The gamer’s dilemma offers three plausible but jointly inconsistent premises:

  • Virtual murder in video games is morally permissible.

  • Virtual paedophelia in video games is not morally permissible.

  • There is no morally relevant difference between virtual murder and virtual paedophelia in video games.Footnote 1

Since its introduction by Morgan Luck (2009), the dilemma has received a number of different responses. But it has also been construed, or reconstrued, in a number of ways. On some construals, the dilemma has a wide scope of application. It is thought to apply to all instances of virtual murder and paedophelia, and sometimes more than just virtual instances (e.g. by construing virtual murder and paedophelia as representations of murder and paedophelia, and construing the dilemma as applying to all representational instances).Footnote 2 By contrast, there are narrower construals on which the dilemma applies to only some instances of virtual murder and paedophelia e.g. those that are interactive, or that occur in particular video gaming contexts.Footnote 3

But what lesson should we draw from this? One possibility is that different construals offer competing understandings of a single dilemma. Another is that different construals reveal the presence of more than one dilemma. In this paper I argue for the latter position. There is more than one gamer’s dilemma. Depending on how we understand two key ideas in the dilemma, we have up to three distinct dilemmas.Footnote 4 Each dilemma is about a different issue, applies to different cases, and is susceptible to different solutions. And each raises an interesting problem that is worth resolving.

The two ideas that give us the different dilemmas are those of (1) occurring in a video game and (2) being a virtual instance of murder or paedophelia. For (1), depending on how strictly we interpret the idea that virtual murder and paedophelia are performed in a video game, the dilemma’s focus changes. One dilemma asks about our intuitions concerning what is permissible in video games, as games. Another focuses on what is permissible virtually. For (2), we can draw a distinction between different types of virtual instances. Some virtual instances acquire their status by only representing the properties of a (typically nonvirtual) original. Others reproduce essential properties of the original. But depending on the type of instance we have in mind when considering virtual murder and paedophelia, the dilemma’s focus changes. One dilemma is about representing the acts in question virtually, another is about partially reproducing their essential properties virtually. This gives us three dilemmas: a dilemma about in-game acts, another about represented acts, and a third about partially reproduced acts.

The argument proceeds in three sections. Section 1 introduces Luck’s original dilemma and argues that the emphasis placed on the video gaming context changes the dilemma’s focus, which gives us the first two dilemmas. Section 2 turns to virtual instances of murder and paedophelia, and argues that like other virtual items, these acts are produced in at least two ways. But each type of instance raises distinct problems, and this splits the dilemma on virtual acts into two more specific dilemmas. Section 3 concludes the discussion by considering how different participants in the debate have interpreted the dilemma, and its solution, in different ways. It also considers whether any of the dilemmas is plausibly the ‘correct’ gamer’s dilemma.

The gamer’s dilemma and video games

Morgan Luck (2009) introduces the gamer’s dilemma as follows:

Most people agree that murder is wrong. Yet, within computer games virtual murder scarcely raises an eyebrow. In one respect this is hardly surprising, as no one is actually murdered within a computer game. A virtual murder, some might argue, is no more unethical than taking a pawn in a game of chess. However, if no actual children are abused in acts of virtual paedophilia (life-like simulations of the actual practice), does that mean we should disregard these acts with the same abandon we do virtual murder? (p. 31)

The emphasis in Luck’s (2009) statement of the dilemma is unclear. Is the dilemma about acts that occur in a video game,Footnote 5 or virtually? Some parts of Luck’s discussion suggest the former. For instance, Luck compares virtual murder to taking a pawn in a game of chess, refers to the subject performing the acts as a ‘game player’, and considers a response to the dilemma (objection 3) that focuses on the competitive element in games. This suggests a dilemma about acts performed in games, in-game acts. But at other times Luck seems to have the virtual context in mind. In the second footnote, Luck (2009) writes “I take the arguments presented in this paper to be able to operate sufficiently under a broad conception of a virtual/game environment” (p. 31). This suggests that the video gaming component is inessential. The conception of video game environments is ‘broad’, and Luck treats ‘virtual’ and ‘game’ as interchangeable. In addition, Luck (2009) discusses the possibility of extending the dilemma to other non-game forms of popular entertainment.Footnote 6 Finally, in Luck (2018), the dilemma is stated in a way that omits reference to video games, which suggests its focus is virtual actions.

