Statebook of the art
The bad thing about the Open Rights Group's new site, Statebook is that it looks so perfectly simple to use that the government may decide it's actually a good idea to implement something very like it. And, unfortunately, that same simplicity may also create the illusion in the minds of the untutored who still populate the ranks of civil servants and politicians that the technology works and is perfectly accurate.
For those who shun social networks and all who sail in her: Statebook's interface is an almost identical copy of that of Facebook. True, on Facebook the applications you click on to add are much more clearly pointless wastes of time, like making lists of movies you've liked to share with your friends or playing Lexulus (the reinvention of the game formerly known as Scrabulous until Hasbrouck got all huffy and had it shut down).
Politicians need to resist the temptation to believe it's as easy as it looks. The interfaces of both the fictional Statebook and the real Facebook look deceptively simple. In fact, although friends tell me how much they like the convenience of being able to share photos with their friends in a convenient single location, and others tell me how much they prefer Facebook's private messaging to email, Facebook is unwieldy and clunky to use, requiring a lot of wait time for pages to load even over a fast broadband connection. Even if it weren't, though, one of the difficulties with systems attempting to put EZ-2-ewes front ends on large and complicated databases is that they deceive users into thinking the underlying tasks are also simple.
A good example would be airline reservations systems. The fact is that underneath the simple searching offered by Expedia or Travelocity lies some extremely complex software; it prices every itinerary rather precisely depending on a host of variables. These may not just the obvious things like the class of cabin, but the time of day, the day of the week, the time of year, the category of flyer, the routing, how far in advance the ticket is being purchased, and the number of available seats left. Only some of this is made explicit; frequent flyers trying to maxmize their miles per dollar despair while trying to dig out arcane details like the class of fare.
In his 1988 book The Design of Everyday Things, Donald Norman wrote about the need to avoid confusing the simplicity or complexity of an interface with the characteristics of the underlying tasks. He also writes about the mental models people create as they attempt to understand the controls that operate a given device. His example is a refrigerator with two compartments and two thermostatic controls. An uninformed user naturally assumes each thermostat controls one compartment, but in his example, one control sets the thermostat and the other directs the proportion of cold air that's sent to each comparment. The user's mental model is wrong and, as a consequence, attempts that user makes to set the temperature will also, most likely, be wrong.
In focusing on the increasing quantity and breadth of data the government is collecting on all of us, we've neglected to think about how this data will be presented to its eventual users. We have warned about the errors that build up in very large databases that are compiled from multiple sources. We have expressed concern about surveillance and about its chilling impact on spontaneous behaviour. And we have pointed out that data is not knowledge; it is very easy to take even accurate data and build a completely false picture of a person's life. Perhaps instead we should be focusing on ensuring that the software used to query these giant databases-in-progress teaches users not to expect too much.
As an everyday example of what I mean, take the automatic line-calling system used in tennis since 2005, Hawkeye. Hawkeye is not perfectly accurate. Its judgements are based on reconstructions that put together the video images and timing data from four or more high-speed video cameras. The system uses the data to calculate the three-dimensional flight of the ball; it incorporates its knowledge of the laws of physics, its model of the tennis court, and its database of the rules of the game in order to judge whether the ball is in or out. Its official margin for error is 3.6mm.
A study by two researchers at Cardiff University disputed that number. But more relevant here, they pointed out that the animated graphics used to show the reconstructed flight of the ball and the circle indicating where it landed on the court surface are misleading because they look to viewers as though they are authoritative. The two researchers, Harry Collins and Robert Evans, proposed that in the interests of public education the graphic should be redesigned to display the margin for error and the level of confidence.
This would be a good approach for database matches, too, especially since the number of false matches and errors will grow with the size of the databases. A real-life Statebook that doesn't reflect the uncertainty factor of each search, each match, and each interpretation next to every hit would indeed be truly dangerous.
Wendy M. Grossman's Web site has an extensive archive of her books, articles, and music, and an archive of all the earlier columns in this series. Readers are welcome to post here, at net.wars home, at her personal blog, or by email to netwars@skeptic.demon.co.uk (but please turn off HTML).