Sometimes simple is not simple enough

Sometime after the sauna last night I found myself thinking about the navigation system that I had cheerfully announced yesterday as finished. And suddenly it seemed obvious that it was very, very wrong.

Conceptually the system looked like this:

        section > topic > category > keywords

which on a typical page might look something like this:

        Writing > Essay > web 2.0 > online, pedagogy

The first problem with this was that the words topic and category were the wrong way round. Grammatically the page should belong to the category “essay” and should be said to deal with the topic “web 2.0”. This was simple to fix, if somewhat tiresome. I had to go back through all the pages I had created and change the block of code that creates the bottom-of-the-page navigation strip.

The second problem was somewhat trickier. Even after correcting the terminology, I realised that, if I had upoladed the essay twenty four hours earlier it might well have been catalogued like this.

        Writing > essay > pedagogy > online, web 2.0

According to my mood, or what was foremost in my mind, any of the keywords could logically have served as the alleged topic, since the essay actually deals with three topics: online developments, pedagogy and Web 2.0.

Topic and keywords were both arbitrary, in that they would be decided on a case-by-case basis and could vary according to who decided them or when they were decided. Two arbitrary levels in the system was one too many. I abandoned the topic part, and was left with a three level system that was genuinely simple (at last).

Topic proved to be an unnecessary addition to the classification system because it complicated the scheme in a way that was guaranteed to cause problems later. At its simplest a search for everything covering the topic “Web 2.0” would not turn up anything that had the keyword “Web 2.0”, or if it did it would be the result of a great deal of superfluous coding.

The final classification system has three layers. Sections correspond to the names of the menus. Categories are the menu items under each menu. These are relatively fixed. They might even be permanent.

I might need a new category under Research if I suddenly decided to start investigating the psychic powers of fish. I might need a new section if I became obsessed with baseball and start collecting details of the players, teams and statistics. (Oh not, please not that. Anything but that…) The chances are I will continue to live happily without these.

It then occurred to me that maybe I should rename the divisions as Menus, Menu items and Keywords. Although this seemed like an attractive idea for a few minutes I eventually decided not to.

The current terms describe a logical division in which sections and categories divide the material and keywords provide links between items that are thematically similar across the divisions. The menu system is just one convenient method of visualising and displaying this logic. It is an interface to the system, not the system itself; and one day I might make a better interface.

Finally I invented one simple rule to govern the creation of keywords. They can be anything except a word that is used for the sections and categories. I then promptly broke this rule by deciding that lists should be flagged by a keyword, but they should also be displayed under the Index menu, so they can be found quickly.

Until the next time I need to break this rule, though, that is absolutely the only time I will ever break it.