A recent segment on NPR discussed with New Yorker writer Peter Hessler, who has lived in China for the past 15 years, what it was like to re-enter life in the United States and how United States looks to Chinese citizens. Hessler discussed how hard it is for the rest of the world to understand our complex system of check and balances, of federal, state and local power, of influential groups with non-governmental status. So that raised the question of what governmental websites do to help orient visitors to what the basic organization and framework of government.
What if we were visiting from Mars? What would we learn from our governmental websites about how the United States is organized. The Mars test, in taxonomy and information design, is also called the ‘mental model.’ A mental model uses common knowledge or frameworks for creating website navigation. So a good place to start design a US Government website might be with 4th grade civics, which distinguishes Executive and administration, Legislative, and Judicial Branches and explain responsibilities of federal government and those functions reserved for state government.
Here is the US Government portal called USA.gov. Does it pass the Mars test?
It is a directory like interface that is organized, it seems to me, based on arbitrary topics with no association to government agencies. Where would I even begin to find out about the President of the United States, the new health care bill, the Supreme Court? How do you find a local office of a government office like my legislator’s office or the social security office. In a week where a United States Supreme Court justice retired and volcanic ash disrupted air travel, there is no acknowledgment of these events or links to related website. The site in fact gives an impression that lights are on but nobody is home.
USA.Gov.com is actually experimenting with some sophisticated clustering software such as Vivisimo (vivisimo.com). This clustering application illustrates how clustering results can be customized in this case by topic, by agency and by sources. While the topic clusters are automatically generated on-the-fly, the agency and source filters are generated based on HTML metatags.
The United Kingdom is experimenting with its own clustered interface but the site also uses RDFa and shared metadata. This system has the advantage of having a reusable metadata model that can allow state and local agencies map their content to the governmental model. This promotes “harmonization” and cooperation in supplying data between federal and state government. Because of this harmonization through use of shared metadata, directgov.uk can enable features such as search by zipcode for local offices that deliver state and local services. Even better, the interface looks like someone is minding the store and cares what content appears on the website.
I am not opposed to clustering. Clustering promises to be a great technology to quickly retrieve masses of documents and content, but a little upfront work is needed to filter automated technologies into useful categories that reflect our shared knowledge and common sense. This work would help in creating automated systems that sort results into useful buckets that clarify content and help users find government assistance and solutions.
Search.usa.gov is actually an exciting engine that has clustered over 50 million government documents. However it needs a friendlier, warmer interface to the experience. For example search for Supreme Court, and results mixes state courts with the United States Supreme Court. Wouldn’t search experience be improved if the portal to the search engine helpe users understand and filtered searches to distinguish between by federal and state courts.
Using common models through taxonomies and shared metadata might not only help the visitors from Mars. It might also help citizens of the United States find a clearly navigable path based on stuff they learned in 4th grade.