An email to the Sunlight Foundation's OpenGov listserve from Jim Harper of the Cato Institute points to his blog post, chock full of links and information, about the upcoming House conference on legislative transparency. Among the references are a number of quite detailed and on-target recommendations that Cato is making for the drafting process and content.
In particular, Cato proposes a legislative data model that would include a great deal of useful metadata at every level of a bill. The model is quite similar to the California legislative model described at legix.info, parts of which I worked with in the California Law Hackathon. Although the model requires adding a lot of different kinds of metadata to the text, all of that data is easily available when a bill is being written (e.g. the bill sponsor). It is much more difficult to extract by parsing after the fact. As I've discussed in earlier posts, a richly marked-up legislative text would be very valuable, and goes hand-in-hand with the recommendations I make to make the text itself more amenable to automated analysis.