This is a reprint of an article that appeared in the April 2010 article of Game Developer Magazine.  It has also appeared on Gamasutra at this link.

Bill Belichick is regarded by many football fans as a brilliant tactical coach, but in November of 2009, he made a decision that is debated to this very day.

His Patriots were up by six against their hated rivals, the Colts, when his team faced fourth and two at their own 28 yard line with two minutes left. Most coaches in this situation would automatically punt. Going for the fourth down and failing would give the Colts’ Peyton Manning, one of the top quarterbacks in the game, a short field of 30 yards to score a touchdown and win the game. Punting would make him travel at least 70. The Patriots went for it. They failed, and then lost the game.

After the game, Belichick was defensive. He argued that going for it had high odds of success, and getting the first down would have effectively won the game. On the flip side, the Patriots’ defensive line was exhausted, and Manning was cutting through it like butter—in that particular situation, the difference between 30 yards and 70 was relatively insubstantial. He argued that the upside was infinite and the consequences of failure weren’t all that different from punting. If he’d succeeded, people would have called him a genius.

Balancing Failure and Rewards

The relative chance of failure compared to the upside of success is what made Belichick’s dilemma the fodder of debate for months after the game. Key decisions in the best games provide well-balanced decisions like this to the player. If the balance is askew, the experience can be weakened. Consider, for example, the myriad Facebook poker games where players have a near-infinite supply of fake cash. Since players face no real risk for their bets, they constantly make monster bets with abandon, making it almost impossible for purists to bet and bluff in a realistic way.

As designers, we create the carrots and the sticks that drive players through our simulation. Most designers typically think first about the carrots — the rewards and bennies that encourage players to pursue “good behavior,” but equally important are the consequences of failure and tough decisions. It’s the consequences that give these decisions real weight; they provide the emotional lift for the greatest rewards.

Sometimes, these choices are fairly black and white. In BioShock, the player is given the choice to harvest or save the “Little Sisters” (young girls who provide the psychic energy that unlocks the player’s latent power). Harvesting the sisters feels horrific but grants the player more power than saving them. The ratio of sisters saved to harvested determines what game ending you will see. Saving the sisters makes for a harder game experience in the short term, but the player receives gifts that nearly equalize the difference, and provide a much more rewarding ending. In short, the path of evil is the path to quick power, but the path of good has greater long-term gains.

The Behavior You Incentivize

Economists are fond of saying that you get the behavior you incentivize. One commonly cited fact from real life is that mandatory seat belt laws have resulted in an increase in pedestrian deaths. One consequence of wearing a seat belt is that the driver himself is safer, which allows him to drive faster and more recklessly.

As designers, we must be careful of the behavior we incentivize — it is dangerously easy to penalize good behavior, or reward activities that actively destroy the player’s own game experience. If you make a game where jumping is faster than running, players will jump everywhere they go, no matter how silly it looks.

Sometimes, the consequences of a design decision are more insidious than they seem at first glance. It may make sense for quest decisions to affect the player, but consider that if you immediately slap an alignment penalty on a player as soon as he dares talk to the roguish thief hiding from the city guard, you’re punishing him for trying to play your story, and teaching him that some of your content is best bypassed if he wants to take a lawful path. Writers need to be careful of how and where they structure these key choices, and how they ensure that completionist players have an avenue to experience the whole game without trashing their characters.

Facebook games like Zynga’s Mafia Wars and Vampire Wars also create some incidental bizarre behavior. Success in these games often depends on the player having a large number of friends also playing these games — frustrating for players with few friends, and unnerving for those who are uncertain they want to badger their friends into joining them in their little vampire gaming fetish.

The developers saw spamming friends as an advertising opportunity, but these players saw it as a negative consequence. As a result, many devotees of these games set up alternate Facebook profiles, joining groups of like-minded “fake” friends in order to game the system without polluting their real friends list. Whether Facebook or Zynga sees this as something that negatively impacts them remains to be seen.

Being Too Harsh

Over the years, most games have become less punitive to failure. Massively multiplayer games are no exception. Back in the days of text MUDs, death frequently meant the loss of a level’s worth of experience. By comparison, EverQuest‘s penalty was only one-tenth of that — and of course, it seemed ridiculously benign. Ten years later, penalties have been reduced to the point where death is little more than a minor inconvenience.

