measurement in complex systems

Further to the previous post, why do we measure? Operationally the main reason we need information to make decisions (indeed to find out if we need to make a decision in the first place). Quantitative information is needed to work out how much of something we need to do.

Now in ordered systems, that's straight-forward. My car is 2 litres low on oil. So I add 2 litres of oil. Job done.

The issue is that many systems are not ordered and these complex systems have 2 annoying properties:

1. Input-output may not be linear. A small change to the system may have a big (even catastrophic) effect.
2. By measuring the system we may actually change its state - e.g. sending out an employee engagement survey may actually raise or lower employee engagement.

So these 2 properties require us to:

1. Measure the system more regularly when we make changes so we can understand whether our impact is greater or smaller than expected. We need to sense it.
2. Use plenty of indirect measures that are less likely to be disturbed by our intervention.

Now I believe that the social media ecosystem is more often a complex environment than it is a simple one. So we need measurement systems that fulfil the 2 criteria above.

A little

Often

And off to one side.

Let me go off and find some examples of what these might be. One example might be the monitoring of tag clouds.