Machine shop planning
Posted by aogFriday, 26 December 2003 at 08:42 TrackBack Ping URL

One key point in a probabilistic universe is that it is impossible to know the future, and that because of accumulating error the longer in the future you try to predict, the more off you will be. This has implications for planning. The main one is that there is no perfect plan because the more specific the events it is designed for the more likely it is to be inapplicable. A perfect plan that envisions every detail is guaranteed to be either wrong or too complex to comprehend.

The best plans are ones with some moderate level of detail for the most likely series of events, contigency plans for when things don’t go as planned and actions to try to get events back on the favored track. The company I work for, which is in the high tech industry, never used to plan more than about 18-24 months out. This was because it simply didn’t make sense - in those days no one knew what the industry would be like in two years, so why plan for the unknowable? All that was known for certain was that things would be very different on that time scale, but how exactly was very unclear. The rational position was to not commit to specific plans on that time scale, but to stay flexible so as to be able to respond to the expected unexpected shifts.

This is another way of looking at the dynamist vs. stasist viewpoint or even Hayek’s view on central planning. The problem is as always finding the balance, because while nothing is certain many things are effectively certain (such as tomorrow’s sunrise).

Moreover, one can do planning at a higher level. This seems to be the way the US military is evolving, planning for the unplanned. In this style one plans a toolkit of responses to likely events while providing for resources to supply those responses. The goal is to create a planning / feedback loop that can handle the shift terrain of reality. One plans the structures of response, not the responses themselves.

This kind of “meta-planning” is analog in the social realm of tool building in the physical realm. This is the step of building tools to build tools. I.e., rather than building a complete tool set that can handle anything, one builds a machine shop that can’t do much itself but can build whatever tool turns out to be needed. In a chaotic and changing environment, that’s the right approach.