Making a Better Index

Written by glenn mcdonald Tuesday, 17 August 2010 17:46

Endeca, our Cambridge neighbors, have published a very nice library of design patterns for solving various common problems in data-driven software (and not just in Endeca's own systems). Each pattern is illustrated and explained in useful detail. Ironically, though, the interface to the set of patterns is kind of awkward. You can "browse" them by Industry, Topic or Usage, but there's no complete list of the patterns themselves, so it's hard to tell, without clicking through the entire list of tags taking mental notes, whether there are 100 of them or 10! The individual patterns also don't show the industries or topics or usages for which they are recommended, so it's hard to make sense of the tags.

So here, in one of the smallest Needle projects we've done yet, is Needle's bonus index to this library. We sent Needle on three passes through this small site, following the Industry and Topic and Usage paths individually and then automatically merging together the resulting pattern lists under each tag. This lets us look at the data from any perspective, including the more helpful one from which it is just a simple list of patterns (there turn out to be 15), with all the tags visible for each one. Click on any individual pattern to see a little more detail about it, including the link to its page in the Endeca library itself.

This is hardly rocket science, or even UI science, and no doubt Endeca will fix their UI themselves by 10 minutes after they notice this blog post (hi!). But however long it takes them, the moral is that if your data software requires you to do work to reveal the data's connections to the audience, some of that work probably won't get done, and thus some of those connections will be mysterious. Data software needn't foster mystery, we have plenty of that already. In Needle we try to have the data at least take care of showing you its connections itself, by simple virtue of its inherently connected nature.