6 Regret and Competence
One of the primary motivators of human behavior is avoiding regret. Before Kahneman and Tversky formalized prospect theory and loss aversion, they believed that regret avoidance was at the root of the human behaviors they were studying.1 They learned that there are behaviors that regret avoidance could not explain and were led to a broader picture. However, it is worth delving deeper into the dynamic of regret, which has direct applicability to game design.
Legacy Games
The box for Risk: Legacy2 is sealed with a sticker that says, “What’s done can never be undone.” (See figure 6.1.)
Risk: Legacy box sticker.
Then, when setting up the game for the first time, one of the first things you do is choose between two special faction power stickers to place on a faction card. The rules then tell you to destroy the other one, and they are quite explicit about what destroy means: “If a card is DESTROYED, it is removed from the game permanently. Rip it up. Throw it in the trash.”
This begins a series of one-way decisions the players need to make. They write on the board in marker, add stickers to spaces, and tear up cards. In the end, they are left with a game that is unique to their group, but it is also littered with the ghosts of other choices the players could have made but did not. Forcing the players to make physical changes to the board ups the emotional stakes—and also slyly subverts our expectations about how we interact with a board game.
When people make a choice, they are frequently left with unease about the choices they leave behind, the roads not taken. In psychology, this is formally called regret, and it is a form of loss aversion.
Permanent choices, even simple ones like tearing up a playing card you didn’t select, can make people uneasy, and they may agonize over the decision. Risk: Legacy and its progeny (collectively called legacy games) are designed to be played a finite number of times before the story or journey is complete (fifteen times for Risk: Legacy; from twelve to twenty-four times for Pandemic: Legacy3). There is a subculture of gamers who go to great lengths not to destroy cards or mark up the board, using clear overlays, numbered baggies, and other techniques to make it possible to reset the game back to its out-of-the-box configuration so that it can be played through again. Although they maintain that it is about preserving the investment they made in the game (so that it can be enjoyed again by another group), the discomfort of cutting off potential futures and the possibility of rectifying mistakes certainly plays an equally large, if not larger, role in their decision.
Except for the simplest ones, like Candy Land,4 there is always an element of choice in games—and wherever there is choice, there is the possibility of regret. But there are many types of choices in games, from those with great strategic impact on the game, like moves in chess, to choices that are purely luck-based but may have an impact on the player personally, like deciding what number to bet on in roulette.
There is a taxonomy of choices and how they lead to different levels of potential regret. But before examining these, let’s take a look at a simple game to shine a bit more light on the psychology of regret.
Regret Game
I have two dice—one red, one white—and two identical cups. I secretly place one die under each cup (no trickery), mix them up, and ask you to select the cup with the red die. You did not see me put the dice under the cups, so you have absolutely no information on which to base your decision. If you make the right choice, I give you five dollars. If not, you gain nothing.
Go ahead and select one of the cups. Let’s say you pick the cup on the right. I slide it toward you but don’t let you look underneath.
I then ask you if you want to switch to the cup on the left, to change your choice.
Would you switch? The majority of people do not; about 90 percent do not switch, according to studies.5 Personally, I do this experiment with my class and at other presentations, and I have yet to have someone switch when offered the opportunity.
Why is that? The odds of being right are fifty-fifty, so why not switch?
Experimenters point to two reasons. First is the endowment effect described in chapter 2. Once you select a cup, it subtly becomes yours—and it becomes more valuable to you than the other cup. Moving the cup so that it is in front of you amplifies this effect, as the cup is moved into your personal space.
The second reason is that, psychologically, one of the drivers of our actions is our effort to minimize regret. If we make a choice and it turns out to be wrong, we feel bad. But what if we make a choice, switch, and then find out our first decision was actually correct? We feel worse. We know this about ourselves, and so, when presented with the option to switch away from our cup, it is not very enticing to do so.
We can explore exactly how much worse we feel if we switch and are wrong through a simple experiment.
In our game, if you pick the cup that has the red die, you win five dollars. Now let’s say I change the rules slightly. After you’ve made your cup choice, I give you the option of switching to the other cup as usual. But now, if you switch and are correct, you win six dollars. If you stick with your original choice and are right, you still only get five dollars. Would you switch now?
Again, most people would not switch.
