Patrick Logan points to a great article on how to train software developers. I know from first hand experience that this is a model that works.

Tom Coates notes that just about everything is becoming amateurised. I think that there are parallels with open source. There are a couple of quotes in his post that are relevant to open source as well. The first is about access to equipment.

Fundmentally it's because the gap between what can be accomplished at home and what can be accomplished in a work environment has narrowed dramatically over the last ten to fifteen years.

For the production of much computer software, this is definitely true. In fact, at many of the jobs or clients I've worked at, I actually had better equipment at home than I was given "on the job". Open source tools for developing software have also improve substantially, although I thing there is some disagreement on whether or not we've reached parity with "professional" setups. On the other hand, I know that I've saved a lot of people time and money by demonstrating Eclipse for them.

The second quote has to do with information.

But it's not only equipment that separates the professional from the amateur, it's also access to information.

This is the other big thing that has enabled open source, is the availability of information, specifically, the source code. Lots of other information is being exchanged as well.

The thing that bothers me the most about his post isn't really his fault at all. It's the categorization of amateur and professional. I hear this one a lot as a criticism of open source projects. I don't think that the labeling is particularly helpful. Just because someone gets paid to do something doesn't automatically make them better than someone who does the same thing as their hobby. There are lots of reasons why people choose not to perform a creative activity professionally. That shouldn't be taken as a knock on their abilities. I wish that we could find better terminology. After all, the software running 64% of the web servers on the Internet was written by "amateurs".

Jason Marshal thinks that we should be using byte code manipulation to achieve generative programming. I think that's an interesting idea. Certainly libraries like BCEL, jclasslib, Javassist, JOIE, and serp could use improvement.

What's more interesting to me is that Jason is advocating this API based approach to code generation, but thinks that macros are more trouble than they are worth. Yet macros would give you much of the same functionality as a byte code library, although not all of it, and do it in a structured, disciplined way. So what am I missing here?