June 20, 2007

Default Options

In April, Swarthmore psychologist Barry Schwartz published an NYT op-ed called "Unnatural Selections," about default options in practically every sphere of life.

People have to make a lot of choices — too many choices. If we considered all the options, we'd overheat, burn, and then die of choice overload. To economize on brainpower and time, we generally go with the default. Schwartz's example: in New Jersey, where no-fault car insurance is the default option, most consumers get that, but in Pennsylvania, which began offering no-fault around the same time, consumers have to choose it explicitly — so most people don't. This goes against their economic interests, since no-fault insurance is a lot cheaper.

The cartoon accompanying Schwartz's article makes the point better than he does. A scowling diner stares in disgust at a restaurant table covered with food -- literally covered, with gravy oozing into the tablecloth, because he "didn't choose the 'plate' option." A leisure-suited salesman shows off a new car with no wheels, because "you didn't choose the 'wheels' option." Etc. etc.

There's a real lesson here. It's related to, but not identical with, ideas like "path dependence" and "configuring the user." All infrastructures create default options; in fact, that's half the point of having an infrastructure at all. Yet the bigger, more widely accepted, and more deeply entrenched an infrastructure becomes, the harder it becomes to make use of other options, the ones that have to be deliberately chosen and configured (if they still exist).

I'm living in Amsterdam this semester, where bicycles are (almost) the default option for short-distance transport: 40 percent of all trips inside the city limits. A standard conversation you can have with Dutch people involves the total dominance of cars in the USA. Like most Europeans, they assume you can go almost anywhere on public transport. They're not idiots; they know that many trips may be slower on trains, buses, and trams than in a car. What they're not prepared for is the "you can't get there from here" answer they'll hear across the USA, at least outside the downtowns of large Northeastern cities and a handful of others. A Dutch colleague took a one-semester job at a US university. To her complete amazement, even for such a short period there was really no way to do without a car. Not only that, she would be expected to pay for the car, insurance, parking, and everything else the four-wheeled beast needs to live. Her employers and colleagues were equally amazed by her consternation, since from their point of view, what else would you do? It's the default option.

In the IT world, this lesson gets learned again and again, a bit better over time, perhaps, but never once and for all. In the 1980s Terry Winograd and Fernando Flores developed a program for distributed workgroups called the Coordinator. Based on a speech-act theory of interaction, the Coordinator attempted to help people complete multi-step interactions. To begin, they had to explicitly categorize each communication as a form of action: make a Request, reply with a Promise, Cancel an earlier request, make a Counter-request, etc. "Request" was the first action in most sequences, so it was the default option, and sure enough, practically every transaction -- including responses to requests -- were soon identified as "Requests," reducing the capacity of the program to channel and track action and turning it basically into glorified email. Another example: Microsoft Word users, probably including you, suffer the plague of Word's "auto-recover" feature, which sounds a lot like the "auto-save" option most programs have -- but isn't. (Word keeps a file of changes to the document, but does not actually save the document itself, so that if you crash and then make the wrong response to the auto-recover dialogue, your changes are gone forever. I recently spent an entire day trying, and failing, to drag some of my work out of Autorecover's jaws.)

This lesson applies to almost anything: privacy (opt-in vs. opt-out), law (the power of precedent), social interaction (default greetings, goodbyes, unspoken codes of behavior surrounding argument), software (default document and data formats), and on and on.

So designers, choose the choices carefully. Whatever users get when they don't make a choice is probably what most people will end up with. And contrary to popular belief, most people don't actually want to make more choices.

No comments: