A long time ago, I was playing Age of Empires 1 and used cheats to peek at the computer’s base and saw that cheating bastard had more resources than I do and was multi-tasking. Fine! I accepted the challenge and barely beat it. If I fought the computer on equal footing, it wasn’t much fun since I could predict more or less what it is doing, especially after several rounds. But I don’t mind having a cheating A.I. since it felt like being almost overwhelmed or rather getting the right amount of challenge. Alas tis’ was short-lived when I played skirmishes in Company of Heroes, when the battlefield is tipped in my favour, the outcome has rapidly decided.
The computer is a cheating bastard is an identified trope (see tvtropes). As I recall from a game developer whose name I don’t remember is that A.I.s are not very smart and to achieve a desired level of difficulty, game developers instead gave the A.I.s cheats rather than improving them which is something of a cost saver.
J.J. De Simone (University of Wisconsin-Madison) and colleagues examined how that cheating bastard affects participants’ feelings of presence, hostility and enjoyment.
In sports and board games, when an opponent cheats, the other players typically greet it with disdain, anger, and disengagement. However, work has yet to fully address the role of the computer cheating in video games. In this study, participants played either a cheating or a non-cheating version of a modified open-source tower-defense game. Results indicate that when a computer competitor cheats, players perceive the opponent as being more human. Cheating also increases player aggravation and presence, but does not affect enjoyment of the experience. Additionally, players that firmly believed that their opponent was controlled by the computer exhibited significantly less state hostility compared to players that were less certain of the nature of their competitor. Game designers can integrate subtle levels of cheating into computer opponents without any real negative responses from the players. The results indicate that minor levels of cheating might also increase player engagement with video games.
My colleague, Carlos Cruz, is looking into griefing which is conceptually close to cheating.
We know that the computer is a cheating bastard, but De Simone correctly argued that little hard data is found to exactly tell us if we truly accept it as a necessary part of gaming. Two theories are proposed to explain why we might enjoy playing with a cheating bastard.
Social cognitive theory posits that we learn by observing others and we replicate those behaviours in the learned contexts. The authors argued that social cognitive theory explains that the players’ emotional and behavioural state might influenced by the videogame. Social dissonance theory posits that a disconnect between our attitudes and behaviours creates an aversive cognitive state. For example, an environmentally mindful person has thrown several bottles of beer on the lawn, having done something that is counter to his beliefs creates discomfort in his mind. Thus, the person will need to reason a way for his acts and remove that cognitive dissonance. People knowingly and enjoy playing with a cheating computer by reasoning that is the norm explained by cognitive dissonance theory. However, the authors argued that subtle means of cheating are acceptable, yet overt means have not been examined.
The authors identified four factors stemming from a cheating bastard. Presence, enjoyment, the perception of the opponent and hostility. Presence is defined as “being there” in the virtual environment, either socially in the presence of a social other, physically, or your self being in the world. Prior studies on enjoyment found that playing against human competitors is more fun than a computer because of the perceived social relationship with a social entity. These two are tied in with the perception of the opponent, people who are convinced that their opponent is human are likely to enjoy and feel more present in the game. The authors argued that overt cheating is more likely to make people think and feel that their opponent is human. IMO, it is perhaps that as players are uncertain of the opponent’s status as human or not, signs of cunning or perhaps unpredictability might be salient cues of humanness. Of course, when you unambiguously know that the other person is a cheating bastard, you will not take it kindly to this immoral act, thus you are more hostile.
Participants: 37 undergraduate students… that’s a small number and a limitation to consider. But then, he did went around campus asking students for their participation, did the experiment in naturalistic settings, such as frat houses or dorms, and gave $5 for completing the experiment. Average age is 21.73.
Manipulation check: two items to check whether the participant felt that the opponent was being fair and was cheating. And yes, participants do recognize that the opponent is a cheating bastard.
Perception of opponent: one item assessing whether the opponent is human.
Enjoyment: 6 items from the media enjoyment scale using a 7-point agreement scale.
Presence: 4 items from scales used for presence and transportation using a 7-point agreement scale.
Hostility: the state hostility scale consisting of 35 items using a 5-point agreement scale. There are three subscales: lack of positive feelings, meanness, and aggravation.
Videogame used: The author made his own videogame from an open-source software (see ASD-Tower Defense). The videogame is a tower defense game. This allowed the author to implement his cheating schemes. In the fifth wave, there was a 25% chance that the information on the incoming wave is faked and a 25% change that the enemy is stronger or an additional 25% that the enemy is unkillable. The opponent sent text messages revealing the deception, the author argued for realism in the gaming session versus higher experimental control and given his field trip around campus, maintaining realism is key. Play time last between 15-17 minutes.
After informed consent, they played the videogame for 15-17 minutes. After play time, they completed the questionnaires. The participants were randomly assigned to one of two conditions: cheating (n=19) versus normal (n=18). The participants were told that either a human or a computer is controlling the opponent.
The manipulation check revealed that those in the cheating condition did indeed reported that the opponent was being unfair and was cheating than those in the non-cheating condition.
The authors conducted a series of ANCOVAs with age as a covariate, as its range is quite large with 18 to 42, to test their seven hypotheses (oh just go read the article).
Hypothesis 1: they found that those in the cheating condition rated their opponent as more human (M = 2.11, SD = 1.05) than those in the non-cheating condition (M = 1.44, SD – 0.62), thus supporting the hypothesis that a cheating opponent makes people think that their opponent is human at a modest effect size.
