2 ms·
I agree with the author's main point (software engineers should take another look at how Alexander frames the patterns in "A Pattern Language"). Obviously we sh
by rando_2458976 10y ago
I agree with the author's main point (software engineers should take another look at how Alexander frames the patterns in "A Pattern Language"). Obviously we should all look for common patterns and problems in our work and build up a shared "vocabulary" of approaches to those situations.
But these pattern catalogs present a risk of seeming complete. To believe that you're near completion on a list of patterns in any field is utter hubris. Alexander's own hubris at having discovered objectively "correct" architectural styles led him down a path towards obsolescence while the rest of the architectural discipline moved on. The hand wavy "scientific" surveys Alexander conducted wouldn't pass a basic social science methods review these days. Pseudoscientific methods of assembling anecdotes or interviews from small samples that don't control for sampling bias will always result in some interesting insights that can't be reliably generalized beyond the narrow scope from which they were gathered. That's what both of these pattern languages are. At best they are a handful of neat ideas that other practitioners can reuse. But these are not some sort of reliable canon. We're all going to have to do more rigorous research if that's what we want.
There might be another parallel in here about the fetishization of anachronisms. Alexander, who is an architect not a carpenter, goes out of his way to glorify hand-crafting things (literally avoiding power tools). Unless you're a carpenter who specializes in make "artisanal" works of art for wealthy patrons, I guarantee you'll enjoy the productivity gains of a pneumatic nail gun. So if you ever find yourself insisting that some old "pattern" is more pure at the expense of everyone else's productivity, double check whether your dogma is worth it.
TL;DR. Don't take patterns too seriously. Always look for new ones.