`3 Framing`

`Disease`

When performing their original studies on loss aversion, Kahneman and Tversky did not just ask questions about money. They also tested people’s reactions to different types of emotional and nonemotional stimuli.

Here’s another set of choices they posed to subjects:

A deadly disease is rapidly spreading! If nothing is done, six hundred people will die.

You are the leader of the Pandemic Response Team and are presented with two options:

Option A

Two hundred people will be saved.

Option B

One in three chance that all six hundred are saved.

Two in three chance that no people are saved.

When presented with these choices, 72 percent chose A—guaranteeing saving the lives of two hundred people.

Here’s another set of options:

A deadly disease is rapidly spreading! If nothing is done, six hundred people will die.

You are the leader of the Pandemic Response Team and are presented with two options:

Option C

Four hundred people will die.

Option D

One in three chance that no one dies.

Two in three chance that six hundred die.

What would your choice be? When presented with these options, 78 percent of people choose D.

I’m guessing that you see where we’re going with this.

Choice A and choice C are exactly the same, as are choices B and D.

Take a look at it again:

Option A	Two hundred people will be saved.
Option B	One in three chance that all six hundred are saved. Two in three chance that no people are saved.
Option C	Four hundred people will die.
Option D	One in three chance that no one dies. Two in three chance that six hundred die.

With six hundred lives on the line, “two hundred people will be saved” is the same as “four hundred people will die.” And a “two in three chance no people are saved” is equivalent to a “two in three chance that six hundred die.” However, people react to these very differently. This is called the framing effect. The way that choices are presented can have a dramatic impact on how people answer.

One person I showed this to realized what the trick was, but she still couldn’t bring herself to pick choice C—even though she had already picked choice A. “How can you pick a choice that says four hundred people die?” she said.

These results map very neatly onto the monetary loss aversion rules that we explored earlier. Two of the key findings are as follows:

People will take a sure gain rather than gamble for a larger gain.
People will gamble to avoid a sure loss.

And that is exactly what is happening here. In the first example, the choices are all expressed as gains, in a positive sense: two hundred people will be saved, or a two in three chance that no one is saved. Saved has a very positive connotation. Accepting the sure thing of A appeals to us more than the chance that no one is saved. And, conversely, when expressed as losses, or people dying, the sure loss is not as appealing as taking a chance to avoid losing people.

Manipulating framing is seen constantly in the world of sales and marketing. As a simple example, in California, Colorado, and several other US states, it was made illegal to charge a surcharge to a customer paying by credit card. It was, however, legal to give a discount for paying cash. The results were predictable. Merchants increased their prices slightly and offered discounts for cash. The same exact prices are paid as were paid before (with the credit card surcharges), but people and lawmakers felt better about it. It’s more palatable for some people to receive a discount than for some to get a surcharge, even if the actual dollars changing hands are the same.

When laws regarding surcharges versus discounts were being debated before the US Senate Banking Committee in 1975, Jeffrey Bucher from the Federal Reserve Board maintained that there should be no distinction between discounts and surcharges—that they should be treated equally under the law. He did note, though, that “critics argued that a surcharge carries the connotation of a penalty on credit card users while a discount is viewed as a bonus to cash customers. They contended that this difference in psychological impact makes it more likely that surcharge systems will discourage customers from using credit cards.”

`Offsets and Isolation`

Framing can be a highly effective way of affecting player emotion during game play. There are a variety of ways that it can be integrated into different mechanics, but perhaps the simplest is achieved simply by adding an offset to the gains and losses.

For example, here’s a game:

You are given $1,000. Then you are asked to choose between:

Option A	A 50 percent chance of winning $1,000.
Option B	A guaranteed $500.

As you should predict by now, 84 percent of people chose option B.

Here’s another game:

You are given $2,000. Then you are asked to choose between:

Option C	A 50 percent chance of losing $1,000.
Option D	You must lose $500.

This time, 70 percent of people chose C.

