ndanger.organism

I’ve finished going through all the OOPSLA 2007 podcasts, and was struck by the keynotes from a pair of CS classics. David Parnas talked about the need for documentation while Frederick Brooks talked about collaboration and conceptual integrity. Parnas’ talk is clearly against the current hip “we don’t document because we write clean code” meme. Brooks’ love of conceptually integrity is uncontroversial. But he reiterates an idea he mentions in The Mythical Man Month, a development team with a clear leader, which is against the agile idea of a flat team that delegates design to individual coders (with design reviews, etc. for quality control).

It’s interesting, and kind of nice, hearing people I respect diss a development methodology I enjoy. My first reaction was “what?!”, which was a clear sign that I hadn’t been thinking critically about Agile for awhile. I’m also curious how their views are due to their age, or more specifically due to their history. Parnas’ quotes frequently from his work on airplanes for the Navy and the Airforce while Brooks is known for IBM’s OS/360. Both are the massive systems that are difficult–well, or incredibly easy, depending on who you ask–with Agile. Likewise, while both Parnas and Barnes state that they’re anti-waterfall, I wonder how they would imagine the development of something like a 37 Signals web app.

And then there’s John McCarthy…

…whose discussion of Elephant 2000 was complex enough that the only question was of the “I’m smart and know of this related research… do you?” format. In the spare time left over he went through a history of Lisp. Everyone loved that part.

So is John McCarthy bad at explaining things, or is he still so far ahead of his time that people will be talking about Elephant at OOPSLA 2057?