But whether the dilemma is about in-game or virtual acts is a significant difference. First, this changes what the dilemma is about. One dilemma is about what is permissible in games (we can call this the gaming dilemma), another is about what is permissible virtually (we can call this the virtual dilemma). The gaming dilemma asks about the types of performances that are morally permissible in games. Specifically, it asks whether virtual murder and paedophelia are permissible in-game acts. This can be compared to asking if in a board game like Monopoly, the in-game act of purchasing property is morally permissible.Footnote 7 The virtual dilemma, by contrast, focuses on virtual performances, not gaming ones. Here video games, and games more generally, are incidental. The dilemma is about the values we place on virtual murder and paedophelia, in video games or otherwise. Video games happen to be the place where many virtual acts first emerged, and maybe continue to exist, but virtual acts are not constrained to video games. The Unity game engine scene viewer, a physics simulator, and an online social world all offer virtual worlds that enable virtual murder and paedophelia.

Second, the two dilemmas focus on different instances of murder and paedophelia. The gaming dilemma focuses on acts that occur in video games. But the dilemma can be extended to other types of games. Presumably if there is a worry about committing virtual murder and molestation in video games other game acts will raise similar worries. For instance, acts in card and board games are not (typically) virtual, but they are symbolic in a way that prevents them from being instances of the original. A board game may revolve around murder or paedophelia, as could a card game. The gaming dilemma can be raised for these cases: is a player more blameworthy for choosing to play a pedophelic board game, and are they more blameworthy for using paedophelia rather than murder cards to progress? The gaming dilemma also excludes certain cases. Consider virtual murders and molestations performed in VRchat, a social virtual world. The acts can be performed by an adult, in private quarters, to nonhuman virtual characters, and where no one is murdered or molested.Footnote 8 This would meet Luck’s requirements outside the gaming context, and seems to elicit the requisite intuitions. Still, such cases are not part of the gaming dilemma, since VRchat is not a game.Footnote 9 By contrast, the virtual dilemma includes virtual murder and paedophelia in VRchat, since they are virtual. But it excludes acts in card and board games, since they are not (typically) virtual.

Finally, the two dilemmas are resolved in different ways.Footnote 10 The gaming dilemma is resolved by showing that in-game performances of virtual murder and paedophelia are equally permissible, or that there is a relevant difference between the two. But the virtual dilemma is resolved by showing this for virtual murder and paedophelia, whether or not they occur in a game. So for instance, thinking that what happens in a game is morally insulated because ‘it is just a game’ will undermine the gaming but not virtual dilemma.Footnote 11 But thinking that the virtual is not real and therefore less valuable will undermine the virtual but not gaming dilemma.Footnote 12

Because these dilemmas are different they are also related to different philosophical problems. The gaming dilemma focuses on how ethical features of in-game acts affect a game’s value, and one’s engagement with it. This makes the dilemma similar to problems raised by an artwork’s ethical features,Footnote 13 which can also impact a work’s value and one’s experience of it e.g. an artwork may lose value because it endorses an immoral point of view. The gaming dilemma may be subsumed under this broader class of problems—about ethics and art—since video games are thought to be an artform,Footnote 14 and the dilemma focuses on their (potentially) immoral contents. By contrast, the virtual dilemma is not related to morality in games, or art, since not all virtual acts occur in games or art. Instead, the dilemma is related to problems like those raised by Nozick’s experience machine and more recently, virtual reality.Footnote 15 Responses to this dilemma will be influenced by background views like virtual realism or irrealism, which focus on the reality and value of virtual items.Footnote 16

Video games include both game and virtual acts, so a token act can raise both dilemmas. But the reason a token act counts differs across the two dilemmas. Are both dilemmas significant? I think so. Both raise interesting problems. There are substantive questions we can ask about the sorts of acts that are permissible in games. We might wonder whether games that involve animals fighting are permissible, whether a given sex game is permissible, and whether a video, board, or card game’s acts are permissible. The answer is not immediately obvious. And knowing the answer matters because all people play games, and the modern world is increasingly ‘gamified’. Similarly, there are substantive questions about the value of virtual acts. Our access to the virtual is recent, so our moral intuitions about it are hazy. But the virtual offers new room for exploring human freedom, and the contemporary world is increasingly dependent on it. So we can expect moral questions about the virtual to become increasingly pressing as more acts are performed virtually.

So far we have two dilemmas. But one may think there is also a third dilemma, focusing on the value we place on acts performed in games that are also virtual. But this third dilemma seems less interesting. This is because resolving it seems to involve no more than resolving the two more basic dilemmas. If we know whether virtual murder and paedophelia are permissible in games, and we know whether they are permissible as virtual acts, we should know whether they are permissible in games that involve the virtual.