What if I offered you ten dollars if you switch and win? What about fifteen dollars?
Researchers have performed these experiments, and the results are illuminating. At triple the value, 50 percent of people switch. You need to get to ten times the value before 90 percent of people switch.
Experiments like this, by the way, are one of the methods that researchers use to figure out how much worse loss aversion makes things feel. This puts it at around three times as bad, which is in line with the two to three times values seen in other experiments.
When I was in school and taking multiple choice exams, the advice was always to “stick with your first instinct” rather than to change to a different answer. I am now convinced that this advice does not give you a better chance at getting the correct answer—but it does make people feel better than if they get it wrong and find out their original choice was correct. I would much prefer getting it wrong with my first guess than changing away from the right answer!
Regret as a Game Design Tool
As a designer, depending on the emotional stakes you want from your players, choices can be tuned to different levels. Here are the characteristics of choices that may be prone to inducing more regret:
- Being irrevocable
- Having a major impact on the game
- Offering fewer options
- Having a clear cause and effect between decision and outcome
The first two should be clear, but let’s take a closer look at the last two. If the player has to choose between two options, or many options, the chances for regret vary. They are much greater with only two options. As a thought experiment, think about placing a bet on a roulette table. Let’s say that on one spin you place a bet on the number 17, and it comes up 31. Then on the next spin you place a bet on red, and it comes up black.
Most likely, you would feel more regret at losing the latter bet than the former. With thirty-six possible numbers to bet on, your responsibility for making the “correct” choice is diffuse. It doesn’t sting as badly when your choice goes awry. However, with a red/black bet, there are only two options—so if you lose, it’s easy to picture yourself as having made the correct decision. In psychology, this is known as a near miss, and it is much more emotionally impactful than being far away from the correct choice. If I bet on 17, I feel much more regret if the spin comes up 16 than I do if it comes up 21. With only two options, you either win or get a near miss.
Near misses in games that you can play repeatedly have actually been shown to motivate continued play.6 They generate a negative emotion (regret) along with a positive one (motivation). Video slot machines are specifically designed to take advantage of this effect. Because the animation of the wheels is strictly under computer control, once the random number generator decides the player has lost, the animation of the wheels stopping is designed to show the player that they almost won—just being one symbol away from winning.7 This is such a powerful effect that some jurisdictions have started to outlaw machines that do this, but it’s still legal in most places.
The last criterion on the list for having more regret is a clear linkage between the decision and the outcome. With many decisions in games, it isn’t clear what the actual impact on the game is. This is particularly true as the number of decisions players make increases and the number of options for each decision increases. Both will create a disconnect between the decisions and the final result. Many games feature a series of microdecisions. These do not tend to be games that engender regret.
Let’s look at a game that does create a strong sense of regret in the player. Lost Cities,8 by Reiner Knizia, is a card game for two players (figure 6.2). In it, players are running expeditions to reach archeological sites represented by five different suits. Players begin an expedition by placing a card down on the table. In future turns, they may add to one of their expeditions by adding a card of the same suit but a higher number than the last card played to the expedition. At the end of the game, each expedition is worth the sum of all the cards played to it, minus twenty points.
Lost Cities.
Knizia is a mathematician by training, and many of his games, particularly the early ones, have a simple mathematical twist that makes the game. In Lost Cities, the twist is that twenty points are deducted from each expedition. If you start a new expedition with a 1 card, you now have an expedition worth negative nineteen points. You’re already in the hole. But if you don’t have an expedition of a certain color, it’s not worth negative twenty points; it’s simply worth 0.
The crux of the game is taking calculated risks—specifically the risk that you will be able to add enough cards to an expedition to climb out of that hole and earn the points you need to win the game. Early in the game, it’s likely that you will be able to get enough cards played to an expedition to score positive points. But as the game progresses through the midgame, starting a new expedition becomes more fraught.
Ultimately, this can lead to a decision the player regrets. This decision has all the hallmarks noted earlier:
- It is irrevocable. Once you start a new expedition, that card is locked down. There is no opportunity to cancel it. It is on the table and will be there for the remainder of the game.
- It has a major impact on the game. Scoring negative nineteen will almost certainly cause you to lose the game.
- There are only two options. Either you start a new expedition, or you add to an existing one.
