On writing systems for collective intelligence

Note-taking systems (wikis, Zettelkasten) suit well for personal research and reasoning/sensemaking in writing but not so much for memoising stuff, as I observed here: Evergreen notes and Zettelkasten are good for "frontier thinking", but not for learning a practice or a domain, and neither for collective reasoning. This is an empirical statement Andy Matuschak observed this here: “Collaborative knowledge-management systems tend to decay”.

Matuschak’s arguments of why wikis are not working well for developing collective knowledge (difficulty in maintaining atomicity and concept-orientation of notes) seem to me somewhat made up, though. In A Thousand Brains, Jeff Hawkins explains that Humans learn only within reference frames. Human brains put concepts into their own, idiosyncratic reference frames. Collective interpersonal intelligence clearly works differently (though unclear, how exactly), so people don’t feel they can “fit” their notes into a system of notes representing concepts from someone else’s brain.

Therefore, it seems there is still no better way for developing ideas collectively than good old written conversation. The forms may vary from a web forum to a blog of essays (or RFCs, in an engineering setting). Other participants comment on these posts or write their own posts in response. A small insight here is that it’s probably better for a participant to rewrite an RFC completely and publish it separately instead of (or in addition to) commenting on an existing RFC if they disagree with it on a lot of points.

Here’re two posts testifying that this mode of collective sense-making is successful at tech companies: 1, 2.

When a group of people converges on some model (theory, design, explanation, ontology, etc.) it should be extracted from the conversation as a set of architecture documents + code + documentation (in case of engineering projects), or, in case of scientific or philosophical research projects, as a wiki (finally!), or a curriculum, or a course, or a book.

Ward Cunningham, the inventor of the original wiki, created Federated Wiki 10 years ago, a system that could work basically like a personal wiki for people who can copy each other’s notes and modify them. Over time though, people may converge on using certain versions of notes, which will become the “reference ontology” through consensus. Federated Wiki didn’t take off and it seems to add unnecessary complexity (compared to the approach suggested above) for small, relatively cohesive groups of researchers.

Productivity tools such as Notion, Coda, or Confluence are excellent for documenting models (of collective action, or engineered systems) according to existing meta-ontologies, but not for developing new ontologies.

I don’t know whether they work well for collective meta-research though because I participated in such projects.

Developers of Roam Research and numerous similar note-taking systems apparently disagree with my argument laid out above, because they declare that their systems make “collective thought” possible.