"Focus on creating a simple, reliable, and publicly accessible infrastructure that 'exposes' the underlying data."
"Require that federal websites themselves use the same open systems for accessing the underlying data as they make available to the public at large."
This would be analogous to requiring U.S. State Department officials to get their official government passports through the same process that regular citizens use to get passports. In theory, it would require the State Department to improve its passport process in general.
According to Robinson et al., their plan will spur a web renaissance in the private and non-profit sectors that will make it unnecessary for the government to do anything but publish the raw data online. The marketplace of ideas will take care of the rest and create all the necessary applications of government data.
It is a good idea in theory, but there are at least two major barriers to their vision: privacy and cost.
Most agencies with really useful data have strong privacy restrictions on the underlying data. For example, the IRS would never intentionally expose your personal tax records to public eyes. Probably all of the statistical agencies face similar restrictions on at least some of their data; many agencies are entrusted with either proprietary business data or some kind of personal information.
Not surprisingly, reworking every government web application would also be very expensive.
Thus, at a minimum, I would recommend the following revisions to the authors' proposal:
The same framework of separating data from interaction can still work, but security and privacy controls must be in place for protected data so only authorized applications (like the agency's website) can access the protected data.
A measured approach to implementation is necessary; start with pilot projects then slowly phase in compliance requirements over multiple years.
Perhaps the U.S. Patent and Trademark Office would provide a good pilot project. They already have some experience with Web Services, and patents are a great source of data for third-party applications.
This means that we the citizen get to pay for the original government information through our tax dollars and then to pay again though a commercial site. It means that even more people will be coming to public libraries expecting to get free government information on the internet with the libraries picking up the cost.
Great post. I'm in agreement that the paper poses some good ideas, but I think it goes too far in cutting out the role of government Web managers. I posted my thoughts here and would love your feedback.
I think people tend to read too much into the recommendation. As I understand it, the idea is that if information is going to be displayed on a website, then that information should come from the same repository of data the public has access to. If there are privacy concerns, etc., then it probably won't be on the website in the first place. The authors weren't oblivious to these concerns --- I don't think they saw themselves proposing absolute government-wide policy. There are plenty of cases where these concerns simply don't come up in the first place, and we can use this standard there.
-Josh (GovTrack.us and, full disclosure, friend of the first author)