- The connection between you starting, or not starting, an expedition and the points it scores or loses is clear. At the end of the game, it’s very easy to point to the decision that led to your victory or defeat.
All these factors combine to turn what is, at its heart, a very simple card game into an emotional roller coaster as you wrestle with simple, but not simplistic, choices.
Regret and Endowed Progress
A decision that presents the opportunity for regret slows down choice. People need to think harder about it as they try to disentangle the rational from the emotional response. Some features that make decisions more prone to regret are importance and permanence. If you can’t go back and change things later, it’s harder to make that choice.
To keep the game moving and lighten the emotional stakes, game designers typically avoid including decisions that are prone to triggering regret. However, there are certain games in which the designer wants to bring forth these emotions, like Risk: Legacy. There, the emotional engagement of permanent decisions immediately causes players to adopt a certain mindset.
In one sense, regret is the flip side of endowed progress and the techniques we discussed in the last chapter. Endowed progress is about making positive effects more likely to happen by moving the player toward the goal. In addition, these are outcomes that, if they occur, will definitely be positive. If you get eight additional stamps, you will get the free car wash. If you get eight more victory points, you will win the game. If you get to three thousand experience points, you will go up a level.
Choices that lead to regret are negative, even if the effects are positive. In Risk: Legacy, the players need to choose between one of two special abilities. Both are positive, and both make that faction more powerful. But the other needs to be ripped up and destroyed, and it’s impossible for the players to know what the future would have held if they had chosen that other ability. The loss, in this case, is not for something tangible but for a potential future—which mirrors some of the most difficult choices we make in life.
Endowed progress is the opposite of this. Other options are not eliminated; endowed progress is about moving toward a singular goal. Regret typically comes from big choices made at a crossroads, whereas endowed progress breaks things up into small chunks that are easy to deal with. It is not surprising that one creates angst while the other motivates.
Massively multiplayer online (MMO) games, like World of Warcraft,9 create regret in their structure. They often force players to make certain decisions about skills and other options when new levels are reached, and these are decisions that have far-reaching and long-term implications for characters that may be played for years. This leads to players stressing over how to advance their characters, agonizing over the distribution of each point. Two mechanisms have arisen in an attempt to alleviate the regret that these choices can generate: one from the community and one from within the games themselves.
Because this is the age of the internet, there is a plethora of resources available to players on how to design their characters (called builds or specs). These range from calculators that show how your character will perform when certain skills and spells are chosen to complete recipes for what the “best” character design is to fulfill certain roles or play styles. Selecting one of these gives partial immunity to regret. It gives players a glimpse into the future, more visibility into what they can expect. Psychologically, it allows for shifting blame to the author of the build guide if the player doesn’t get the satisfaction or performance they were expecting. This ignores the fact that in-game performance relies on more than just the skills chosen; it also relies on the skill of the player. But again, it gives psychological cover to the player and one less thing for them to worry about, which reduces prospective regret. These types of guides go well beyond MMOs. They also are very influential in the world of Magic: The Gathering deck creation (net decking) and miniatures games, in which army composition lists on popular sites are very influential.
The other element that helps ease regret is the ability in some games to perform a respec, via which a character can be completely rebuilt from the ground up. These are usually earned at certain points in character development or can be purchased or are sometimes given when character abilities are modified in their ability or power. Regardless of the source, this is a powerful antidote to regret. Making decisions about your character development with the knowledge that they are not irrevocable does wonders to quell regret, even if the potential for such a respec is well in the future.
The rise of battle royale games like Fortnite and PlayerUnknown’s Battlegrounds (PUBG)10 present an interesting case study on regret, and loss aversion more generally.11 These games feature permadeath, which should increase player anxiety and works against these being for casual players. However, there are several factors that mitigate the psychology—but first, let’s take a brief look at the format: Battle royale games have up to one hundred players in the same game, playing either individually or as small squads of two or three people. When a player is killed, they are eliminated from the game, and may not return. When only one player (or squad) remains, they win the game. To force player interaction, the play area shrinks over the course of the game, driving players together into the same area.
Now, the first mitigating feature of a battle royale is its length. The games typically take fifteen to twenty minutes in to complete, thanks to the ever-shrinking field. Many players are eliminated much earlier, giving them a minimal time investment. There’s a big difference between permadeath after five minutes and after five hours.
