Global optimization vs. Local optimization

Donald Reinertsen @DReinertsen

Global optimization vs. Local optimization. Compare the promise and reality of centrally planned economies vs. markets. 1/3

Donald Reinertsen @DReinertsen

Global optimization can be exponentially more difficult and is more perishable, only briefly valid 2/3

Donald Reinertsen @DReinertsen

Local optimization should be in the toolkit of any lean thinker. It can be closer to the true spirit of lean than system thinking. 3/3

Al Shalloway. Biz Agility &Value Stream Management @alshalloway

@DReinertsen power of flow is you can do local optimizations w/ confidence you are helping the whole

Al Shalloway. Biz Agility &Value Stream Management @alshalloway

@DReinertsen w/o understanding principles local optimizations can have adverse affect on entire Value stream

Al Shalloway. Biz Agility &Value Stream Management @alshalloway

@DReinertsen hopefully clear i'm in strong agreement

Akbar Zamir @akbarzamir

@DReinertsen What about insights from Op Research, the beer game and such? Local optimisation in that supply chain seems problematic.

Akbar Zamir @akbarzamir

@DReinertsen Or is it rather a question of *which* local property we choose to optimise?

Donald Reinertsen @DReinertsen

@akbarzamir I am in favor of having both tools available and fitting to context. Beer game is simple but illustrates time lags well.

Alina Hsu @alinahsu

@DReinertsen IME global optimization is more a matter of using balanced whole-system metrics than of central planning.

Donald Reinertsen @DReinertsen

@akbarzamir IMO it is smart, but not really global optimization, when you start dropping parameters from your target performance function.

Akbar Zamir @akbarzamir

@DReinertsen I guess I'd expect a greater lag in feedback when monitoring system rather than local parameter.

Niklas Björnerstedt @smalltalk80

.@DReinertsen Global optimization by top down management is not Systems thinking.

Donald Reinertsen @DReinertsen

@alinahsu In world where local optimization is unfashionable we could call choosing a subset of metrics to optimize "global optimization".

Donald Reinertsen @DReinertsen

@smalltalk80 It is probably not really optimized if top down. I see global optimization as state variable, ie path-independent.

Donald Reinertsen @DReinertsen

@akbarzamir A good example of this is how local WIP control loop of kanban can be 100x faster than the more global WIP constraint of TOC.

Niklas Björnerstedt @smalltalk80

.@DReinertsen The point is that systems thinking is about more than a systems optimum. It is also about the right way of achieving it.