good and that Perl books are often very generalized. Yes, this can be a handicap for the rawest newbies, but anybody with any exposure to programming in the Cold Cruel World^TM quickly wants to do more than say 'Hello, World' in FONT SIZE='7'. It is very true that PHP has benefitted from the fact that its TIOOWTDI constraints make books easy to write clearly. It has a very limited problem domain, and thus there aren't many options to write about.

I don't think we need to focus on one templating system. I don't see books on TT being in conflict with books on Catalyst or Embperl. Better to have books on all three!

A good example of why having multiple options is a positive virtue: OpenBSD has probably the smallest user community of all BSD and linux variants, but every single one of the others now includes OpenSSH, and the same thing is happening with the pf security construct. Cross-pollination between open source projects means that good features get propagated in many directions.