It doesnt take so much “scale” if one has deficient short term memory/recall/adhd or is (as youve elegantly put it) “too retarded”. (hey - samesies)
Tags/content classifiers/ontologies are I think the solution here, but require continuously grooming your data to ensure it’s classified correctly - a time investment.
My opinion is that modern ML classifiers are helping,l - Ive found some help with tools that recently added auto-tagging - and I think the real magic bullet will be augmenting this capability with relevant personal/activity context. An algorithm can infer much of the contextual relevances that are missing from the current tools if it can match some incoming information to any or all of the areas/topics/projects/horizons/decimal-things that users of organizational tools have decided are important to them.