Item — HN Reader

I've thought about something similar before in the context of explainable machine learning [0]. The author alludes to it - explicit knowledge is like an expert system. And I would argue that tacit knowledge is closer to a modern neural network that has learned from lots of examples. Asking such a system for an "explanation" - just like trying to encode knowledge in an expert system, inevitable doesn't give a satisfying result because of all the judgment and caveats involved in real live. A better approach is to judge competence based on past experience.

As an additional corollary, the tacit knowledge concept is a good argument against decision frameworks generally, which destroy information by trying to capture experience in a rubric.

[0] Getting more out of experts by focusing on results and not process: observations of people and neural networks http://marble.onl/managing_ml.html

1 comments

This is spot on; I wish I could upvote it twice.

That said, I believe it is possible (and useful) to slowly extract parts of tacit knowledge into decision framework. Kind of "best practice" lists in respective fields. This will help new comers avoid past mistakes and use that time/cognitive capacity to push the boundary of knowledge/skill to uncharted territory. Otherwise we will be stuck in a status-quo.

Take chess. There are advises like don't play such-and-such a move in this position. If you follow the rabbit hole you'll see that hundreds of chess masters had played those lines to weaken their position. And later scores had analysed those games to codify it into "don't play foo in baaz middle game". Of course GMs sometimes go against those advises because they are GMs.

It's almost impossible to advance a field if we all relied on tacit knowledge alone. Because it doesn't scale and we'll spend all our cognitive capacity in repeatedly committing same mistakes.

As an expert in my field I repeatedly force myself to question the source of my judgement when someone asks me "why" when I say "this API doesn't look right". When I do so I indeed realise that it violates some foundational principle and explain it. An expert who can't describe the source of their wisdom, once in a while, end up getting stuck IMO. Because they are not exercising that deductive part of their brain and come to rely more and more on their "wisdom". Over time they fail to adopt to changing conditions and become dogmatic and less relevant.