Monday, March 14, 2016

Analysis, explicable systems, and inexplicable outcomes

A thought.

Root cause analysis only has pertinence to explicable and replicable systems. Root cause analysis depends on the capacity to change individual variables in order to understand the relative impact of those individual variables. Not all outcomes are the result of explicable systems. Many outcomes are the result of a mix of explicable, inexplicable and random systems.

Examples of events arising from non-explicable systems - the popularity of an individual book, movie, song or TV show; the rise and fall in popularity of individuals; the precise weather conditions at a particular location, at a particular time in the distant future; prices; romance; children; many biological systems; many health outcomes from medical or public health policies; etc.

We mix explicable systems (mechanistic systems which we understand and about which we can make precise, accurate and reliable predictions), inexplicable systems (complex systems which we understand in outline but not mechanistically and about which we can not make precise, accurate and reliable forecasts), and random systems (systems about which we have little or no understanding).

Contemporarily inexplicable and random systems, through research and analysis, can yield information that moves them into the explicable systems column. Think about the movement of planets and stars - once inexplicable but now largely explicable.

Inexplicable and random systems are subject to fruitful speculation and conjecture without yet being explicable. Root cause analysis is an invaluable tool with explicable systems. The problem is that it is inherently less reliable when applied to inexplicable and random systems.

You can speculate about how an inexplicable system generates a given outcome and then seemingly apply root cause analysis based on those predicate assumptions and demonstrate that the root cause analysis is consistent with the speculation. But that is essentially a tautological exercise. More precisely, however, all such an exercise does is increase the plausibility of the speculation (if it is indeed consistent with the hypothesis). It cannot, without explicability and replication, actually provide an affirmation. Alternate speculations can simultaneously be equally plausible.

The challenge is to disentangle what type of system with which we are dealing. If it is explicable, then it is possible to speak with authority and confidence.

Most outcomes are the result of inexplicable and random systems. In that case, we can speak with some rationality about plausibility but we cannot speak with authority.

It seems to me that too often, in public discourse, we are blind to these distinctions. Opinions are offered as facts based on a primal/emotional belief in one class of speculation over another.

One question sorts the wheat from the chaff - Is the system reliably explicable? If yes, then great, it is just a matter of making sure the data is correct and the right functions have been performed. Checking the math as it were.

If the answer is no, the system is inexplicable or random, then all conclusions can be usefully parked as speculative.

No comments:

Post a Comment