What does HL7 do?

When thinking about what HL7 does, what it is paid for, who gets the value, and who provides the resources, I have come up with the following headings for the things that HL7 does. 
The last two are delivered through meetings that are funded by registration and sponsorship, and through electronic services to members (telcons, website, mailing lists, newsletters etc).
The first three are far more interesting - while HL7 does create a lot of value in these areas, there is no real link between who pays for that value to be created, and who benefits from it.  Only recently has work stared to look at quantifying what the value is, who realises it, and how HL7 can capture a portion of that value for reinvestment.  Something for a future post...

The headings are:

•Publish and distribute standards

•Facilitate standards development

•Maintain a Specification Development Methodology

•Networking for healthcare interoperability and informatics business

•One stop shop for healthcare interoperability issues

Promoting standards

It is clear that writing standards is not enough - they need to be used, and to be easy to use.  That means they have to be easy to find, and easy to digest.  A great example of this is http://www.rfc3881.net/, a healthcare audit information standard.  The documentation is freely available, there is a summary table view that shows you quickly what information items need to be supported, and there is even a facebook link to help promote in social media.

Semantic Interoperability

As a phrase this "Semantic Interoperability" is charmingly incomprehensible and empty. 
The best definition of "Semantic" that I have found is "All that there is in language except syntax" (http://bit.ly/c44hvy).  This is a negative definition -- and so we are little better off than if we had just talked about language.  What we have done is lost the half of the audience who admit that they dont know what it means, and want to live a life without such complex vocabulary.  I suggest that wherever "semantics" is said "language" should be used instead
"Interoperability" really needs to be qualified to be clear what or who are interoperating, and to what end. While you can interoperate without language, you cannot use a language without interoperating (private languages being of doubtful utility).
So - those of us who have been saying that we are interested in "semantic interoperability" would do better to say that we are interested in "language" and be done with it.
Those of us interested in semantic interoperability of healthcare information between IT systems, would be better off saying that we are interested in "machine readable languages for healthcare".

Language Implementation Patterns

Having seen that HL7 specifications look and act very like Domain Specific Languages, it seemed sensible to look at the book "Language Implementation Patterns" by Terence Parr to see whether it could help with the implementation of healthcare interfaces.
After a quick scan things are looking positive - a discussion of language parsing strategies reminds me of discussions within HL7 on schema simplification.
The benefit of talking about languages rather than models is that languages ask to be used, whereas models ask to be understood. I am hopeful that the switch to talking about making useful languages will sidestep some of the rather religious discussions about modeling style.