Architecture not application

I’ve been thinking and reading around the subjects of architecture and design recently. I have been arguing for some time in the context of software that we need to focus on broad architectures into which objects (people and software) and be placed, evolve, mutate and form time to time die. I’m increasingly starting to take that idea across into wider organisational design. It fits in with the complexity theme of partial and mutable constraints. By coincidence I got a linked in request from one of the best designers I have worked with in my life (ands its been a lot) Tracee Wolf who is still with IBM. Her web site has some great examples of her work, including some projects I was involved in. I also got a request from a good friend who’s daughter is looking for a placement after three years at the Mackintosh School of Architecture in Glasgow. An example of her work illustrates this post, its all about the manipulation of light in a small area (Incidentally if anyone knows of a good placement opportunity let me know).

Interestingly architects often make the best user interface designers. It’s that capacity to represent three dimensions of space, together with light, time and other aesthetic properties into two dimensions with a deceptive simplicity of form. As in software, so in organisations. Creating a loose structure that can coevolve with the needs of your people and markets creates a resilient capacity to cope with change and tolerance of failure. RIgid structures in which all function (for which read application) is determined are fine within the context of their creation. However when the context shifts their robustness becomes brittle and they fail catastrophically. Part of the evolutionary success of the human race (well OK the jury is out on that for the moment but stay with me here) is that we were never optimising for any particular environment, and while we lived within constrains we learnt how to modify those constrains by interaction with them. We co-evolved, but not at random, we did it within constraints.