[ This is so 2015, but I feel like I have to clarify and narrow these claims, even though they would strengthen my personal stance. A narrower claim,  that the ability to think and talk about claimed facts about things in the bibliographic universe, rather than exchanging machine readable  bibliographic records,  makes it easier to think and talk about the relationship between  those things, to make additional statements about them, and to set up procedures and infrastructures  for cooperative work where incorrect assertions can be corrected ] 

On Feb 2, 2017 10:11 AM, "Martynas Jusevičius" <[log in to unmask]> wrote:
"Errant standards"? Do you even know what Linked Data and RDF is?

Knight Errant:  a knight traveling in search of adventures in which to exhibit military skill, prowess, and generosity (–errant

Standard Errant:  a standard doing likewise? 

At the core, RDF provides a generic, domain-agnostic, Web-native data model with a zero-cost merge. It is future-proof: you are guaranteed to be able to convert, interlink and merge your (bibliographic) data with any ther RDF dataset, and publish it on the web.  Neither MARC nor RDBMS nor XML have these features built-in.

1) The RDF model is roughly equivalent to a relational database model where all tables a restricted to two columns. The model has several quirks caused by early infection with XML (curse you, language tags). RDF  graphs are simply sets of triples, and the metaphor of graphs is not essential to the interpretation. 

2) Merging two RDF graphs may not be zero-cost, formally, because blank nodes. But more importantly,  merging  two RDF graphs may not be zero-cost informally, because they may use different vocabularies, or the vocabularies may have been improperly applied.  
Because of the choice of logics used for the standardized semantic web, the results of merging a second graph aren't allowed to override a conclusion derivable from the first graph.