EMedia Professional - Author, Author, May 1997

author
author

Intelligent Interactivity:
Authoring's Best-Kept Secret

Kathy Kozel

EMedia Professional, May 1997
Copyright © Online Inc.

Although today's average PC has a processor that my "smart" bean-grinder would die for, today's average CD-ROM title uses that power for mindless brute-force media effects.

I have a "smart" toaster. My electronic daytimer knows my handwriting. My hair conditioner "senses" my hair type. So-called "smart" technology is popping up in everything from appliances to after shave. In everything, that is, except multimedia.

Although today's average PC has a processor that my "smart" bean-grinder would die for, today's average CD-ROM title uses that power for mindless brute-force media effects. The typical marketing, training, and even consumer title responds to clicks without the slightest attempt to learn and adapt based on user input. And yet this ability to "think" is precisely what gives interactive multimedia its stunning advantage over incurably comatose media like print and video.

And the good news is, thinking is well within multimedia's current ken. Using off-the-shelf tools including Director, IconAuthor, and mTropolis, authors can build programs that subtly shift in just the right places to help motivate the user to buy, learn, or play some more.

Using any tool that supports variables and logical branching, authors can build a kind of entry-level artificial intelligence. The real question is why this power has been so under-exploited when it may be the one technique that finally gives interactive media the glory it deserves.

ONE SIZE FITS ONE

As an interactive author, imagine the following scenario: You're authoring a marketing CD-ROM for a new luxury model car. You're on a limited development budget, forced to repurpose existing media assets for video and still photos. The toughest news comes from marketing, where research shows diverse, almost conflicting reasons why even your narrowly defined target market might buy this car. Some only care about high performance. For others, safety is number one. One large group cares only about quality and reliability. Then there's the style crowd...

So how do you craft a program that can work for all those interests? A television ad campaign might include a range of commercials, tailored to the demographics of the sponsored show. But you don't have the time or funding to create multiple versions of your program.

The conventional wisdom of multimedia has been that giving the user "choices" in the form of menu buttons takes care of the problem. The user chooses only what he or she is uniquely interested in, right? But we now have enough experience (and research data) to know that this was always a flawed concept. Even if users can be expected to know exactly where their interests are, they can't always relate those interests to the topics on a menu.

So you decide to make your program clever enough to do what a great salesperson would do--adjust to the individual rather than just spit out the same pitch regardless of who asks. You interview a dozen of the best salespeople for this car, and they share a similar formula: ask questions, adjust what you say (including tone, content, vocabulary), and--perhaps most importantly--know when to shut up.

A LITTLE LOGIC

But now that you have a strategy, how do you transfer it into your program? If Fred is your ace salesman, how do you author a virtual "Fred"? Using variables and the if/then logic in Director, you can build a "Fred-on-a-disc" (and one that never asks for a raise or vacation). Your program won't replace the real Fred. It can't begin to match his intelligence, shares none of his common sense, and can't quite manage the whole test-drive thing. But your electronic Fred may help the carbon-based one stay booked with appointments.

The first step is to clarify the key differences that really matter. In other words, what are the different goals or interests that would affect the way an end-user wants, needs, and reacts to the content in this program? Assuming Fred confirms your market research, you decide to go with five categories of user interest: power/ performance, quality/reliability, safety, price/value, and style/image.

Using Director, you declare a global variable at the beginning of your program called userInterest. Now you know that your program needs both to determine this user interest and then use it in some meaningful way. The real Fred, or course, uses a highly sophisticated technique for deducing the prospective customer's interests: he asks.

So you ask. Or rather, your program puts up a screen with the five key priorities across the bottom, and then asks the user to rank them in order of importance, by rearranging the words on the screen and placing them on a grid. The user is immediately involved physically, and is already beginning to think just a little bit more deeply about buying a car. Now you have a crucial piece of intelligence on this customer/user, and you just have to use it to customize the interaction the way Fred would.

Fred knows all about the brakes on this car. But when he talks to Joe Performance Guy he frames the key features and benefits differently from when he talks to John Safety-First Guy. So that's what your program does. You have only one set of stills and video on the brakes, but with an extra hour of art production, you come up with five different headlines for the section on brakes. "America's Safest Brakes," "Engineered for Reliability," "Sixty to Zero in... ," and so on.

When users get to the section on brakes, each one sees and hears the same media assets, but with one very powerful difference--the content is now framed in a way that's most meaningful to each unique user. And Madison Avenue has already proven that the suggestion made by the headline colors the rest of the message, often in a profound way.

You do this, of course, using Director's ability to swap a sprite--an art asset--on the screen, based on a simple script that goes something like,

Set the memberNum of sprite HeadlineSprite = member (userInterest & "HeadlineArt.pict")

Now, that's just with an extra hour of work. With an extra week, you could really up Virtual Fred's IQ. Fred tells you that besides framing the topics differently for different people, he tailors his choice of the topics themselves. "Not everybody even needs to hear about the carburetor," he says, for example. "A benefit for Customer A isn't necessarily a benefit for Customer B." In fact, Fred takes an entirely different path through the topics with each of the five customer categories you defined. Some paths include the same topics but in a different order, while others reflect changes in the topic choices themselves.

Some paths offer similar topics but with a completely different level of explanation, where one might be highly technical or another more practical. You still don't have a budget for shooting new video and stills, but you can certainly define some different paths through the existing content with only a few more lines of scripting, and you can spring for a few extra variations on the text that accompanies each section.

FINDING FRED

Making a smart program is based on authoring your program to do what Fred would do. Of course, finding a Fred isn't always easy, but virtually every domain has experts. The trick in finding a Fred is not to look for content experts, but rather folks who are experts at achieving the goal of your program. Fred might not have the most encyclopedic knowledge of the car, but if he sells the most, he's the one you want. You don't need what he knows about the cars, you want to author what he knows about the people who buy them.

A training program can draw on the expertise of a successful tutor or trainer for how that content should be framed and experienced by different classes of users. A game can draw on user-testing with prototypes or similar games to find out how the challenge level could be adapted at run-time so that the game would provide exactly the right gaming experience for the widest range of users.

It's up to authors to take interactivity to a smarter level. Enhancing interactivity is the next frontier for authors, since they can pursue it with existing tools, while we all wait for bandwidth and processor speed to catch up with our media-delivery dreams. The authoring tool vendors have given us all we need to build smart programs; they've just been wondering when we'd all finally wise up.

Kathy Kozel, an EMedia Professional contributing editor and regular columnist for Author, Author, is a multimedia developer based in Thousand Oaks, California.

Comments? Email us at letters@onlineinc.com.

Home Page

Intelligent Interactivity: Authoring's Best-Kept Secret

ONE SIZE FITS ONE

A LITTLE LOGIC

FINDING FRED

Intelligent Interactivity:
Authoring's Best-Kept Secret