Tacit knowledge is more important than deliberate practice

Are you familiar with the Recognition Primed Decision Making Model, or the field of Naturalistic Decision Making? NDM is a field of research that focuses on taking expert intuition (which I refer to as tacit knowledge in this piece), extracting it out of the heads of professionals and then creating training methods to enable less skilled operators to learn these tacit skills on their own. The methods are primarily associated with and deployed in the military.

I think the most famous paper from this tradition is Gary Klein and Daniel Kahneman's A Failure to Disagree, which lays out the conditions necessary for expert intuition: https://www.fs.usda.gov/rmrs/sites/default/files/Kahneman200...

(The two are intellectual rivals, but they put aside their differences to write this piece, over a period of about 6 years, IIRC.)

If you want to get at a definitional debate about the computing applications, that is fine by me. If you want to argue about the philosophical problems with Polanyi's formulation, that is also cool. But understand that I don't particularly care for either debate. I am interested in instrumental outcomes. The discovery of tacit knowledge and the methods found in NDM have been incredibly useful to me in the pursuit of skill acquisition. This was mostly what I was getting at in the post.

So many amazing thoughts in this article that apply to the world of software engineering.

If tacit knowledge is as important as the author makes out, then hiring "young guns" is always going to have a cost over hiring people that have done it before. As an older developer, that's interesting to me. A contrary point, however, is that maybe all businesses are new domains of exploration, so you might as well pay the cheapest person to develop your domain specific tacit knowledge.

If tacit knowledge is not something you can get by reading, then training and mentoring is much better (dare I say vital!) than documenting it all in a wiki, or just making sure everything is done asynchronously in a well written pull request.

If tacit knowledge is that important, then getting it is a privilege. If you are in a position where you can be paid to develop those skills and don't just have to read up in the evenings after your kids have gone to bed. Many people don't have that privilege but the software industry prides itself partially on the idea that you just have to self teach yourself; if you aren't doing that, you won't be successful in your job. Tacit knowledge is not that.

By this definition, chess used to be an art, but is now the domain of science.

I kinda think he's talking about intuition. It's the non-verbal understanding that underpins mastery in any domain. If deliberate practice is the means, then tacit knowledge is the end.

> only knowledge that is tacit for the typical person is the knowledge of how to move muscles.

Thats's not really true.

If you are a native English speaker, it might surprise you to learn that English adjectives have a default order. We normally say "the big red ball" but "the red big ball" sounds decidedly odd unless there's an explicit contrast ("No, not the blue one. I want the red big ball"). This is apparently taught in ESL classes, but neither I nor any of my native-speaker colleagues remember learning it in school.

You have an "intuitive" model of physics in your head that allows you to rapidly determine if piles of objects are stable without grinding through balance-of-forces calculations. It's not exact, and deviates from reality in interesting ways, but it's there. https://www.pnas.org/content/110/45/18327.short

You have all kinds of implicit priors that help you make sense of sensory input. You have some idea of the speed and smoothness with which things move, how an object's apparent size changes as you move towards or away from it and how it might look under different lighting conditions. People share some, but not all of these priors, and they're hard to elicit, hence the endless discussion about The Dress that was either black/blue or white/gold.