“The successful development of science requires a proper balance to be maintained between the method of building up from observations and the method of deducing by pure reasoning from speculative assumptions.” — Paul Dirac, quoted in The Strangest Man by Graham Farmelo
There are two ways of arriving at an understanding of something. The first is to observe carefully, accumulate evidence, and let patterns emerge. The second is to start with a principle or hypothesis and reason forward from it, testing the conclusions against reality afterwards. Science, at its most productive, moves between both. Neither works well alone.
Pure observation without theory produces catalogues, not understanding. You can describe everything you see and still not know what causes what, or what to look for next. The data does not interpret itself.
Pure reasoning without observation produces elegant systems that may have no contact with the world they claim to describe. The history of ideas is full of internally consistent frameworks that turned out to be wrong about reality while remaining impeccable on their own terms.
Dirac’s point, made in the context of quantum mechanics, applies beyond physics. The productive position is not to choose between observation and reasoning but to know which the situation calls for, and to move between them without losing hold of either.
In organisations, the failure modes are recognisable. The institution that only trusts what it has measured cannot generate novel responses to novel conditions. It is always one step behind, because the data it is using describes the world as it was. The institution that reasons entirely from first principles, applying a framework without close attention to what is actually happening, produces recommendations that are coherent but wrong. Both are common. Both look, from inside, like rigour.
The engineering posture tends toward the second failure: it assumes that sufficient reasoning from known principles will produce the correct answer, and that execution is mainly a matter of applying it correctly. This works in closed systems where the variables are known. It fails when conditions are changing faster than the framework can account for.
The combination of modes is harder in practice than in principle, because each has its own culture, its own standards of evidence, and its own language. People trained in one tend to be sceptical of the other. The instinct is to resolve the tension by picking a side. The actual requirement is to hold both without collapsing one into the other.
Related
- Mixing art and science — what genuine synthesis of different epistemic modes requires in practice
- The NASA mindset and its limits — the engineering posture as deductive reasoning applied beyond its appropriate range
- What certainty proves — what happens when deductive reasoning becomes its own end, and stops pointing outward
- Informed intervention — the practice that emerges from holding both modes: understanding a system well enough to intervene, then stepping back
- The Right Question — the operational research tradition as a historical case of this disposition applied to institutional problems