(Warning: this is rather incomplete and more abstract than usual)
In logical reasoning we have the Premise, the Argument, and the Conclusion. In order to arrive at the correct conclusion, we need both the right premise as well as the right argument. Arguing vehemently from faulty premises will get us nowhere, and having the right premise without argument will not take us to the conclusion either.
Here I want to map the following notions to decision making under uncertainty:
- Premise = Exposure or Tinkering
- Argument = Execution
- Conclusion = Outcome
In uncertain domains, one can’t force a desired outcome (conclusion). No amount of execution (argument) is sufficient to guarantee an outcome. It takes the right exposure (premise) as well as the right execution (argument) to arrive at a desired outcome (conclusion).
In highly uncertain domains where power-laws rule, the exposure (premise) becomes more important than the execution (argument) simply because of the nature of the power-laws (rare hits). If one spends an inordinate amount of fixed resources on execution, it is easy to miss an outcome simply because of the scarcity of the right premises. So it becomes important to focus on searching for the right premise. So resources need to go into maximizing exposure. Which is what Taleb calls “tinkering.”
An efficient and lightweight execution framework that will take a premise, quickly generate an argument, and spit out a conclusion (good or bad) is ideal - but there aren’t that many. It needs to be built.
It is necessary (but not sufficient) to first build an efficient execution framework before beginning to search. Tinkering (exposure, premise) alone is not sufficient. And I do see a paucity of such execution frameworks. Good generalized execution frameworks don’t exist as is and are not easy to build. It is highly inefficient to tinker without building an execution framework first, because tinkering requires the ability to fail quickly and test quickly.
This means one can only tinker in domains where one has access to an execution framework. One can’t run around building an execution framework every time a good premise comes along. If you are a scientist with exposure to lab equipment but with zero access to hollywood executives, it makes sense to tinker within the lab environment even if you think you can churn out a large number of good screenplays.
So now we’ve kind of rocked back to saying that first identifying or building an execution framework is important. Tinkering follows only after that.
To sum up: Tinker (buy exposure) within an execution framework to which you have access. If you don’t have access to one, it may be useful to build one before you start stochastic tinkering.
On a side note, one great execution framework we are all born with is the human body. All it needs is the exposure to the right set of premises and the outcomes will follow without any need to force anything.