Human readability is the first test#
A machine can process a lot of structure that a human reader cannot easily follow. That is not a virtue if the text is meant for public understanding.
If thought is going to be machine-readable, it should still be human-readable. Otherwise the structure may serve indexing or extraction while failing the reader it was written for.
Why balance matters#
Some structures help both humans and machines. Clear headings, labeled sections, explicit boundaries, and visible hierarchy are useful because they reduce ambiguity without hiding the argument.
The failure mode appears when structure becomes optimized only for automated parsing. At that point, the text may look organized but no longer feel like a genuine explanation. It has become a machine surface first and a human surface second.
SROW matters because it keeps the public reading experience central. UKM matters because machine-readable structure still has to belong to a coherent knowledge body rather than a pile of tags.
Practical takeaway#
If a document is machine-readable but awkward for people, the design is incomplete.
The best structures support both.