Look out for those moments where you find yourself thinking "There's got to be a better way", whether you have in a mind a 5 second task or an application architecture. There is a probably a better way. And chances, at this point, someone has probably already figured it out and published it somewhere.

This at least feels like a primary way that I improve. I start with the sense that there is a better to approach a particular problem, and I start to research and discuss that.

Some things took a lot longer to figure out and sink in. Reading "Code Complete", as mentioned above, has really helped with my understanding of appropriate use of global data. In retrospect, I would have read more books on the theory of programming earlier on. It's easy enough to find a helpful module on CPAN for a particular task. It's harder to find a good resource that discusses the merits of one software design philosophy versus another in a useful way.

Perlmonks is useful, but you aren't likely to get a 15 page response examing the details on how to best name subroutines. :)

Mark