Understanding Agency

I've been thinking a lot about the concept of agency in interaction design. It's an issue that comes up every time people start talking about the future of computing. The Apple Knowledge Navigator from the late-eighties is a pretty good example of the digital butler concept. It's a long-standing goal, but it's never really taken off. Is this a problem of technology? Even if we could throw maximum computing cycles at the problem, would people really want a computer with that kind of autonomy?

To get at that question, I've been thinking about the assignment of agency in real life. Some things are trivial. For example, most people have no qualms about asking a total stranger to push the button for their floor on an elevator. And we also don't mind taking on that chore when asked. It's an effortless assignment of agency with little obvious risk.

A little way up the scale is the idea of a minor asking someone else to buy cigarettes or beer. This is more socially nuanced. First of all, it's illegal. There are issues of trust involved requiring both parties to take a nominal risk. The money involved and the severity of the consequence are low enough that it's still socially viable in a way that probably doesn't scale up to, say, buying drugs. What's interesting here is that it's still largely ad hoc. You might have an older brother who buys you beer, but that's not actually his job.

Farther up the scale is valet parking. I remember having a problem with handing over my keys at the parking garage in KC. It's a much bigger risk. Regardless of what kind of car you drive, it's a pretty expensive piece of hardware to hand over to a stranger. I still never opt for valet parking unless it's required. It's my car and I don't really believe they can drive it much better than I can.

Finally, there's the concept of service qua service. Here, agency is built into the offering. Think brokerage or realtor. Complex or time-consuming tasks that we're not really qualified to do. For agency to work we require expertise on the part of the agent. We have to trust that they do it better than anyone.

As designers, we've got to solve the problem of instilling and communicating expertise in a convincing way for computer agents to take off; it's a human problem of trust, but the technology isn't nearly there yet. That doesn't mean we abandon the concept of agency, it just means that we refocus our efforts on people, not computers. Services are all about convincing people to abdicate their agency to experts. Figuring out how to do that is what Service Design is all about.

Really great thoughts!

Right now, I can think of a few instances where I have agents working on my behalf...not a cyber butler, but certainly dumb robots who do their jobs. For example, I can't even remember the last time I sat down to write checks for all the bills. I have my bank's website setup to do that for me. I get traffic sms alerts right around the time I usually drive home. And TIVO's always looking out for things it knows I'll like!

I agree that it's a big job for the envisioned future to become reality...the C3PO one anyway. It's certainly trust...and probably also a little bit of suspension of disbelief. When we can begin to anthropomorphisize our agents, then I think we'll be on to something.
Maybe that's the middle ground. Agents that focus on one thing and do it well. Experts really. Maybe the problem with the concept of digital agents is expecting there to be one agent that does it all.

Post a Comment

URL (or E-mail Address)

Your name