Iteration
|
The design of organisations and products
|
A key, and much underestimated, quality in our crazy sugar-coated world. Where you are confronted with imperfect, incomplete or conflicting information, variables that are beyond your control (children, animals, opposing ISDA negotiators), unknown unknowns — in short, complex systems — then your decision making process should be iterative.
To iterate is to build a heuristic. It is hypothesise; to guesstimate, to test; to tweak; to rerun. To accept that, since there is imperfect, incomplete information, any decision is to some extent uniformed, but since some action, probabilistically, is better than none — “doing nothing” being no more than a special case of “do something”, it has no inherent logical priority do anything other single action — your best bet is to take as informed a decision as you can, based on what you know, for now, but be ready to re-test that idea and change your action as the situation, and the information you have to hand about it, changes.
That is, you iterate. The decision process is not static, it is not preordained — it is an ongoing dynamic process.
This principle applies whether you are solving new problems, dealing with an unexpected crisis, or building out your system — the end-to-end principle allows maximum iteration. Don’t be wedded to the way you’ve been doing things — I know, I know my little eaglets, it is so hard to let go of the comfort blanket of precedent, but you must — try, and expect things to fail. Don’t commit. Scrub them out and try again.
Some uncomfortable truths:
- Successful iteration is hard. The more practice you have at it, the better you will be. The better you understand the systems and subsystems comprising your environment, the better. Expertise, skill and experience matter. Your itinerant school-leaver from Bucharest might be cheap and fungible, but she won't be good at iterating. Though she will get better.
- Iterating won’t always work. The thing about tail events is they’re hard to predict. On the other hand, an iterative process will almost certainly be more effective than a chatbot. And trying something that doesn’t work still yields you information: it is a falsification: now know that that isn’t the answer. You needn’t fret about what might have been.
Other articulations
- Think of the world in terms of systems, not units — Donella H. Meadows
- Prisoner’s dilemma — the payoffs are totally different if you play an indefinite-round game of prisoner’s dilemma (hence the so-called “iterated prisoner’s dilemma”). But note the impact of convexity, that can turn an iterated game into a single round.