So What Is Interactivity?

I have written an entire book on the subject of interactivity (The Art of Interactive Design, No Starch Press, 2002), and I suggest that you consult that book for more material on interactivity. However, I will present here a quickie synopsis of the core concepts.

There is one common experience we all share that is truly, fundamentally, interactive: a conversation. If you take some time to consider carefully the nature of conversations, you'll come to a clearer understanding of interactivity. A conversation, in its simplest form, starts out with two people. I'll call them Joe and Fred. The conversation begins when Joe expresses something to Fred. At this point, the ball is in Fred's corner. He performs three steps in order to hold up his end of the conversation:

1. Fred listens to what Joe has to say. He expends the energy to pay attention to Joe's words. He gathers in all of Joe's words and assembles them into a coherent whole.

2. Fred thinks about what Joe said. He considers, contemplates, and cogitates. The wheels turn as Fred develops his response to Joe's statement.

3. Fred expresses his response back to Joe. He formulates the words and speaks them.

Now the tables are turned; the ball is in Joe's court. Joe must listen to what Fred says; Joe must think about it and develop a reaction; then Joe must express his reaction back to Fred. This process goes back and forth until the participants terminate it. Thus, a conversation is a cyclic process in which each participant in turn listens, thinks, and speaks.

We can generalize the example of the conversation to get a definition of interactivity:

A cyclic process in which two active agents alternately (and metaphorically) listen, think, and speak.

Of course, the task of the game designer is to automate interactivity, to replace one of the participants in the conversation with a machine. We can therefore rephrase the problem of designing interactive entertainment as follows: "How can we program the computer to be an entertaining conversational (metaphorically speaking) partner?"

The overall answer is simple: In order to be a good conversational partner, the computer must perform all three steps in the conversational sequence and it must perform them all well. It must listen well, giving the player the opportunity to say anything relevant to the situation. It must think well, coming up with interesting and relevant reactions to the player's input. And finally, it must speak well, expressing its reaction clearly. It's not good enough for the computer to perform one or two of the steps well, as compensation for performing a third step poorly. All three steps must be performed well in order for the computer to achieve entertaining interaction.

To substantiate this, I need only refer you to your own experience with conversations. How many times have you had a conversation with somebody who could not perform one of the three steps well? For example, have you ever had a conversation in which the other party did not listen to what you were saying? Perhaps this person could think very well, and was quite articulate in expressing his reactions, but if he didn't listen to what you were saying, was the conversation not a waste of time? And how many times have you had a conversation with a person who listened well, but just couldn't think well in other words, a dummy? Don't you find conversations with dolts to be a waste of your time? Or how about the conversations with people who just can't express themselves? They stammer and struggle to articulate their ideas, but they do such a poor job of it that the entire conversation isn't worth the effort.

Thus, in order to have a good conversation, both parties must be able to perform all three steps well. This rule can be generalized to all forms of interaction. Thus, if the computer is going to engage in something like a conversation, then it must perform all three steps. It must listen to the user, think about what he has said, develop an interesting and entertaining reaction to the user's input, and then it must express that reaction back to the user. And it must perform all three steps well.