Contrast this with the first two games we looked at in chapter 1:

Option E	An 80 percent chance of winning $4,000.
Option F	A 100 percent chance of winning $3,000.
Option G	An 80 percent chance of losing $4,000.
Option H	A 100 percent chance of losing $3,000.

Unlike the games from chapter 1, the new games (choices A–D) are exactly the same. In both options A and C, you have a 50 percent chance of ending with $1,000 and a 50 percent chance of ending with $2,000. In options B and D, you have a 100 percent chance of ending with $1,500.

However, by changing the framing—by offsetting the initial conditions of how much money you start with—the designer of these questions was able to express the first question in terms of gains and the second in terms of losses. In the first example, you are given $1,000 and can gain. In the second, you are given more money—$2,000—but can lose. And the natural inclination to want guaranteed gains, and to gamble to avoid losses, asserts itself.

Rather than simply switching terminology for losses and gains, this effect can be exploited by dividing a decision up into smaller decisions, a process called isolation. For example, people were asked to decide between these two choices:

Option A	A 20 percent chance of winning $4,000.
Option B	A 25 percent chance of winning $3,000.

Sixty-six percent said they would pick choice A—which is actually the correct option mathematically. Expectation value for A is $800, and for B it is $750. Regardless of the math, this is in line with studies that show that people basically treat 20 percent and 25 percent as the same chances, which drives the selection of A, which has the higher dollar value.

Now let’s look at another game:

This is a two-stage game. In the first stage, you have a 75 percent chance to simply lose the game and a 25 percent chance to move to the second round.

If you advance to the second round, you have a choice:

Option C	An 80 percent chance of winning $4,000.
Option D	A 100 percent chance of winning $3,000.

People were asked to choose either C or D before they knew if they were going to advance to the second round. In this case, 78 percent of people chose option D.

Mathematically, these two problems are exactly the same. Choice A in the first game and choice C in the second game are identical.¹ Remember, in the first game, 66 percent of people chose the $4,000 option, and in the second 78 percent chose the $3,000 option. Unlike our first loss aversion questions in chapter 1, in which the first game was gaining money and the second losing, the payouts on these two games are the same.

The point here isn’t that people are bad at math. Well, in a way it is, but the important factor is that people rely on rules of thumb or gut feelings to make these decisions. In this and similar cases, people ignore the portion of the problem that overlaps. Because it applies to both choices, it can simply be factored out and ignored. Creating the first stage returns choices C and D to one of our original loss aversion questions. Specifically, the reintroduction of the word guaranteed has a huge impact on people. This problem was specifically partitioned to give people a guaranteed pick with choice D, which they predictably flocked to. As we discussed in the first chapter, people give much more weight to choices that have a 100 percent chance of happening.

An offset manipulates the payout values, whereas isolation manipulates the payout probabilities. The effect of both is to make it more likely that people will select a certain option or perceive the option in a certain way. Poker has certain elements of this feature: players may treat the pot as an abstract, common pool of chips and overvalue their current bet. Sophisticated poker players lean heavily on the concept of pot odds, which explicitly links the immediate bet decision with the full mathematical picture.

`Framing in Board Games`

I ran into the phenomenon of framing with one of my tabletop game designs. In Pit Crew,² players play on teams and try to get their race car back out onto the track quickly while making as few mistakes as possible. When teams do good things, like getting out onto the racetrack first, their score marker moves up on a track. After three rounds, the team that has moved furthest along the scoring track is the winner. As mentioned, teams may make mistakes in fixing their car, either accidentally or intentionally, to get out faster. Those mistakes generate penalty points. The question we needed to resolve was this: What should those penalty points do?

We tested it two different ways. In the first iteration, penalty points moved you backward on the score track (see figure 3.1). In essence, penalty points were a negative score.

We also tested with several groups a setup in which penalty points, rather than moving your score marker backward, moved all the other score markers that number of spaces forward. So if your team got two penalty points, all the opposing teams moved forward two spaces.

From a game effect standpoint, this is exactly the same. Subtracting points from one team is the exact same thing as adding points to all the other teams.