Some hardcore players long for the old days, and some armchair designers even push the idea of permanent death as a way to create more tense and dangerous online worlds. What these designers lose track of is how these penalties affect the risks that players are willing to take. Most designers want players that take big risks, try odd and unusual things, players that will test the limits of the simulation to discover emergent gameplay. It’s here that the experiential magic of interactive fiction really shines.

Some players crave harsh penalties, of course. They like to play Diablo 2 on Hardcore mode or Demons’ Souls, where the tension of ultimate failure provides an emotional edge, and minimizing the risk of failure is a key strategic decision in the game. Make no mistake, though, this is a hardcore game direction.

However, if the penalties are too harsh players won’t take chances. They’ll seek out lesser challenges. In MMOs, they’ll hunt below their level, avoid grouping with strangers and not show up for player vs. player situations in which they are clearly the underdog. In short, they’ll bore themselves to death, and then blame you, the designer. Correctly, I might add.

High Risks, High Rewards

There exists in baseball a huge subculture of fans that specialize in the statistical examination of the sport. Devotees of the science of sabermetrics attempt to upend any number of common conceptions of baseball. One of these misconceptions is that attempting to steal a base is ever a good idea. Statistically, the consequences of failing to steal a base (losing one of your precious 27 outs) almost always outweighs the potential benefit.

Despite this, managers still try to steal bases. The interesting thing is that, despite the fact that it’s usually a terrible idea mathematically, it genuinely makes for a more entertaining baseball game. The apparent risk vs. reward does not match the actual risk vs. reward (which, fortunately, results in managers making decisions that make for better television).

As designers, it’s important that the risks a player takes have rewards that correspond to their penalty—or at least feel like it. It’s okay to have high-risk choices for the player, but the rewards have to provide an emotional high that matches.

These high-risk, high-reward choices are excellent ways to provide an additional layer of gameplay for more advanced players. Killing 10 players with your bare hands is, for many players, worth the achievement. Attempting to throw an opponent out of the ring in Soul Calibur is hard to pull off, but completely worth your opponent’s humiliation if you execute it. Rocket jumping in Quake to the perfect sniping place is worth occasionally blowing yourself up or shooting yourself into the lava.

The important thing about these high-risk, high-reward choices is in fact the element of choice—they’re alternatives. Making these optional gaming avenues leaves a safer, more predictable path for more casual, less-skilled players. Perhaps just as importantly, though, they provide a way for your more dedicated gamers to demonstrate and declare mastery over the game.

Unclear Consequences

Any first-year psychology student will tell you that the best way for rewards and punishment to work as a modifier of behavior is to ensure that the consequences are swift and directly related to an action. Whacking your puppy on the nose with a newspaper won’t work unless you catch her in the act of piddling on the rug. While game players are a bit more self-aware than the average pup, designers can make their games much more powerful by making the consequences of failure explicit.

In the early days of MMO development, designers theorized about a virtual ecology. In this model, the virtual dragons would feast upon electronic sheep. If no sheep existed in their hunting area, the dragon would have no choice but to seek food in the village where players live. The moral of the story is to not destroy your own ecosystem.

The problem is the average player’s inability to see the whole system. He doesn’t know that killing sheep makes the dragon relocate — he may not even know that there were ever sheep around at all. All he knows is that he was one-shot by a dragon while trying to sell his gear to a vendor in newbie village. The idea that another person he’s never met could kill him by making mutton chops half a mile away doesn’t feel particularly fair.

This is a difficult problem, but not an insolvable one. MMOs with complicated realm vs. realm combat such as Eve Online and Warhammer: Age of Reckoning use political maps to show players’ land control, giving them a global view of the actions of the community. World War II Online goes further — that game’s website allows players to see a history of how the front of the war has shifted over time.

Big, Bold Choices

Interactive games are at their best when players learn by doing. For this to work, though, players need a clear cause-and-effect for their actions. Subtlety here is often wasted. Give players a choice between two relatively minor stat penalties, and he will be left wondering if it was the right choice — or whether the choice made any difference at all. In games like Dragon Age: Origins and Mass Effect, however, the big choices are black and white. He doesn’t wonder if he made the right or wrong choice — he knows, as the consequences of his actions have big, bold effects on the game world.

Too often, though, subtlety is lost within the noise of the simulation. Some may decry this lack of subtlety of consequences, but I think that most players approach games coming from a real world where their choices often have no visible impact, or where the risks are simply too high for them to follow their hearts. Providing an avenue for experimentation and release is one of the things games do exceptionally well. Making the games that best provide this for the player requires a well-designed carrot — and an equally well-designed stick.