How do we want to make our games? Do we want to be challenging the player? Empowering them to explore? Providing them with toys to play with? These decisions can make a vast difference to how a game plays, and what the focus of play will be. Much depends on what proportion of the workload of control is assigned to the avatar, and how much lands at the feet of the player. How this issue is resolved determines the interaction model of the game.
Perhaps there is an unspoken assumption among the so-called Hardcore players, those for whom gaming is a lifestyle choice, that the player will shoulder the entire burden of control. This would in part explain why so many games require the player to master myriad control mechanisms, which often rival in complexity the most esoteric machines in our modern culture.
When the difficulty of acting falls on the shoulders of the player, the model of interaction might be considered the empty vessel. The avatar does little or nothing without the player expressly triggering it with the correct control combination.
There is no doubt that some players enjoy mastering the complex control systems that result from the empty vessel model. Some enjoy the payoff of fiero achieved by overcoming difficulty; some feel a sense of satisfaction from achieving mastery of systems. There is also little doubt that when taken in the context of a wider audience, these desires represent a minority viewpoint. For every player who would like to get to grips with complex controls, there are a dozen others who would prefer to be playing rather than learning controls.
The comparative success of the First Person Shooter genre (in the West, at least - the Japanese and Asian market remains comfortably resistant) creates the illusion that we can get away with complex control mechanisms in games, and to a certain extent this is fair. After all, Goldeneye (1997) and Half-Life (1998) sold some 8 million units each we're told, which is an impressive audience. However, notice that both games sold these numbers almost ten years ago - and in the case of Goldeneye, the game featured a much simpler basic control mechanism than its decendents.
FPS games are a very successful niche market. But make no mistake, they are not a mass market genre in any normal sense of the phrase, typically pulling in no more than 4 million players, and more commonly bombing completely and making a loss for their developers and publishers. The players who enjoy them take from game to game legacy skills that make each future FPS trivial to learn to play - provided the new games don't add too many bells and whistles. Put such a game in the hands of a non-gamer, however, and they usually cannot manage to turn or move at all.
At the other end of the scale are genres such as point and click adventures (which survive as a much smaller niche market, with typical maximum sales closer to perhaps a million units, and more commonly under 100,000 units). In these games, the control elements are minimised - practically eliminated in some cases. The player just has to point the avatar to a particular place, and the avatar takes actions automatically. However, the point and click adventure also eliminates something else: it tends to eliminate play. Not entirely - puzzle solving and narrative play remains - but the capacity to pick up and play in a direct fashion has been lost. The mass market audience appears to be strongly engaged by games in which they can be amused by what they can make their avatar do immediately. Adventures are more like interactive novels: they require a certain commitment from the player before they pay-off; gratification is delayed.
The interaction model at work in the adventure game genre is grossly simplified, and could be considered the distant voice model. The player does not so much take control of the avatar as act as a voice in the avatar's head, guiding the next step. It is a very strategic mode of control, less concerned with the nuts and bolts of moving and acting and more concerned with high level progress. Indeed, the adventure game is largely about guiding the story forward. (Note that this interaction model applies to both text-driven adventures, and point and clicks).
Both these genres of games - FPS's and adventure games - largely avoid a problem which lies at the heart of modern games: camera control in a 3D world. The player needs to see what the avatar is doing, and this requires the camera to behave in an intelligent fashion (for interaction models approaching the distant voice) or for the player to be put in control of the camera (for models approaching the empty vessel).
This is a serious issue. To reach a mass market audience, we require simpler controls than that implied by the empty vessel model.Put the player in charge of controlling the camera, and you increase the burden of control on the player. Put the game in charge of controlling the camera, and you are likely at some point to frustrate the player with a poor view. Unless, of course, you spend a fortune creating custom camera systems from level to level, which certain (vastly expensive) games such as God of War choose as their approach. Since I do not want to see game development slow to a trickle of sequels, I'm not keen on this as a solution.
Sadly, I do not have the answers for the camera dilemma. There are many options - designing games to use views which eliminate camera problems (top down and side on views), game designs and level designs which are adapted to a specific camera system (using only cavernous interior spaces, for instance), transparency effects and cutout views to preserve the players line of sight. But it is already clear that leaving the player as the sole camera operator (as generally happens when the empty vessel model applies) is insufficient for reaching a wide audience.
Let us pretend, however, for the purposes of considering possible interaction models, that we have a perfect, reliable and inexpensive camera solution. We are now in a position to consider an interaction model which lies between the complexity of the empty vessel model, and the extreme simplicity of the distant voice model.
What if the avatar, rather than being an empty vessel, requiring every action to be initiated by the player, is quite competent. What if the avatar automatically works out what to do in any given situation? She reaches the edge of a ledge, but there is another one ahead within jumping distance, and it will definitely be possible to jump back when across: merely pushing ahead in this instance will cause her to jump to the other edge, because this is the obvious action to take when you are guiding someone to continue forward in this situation. Similarly, imagine a situation where the avatar is being fired upon by a foe; you run her towards a low wall which can provide cover. When she arrives, she automatically takes cover, because this is the sensible action to take in this situation. If you keep the stick moving, she will run along behind the cover.
This is what I have whimsically termed the blind captain interaction model. ('Captain' to imply the competence of the avatar; 'blind' to imply the player's role as the avatar's guiding eyes). Partly, this model depends upon the use of intelligently defined context-sensitive transit mechanisms, which are becoming very common in games. For example, in Grand Theft Auto: San Andreas, if you jump near a wall, the avatar will automatically grab hold of the wall. Push up, and he will climb on top. This is the essence of a context-sensitive transit mechanism. In the blind captain model, the avatar would automatically scale the wall when reaching it, though - after all, why would the avatar run on the spot standing next to a wall? This is only part of what is required for the blind captain model, however. The other side of the coin is defining responses to situations according to what would make sense for the avatar, rather than relying on the player to specify the interaction.
The essence of the blind captain model is that the avatar has a simple (rule-based) artificial intelligence which determines its behaviour, and the player is freed to guide the avatar, rather than to initiate all actions with specific control combinations. Two way communication between the avatar and the player is probably required. If the avatar can see a course of action that might not be obvious to the player, a signal (a '!' above the avatar's head would suffice) would be required to inform the player. In these situations, a (single) button would be used to initiate this avatar-spotted course of action. For instance, at a ledge with no platform to jump to, the avatar may be able to see that the ledge could be negotiated if she hangs from it. This is signalled to the player, and if the player wants to explore that option, they press the action button to do so.
What is the purpose of this model? It is to empower the player to play, rather than to challenge the player to learn to carry out complex tasks. In a game such as Prince of Persia: The Sands of Time, the player is afforded an impressive number of actions, all of which are beuatifully animated - but the control skills required to complete these actions are prohibitively hard for anyone without sufficient legacy skills to master. It follows the empty vessel model. Imagine the appeal to a wider audience of a similar game which made it easy for the player to guide the avatar through such impressive feats. Just watching the animation would be engaging.
I am not proposing that the blind captain model would replace the empty vessel and distant voice models - these models already have loyal audiences, and will be with us for a long time to come in one form or another. Rather, I am suggesting that exploring the blind captain model might give us a new and wider audience. Hardcore players may be intellectually resistant to sharing control with the avatar; many would like to be given absolute control, and thanks to their legacy skills have no difficulty with complex control mechanisms. But such people are definitely a minority.
Imagine, in the near future, a 007 game in which the player was actually empowered to become Bond instead of being challenged to perform at his level of competence. Failing over and over again until the player can master the challenge presented is not a game experience which resembles watching a Bond film. Bond (or any other action hero - male or female) is naturally competent. He finds ways to survive being shot. He finds ways to survive jumping of roofs and cliffs. He finds ways to survive anything that is thrown at him. (He finds a way to seduce practically every non-lesbian woman he can find). A blind captain 007 game would give the player these survival skills for free, thus empowering the player to investigate master criminal plots and battle enemies using their own skill and judgement. It would allow the player for the first time to play at being Bond, just as a child would play in their own imagination. That's what Bond is: an adolescent power fantasy. Shouldn't Bond games deliver this same experience?
There is one thing I am confident that players in the nebulous mass market want: they want to play. The easier the control mechanism, the more likely this is to happen.
The opening image is Movement by Melanie Weidner, taken from the Listen for Joy art gallery. As ever, no copyright infringement is intended, and I will take the image down if asked.
Footnotes for Regular Readers
The empty vessel model supports Tactical play only when the player already has the legacy skills to be able to play without mastering a lot of additional controls. In general, it supports Logistical play - the player is required to repeat actions over and over again until they work out what to do and/or successfully implement it. Because challenge is a key focus of games favouring this interaction model, games using this model seem to appeal primarily to players who enjoy the Type 1 Conqueror play style in DGD1 or Lazzaro's "Hard Fun", i.e. players for whom fiero is a key emotional payoff.
The distant voice model is chiefly a Strategic approach. Players generally require the ability to hypothesise and solve problems abstractly to enjoy this means of play. In this context, it is not suprising that games employing the distant voice model have a smaller audience than those emplying the empty vessel model (Strategic skills are assumed to be strongly expressed in about 10% of the population, whereas Logistical skills are assumed to be strongly expressed in about 50% of the population).
The blind captain model has been conceived with an eye towards favouring paidia over ludus. I believe this is a key to mass market appeal. The type of play supported by the game could match any skill approach - Tactical, Logistical, Strategic or Diplomatic - and ideally, would support multiple different approaches in order to have maximal appeal.