The groups that played with the second system enjoyed the game much more. They tended to play with more abandon and were more willing to accept penalties to finish early and gain finishing bonuses. The groups that moved backward on the scoring track were much more concerned about the penalties. They did not want to move backward on the track and geared their play to avoid that happening, even if it hurt their overall position.

When we discussed both options with some of the groups, the sentiment was overwhelmingly in favor of the system in which penalty points moved your opponents forward. You always kept progress you had made. It was much less emotional to give a boost to your opponents than to have to give something back. For a family game, young gamers were also much more attracted to the “positive” system than the “negative” system.

The impact of framing is clear here. When the system is set up around gains, and somebody gains regardless of what happens, it makes for a more enjoyable experience, especially for casual and younger players. When you actually move backward for mistakes, it distorts the way players approach the game and makes for more infighting within the team if someone messes up. A happy bonus is that advancing on a track fits very well with the theme and metaphor of racing.

Around the same time that I was working on Pit Crew, I was developing another game that faced a similar choice: The Expanse,³ based on a television series (see figure 3.2). In this game, players represent factions trying to spread influence and power throughout the solar system. They win by having the most control points (CP) at the end of the game. At certain points, I wanted players to have to spend CP to be able to do certain things—spend CP now in the hopes of a bigger reward later.

`Figure 3.2`

A portion of the scoring track in The Expanse.

I seriously considered the option of having the other players gain CP in these cases instead of the acting player losing them, for similar reasons as in Pit Crew. It would have less emotional weight and make the players more comfortable. But The Expanse is a much heavier game than Pit Crew, and not really light-hearted or aimed at families. And while in Pit Crew I wanted to encourage players to play fast and loose with making mistakes, intentional or not, in The Expanse I wanted the decision to give up CP to feel emotional. I wanted the players to feel more invested in those decisions—to make it a bit more agonizing and difficult for players to make that choice. So in this case, to emphasize that difficult choice, I chose to force players to lose CP.

There are certainly many other games in which moving backward (more properly “getting sent back”) is a key part. This is a feature of games ancient and modern, beginning with the five-thousand-year-old Egyptian game senet, through backgammon, pachisi, snakes and ladders, and, in a more modern incarnation, Sorry!⁴

Over the last thirty years of game development, however, this reliance on punishing other players has diminished. As an example, the prototypical conquest game of the last century is Risk.⁵ In this game, players battle over territories across the globe. However, it is essentially a zero-sum game. All the territories begin the game controlled by players, and the only way to progress in the game is to eliminate all the other players from the board. Games like this, with player elimination as the goal, including Monopoly,⁶ have been the cause of many bad feelings during family game nights. This design methodology led to a certain marginalization of tabletop games through the mid-twentieth century.

Many games of this time, like Monopoly and Risk, follow a similar pattern. Players start with few or disconnected resources, build up better positions and gain stuff, and then have that position or their stuff destroyed. In Monopoly, this dynamic is expressed as starting with money but no properties. Players accumulate properties, build houses and hotels, and elevate their position. However, for all players except the winner, there is a turning point where the houses, hotels, properties, and money are all lost. Similarly, in Risk, players start with a few armies scattered all over the world. Gradually they build their forces and form perimeters to take and hold continents. However, again, for all except the winner, every army you have is going to be destroyed and every territory you control lost.

As we’ve demonstrated, via loss aversion and the endowment effect, getting things and then losing them is an emotional trigger point. Games that are set up like this engender a strong response and often lead to arguments and bad feelings.⁷ This trend was turned around by the European school of game design—particularly in Germany, where the emphasis shifted away from winning through elimination to winning by advancing fastest, almost a race condition. A key turning point using this design methodology was the publication of the tabletop game the Settlers of Catan⁸ (now simply called Catan), designed by Klaus Teuber. Catan offers several innovations, and we will revisit them in later chapters, but at this time I would like to focus simply on the victory condition. The theme of Catan is settling an uninhabited territory. The goal is not to gain dominance by eliminating or attacking the other players. It is to be the first player to reach ten points by building settlements and cities and achieving certain goals (see figure 3.3).

