6 ideas
Organizations learn from experience, but experience presents too few cases, too much noise, and too much causal ambiguity to support reliable inference. The history that actually unfolds is one draw from a distribution of possible histories, so the lessons drawn from it are frequently wrong even when the learning process is intelligent.
Low-intellect learning replicates past actions that were followed by good outcomes without understanding why, while high-intellect learning tries to build causal models and stories of how outcomes arise. Both are unreliable: the first reinforces superstition, the second produces plausible but often false narratives.
When actions are followed by good or bad results, organizations infer a causal connection and adjust behavior accordingly, even when the link is coincidental. Because the connection is reinforced regardless of truth, organizations confidently accumulate beliefs that have no basis in the actual mechanics of success.
Any observed history is a single sample from a probability distribution over events that could have happened, with luck heavily shaping which outcomes occurred. Viewing experience this way reveals that successful actors may have simply drawn favorable outcomes, making their practices unreliable guides to imitate.
Organizations encode experience into stories and models that are valued for coherence, dramatic structure, and conformity to prior beliefs rather than for accuracy. These narratives spread and persist because they satisfy psychological and social needs, not because they correctly capture how events were produced.
The pursuit of efficient adaptation to recent experience tends to eliminate slack, exploration, and the deviant beliefs that occasionally produce breakthroughs. Genuine wisdom requires tolerating ambiguity and the comprehension that experience cannot be fully understood, accepting limits rather than maximizing apparent learning.