You might call it challenge, difficulty or a scenario, but a universal trait of great games is that they test players in some way. Games have a learning value (as Raph Koster so memorably chronicled) and a huge part of their fun comes from mastering them. From the simplex crossword to the massive complexity of Battlefield 3, games push the player to be better in some way. Even creative games like the Sims are test driven.
Yet it is a major fallacy to conclude that all types of test consequently make for good games. They don’t, and there are good reasons as to why. The psychology of play and boredom gets in the way, the lack of clarity in some kinds of test makes them frustrating and the overall opacity of their results leave players nonplussed.
This is an essay about optimality, the play brain and why successful games need to be far more abstract than they might appear.
The Play Brain
The more artistic, refined and human enjoyment within a game comes from a high order of consciousness. Players are not zombies, so they can get scared, appreciate humour and get swept up in the thauma and numina of the experience. Players are perfectly capable of finding a game meaningful, sorrowful or mind-expanding just as with any other form of art.
However videogames struggle with subtlety. While they often have the appearance of being layered simulations with complex underpinnings, all games really are just planes and pieces with rules that govern how they behave. And in all great games, the behaviours of pieces are predictable. It is necessary that they be so, in fact.
The reason is that underneath the player’s artistic and experiential enjoyment of the game lies a learning machine that I call the play brain. The play brain is the part of the mind that only sees the game in terms of levers and buttons, actions and reactions, and its only interest is in self improvement. It sees the frame of the game rather than the richness of its detail, and forms a purely literal understanding of it. It is symbol-driven rather than linguistic, deductive rather than dramatic, and it has little interest in anything that isn’t speaking to it on its level.
It realises that all first person shooter enemies are just cylindrical objects that take three shots to kill. It sees through the pomp and ceremony of Shogun Total War 2 and grasps that all the troops on the battlefield are just dots that move on command. It figures out that all crops in your virtual farm are just clocks that will spit out coins at the end of a cycle.
It is also the part of the player that feels bored, is relatively sociopathic, reacts negatively to unfairness and derives excitement from the learning of epic wins. The play brain is, at heart, what drives players to play.
The foundation of successful tests, and thus successful games, comes from remembering that the play brain sees through the disguise. It only understands the mechanism, and it is it that decides whether the mechanism is interesting or not. If your game doesn’t speak to the play brain first and foremost then you have big big problems.
Designing for experience is a popular way to think and talk about games at the moment, but experience comes more as a by-product of play rather than the other way around. The reason why tests are the basis of great games is that the play brain is wired to look for optimality rather than experience.
We instinctively regard any situation in life as one which could be better if solved, and we look for the levers to allow us to do that. We all want to have better jobs, taxes, standards of living, gadgets, experiences, bodies, stories to tell and so on and we either strive to achieve those or become frustrated and depressed when we can not. We want to be optimal.
We do not all have the same instinct for optimality in the same areas. A gardener trying to figure the best way to lay out their flowers and bushes for maximum satisfaction and efficiency is just as driven by optimality as a rock climber trying to scale the Eiger. Each is hard to achieve because real life is infinite and opaque. It is complex to understand how you would get a better job or a more rewarding love life, create a great work of art or move to live in Fiji. Often too complex in fact.
Where these real life situations are often frustrating, in games they are easier because games are finite: The levers are more readily apparent, as are the consequences of action. The play brain likes games because it finds optimality more directly achievable than in reality. The play brain wants to master the mechanism and then dispose of it as it feels that it has seen either all that the mechanism has to offer or that it never will.
Every player has their personal maximum mastery with a game, the point at which they realise that they are playing no better, or that it is now too easy, or that improving within it has started to feel like labour in the infinite of reality rather than fun in the finite of ideality. So the task for all worldmakers is to make great tests that speak to the play brain and keep it engaged.
This is not easy for two reasons. The first reason is that game developers often struggle with the idea that they need to make their game basic. It seems counter-intuitive to make something so raw when the ambition is to make rich worlds and enchant players. It seems retrograde, or a throwing in of the towel even, to allow the game to be reduced to simple puzzles and action tests, and both narrativist and simulationist developers believe that to do so is to admit failure.
The second, and more important, reason is that while some game loops just naturally produce compelling tests that can extend by virtue of pressure, most games need to extend their actions through the player, objects or the environment. The core tests that such extension is based upon thus need to be able to support that ambition.
And this is where many games run into trouble. There are some kinds of test that work well for games and some that simply do not. The ones that do share a common trait, and the ones that don’t also share a common trait. Great tests are fair, while poor tests are unfair.
Obvious levers, clarity of action / response and clear extension lines are what fair tests need. What this means is that fair tests tend to be abstract and literal. The rules are very simple, the consequences obvious, and the actions you can use to try and win the test are clear. What’s not clear (and why this is a test) is the exact pattern of actions required to win. You know what you can do, so your challenge is to figure out how to do it better and in what order to do it to win.
Some types of fair test are:
Combat Tests: The bulk of hardcore console games are action games that involve hitting and shooting. They are so for a reason. Combat games are highly master-able, often extensible and so players can immediately understand what they need to do in order to win. The test is whether they have the skill to do it.
Agility Tests: Platform games, from Mario to Ico and back, test the player’s ability to work out a series of actions to get from a start point to the gold star without getting killed. The single biggest cause of unfairness in agility tests is bad camera placement, which is why Mirror’s Edge is an unfair platform game.
Reaction Tests: Racing games keep the scenario pretty simple, so the player’s test is obvious (win) and easy to comprehend. The question is whether he can react fast enough or not to get there. Beat em up fighting games also test reactions, but in a different context.
Mechanical Tests: The game poses an abstract puzzle based on tangible objects that the player has to align correctly. Tetris, Bejewelled, Peggle and Diner Dash are all games made in this vein. The bulk of successful casual games are in essence mechanical tests.
Perception Tests: Some perception tests are fair, others are not. A great example of fair perception tests are games like the Agatha Christie hidden object series where it is clear what the player has to find, and her job is to find it.
Logistic Tests: Roleplaying games like Final Fantasy and Diablo are logistic tests. The player needs to figure out the right combination of characters, items, heal potions, special attacks and defences to be able to optimally take down enemy characters. So he spends a significant amount of time looking at numbers and stats and working out winning combinations.
Geographical Tests: Strategy games like Chess are pure geographical tests. The player has to work with the board and figure out a strategy of attack lines, reacting to the opponent and so on. Games in the Starcraft, Advance Wars and Total War vein combine geographical and logistic tests.
Memory Tests: The player must learn a pattern sequence and be able to deploy it, often quickly. This can be as static as a Simon Says game or the more complex task of trying to remember big moves in Street Fighter 2. Memory tests are also a part of games like Rock Band where the song pattern is always the same – and the player can anticipate it after a few plays.
Unfair tests, on the other hand, are opaque, random and do not extend well. The play brain cannot see how they resolve, cannot deduce how it could be playing better, and ultimately is left guessing at what to do. Unfair tests either do not extend at all, or if they do it is not perceptible.
There are plenty of games that are based on unfair tests. They sound fun, but when it comes down to it the play brain will tend to move away from the unfair parts of the game to the part of the game that feels fair if possible, or simply stop playing at all.
Here are some examples:
Intuitive Tests: This one is in my mind most because I’m playing LA Noire at the moment. Intuitive tests rely on the player to perceive subtle information (in LA Noire’s case: facial expressions) in order to read the signs and choose wisely. Intuition is not a quantity that can be expressed in an objects-and-rules fashion, and the direct relationship between cause and effect is not obvious. As a case in point, the LA Noire conversation engine does not produce predictable results, so the play brain has no way to know how to play optimally. The ambition is complex and high minded, but the actual experience feels quite arbitrary.
Cultural Tests: Riddles and other tests that rely on semantic or cultural knowledge are also unfair. Riddles rely on the player having a similar set of semantic language understanding to that of writer, and if the player doesn’t have that then he is unlikely to improve. Quizzes based in a cultural frame of reference with which the player is not familiar invite cheating (if possible) or hit the play brain’s maximum mastery almost immediately.
