Excellent summary of the state of affairs!
Do you know the Underlay Project? https://underlay.mit.edu/ Made by one of the creators of the original Google Knowledge Graph (well, they bought his project), they are trying to capture the one thing we can prove (more or less…) is true: what people have said about things. They are trying to create a complete knowledge base of what (notable) people have said, and where, with provenance. The idea would then be to allow different entities to implement a "trust layer" on top of these "facts" in order to help understand things better — in order to provide a more solid ground upon which to build our "constitution of knowledge".
I would love to chat with you about my group's approach, which is a different take on things. Rather than base knowledge upon who we trust, the goal is to build up our knowledge based on the things we believe, and then examine the inconsistencies. Our group is the Canonical Debate Lab, and true to that name, the goal is to create a single canonical debate graph, which includes all the arguments and evidence humanity can muster, in a single well-structured place. We believe that with the proper structure, we can resolve the paradox you have revealed: "this gain is canceled out by today’s increase in complexity."