in which the OPAC was wrongly provided

Perhaps you have heard this expressed already, but I haven't seen it put this way, and it helps me to frame the issue.

Libraries used to do two things:
- librarians talked to patrons
- librarians used the catalogue

then the Internet came along, and they decided patrons should use the catalogue online

MISTAKE

There are two things libraries need to support in the Internet environment
#1 patron access to library services
#2 advanced patron and external system access to library data and services

#1 is belatedly being attempted with layers on top of the catalogue, like Endeca
Personally I would have just said to Google "make me a website that will return relevant results to my patrons from my catalogue data, and link to the appropriate ordering pages", but since libraries have a hate on for Google, that's basically what they're cooking on their own with Lucene.

Anyway, #1 seems to be understandably the current number one library obsession, and I'm not going to say it's trivial, but it certainly appears technologically straightforward. Step A) rescue your data from its catalogue prison - suck it out somewhere that modern technology can be applied to it. Step B) apply modern off-the-shelf search technology to it.

I'm not even sure you need to do Step B. Step B could equally be "let every search engine in the world index it with appropriate links back to my library, and some way for patrons to identify what geographic location or specific libraries they are interested in".

#2 gets more into the Service-Oriented Architecture areas that interest me. I don't think libraries ever really thought much about machine-to-machine communication of data and provision of services. Even Z39.50 works in the library world something like a) librarian formulates search b) search sent by Z39.50 c) librarian analyzes results of search and e.g. uses for copy cataloging. That is to say: humans. Always humans.

So while I think getting data out of the catalogue is an important first step, there is much more to making a full SOA for libraries, at least one that won't be DOA. Think about

* how can you expose your data for use by other people and machines - microformats? REST interfaces? Web Services?
* how can you expose your library services ("request ILL", "please purchase this book", etc.) for similar use by other people and machines?

I think considering the above issues will lead you to conclude that either your library individually, or all libraries in general, need to do a service-oriented analysis of their business functions and data.

Previously:
December 5, 2006 but how does SOA fix my ILS?