`Figure 3.3`

Settlements, cities, and roads in the Settlers of Catan.

Settlements and cities allow players to gain more resources, as well as victory points. Importantly, these structures are permanent. They can never be lost. Although their effectiveness can be reduced by a robber mechanism, this can affect multiple players and is only temporary. Even so, a video from game review site the Dice Tower had the three hosts name their top ten most annoying board game characters. Two of the three included the Catan robber on their lists.⁹

There are only two ways that points can be earned in Catan, and subsequently taken away: having the largest army, which is represented by soldier cards, and having the longest road. However, you lose these points only by another player getting more soldier cards or a longer road than you have. There is no way to “attack” another player’s army to reduce its size. You simply have to build yours up to be larger.¹⁰

Although only one player is the winner at the end of a game of Catan, all players will end up with more than what they started with. This shift to emphasize growth rather than destruction became the hallmark of the European game design aesthetic.

Modern games that build on the theme of Risk (games of conquest) have embraced this as well. Although there is destruction, as these are conflict games, it is not about ultimate destruction. Nexus Ops,¹¹ Kemet,¹² Eclipse,¹³ and Cry Havoc¹⁴ are all about attacking and growing more powerful than your opponents, and all involve the collection of points to win the game rather than wiping other players off the map.

Nexus Ops is probably the clearest Risk descendent, both in the way it plays and its publisher (Hasbro). Via cards, it assigns missions to players to take certain territories or win battles under certain conditions to gain points. Winning battles also awards points. The other games work in a similar fashion. Points are awarded for territory, technology, and other factors, in addition to battles. Attacking players is a means to an end but not the objective, and players cannot be eliminated from the game. All of this leads to a less emotionally fraught conflict game in which players always have the opportunity to avenge a lost battle.

`Framing in Video Games`

This sensibility has been embraced in the video game world as well, although video games often combine destruction with increasing power by all players. To go back to our original board game examples, Risk and Monopoly, when you lose stuff you also lose the things that allow you to achieve victory. For example, in Monopoly, when you lose money to other players or have to sell off houses or hotels, you are losing the very tools that you need to achieve victory. In Risk, the loss of territories and armies makes it harder to conquer your opponents. Risk does have the “card set” mechanic via which you can gain bonuses that increase in value, but this is a patch to cover up a system with a strong negative feedback loop.

Modern video games typically incorporate destruction on a separate “track” that doesn’t directly penalize the team. For example, in League of Legends, two teams of players battle to destroy their opponents’ Nexus. The first team to destroy the Nexus is victorious. To destroy the Nexus, teams must also destroy towers and inhibitors, which are defensive structures. However, the loss of these structures doesn’t penalize the team that lost them. Either the effect is neutral, or it gives a bonus to the opposing team. All teams in League of Legends end the game more powerful than when they started.

There are several connected parameters in League of Legends. Each player chooses a champion they will play during the game. That champion starts with few abilities and is fairly weak. Computer-controlled minions are spawned at regular intervals, and by defeating those minions, champions earn gold and experience. Experience can be used to gain levels, increasing strength and allowing for improvement of each champion’s special attacks, eventually unlocking their ultimate ability at level 6. Gold is used to purchase items, like swords, armor, or potions, that give bonuses and special abilities.

So what happens when a champion is killed? Are they permanently out of the game? Do they lose gold or experience, or reset back to level 1? None of these. The champion is out of the match for a few seconds (the exact time depending on how far into the match it is) and then respawns exactly as it was, with full health and mana. While waiting to respawn, players can purchase additional items from the shop, so even that wait time is mitigated somewhat.

Although being out of the match for a few seconds (plus the time required to walk back to the front lines) can be critical in a close match, Riot has positioned the other effects in a positive frame rather than a negative one—as bonuses instead of penalties. If a champion dies, the player who landed the killing blow earns an experience and gold boost. This allows for better abilities and items and makes the team more powerful. However, it does not penalize the champion who was killed.

