Earlier quoted context omitted.
I have been thinking about something similar. I think its usability will drastically improve once you are able to find overlaps and differences in meaning between the perspectives of different people/ideas. I would also suspect that it would be uncommon to not at least fall back on some ingroup definitions, even just for practical reasons. After all, it will likely be rare to learn new vocabulary for individual peopl…
That's interesting. With sufficient data we could probably train a model to come up with an embedding for the definition, and another model to identify which particular passages contribute most to a difference in the embedding (in order to highlight). If you would be interested in providing feedback on early versions of the project, drop me an email. Likewise, if you pursue a project like that, I'd love to hear about…
I am still considering the benefits and possibility of upfront structuring of information, knowledge, certainty and errors but information- and complexity overload is a real pain. Which also means i am keeping a bit of a distance to committing to how the end result should look. Especially as this is one of these projects that needs to be effortlessly integrated into an individuals existing process to be usable. At which point you might as well improve that process while you are at it.
Your approach on the other hand sounds quite practical, lets say combined with some hypertext structure like project xanadu or the hover boxes of lesswrong. The question would be how good the results get. Anyway, good luck, i think this could be quite beneficial.
Projects/Concepts you might (not) find interesting:
* Taiwan has a process called gov0 (gov zero), a communication channel aimed at re-framing proposed laws in a way that allow for reaching consensus. Currently done manually but consensus finding seems to be a key part of communication.
* Cognitive sciences / biases / warfare / behavior therapy. As well as processes aimed at capturing the human error, like crew resource management for pilots. All concerned with errors in our own perception and thinking * Juliane Gallinas TED talks
* Ontology vs epistemology, quite important for people with a CS background in my experience. I like to think of it in terms of complexity reduction in the context of information theory similar to an analog digital converter.
* The 5 level explanation series from WIRED https://www.youtube.com/watch?v=raHM3k-uR0E
Especially the last one is a great example that communication is likely not a matter of finding THE description, but one of connecting your perspective to a topic/concept. Here it works through iterative reduction/increase of complexity, but working from related concepts should also be possible. Lets say one explanation for an engineer and one for a psychologist.
The post itself is also a good example for the the difficulty of communicating (after identifying and isolating) what is relevant and the necessity to cut what isnt. Which will differ depending on ones perspective and goals. I suspect the same effect does occur if you ask people to define the vocabulary they use themselves. Thus the focus on agreeing on the meaningful difference/overlap instead.