Qualification Tests: You can have fun with the game, but only after you have qualified to play it. This means that the play brain has to see the levers without the benefit of any context, and often all at once, and it immediately feels stupid. An infamous qualification test is the opening level of Driver, where the player had to prove that they could pull off a series of driving stunts in a garage before being allowed into the rest of the game.
Tolerance Tests: As I wrote about recently, busywork masquerading as gameplay is generally not much fun. Games that use busywork, appointments and other delaying tactics are only really testing the player’s tolerance to put up with their crap, and if they miss out for some reason (for example failing to collect your crops because you had to go to a funeral that day) it feels as though you lost for reasons beyond your control. And that’s unfair.
Arbitrary Tests: For a few years it seemed like every action game from Resident Evil 4 to God of War 3 was using quick-time events as a way to spice up their gameplay. The game would suddenly flash an icon on the screen, and if the player did not react instantly he would lose. Quick time events are an example of an arbitrary test, one which just randomly appears in the sequence of play and offers a simple X/Y or fail/win choice. They are highly unfair.
Luck Tests: They may be compulsive because they offer potential big cash rewards, but luck tests like slot machines offer the least agency to the play brain of any game type. The player understands that they are unfair, but takes a punt on them because they might win money.
Morality Tests: Another recent fashion was games that ranked the player’s behaviour on an ethical scale. Are they a good citizen or a bad egg? Do they get to be a Sith or a Jedi? Rule in tyranny or in light? What these tests actually amount to is a hidden table of plusses and minuses for player actions that are arbitrary and often inconsistent. The play brain does not have the table to hand, and so does not know what is optimal and what is not. Again, this is unfair.
Character Tests: The player’s presence in a game world is a doll, but every entity that he does not control is a character. A character test requires the player to teach, impart, influence or otherwise be nice to a character in the hope that they will improve or behave. Character tests as a part of games, where you basically need the character to do something for you in order to get on with the game, are often maddening because the sequence of actions required is hidden from view.
There are also some games that try to be non-tests. In the casual and social game markets in particular, a kinked logic has emerged where developers are so terrified of putting off the female market by making anything seem difficult that they end up creating applications which are just supposed to be clicked to pass the time. (Which Ian Bogost rightly satirised with Cow Clicker.)
Quite aside from the pretty dismal interpretation of what women want that this view encompasses (and in my experience it tends to be male developers who fall into this trap), the basic problem that these designs have is that they are actively trying to be boring. It’s mystifying to wonder how their logic ends up at that point, but it tends to come from woolly thinking about the value of experience combined with a very facile reading of books on behaviourism and market trends.
The idea that such developers usually end up with is invariably some kind of dress-up doll or pet caring application, where the pet doesn’t even have the decency to get hungry or dirty if not maintained, but is not rich enough in content to qualify as a virtual toy in its own right. Why women are supposed to care about something so patronising is a question that goes unanswered.
My advice: If you want to make a toy then do so. Don’t dress it up with parts of a game and spectacularly fall between two stools in the process. Making digital toys has always been an exciting part of software creation, so own it and make it your own if that’s what you really want to do.
Be not afraid of tests, and don’t fall into the trap of making tests that are too subtle, or not tests at all. Literal is good. Abstract is good. Clarity of action and response is great. The foundation of crafting that great experience that you want to make is to understand that boiling your game down into see-target-and-hit-it or square-peg-into-square-hole symbolic scenarios is a good thing rather than a design failure.
Too often worldmakers get trapped by the idea that making a simple thing that works great is somehow admitting defeat, or are spooked by the mad idea that the customers don’t actually want to play and would prefer to just exist in a game world in some sort of stasis.
Neither is true. The play brain wants to play and it only sees the world through very literal eyes. If you want to be a great artist making an awesome world then go for it, but understand who you are dealing with first. If it has scope for optimality then it’s flawed, and nobody wants to play a flawed game.