The second, and perhaps more significant, factor is the structure of the game itself. The longer that players last in the game, the better they are doing. The number of players remaining is prominently displayed in the corner of the screen in all these games, and merely surviving as that number ticks down is gratifying. Length both increases the stakes, as you get closer to possibly winning, and decreases them, as you have already done better than so many others in the game. And given that one hundred players are in each game, casual players go in with realistic expectations about their chances of winning.
Competence
In addition to thinking about possible future regret, decisions are also impacted by how much information people feel they have available to make the decision. This is called Competence. Before getting into the details of exactly what that means, let’s look at a simple game. Two games, actually; you can decide which one you want to play. The objective in both games is the same: guess if a die roll is even or odd. Here are the choices:
Game 1
You guess whether the roll of a die will be even or odd. Then I roll a die, and we see if you’re right.
Game 2
I secretly roll a die and put it under a cup. Then you guess if the roll will be even or odd, I lift the cup, and you see if you’re right.
When presented with this choice, most people—almost 67 percent—select game 1. This seems peculiar at first. Why should there be a preference for one over the other? The chances of winning the game are 50 percent either way. What is the difference?
This illustrates a concept that is referred to as “competence”. Competence is defined (semimathematically) as the amount of information you have before making a decision divided by the total amount of information that can possibly be known. In the game 1, you know everything there is to know about the situation. But in the second, there is a key fact that exists in the universe—what the die result is—that you don’t know. So, your competence in game 2 is much lower than it is in game 1.
Studies have shown that people much prefer high competence situations, even when competence has no bearing on the outcome. Low competence makes people uneasy and ties into the same psychological circuits that make us avoid regret and loss.
Here’s another game choice:
Game 1
I give you a box with fifty red balls and fifty green balls. You name a color and then blindly pull out a ball. If it matches your chosen color, you win.
Game 2
I give you a box. It is guaranteed to have at least one ball but could hold any number. And all the balls in the box are either red or green, but I’m not telling you the distribution. They could all be red, or all green, or a fifty-fifty mix, or anything in between. The game is the same as game 1: you name a color, blindly pull out a ball, and if it matches your chosen color, you win.
Which game would you rather play?
The chances of winning either game are 50 percent. This is obvious for game 1 but might need some explanation for game 2. How can it be 50 percent with so many unknown variables? The key is that the player gets a choice of which color they are trying to pull from the box. Because the contents of the box can’t be changed after this choice is made, there’s no way to set up the balls in such a way that overall there isn’t a 50 percent chance of winning.
As an example, let’s say I put five red balls in the box—that’s it. Assuming that you, as the player, have an equal chance of announcing that your target is red or green, then 50 percent of the time you pick red, giving you a 100 percent chance to win—and 50 percent of the time you pick green, with a 0 percent chance to win. So your overall win rate is (0.5)*(1) + (0.5)*(0) = 0.5. The general equation for a box that has X fraction of red balls and (1-X) fraction of green balls is (0.5)*(X) + (0.5)*(1-X), which is always 0.5 regardless of X.
Even when this is explained to people, over 90 percent pick game 1. The uncertainty makes the second game much less attractive. People are more likely to think that there is some trick they aren’t understanding, that they are being duped. Game 1 is clear: fifty balls of each type, and you have a gut feeling for what your options are.
But even in less extreme cases, research has shown that people gravitate toward the option in which there is the least amount of hidden information, as illustrated in the first “die under a cup” example.
Attack/Defend Example: Who Chooses First?
Let’s take this principle and apply it to a simple fighting game.
There are two players: an attacker and a defender. The attacker has two choices: punch and kick. The defender also has two choices: block high and block low. Block high counters a punch and block low counters a kick. You have cards representing the choices, and you have to pick one and place it face down in front of you.
Would you prefer to play first or second in this situation? From an outcome standpoint, it doesn’t make a difference—unless you have some sort of tell that helps your opponent figure you out. But when surveyed, 65 percent of people preferred to go first. This preference is what we would expect based on competence theory. If you go first, you know all the information there is to know in the world. But if you play second, there’s a big piece of knowledge that you don’t have access to—what your opponent has chosen. This makes it a much more emotionally fraught decision for many people.
