This is a "publicistic" title, but this problem is related to a fundamental design decision.
The role of votingEdit
If we allow voting, then “the more people agree the more true it is” principle will rule the system. It then will be no different from any existing one. However, only voting system can be applied to the structure design and rule denfinition. On the meta-level for the system, so to speak, where we, as creators of the systems, discuss and agree on the features. Otherwise we'd go recursive :)
So if we agree on the fundamental principles one by one, we can build knowledge together.
I propose the first principle that data Y is better than Z, if it’s more simple (occam’s razor). (has less axioms and more connections, has less entropy) and more proving power. These criteria should be obviously formalized. This is part of work to be done.
The resulting knowledge, therefore, will have these qualities and can be named "better" or "worse" by said criteria.
So data validity (“X is more true than Y if..”) measurement should be one of the foundational design decisions. It then can be constantly calculated, automated, and built-in in the interface. Therefore each addition of data to the system can be measured by “does it increase entropy or decrease”, which is pretty cool.
There should probably be local entropy coefficients and global EC, for example:
Addition of argument A decreases the EC of the subsystem X, but increases the overall Entropy Coef. of the sytem, whereas the counterargument B increases the EC of X, but decreases the global EC, therefore B is better and more valid. Or a local comparison example:
if A and C are contradictory, & argument A decreases the EC of the subsystem X & argument C decreases the EC of subsystem Y subsystem X is larger than subsystem Y therefore argument A is better (more valid) than C
I'm assuming all the parameters here will have measurments (decreases by 5%, better by 23%, contradictory by 20%, smaller by 50% etc)