Near the end of his presentation at the FedWeb
conference (being held in San Diego this week), Stuart Shulman
, an assistant professor at the University of Pittsburgh's School of Information Sciences, remarked on the potential usefulness of collaborative filtering technologies for e-rulemaking.
You know about collaborative filtering, though you probably don't know you do. You know about Amazon.com, which serves as the best example of the concept. There, you can read reviews of a product you're interested in purchasing. Most of these reviews are actually written by other visitors
. Anyone with an Amazon account can add a review. Morover, users also rate how useful these reviews are (this is the why you see '2 of 5 people found the following review helpful' in small type before each review). Amazon itself does not need to hire scads of book or music critics, instead the public does the work. The Web site presumably pushes those reviews considered most helpful towards the top of the page.
The computer news site Slashdot.org
also uses this user-based technique to great effect. Slashdot editors post links to stories, along with summaries, which are then commented on by hundreds of readers. A revolving group of volunteer moderators read comments and rate them from one to five, five being the most useful, informative or humorous. Users can set up their pages so that the highest ranking comments are placed at the top of the page, while lower-rated ones get pushed to the bottom or not even displayed at all. The Slashdot folks have made basic code for this site, called Slashcode
, available as open source.
In fact, look around the Web and you will find a whole passel of collaborative filtering sites. Digg
allows visitors to post links to other sites that may be of interest, and other users rate the value of such sites. With Flickr
, people upload, share and judge digital photos. Wikipedia
extends the collaborative model to creating an entire encyclopedia.
Using the same basic techniques of participating and user moderation, the nextgeneration of e-rulemaking software could offer a way to sort out the most meaningful comments, bringing them to the attention of the rule drafters.
Public participation in government rulemaking has always been more than simply giving a proposed rule a thumb's up or thumb's down, Shulman said. The best rulemaking is a deliberative process, where the policy drafters modify the proposed rules after taking in new perspectives from thoughtful opinions.
Unfortunately, the first round of e-rulemaking technologies has not helped encourage this sort of thoughtful deliberation. GCN has covered these shortcomings in the past
, as they were described by Shulman and others. Many agencies have started accepting comments by e-mail in the past few years, only to be flooded by thousands of identical form letters dispatched from advocacy group sites. While the sheer number of missives certainly identify the public agreement or displeasure with a certain proposed rule, thoughtful and potentially useful replies get lost in virtual pile of form letters.
With collaborative filtering, 'You could still have a 100,000 people engage in a process but have it be hierarchical, so they are automatically broken up into groups into 5 or 10,' Shulman said. 'Have them talk to each other in their little groups and have them put forward their best ideas. Create a system where the best ideas rise to the top, and gradually you will have a core set of comments that reflect deliberation across 100,000 people.'
While we suspect interested parties may still game the system, this approach may be an improvement over e-rulemaking responses today, which can be, as Shulman said, 'just 100,000 ways of saying the same thing.'Posted by Joab Jackson
Posted by Brad Grimes, Joab Jackson on May 23, 2006 at 9:39 AM