hierarchical documentation & reference trees


originally done with filing cards


(because that was all there was)


computers can do the same, now


my suggestion:


flat factbase of notes




queries on tags, backtracking (e.g. PROLOG)


there are many ways to traverse a knowledge base


one query for each traversal


author supplies “starting” queries


users create “extra” queries


extra queries can be contributed to the community


new tags created by queries


tags are just an “optimization” for often used queries (PROLOG rules can build up new tags, then query them, at the expense of compute time)


Arrowgrams + backtracking matcher


one Component for each query