In essence, League of Legends is positioned as a race game. The teams never move backward. When you complete a game, whether you have won or lost, you always end up more powerful than when you started. This parallels the core gameplay of Catan, in which you end up better than you started, win or lose. So players experience the game in a positive way, even if they lose. They will get to unlock better abilities and explore different item combinations regardless of how the game goes. They do not find themselves in a death spiral of decreasing ability or resources. They may have fewer resources than their opponents, but they always have more relative to their starting position. Even if players do absolutely nothing in League of Legends, the champions constantly accumulate gold. This is an economy of abundance in the midst of destruction.

The game Dota 2,¹⁵ a competitor of League of Legends, does penalize players directly for deaths. When a champion is killed, the opposing team gains gold and experience (as in League of Legends), but the defeated champion also loses gold. In comparisons between the games, this is specifically highlighted as a feature that is more “punishing” for the players and makes the games less “accessible.”¹⁶ Although there are many features in Dota 2 that lead to its reputation as a more complex, demanding game, the loss of gold upon death gives the game a more ruthless reputation.

This could be modified by shifting the framing. Eliminating the gold loss to the champion who was killed and increasing the gold gained by the team that scored the kill could sidestep this problem. But one advantage of the current system is that it adds a push-your-luck element because once gold is spent on items, they are yours to keep. Players need to balance the loss of time spent going back to base to cash in their gold with the risk of losing it if killed. Again, the purpose of this discussion of framing and loss aversion is not to say that one way is right and another wrong. It is to understand the emotional impact of game design decisions.

As mentioned earlier, early twentieth-century tabletop games like Monopoly and Risk followed a pattern of growth and building up, followed by destruction. In video games, there are many games that follow this pattern—one of the most famous being Starcraft.¹⁷ In Starcraft, each player begins with a base that produces minerals, which can be used to create basic units or buildings—which, in turn, can be used to produce more advanced units. The goal of the game is the complete destruction of the other player’s base.

Starcraft has phases similar to those in Monopoly or Risk—an early growth period, during which players develop their strategies, marked by small clashes, followed by larger and larger skirmishes until one side is destroyed. There is absolutely no positive framing in Starcraft. As your buildings get destroyed, you lose your ability to build new units to counter the next attack. If your mineral-gathering units are destroyed, it can set you back quite a lot because you will fall behind on building new units. It is very much an unstable equilibrium: a small advantage snowballs into something larger and larger.

In addition, only the winner emerges with a base. The other players all end in destruction, their buildings in ruins. Although it is possible to win the game simply by destroying the enemy headquarters and not necessarily all of their production buildings, destroying the infrastructure of the opponent is typically a key prerequisite to victory.

One of the most popular streaming and eSports games is Counter Strike: Global Offensive (CS:GO).¹⁸ CS:GO is a multiplayer, team-based shooter played over a series of short rounds. At the start of a round, each player gets money to spend on weapons and equipment. Players start with a base of $800 and get additional funds based on what they did the previous round. Scoring kills and achieving objectives give a player additional money in the next round. Although the penalty for death is fairly steep in the short term—being eliminated from that round of play—it does not affect the player monetarily for the next round.

On the other hand, there are some negative monetary penalties that can be applied to the players. In some missions, one team is trying to rescue hostages. Shooting hostages results in loss of funds for the next round. Shooting teammates (friendly fire) also results in lost funds for the next round. In this case, the designers are using loss aversion to their advantage by disincentivizing the players from performing activities that are inarguably bad within the context of the game. Friendly fire, in particular, fits with the type of shooter that CS:GO presents—one that focuses on stealth, positioning, and realism. In contrast, other shooters such as Overwatch¹⁹ do not include friendly fire effects, keeping in line with the cartoonish play style and casual audience.

In terms of framing, the money system in CS:GO is reminiscent of gold in League of Legends. Both are gained (to a limited extent) even if you do nothing. Both are awarded for doing good things in the game and allow your character to become more powerful. And both express competition not by having one team take gold away from the other team (or take more from a central pot than the other team) but by having a winning team accruing it at a faster rate, which the team needs to parlay into an on-field advantage.

