I do wonder though why Mr. Nazer asks the right question by saying "Ultimately we should rethink whether software patents should be allowed at all." A position to which I fully agree, but OTOH the official EFF position at defendinnovation.org is not asking that question and instead promotes a set of policy changes to somehow save software patents but make them less of a problem.
This inconsistency has always puzzled me.
Agreed, PDF is not your perfect solution when it comes to indexing and search. But IMHO that is not what PDF should be used for. PDF is, as its acronym clearly states, a Portable Document Format. So it is of great use to reproduce existing documents or to store canonical versions of a document in a way that is secure and long-term.
Search and indexing etc. on PDF should be done on the metadata level, not on the actual content.
But to store documents that do not change after creation it is a good format. I would propose the distinction between living and dead data. Dead date being stored solely for the purpose of reproduction when needed but without the need to interact with the content. For such "dead" data PDF is a good format.
Living data OTOH - Data that you want to remix, re-use, mesh and re-combine should be stored different.