The reality of books on shelves

We had a useful discussion today around the relative benefits of simplicity and complexity (of the LibraryLocator). It's very tempting to get sucked into the complexity, built up by several generations of librarians, of cataloguing one particular library collection; and then design a data model and a system that reflects this complexity - which will be very hard to adapt to any other library, or to some future change in our own complexity by the next generation.

Alternatively we can generalise (and simplify) as much as possible, whilst checking that any assumptions we make (e.g. "All classmark systems are (mathematically) continuous; and linear") don't break any desired functionality for our own implementation at LSE, and clearly stating these in the "Is this suitable for your library?" section of the re-use guide. (For example, the Bodleian and the British Library both use LMSs that have inherent fetch locations for all items, so may not fit the above assumption.)

The original SHerLoc system used a very simple 'longest possible' match of the desired classmark against a simple Perl hash, so that location data could be maintained at whatever granularity the library staff could manage - maybe just locating an item down to a particular floor or room. It allowed for just two distinct 'collections' (books and journals) because that's how most things were organised then. Can we reflect all our current requirements, just by defining a larger number of collections to cover the different things with different identification schemes?

We set off into the bowels of the LSE Library to check-out two things: How many shelves does the LSE Library have? and How are 'government publications' organised? (apparently these are tricky.)

I'll add some pics from Flickr to a further post.