GDC 2000

Creating Emotional Involvement in Interactive Entertainment

Moderated by Jurie Horneman & Mark Barrett

Attendance
The round-tables were well-visited: the rooms always filled quickly to standing-room only. Many of the attendees were game designers or writers, others were programmers or graphic artists; as many as half of the attendees were involved in the development of multi-player games. In general the attendees were polite and engaged, and made thought-provoking contributions.

Key points
The key points we tried to make were:

Emotional involvement is a key component of any form of entertainment.
Emotional involvement is not being addressed by game developers as a central issue of interactive entertainment.
Creating and sustaining emotional involvement requires developing and harnessing effective techniques. These techniques will be developed by:
1. Adopting effective techniques from passive entertainment. (Linear narratives structures are not techniques: they are themselves the sum of a number of techniques.)
2. Identifying or inventing techniques that spring from interactivity and gameplay.
3. Experimenting with any techniques which might create an emotionally involving experience. (Effectiveness and integration are more important than any particular methodology.)

Day One
We kicked off the first day by asking whether there was anybody who was confused by what we meant by ‘emotional involvement’ or ‘interactive entertainment’. No one raised a hand, so we then asked if there was any disagreement that the industry as a whole needed to pursue emotional involvement much more diligently. Again, no one replied, giving us a clear sense that the audience was with us, and ready to address this pressing issue.

A quick poll of the thirty-five or so participants showed that virtually everyone in attendance considered themselves to be veterans of the industry. The discussions proper began with the straightforward question of just how we should go about creating and sustaining whatever it was that people thought of as emotional involvement.

One of the first points made was that emotional involvement hinges on attachment. This was generally agreed upon, and some terms/techniques from passive forms of entertainment were mentioned as being key, including sympathy and empathy.

One note added toward the end of this thread was the idea of ‘investment’ as an agent of attachment which does not hinge on NPC’s or narrative, but instead comes about as a result of having the player do/build something. Examples included cities in SimCity, frag totals in deathmatch games, and items/spells/levels in RPG’s.

On a related tangent, an old debate about whether games (like chess) do or do not equal emotional involvement sprang up, with supporters on each side of the issue. The crux of the matter seemed to be that a pure game didn’t offer an opportunity for emotional involvement in the same way that an RPG did, which led us to prompt the participants as to whether there was a difference between the emotional involvement that comes from playing a game, and that which comes from watching a movie.

After some discussion the general consensus was that games are about winning and losing, where stories involve the imagination, but both can cause emotional involvement. It was also agreed that the more ways we have to involve the player the better, so neither approach was deemed to be the weaker of the two.

Tying up the idea that investment equals emotional involvement was the point that killing the player destroys both, and there was some question as to how to get around this problem. Planescape: Torment was held up as an example of a game which successfully solved the problem, although there was agreement that the specific solution used had limited applicability. Still, the point was made and agreed to that not killing the player did preserve involvement and investment.

At this point the conversation turned to whether multiplayer games are substantively different from single-player games as regards emotional involvement. Some people felt that the addition of real humans taking on what would be NPC roles in a single-player game gave multiplayer an edge, but there was some doubt as to whether there was that much good role-playing being done by human players. Happily, the discussion moved away from a polarizing referendum on which was the better type of game, and in the end there was again a general agreement that whatever kind of game was being designed, emotional involvement needed to be the focus from the beginning.

This consensus prompted a strong sales-based challenge to the idea of going for something beyond pure gameplay, to which we replied, after some debate, that sales-based arguments are almost impossible to argue around. If something isn’t working and isn’t selling, that only says that it isn’t working and isn’t selling now – not that it will never work or never sell.

(Later, the person making the sales-based argument admitted that perhaps they were a bit disillusioned because they had, in fact, spent the better part of a year trying to infuse a product with emotional involvement, and the experience had turned out badly. The subtext of this point – that failure in a competitive marketplace can discourage good people – should not be overlooked. The key to success may lie as much in protecting the people making the attempt as it does in making the attempt itself.)