Hypothesis 2: Unfortunately, both groups reported no differences in enjoyment (cheating: M = 3.27, SD = 0.61; non-cheating: M = 3.23, SD = 0.65).
Hypothesis 3: On the other hand, those in the cheating condition reported greater feelings of presence (M = 4.69, SD = 1.25) than those in the non-cheating condition (M = 4.02, SD = 1.37) with a quite moderate effect size.
Hypothesis 4: Playing with a cheating opponent would lead to greater feelings of hostility, but they have found non-significant differences for overall state hostility. There were no differences for the subscales of lack of positive feelings and meanness. But they found a marginally significant difference (p = .059) for aggravation where the cheating condition marginally reported greater aggravation (M = 2.3, SD = .85) than the non-cheating condition (M = 1.78, SD = .71).
Hypothesis 5-7: these hypotheses examined the perception of the opponent and its effects to enjoyment presence, and state hostility. The authors did a median split to create two groups: those who strongly believed that the opponent was not human (N = 18) and the rest who had some believed (from a little to strongly) that the opponent was human. Given this skewness, I think it’s reasonable for a median split, if there was a wider distribution… In any case, the two groups are called “clearly non-human” and “possibly non-computer”, respectively. This is a study limitation to consider as this is not manipulated by the researchers (i.e., making it clear to the participant that they are playing with a human or computer).
The authors conducted a correlation to disambiguate the relationship between the perception of the opponent and cheating behaviors and found none (p = .15). wah? Wait what about hypothesis 1? Are the authors talking about the correlation between two categorical variables?
They conducted ANCOVAs with cheating and opponent perception as independent variables and age as a covariate. There were no significant interactions between cheating behaviors and opponent perceptions for enjoyment, presence or state hostility including the subscales. What they found are main effects. The main effects for cheating are pretty much the same as reported earlier for hypothesis 2-4.
Both groups reported no differences in the levels of enjoyment and presence. However, state hostility scores were significantly different. Those in the “possibly non-computer” group reported greater overall state hostility (2.19, SD = .42) than those in “clearly non-human” group (M = 1.70, SD = .4). As for the subscales, those in “possibly non-computer” group reported greater aggravation (M = 2.4, SD = .73) than those in the “clearly non-human” group (M = 1.7, SD = .78). The same for meanness (M = 1.92, SD = .77) and (M = 1.37, SD = .52), respectively. A marginal difference was found for lack of positive feelings (M = 3.28, SD = .75) and (M = 2.83, SD = .78), respectively.
The take home message is that our experiences with unfairness is dependent upon “what” we are playing with. We don’t feel as much hostility when playing against a computer in contrast to those who hints signs of humanness. Nevertheless, the cheating bastard can aggravate us, but we do feel more present or perhaps are more likely to feel flow experiences since “he’s” trying to win with little dirty tricks.
The authors discussed the implication of cheating as a human function and it would serve well for videogame designers by integrating less subtle instances of cheating. I would like to go back on their literature review after reading the article and I would like to include theories of human computer interactions, such as “Computers as Social Actors” from Clifford Nass (Stanford University). I reason that research on human computer interaction might provide a greater context and insights on gaming behaviours as Nass & Moon (2000) would argue that we mindlessly respond to computers as social actors. I would argue that these works would explain hostility and enjoyment via social presence. It would be interesting to conduct a focus group on how gamers detect cheating behaviors and differentiate between bots and human players in MMOs.
Another interesting future study is how computers’ textual communication with players affect gaming experience as I remember my older brother playing with bots who seemingly converse with each other through text chat and respond to my brother’s commands… but then again I think some videogames integrated such features through voice chat, for example Unreal Tournament 3. This is a study limitation that many people think the opponent as a computer possibly being too simple.
The authors discussed the non-significant finding for enjoyment. They reasoned that due to the videogame having some invincible opponents might have proved too challenging for the participants, thus the level of enjoyment might have been affected, but at least it did not make the experience less enjoyable. For future research, I suggest spamming the opponent, I’ll have to talk to Carlos since he did that in one of his projects. The authors suggested whether wins after overwhelming cheaters’ advantages would result in greater enjoyment.
The interesting findings on opponent perceptions seems to relate with “computers as social actors”. Those who had some thoughts that they may be playing with a human reported greater hostility than versus a computer opponent indicates to me that we hold different behavioural standards to two different social actors. I argue that we hold the human opponent to play in a fair manner similar to playing with strangers in a basketball game, something that interpersonal and computer mediated communication researchers might want to explain for me. In contrast, we don’t hold the computer that social standard as it lacks many recognizable human features… unless we play with Siri, I wonder how that would work out…
I must warn that cheating does not mean that players can do so without damaging enjoyment, aside from causing greater aggravation. The study is aimed at videogame designers to allow a greater degree of cheating, it’s likely there’s a line between acceptable and unacceptable cheating for computer opponents and this study has not breached that line for a tower defense game. It is also another study limitation in that tower defense games are usually interpreted human-versus-computer game and does not contain much violence. The authors suggested trying card-based videogames, I think there’s at least one card game on Kongregate.
On a final note, the authors argued to explore the role of competition in videogaming on people’s affective state. I will stop here and start playing skirmish on Company of Heroes against a expert computer.