assumptions about formalizing Math or Physics based on a corpus like arxiv

I'm still reeling from the disconnect between what was proposed in the QED manifesto in 1994 and the current state of formalization of Math (and Physics) research in 2020.

In this post I'll write down what assumptions and what truths I understand about the problem in the context of papers posted to arXiv. In no particular order,

• assumption: There is an intentional meaning for each symbol and each expression. 
• truth: The meaning of each symbol and expression is not stated explicitly in writing.
• assumption: An article is self-consistent in terms of both jargon and math expressions
• truth: The syntax used for jargon and math is not consistent across different papers. 
• truth: There are conventions (often more than one) and which convention is being used is rarely stated explicitly. 
• assumption: Content of a given paper has information contextualized by other references,
• assumption: the context of information is only partially known to the author of a paper. 

The consequence of inconsistent use of conventions and use of implicit associations is that developing a grammar for parsing article text and math will necessarily yield incomplete and incorrect results.