Christopher Alexander: An Introduction for Object-Oriented Designers
patternlanguage.com
patternlanguage.com
We've found a lot of patterns that make things compelling. Next time you watch a movie, set a timer on your watch. At almost exactly ten minutes into the film the exposition and character build up finishes and the plot begins. Why? Because that's been found to be a compelling amount of time to do that. Take more time, and you bore the audience, less and you risk losing them because they don't identify with the characters enough.
But notice I said "most movies". Because it's compelling, it's used often and that makes it unoriginal. Pop songs are all around three minutes and the hook comes at a certain time into the song, every time. So when you make something compelling, you are usually taking away from its originality. If you go for originality, you risk that people won't like it as much.
Genius is when you can make something both unique and compelling. It's rare. And then it's copied to the point where it soon becomes unoriginal. But that moment, when it's beautiful and you've never seen anything like it - that's a wonderful moment.
What Alexander is bringing up is that, while Hollywood films and pop music favors compelling over original, architects usually favor creating something original at the expense of it being compelling. And he is saying that quality without a name is the creation of artifacts that people actually like.
If I remember right, Kent Beck introduced Erich Gamma to Alexander's ideas with the notion of letting users create their experience in software. Erich took it in a different direction obviously, but I can't help but think that Kent had the stronger instinct of what Alexander was trying to say.
We have to remember when creating that startup: It has to be original or everyone will have done it already, but it's additionally about creating something that people will want. And there's the rub. First because that's extraordinarily difficult. But also because the original part is such an extension of your individuality -- it's you -- that then to modify it to adjust it to reflect what to people will like, means taking yourself out of that very personal work.
One thought though: is it really just genius, and the only way to get to that top right compelling and unique state by just jumping there?
Feels like it would be more of a combination: someone tries a unique idea and it's not compelling. Someone else sees this idea, but changes something about it, so it's now unique and slightly compelling. This gets repeated until you end up with something unique and compelling. This new idea now gets copied exactly, and it slowly moves from the top right to the bottom right - compelling but non-unique.
ie. Is this not a continuous process curve innate in all invention? The only real effect might be the speed of movement along the curve - someone (a genius) may be able to iterate across the curve by himself by building on his own ideas?
If this is true, then genius is not really required - all that we require is constant iteration of bad unique ideas until they become something good. Seems to hold up in reality - the more people involved in something, the quicker it gets somewhere good, as there are more iterations happening.
I think there's some important thing to say about the arbitrariness of distinctions like compile-time/run-time, and the architectural implications of "open source," etc, but I don't have time to think it through...
His patterns seek toward satisfaction not pleasure. Pleasure is a byproduct, of respect for the individual"s humanity and the emotional state of satisfaction that results. Pleasure is not a result of response to taste or fashion.
Software, like architecture, usually doesn't have the type of hedonism you're speaking about (as far as I can tell). What people like is what they enjoy using. They often may not even know it, but find they're using it all the time, for example. And that is exactly that quality/satisfaction that you're talking about.
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 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.
Edit - and architects.
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.
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.
"Notes includes presentation of a semiformal algorithmic method that helps automate good partitioning under various assumptions. To use it, one first prepares an exhaustive list of functional and structural constraints...The algorithm takes as input a boolean matrix indicating whether any given pair of constraints interact...The method results in indications of groupings that minimize total requirements interaction and resulting complexity. This statistical clustering algorithm arrives at subsystems by minimizing the interaction of problem requirements that each one deals with."
1. What? — An FP coder
2. Patterns mean "I have run out of language." — Rich Hickey
That's why he called it a pattern language. It's the language of patterns found in life ;)
We just saw a top level post on lessons from Haskell, for example, that while not using the pattern form, expressed a variety of idomatic, reusable solutions to contextualized problems.
Secondly, I think you are misusing the term Cargo Cult. Someone like Doug Lea (the OP) has a lot of experience with good software design - his concurrency libraries, in particular. Suggesting that OOP as a whole is a cargo cult is the height of ignorance.