There is also a broad swath of games that have what can be termed a neutral framing. The abilities of the players remain relatively constant throughout the game. Blizzard’s shooter Overwatch falls into this category. In Overwatch, there is no character progression. The abilities you start with are the abilities you have throughout the game. You can earn an “ultimate” ability by “charging” it through various activities. Once the ultimate is used, the charging resets back to zero, and the player must charge it up again. But death does not affect this in any way. If you die with your ultimate 80 percent charged, you return a few seconds later with your ultimate still 80 percent charged. Other than time out of the game, there are no negative penalties to abilities or character strength.

In the video game arena, games like Starcraft have been declining in popularity and have been usurped by multiplayer online battle arena (MOBA) games and other team games, as well as battle royale games.²⁰ In 2015, MOBAs accounted for 58 percent of watched eSports hours on the streaming service Twitch. Shooters like CS:GO accounted for 27 percent and strategy games like Starcraft for only 10 percent.²¹

Although there is no data suggesting causation, the trend is toward games that use positive framing, emphasizing growth rather than destruction. Battle royale games like Fortnite²² are an interesting contrast, as we’ll discuss in chapter 6. And as we saw in our discussion of the Tracking card in Hearthstone in chapter 1, restructuring the way the actions of the player are described can preserve the game impact while changing the approach the player takes. Changing this:

Look at the top three cards of your deck. Draw one and discard the others.

to this:

Look at the top three cards of your deck. Keep one and shuffle the other two back into your deck. Then discard the bottom two cards in your deck without looking at them.

changes the entire psychology of the player actions.

The key lesson is that game systems can be designed and presented to the players in different ways. Adding or removing the same quantity from all decisions, or from all players, can make a big difference in the decisions that are made and the experience that players have.

Notes

1. Mathematical aside:

For Choice C: 0.25 * 0.8 = 20 percent chance to win.

For Choice D: 0.25 * 1.0 = 25 percent chance to win.

2. Geoffrey Engelstein, creator, Pit Crew (Stronghold Games, 2017).
3. Geoffrey Engelstein, creator, The Expanse (WizKids Games, 2017).
4. Paul Haskell Jr. and William Storey, creators, Sorry! (Hasbro, 1929).
5. Albert Lamorisse and Michael Levin, designers, Risk (Parker Brothers, 1959).
6. Elizabeth Magie and Charles Darrow, creators, Monopoly (Parker Brothers, 1933).
7. On a personal note, my sister and I have not played Risk since the infamous Congo incident of 1980.
8. Klaus Teuber, designer, The Settlers of Catan (Kosmos, 1995).
9. See https://www.youtube.com/watch?v=YJtzAcBStLc.
10. There is a way to reduce the length of a road, interrupting it by constructing a settlement or city in the middle, but this is rare and doesn’t impact the larger point.
11. Charlie Catino and Steven Kimball, designers, Nexus Ops (Avalon Hill, 2005).
12. Jacques Bariot and Guillaume Montiage, designers, Kemet (Matagot, 2012).
13. Touko Tahkokallio, designer, Eclipse (Lautapelit.fi, 2011).
14. Grant Rodiek et al., designers, Cry Havoc (Portal, 2016).
15. IceFrog, developer, Dota 2 (Valve, 2013).
16. See http://www.gamespot.com/forums/pc-mac-linux-society-1000004/dota-2-vs-lol-educated-comparison-29187896/ and http://gazettereview.com/2016/01/league-of-legends-vs-dota/.
17. Chris Metzen and James Phinney, designers, Starcraft (Blizzard, 1998).
18. Uncredited designer, Counter-Strike: Global Offensive (Valve, 2012).
19. Jeremy Craig et al., designers, Overwatch (Blizzard, 2016).
20. For more on battle royales, see chapter 6.
21. See http://fortune.com/2016/04/06/most-popular-esports-games-on-twitch/.
22. Uncredited designer, Fortnite (Epic Games, 2017).