Designing Agentive Technology. Christopher Noessel
Far from being a one-off invention in the form of the thermostat, agentive technologies are appearing all around us, for many long-standing human problems. We can see them if we train ourselves to understand how they’re different.
They are
• Software that persists.
• Watching a data stream (or many) for triggers.
• Performing a task for a user according to their goals and preferences.
They are not
• Tech that assists a user with the performance of a task. That’s assistive tech.
• Conversational “agents,” which are properly thought of as assistants.
• Robots, the software for which is tightly coupled to the hardware. An agent may embody a robot, and a robot may operate as an agent.
• Automation, in which the human is incidental or minimized.
They are different in that
• A valet is the model.
• Design focuses on easy setup and informative touchpoints.
• When it’s working, it’s most often out of sight.
• Touchpoints require conscious attention and consideration.
• The goals of touchpoints are information, course correction, and helping the agent keep on track.
CHAPTER 3
Agentive Tech Can Change the World
They Move Us from Moments to Interests
They’ll Do the Work You’re Not Good At
They’ll Do the Things We’re Unwilling to Do
They’ll Do the Embarrassing Things
. . . and They Will Encourage Discovery Through Drift
They Help Achieve Goals with Minimal Effort
There May Be an Arms Race of Competing Agents
It’s Going to Be Big Enough to Affect Our Infrastructure
Places and Objects Will Need Them
They Will Help Us Overcome Some Human Foibles
Using Them, People Will Program the World
Our Species’ Future May Well Depend on Them
In the first chapter, we walked through the details of one particular example of an agentive technology and deconstructed it bit by bit in the second chapter to better understand what makes this type of tech different. Let’s now look at lots of examples to see what makes them really, really cool.
They Move Us from Moments to Interests
The design of tools focuses very much on the moment of use, as it pertains to some task or goal. That means design attention is given to things like the affordances of the interface, mapping of well-designed controls, and meaningful feedback across many layers of interaction. It’s the see-think-do loop that is the irreducible atom of interaction design.
Much of the benefit of using an agent is that it can persistently look for things the user didn’t even know specifically existed, like a nice shirt, a mention on the web, or a new recording by a favorite artist. For these reasons, setting up a search with an agent isn’t about setting up filters for what’s out there now, but more about what could be out there in the future. It’s about telling the agent what interests you.
Google Alerts: General Interest
To start with an understatement, most people are aware of Google as a search engine. Type “agentive” into its search bar and see the results of web pages, news items, and images on the web now. (And if you’re wondering, at the time of publication, this results in very little, since I’m at the beginning of my quest to rescue the word from obscurity.) But at www.google.com/alerts, you can set up a persistent search where an agent will email you when anything new matching your search terms is published in its news, blog, and web feeds.
Using this, you can set up alerts for almost anything of interest. If it can be found with a basic text Google search, it can be turned into an Alert.
Google even has examples of well-formed Alert searches of possible interest. That gives users an easy opt-in for likely interests, but also shows them examples from which they can learn to construct new ones (even if most are blatant marketing).
Using these tools, you tell the agent what you’re interested in, and it helps you stay on top of it. But interests aren’t just limited to mentions. It can be when favorite recording artists publish new works.
iTunes Follow: Music Interests
If you use iTunes, there are two aspects that are agentive: the smart playlist and the Follow feature.
Regular playlists are dumb collections of songs. (No, no, your taste in music is impeccable. I mean the software logic of this type of list is not smart.) You can edit the list manually, but the list will stay like that until you change it again.
Smart playlists, on the other hand, let you select the features of the song you want in the playlist. Then the playlist acts as an agent when your music collection changes to see if any of the new songs fit the playlist’s definition. If so, Live Updating automatically adds it in.