From a design perspective, this allows the designer to manipulate the emotional state of the players and make decisions harder or easier.
Video Games versus Board Games
A key characteristic of the majority of board games is that the game space is completely known. Maybe you don’t know the order in which the cards will come up, but the range of possible cards is knowledge that players have—certainly after they’ve played the game a few times, but also you can simply pick up the deck between games and look at all the cards. Maybe you don’t know what you’ll roll on the dice, but you know what the possible outcomes of that roll are. In general, board gamers are used to high competence situations.
In the 1970s and 1980s, war games from companies like Avalon Hill and Simulations Publications, Inc. (SPI) were becoming more complex because complexity was equated (often erroneously) with realism in military simulations. However, these games reached a level at which it was difficult for players to manage systems that required increasing amounts of bookkeeping.
When computers became increasingly available in the 1990s, war gamers were excited about the prospect of games being moved to that platform. The computer would handle all the tedious bookkeeping, allowing players to focus on the strategies and tactics. As computers became more powerful, computer war games tracked more and more stats. Carrier Strike12 tracked the fuel in each aircraft in the air, the fatigue level of individual pilots, and how many planes were on each deck of the carrier. The Operational Art of War13 kept track of the exact mix of weaponry in entire battalions—as well as detailed tracking of supplies, fatigue, and disorganization (see figure 6.3). And all this information was available to players. All the data could be accessed and reviewed by the players when making decisions about what to do in the game.
Unit report from The Operational Art of War. The unit itself is composed of various assets, each of which has its own statistics (lower left).
However, the level of detail exceeded the capacity of most players to deal with it effectively. It ended up being a fog of data, and players retreated from it, abstracting it in their minds. It flipped around and became a negative: players were aware that this data existed but were not able to take advantage of it in any meaningful way. Ironically, even though all the data could be known, this triggered a feeling of low competence and all the negative emotions and uncertainty that come along with that.
This type of war game became a dead end, development-wise. There were further attempts to incorporate this level of detail, but they attracted a limited audience. There is a limit to how much information people can absorb and act upon, and giving players more information leads to blow back and reduced enjoyment. The design space shifted toward more abstraction—or at least smarter abstractions that represented realism by focusing on the information and decisions that would be available to actual commanders at the level being simulated. The Combat Mission14 series of games has the players issuing orders to troops and tanks—and then sitting back to watch a movie of the action unfolding. The game gives high-level information about the status of various units but, ultimately, does not give the player access to detailed information about each soldier or how many inches of armor plating was penetrated. Instead, simple status indicators show “disorder” levels (like Good Order or Panic), and the player works with that information.
An interesting feature of these systems is that they don’t reveal to the player exactly what these status indicators mean, for example, to the firepower of the unit. The percentage reduction of strength for different statuses is built into the software, but it isn’t revealed to the players—as it would have to be in a board game in which the players are responsible for resolving combat.
However, this doesn’t feel as much like a loss of competence because in the real world we are used to this type of uncertainty when dealing with people. So we fall back on predispositions and then gradually adjust our mental model within the context of the game as we gain more experience with how those units behave. Being able to map into real-world experience keeps away feelings of low competence.
Many video games take advantage of the players’ feelings of low competence by giving players the chance to repeat encounters and gradually learn the patterns behind what is happening on screen, which ultimately enables them to succeed. This typically happens with boss fights, the biggest challenges of many games.
Dark Souls15 is a classic example of this genre. When players face a new boss, they enter the encounter with extremely low competence. The boss has special attacks, and as it takes damage from the players it will change its patterns and modes of attacking until defeated. When first encountering a boss, players have no idea what to do. They need to be close observers and experiment with different techniques to determine the weaknesses of each particular one.
Normally, one would expect that these encounters, which start with low competence, would result in negative player reactions. However, designers adopt several techniques to avoid this and instead leverage low competence to increase the sense of accomplishment of the players.
First, there is predictability and repetition in the boss’s actions. It performs certain attacks and moves to new attacks in a definite progression, so its patterns can be learned. The learning happens on both a macro level—learning the types of attack patterns the boss moves through as it becomes damaged—and on a micro level by predicting specific attacks.
This leads to the second technique: telegraphing. Through actions, graphical cues, or sound cues, the players begin to learn how to predict which attacks come next. There is also feedback indicating damage levels and when a boss will shift to its next set of attack patterns.
