Made out of people

Three Dimensions of Temporal Software

My workshop has been struggling with "temporality" for a while. We seem to have converged on a conversation about the notion of what temporality is about, and--as far as the conversation has gone--we want to talk about activity¹. Activity is a little tough to design, but it has something to do with the cluster of documents and ideas that are going on "right now." Some of the members are rather frustrated that when you ask a user "what they are doing", they get the answer "doing email."²

(Conversely, a few workshop think of temporailty as visualizations that show how soon two airplanes will crash if the user doesn't redirect them. Air traffic control seems to be our temporary example of True Temporal Research.)

Anyway. Once we get to activity, we were a little stuck. How big is a task? ("Getting a PhD" is a task; so is "answering the phone." They are not really the same sort of thing.)

Tom Moran points out Tom Malone's work on "Process Handbooks". We set up three axes, all of which get summarized as "top" to "bottom":

• From_Abstract_ ("Move the object") to concrete ("pick up the pen.")
• From Big ("clean desk") to little ("move the pen")
• From Tools and user interface to infrastructure.

This last describes the idea that we are sometimes at odds in deciding whether we're interested in user-interface components for the way that "activity looks", or thinking about underlying infrastructure. To me, the infrastructure is more interesting: once you know something about what it means (to the system, not the user) what a task looks like, a user interface for supporting that should be straighforward.

Conversely, the work that I saw demonstrated (Mary Czerwinski's work at Microsoft, for example) was shown as a series of snapshots of your desktop, or a way of storing a cumulative state of what applications and documents are open. While that is valuable information, it doesn't quite map (in my mind) to a task. For one thing, many of my documents are associated with multiple tasks, and vice versa. I'll need to straighten out quite what the taxonomy is.

For another, "doing email" (for example) is--in some sense--of my only two tasks. It's hierarchial, and it triggers lots of other tasks that need to be addressed ("write this document! find this reference!") which in turn trigger the uses of resources and data.

I'll need to give some thought to how these levels stack together.

[Updated 6/9/04 to correct typo in Mary's name]

¹ I admit I don't completely believe them: I think it loses out on collaboration, on counter-scheduling with others. But this is human-computer interaction, and I am a CSCW person. Collective people aren't quite as visible to this crowd---in these talks, we seem to model others more as "(potential) interruptions" than as users or collaborators or even sources and sinks of data.

² Theirs is much like the frustration of a parent whose kid has been talking on the phone for two hours. "What did you talk about?" "Nothing."