Conversational interfaces

I mentioned in my comment on yesterday’s post that “the user interface is a negotiation between the designer and user”, an idea which was definitely inspired by reading Dourish, who makes a similar point in saying that “Computation is a medium”. An interface can also be seen as a conversation, as Suchman describes. So the real question is, what is your interface communicating to the user?

There are a variety of ways to approach this question. One way is to look at it purely from a user perspective, which is the approach Don Norman takes in focusing on affordances – what does the user perceive as ways to interact with this interface? The designer’s intent is lost, and the interface has to stand on its own. This only embraces the surface level of the interface, though, and I’d like to probe deeper.

I’m going to take a step back here to introduce Garfinkel’s concept of “accountability”, which I read about in Dourish. Accountability in this case has to do with an entity being able to account for its behavior, in that it is possible for others to discover its tendencies by observation. Dourish points out that the use of abstraction as a building block in computers has destroyed the accountability of software. He uses the example of networked file systems on an operating system. Because the operating system treats local folders and networked folders as equivalent, both conforming to the abstraction of a “file folder”, it doesn’t warn the user when they start copying a file to the networked folder, even though a network copy will take longer and is more prone to failure. If the user could look inside the folder and discover that it is implemented differently, then the user could make a more informed choice. As he writes, “The features that matter to me as a user are ones that have been hidden by the interface and by the abstraction that it supports.” (p.84) The interface is not accountable; it is not possible to discover through observation what is going on.

How does this apply to the idea of conversational interfaces? An interface which is accountable is open to a deeper exploration than one which is restricted to the top abstraction layer. The user can query it to find out details as needed, much like we query others when they make a conversational reference with which we are unfamiliar. An un-accountable interface, like the file folders, is restricted to the most shallow of interactions, because it can not reveal what is going on underneath. It’s the equivalent of a conversation with a monomaniac, who keeps on repeating the same thing over and over again, regardless of what you ask them. This inability to reframe what it says in response to our query is part of what makes technology alien to us. Humans will try all sorts of different ways to communicate, to get their point across. Technology often only knows one.

So a conversational interface must be accountable in that its internal behavior must be observable and discoverable by the user. Let’s extend the concept of a conversational interface by saying that it should be able to be experienced at multiple levels: a surface level where the user is just getting acquainted (equivalent to sharing nods with your new neighbors in the apartment building), a deeper level where the user starts to figure out some of the more expert details (acquaintances who you share a meal with occasionally), and even deeper levels where the user ascends to guru-hood with the system (equivalent to your closest friends). The user can “converse” with the interface at multiple levels, depending on their needs for the interface.

We can extend this concept further, by introducing the notion of charm. I really liked this quote by Kunstler in his book Home from Nowhere: “charm is the quality of inviting us to participate in another pattern, for instance, to glimpse the pattern of another personality through the veil of manners, customs, pretence.” A charming interface is one that not only can be experienced on multiple levels like the conversational interface, but actively invites the user to explore the deeper levels. Not all users will take the challenge, of course, but none should be so intimidated as to give up because it’s too hard.

I think really great interfaces possess this quality of charm, of invitation. Computer games are a good example of this. They start off with easy levels, building up one’s mastery of the controls until the controls themselves fade to “ready-to-hand” status. Then they continually challenge the user with goals that are just out of reach, inviting the user to explore the world and the interface more deeply. Computer games can be tremendously addictive due to this nature; there’s always one more challenge, one level deeper to go. But it provides a good model for what a charming interface would be.

I’ve wandered a bit afield from my original question: What is your interface communicating to the user? But I think the concepts of accountability, conversation and charm provide a framework for analyzing such communication. An accountable user interface answers the user’s questions as to whether the interface fits the user’s needs. A conversational interface is one that can be used at several different levels, depending on the comfort of the user with the interface. A charming interface invites the user to interact with it, to explore its capabilities, to ask it questions. Each of these lets us go beyond the surface level of affordances, and start to have real conversations with the designer via the medium of the interface.

I think this concept of conversational interfaces is interesting because it frames the question of interface design at multiple levels. How does the interface invite a novice user to try things out without fear of breaking anything? How does the interface indicate to the user what is going on underneath to help them make better choices? How does the interface encourage the user to find more expert methods within the interface? These are all good questions to ask when designing an interface.

I think I’m going to leave it at that for the evening. There’s definitely more to be explored here. Some loose ends I had jotted down from a previous attempt at this article include looking at the Internet and open source software as an example of how open and transparent interfaces lead to diversity and innovation, Jane Jacobs coining of the term “border vacuums” to illustrate how closed, opaque borders shut off diversity, and giving Don Norman more of a fair shake in looking at how affordances can invite interaction. Some neat stuff here. And it’s good that it’s letting me introduce concepts from the Dourish one at a time, because there’s a lot packed into that book and my review’s going to be long enough as it is.