"But needless complexity has made our systems needlessly complex, and therefore needlessly expensive and difficult."

When you say "systems," do you mean mostly back-end processing systems, or the catalog itself. On the catalog level, whatever the complexity, so many of these fields aren't shown at all—pure waste. Take the 008 boolean for Festschrift. I think it's kind of cool, but is there a single OPAC that displays it? Other times, a number of fields get at a question, like whether something is a DVD, without ever solving the question, and nothing is really displayable.

I question, however, the need to think parsimoniously. Libraries look down on Amazon and such for bad data, but the quality is improving and the complexity of the data now, I think, exceeds that of libraries. Anyway, data qua data doesn't make things expensive. WorldCat fits on an iPod these days. And if vendors says the quantity or complexity of MARC makes their systems expensive, I think they're either bad programmers or lying.

What costs is paying expensive people to enter the data, often multiple times and in a very tightly-defined workflow that prevents true collaborative work. There are other cures to that, like drawing data from underused sources (eg., ONIX—I gather OCLC does that now?) new types of sources, like web search engines, access and use statistics, Freebase, LibraryThing and so forth, and changing the workflow so that edits are easier and can be done in a more relaxed, distributed manner. Every time someone writes to AUTOCAT a request for someone with higher access to change a WorldCat record, something very wrong has happened, and its not about the complexity of the metadata. I gather OCLC recognizes this too, and is working to open up access.