Both techniques let players know that they have low competence at the start of the fight—but that with skill and observation, they can increase that competence until it reaches a level where they win the encounter. Having this type of learning and advancement has been shown to be a primary motivator for player engagement in games.
So as a designer, it is important to be aware that low competence itself is not a negative. The problem is low competence that stays low. If the players have a natural path to increase competence as the game goes on, that dynamic can flip from demotivating to highly motivational. However, if handled incorrectly, low competence can indeed be a negative. The game Cuphead16 is a highly rated game in the “run and gun” genre, in which players need to master precise movements and learn to anticipate boss patterns to succeed. Its inventive characters and vintage art style have attracted many players who are not used to a game this punishing, and these players end up frustrated. But even among run and gun aficionados, Cuphead has garnered criticism about two specific design choices leading to low competence in the players.17
First, the levels have random elements that can put players into no-win situations. The game demands precision from players and rewards careful play, but, on some levels, the game builds in randomness that works against player precision by having the chance of random enemy movements that make death inevitable. This takes knowledge of what is to come away from the player and puts it behind a random number generator. Randomness does not foster low competence in and of itself; think back to our first die-rolling example. But random sequences that may create no-win situations emphasize uncertainty about the “facts in the world” (in this case, not the randomness itself, but the way different random elements may interact in the future), giving the player feelings of doubt and discomfort.
The second element is more direct. The bosses in Cuphead do not have a health bar, and most do not have a graphical indication of how close they are to being defeated. If the player loses, the game does show them how close they were to defeating the boss—so players can begin to get a sense of a boss’s health. But in the heat of the battle, this is not shown. This takes away an important tool from the players, and the boss’s health is a definite fact about the world they don’t know. Plus, if it were known, it would (positively) impact players. If you know that one or two more hits will win the battle, you can change your playstyle to take more chances.
Regret and low competence are comparable in the way that they can make it difficult for people to make decisions, but they are typically on opposite ends of the decision spectrum. Regret is caused by having just a few choices with clear possible outcomes. Low competence is caused when the inputs and outputs of the decision are fuzzy and unknown. Regret can be a useful tool in game design, but, in general, sustained low competence (not low competence that can be overcome through learning) should be avoided. It leads to a general feeling of helplessness and will strongly demotivate players.
Notes
1. Michael Lewis, The Undoing Project: A Friendship that Changed Our Minds (New York: W. W. Norton, 2016).
2. Rob Daviau, designer, Risk: Legacy (Hasbro, 2011).
3. Matt Leacock and Rob Daviau, designers, Pandemic: Legacy (Z-Man Games, 2015).
4. Eleanor Abbott, designer, Candy Land (Milton Bradley, 1949).
5. Donald Granberg and Thad Brown, “The Monty Hall Dilemma,” Personality and Social Psychology Bulletin 21, no. 7 (July 1995): 711–723.
6. R. L. Reid, “The Psychology of the Near Miss,” Journal of Gambling Behavior 2, no. 1 (Spring/Summer 1986): 32–39.
7. Natasha Shüll, Addiction by Design: Machine Gambling in Las Vegas (Princeton, NJ: Princeton University Press, 2014).
8. Reiner Knizia, designer, Lost Cities (Kosmos, 1999).
9. Rob Pardo et al., designers, World of Warcraft (Blizzard, 2004).
10. Brendan Greene, creator, PlayerUnknown’s Battlegrounds (PUBG, 2017).
11. I am indebted to Matt Villers, who first suggested these ideas on Twitter. See https://twitter.com/matthew_villers/status/1039313575125217280.
12. Gary Grisby, designer, Carrier Strike: South Pacific 1942–44 (Strategic Simulations, 1992).
13. Norm Kroger, designer, The Operational Art of War (Talonsoft, 1998).
14. Charles Moylan and Stephen Grammont, designers, Combat Mission: Beyond Overlord (Battlefront, 2000).
15. Hidetaka Miyazaki, creator, Dark Souls (Namco Bandai, 2011).
16. Jared Moldenhauer, designer, Cuphead (Studio MDHR, 2017).
17. Snoman Gaming, for example; see https://www.youtube.com/watch?v=G_b3K21ulSQ.