For a short time the discussion stalled, which prompted us to remind the participants that although we can all imagine the goal we’re trying to reach, getting there may involve taking the tiniest of steps for the foreseeable future. The question was then put as to whether the participants had experienced even a glimmer of this idea of ‘emotional involvement’ themselves, and we gave as our own example our experiences with the wandering guards in Half-life. After some discussion it was agreed by all who had played the game that when a guard in Half-life came to your aid, and/or you to his, for a moment there was something going on that went beyond gameplay.

We concluded the hour with a short but sharp agreement that consistency is a key to emotional involvement in almost every way, as it determines whether or not the player can logically rely on the game and game world. It was agreed that inconsistency destroys emotional involvement more quickly than any other factor.

Day Two
Again we got through the preliminaries very quickly, and again there was no dissent. Emotional involvement needs to be put front and center in design, and we’re not doing it.

We also got smarter about gathering demographic info, as follows:

Total Participants – 50+

Game Designers – 20+
Artists – 5+
Producers – 15+
Programmers – 10+

(A few people raised their hand for more than one category.)

Very quickly we moved to the question of the extent to which AI will or will not help our cause. It was clear to everyone that the NPC problem was a big issue, and in that light multi-player designs were simply seen as an effective way to get around the inadequacy of NPC AI.

Again, the connection between empathy and sympathy, and NPC’s, was made clear with examples from games like SWAT 3 and Hidden Agenda. For a moment the discussion seemed to falter on the notion that we need narrative structures to create emotional involvement, but at this point we referenced some of the discussion from the first day regarding games like The Sims or Civilization II, in which investment over time takes the place of explicit narrative structures. The point was made that the meta-technique of getting the player to care about something that is at risk includes both narrative and non-narrative techniques aimed at the same result.

The larger point was then made that the quest for emotional involvement will reach fruition only when we have sufficient techniques (also called “tools” by some). It was noted that this is no different than the process of maturation in any other medium, and probably cannot be rushed, which means that the focus for now should be on finding as many effective techniques as possible.

The single-player/multi-player issue from day one rose again, but after some discussion the conclusion was the same: multi-player solves only the language-interaction component of the NPC problem, but inherently does nothing to bolster emotional involvement beyond creating the possibility of greater depth of interaction between characters. While certainly a plus, this advantage would only be realized by those players with sufficient skill to “tell their own stories” – which, it was agreed, most players do not exhibit.

The remainder of the hour was filled with discussions about how specific techniques might or might not advance the cause. In contrast with the first day, the participants (and the moderators) seemed more comfortable getting down to the business of solving practical problems with workable solutions that are available now, and the tenor of these discussions can only be described as extremely positive.

Day Three
Total Participants – 45+

Designers – 40+ (almost everyone raised their hands)
Producers – 15+
Writers – 10+
Programmers – 5+
Artists – 2
Single-player game designers – 50%
Multi-player game designers – 50%

Again we started off the discussion by asking the basics, and again there wasn’t the slightest hint of disagreement that emotional involvement must be pursued. One subject that came up early was the idea that emotional involvement needs to be planned for from the beginning, rather than nailed on at the end. Making the player feel something, and keeping them in that state, requires organic design choices, not cutscenes and canned narratives.

Also mentioned was the idea that we need to focus on expanding the kinds of emotions we’re currently able to generate. Right now we seem to be able to get people to be mad or excited, but bringing players to tears seems much more difficult.

Several people were quick to point out, however, that being able to elicit any emotion is a good start, and one kind of emotion shouldn’t be seen as better than another. It was agreed that we all simply want a bigger emotional palette to choose from.

Following this we noted that in general the emotions that come from competing, winning and losing, are easy to elicit (through gameplay). It’s the emotions which spring from imaginative involvement/suspension of disbelief that seem to be problematic.

At this point the death of the NPC named Floyd in the Infocom text adventure Planetfall was brought up, as a case in which the player really did feel a compelling emotional involvement and response. This led to a discussion of just how “interactive” the game and sequence were, and whether or not something equivalent could be done with algorithms.

