I find Christopher Alexander to be a much more opinionated architect than Alexander-pattern advocates are about software, at least in A Pattern Language.
For example, one of Alexander's patterns is to make buildings not more than four stories tall. He presents compelling evidence that living in tall buildings increases rates of mental illness and stunts physical and intellectual growth of children, and explains why the 5th floor is one too many. He is not saying, "maybe you should use the four floors pattern here," he is saying, "buildings should not be this tall!"
Another example is that homes on a street should somewhat cluster together and their doors should face each other. What is the equivalent in software of shaping a street to create a few small communities this way, not as an option but simply as the better way to make a street?
I know we don't wish to be dogmatic about things like REST, but the power of Alexander's ideas come from the superiority of his specific patterns over other ideas, not just his power to organize the practice of a profession into patterns.
I've only read Alexander's Synthesis of Form once, and it was a few years ago. Thoughtful reading, no doubt.
I think you may be confusing design requirements with patterns. His "no more than four stories" was a requirement for healthy human development, not a pattern. His 'patterns' were reusable chunks that fit a situation's requirements.
Patterns in programming only add value to the developers and architects.
This seems like the central misunderstanding of Alexander's work.
His patterns are about involving the inhabitant with the work of architecture. This is why Alexander's work is humanistic and beautiful, while stuff like the GoF patterns book is utterly uninteresting for anyone who's not a programmer.
The process described in Notes on the Synthesis of Form is 180 degrees from that advocated in his later work. It is top down and all at once. His later work advocates many hands working incrementally from the bottom up.
I would also suggest that his work is about the user interface of buildings and thus has implications for the design of analogous interfaces for software.
For example, one of Alexander's patterns is to make buildings not more than four stories tall. He presents compelling evidence that living in tall buildings increases rates of mental illness and stunts physical and intellectual growth of children, and explains why the 5th floor is one too many. He is not saying, "maybe you should use the four floors pattern here," he is saying, "buildings should not be this tall!"
Another example is that homes on a street should somewhat cluster together and their doors should face each other. What is the equivalent in software of shaping a street to create a few small communities this way, not as an option but simply as the better way to make a street?
I know we don't wish to be dogmatic about things like REST, but the power of Alexander's ideas come from the superiority of his specific patterns over other ideas, not just his power to organize the practice of a profession into patterns.