This point immediately moved us into questions about AI, and what we can expect to get from AI now, and in the future. Along these lines, Brutus.1 (a program capable of creating stories centered around betrayal) was hailed as a possible advance, but a counter-argument was made that Brutus.1 in fact provided no interactivity for the player. While passable as a story-generating exercise, it was argued that Brutus.1 moves us no closer to being able to provide interactive narratives for the player.

At this point we interjected that one of the key methods for solving the problem of emotional involvement in interactive entertainment is to look at what is portable from passive mediums to interactivity. A case in point was the idea that while Floyd was an entirely scripted NPC, his role in the game/story was equivalent to that of a wingman in a flight sim. Both are characters other than the player, whose roles support the player’s actions. Both can help the player, and sacrifice themselves on the player’s behalf, with the key difference being that a computer can control a wingman’s behavior via algorithms – which is exactly what we’re looking for. (It should be noted that the flight-sim genre is aided by the motivation of the cockpit, which plausibly excludes most language interaction, and all face-to-face communication.)

As the hour wound down, a rather startling question was asked by one participant: How many of the people in the room actually tested for emotional involvement when they were making their games? The deafening silence that followed was a clear indication that we are very much at the dawn of this problem, and have a long way to go. The unanimous resolve of the participants, however, and their zeal to make inroads, left us feeling as if we will make it.

Points of Note

It was suggested that when more complex and subtle emotions are evoked, the term “interactive entertainment” should be preferred over “games”.
AI and multi-player were often suggested as two means of evoking emotions. We argued that although these two aspects can be used to evoke emotions, they are not “magic bullets” that instantly solve the problem of creating emotionally involving interactive entertainment.
There was a fair amount of confusion between “story” and “gameplay” as sources of emotional involvement. This is, in the moderators’ experience, a typical source of confusion when discussing emotional involvement.
Proper definition and introduction of characters was raised multiple times as a technique for evoking emotions that is rarely used in current games.
Floyd from Steve Meretzky’s Planetfall was cited multiple times as the classic example of emotional involvement in interactive entertainment. The key points here are:
1. Floyd’s moment of glory was entirely scripted. In other words, techniques from passive media can be extremely effective.
2. Floyd had been successfully defined and introduced as a character, making the audience care about him.
The security guard aiding the player in Half-Life was mentioned as an example of a strong, but brief, emotional moment, giving an inkling of the potential of emotional involvement that can be created through interactivity.
Consistency was identified multiple times as a key requirement for emotional involvement. Violation of consistency and the subsequent loss of suspension of disbelief in the audience was noted as a classic mistake in current computer games.
Killing the player character and reliance on saved games were also identified as classic mistakes that destroy emotional involvement.
Metagames were identified as a source of emotional involvement. Will Wright told us that players of The Sims had submitted about 3000 Sims stories to the Sims website, and that the players were voting for which story they liked best.
Other sources of emotional involvement that were mentioned were comedy, and nurturing (as in Tamagotchi or Black & White).
It was noted that game developers rely to a great extent on clichés to evoke emotions.
At all three sessions there was a general consensus on the definition and importance of emotional involvement, and on the dire record of the games industry on this topic.

[As a side note, it turned out that the GDC person assisting us at our first session, Mazin Assanie, is a graduate student of John Laird at the University of Michigan. Professor Laird was attending the GDC as well. He and Mazin had a copy of SOAR, one of the world’s most advanced AI systems, integrated with Quake 2 and running on their laptops. Sadly, we didn’t get a chance to see it in action.]

Conclusion
We had a strong feeling that emotional involvement was a topic of great interest to the attendees. In general, there was a large amount of discussion not because of a lack of consensus, but because emotional involvement is a big subject.

Perhaps our perception was biased, but we also sensed a great deal of resonance with other sessions at the conference, notably the tutorials of Noah Falstein and Lee Sheldon, and the lectures of Doug Church, Marc LeBlanc, and David Perry.