Virtual representations and simulation

The two dilemmas may be divided further. We might distinguish dilemmas that occur in different types of games, and different types of virtual contexts. But whether additional divisions are worth making or not depends on their impact. In this section I argue that the virtual dilemma should be divided further. Minimally, we should distinguish two virtual dilemmas. One dilemma is about virtual representations of murder and paedophelia, another is about virtual simulations of these acts, where a simulation of X partially reproduces the essential properties of X.Footnote 17 Each dilemma raises a separate issue, covers different cases, and is resolved in a different way.

To see this, note first that virtual murder and paedophelia are virtual acts, and so share features with other virtual items. Virtual items have been construed in a number of ways.Footnote 18 But whichever view one accepts, one has to account for a peculiar feature of virtual items. In some but not all cases, a given virtual item, virtual X, is an instance of X, the (typically) non-virtual original.Footnote 19 For instance, some virtual money is money. Other virtual money is not money, just as virtual murder and molestation are not murder and molestation (respectively). That virtual items can be virtual reproductions of the original raises a question about how virtual items come to acquire this status.Footnote 20 To answer this, we need to distinguish between two ways of producing virtual counterparts.Footnote 21

When building a given virtual counterpart, virtual X, designers can do one of two things. One option is to reproduce the essential properties of the original, X. For instance, one can virtually reproduce a calculator’s functionality, or a nonvirtual painting’s visible properties. But designers can also adopt a less demanding means of building virtual counterparts. Rather than reproducing the original’s essential properties they may choose to only represent the original’s (essential or nonessential) properties virtually. This can happen in a number of ways e.g. a virtual calculator may represent a calculator by reproducing nonessential properties, like the calculator’s appearance, or by representing its functionality, as when one uses a virtual calculator to ‘calculate’ sums that open locked doors in a video game.

Distinguishing between virtually representing properties and reproducing essential properties allows us to make sense of the fact that sometimes a virtual counterpart is an instance of the original. In some but not all cases, the original’s essential properties are fully reproduced virtually, and this results in a virtual reproduction. When this does not happen, this will be for one of two reasons. Either some but not all of the essential properties will be reproduced, which gives us a partial virtual reproduction, a virtual simulation. Or no essential properties will be reproduced, with properties being represented instead. I call this a virtual representation.

How does this bear on the virtual dilemma? Insofar as the dilemma appeals to virtual murder and paedophelia, the acts it appeals to can vary in whether they reproduce essential properties or merely represent properties of the original acts. This difference changes what the dilemma is about. To see this, note first that the dilemma cannot be about all virtual murder and paedophelia. There is no dilemma when it comes to virtual reproductions of murder and paedophelia, since these virtually reproduce all the essential properties of the original acts, and as such are instances of murder and paedophelia. Our intuitions for these cases are clear: both acts are impermissible.

Instead, the virtual dilemma’s focus is on virtual acts that fall short of virtual reproduction. Such virtual acts are of murder or paedophelia by virtue of virtually representing properties of murder or paedophelia, and/or virtually reproducing some but not all of the original acts’ essential properties (such that the act is not an instance of the original). Examples of the former may occur in the original Grand Theft Auto, which displays its virtual world from a pixelated bird’s eye perspective. Virtual murder and paedophelia in this case will be little more than some changes of colored pixels, designed in a way that allows the pixels to represent the original acts. Examples of the latter may occur in virtual reality settings, where virtual acts more readily reproduce essential properties of the original e.g. virtual reality molestations may accurately reproduce an embodied and enactive first person view of a molestation (assuming these to be essential properties of molestation).

Unlike virtual reproductions, both virtual representations and simulations elicit dilemmas. Consider first the dilemma focused on virtual representations, the representation dilemma. This dilemma turns on the value we place on representations of murder and paedophelia. But what value do we place on representations in general, and on representations of murder and paedophelia in particular? Is it always or only sometimes wrong to represent wrongdoing? What about representing paedophelia in particular? Without clear answers, it is difficult to dismiss or resolve the dilemma. Similarly, consider the dilemma on virtual simulations, the simulation dilemma. Interactive virtual murders and molestations are virtual simulations, since they reproduce at least one essential property of murder and molestation, interactivity. But are interactive virtual murder and paedophelia permissible? What about other instances that partially reproduce murder and paedophelia? Again, without clear answers we can neither easily dismiss nor resolve the dilemma.

We can consider each dilemma more closely. Consider first the representation dilemma. Because the focus is on representations of murder and paedophelia, this dilemma has wide scope. If all simulations are representations, this dilemma covers all cases of virtual simulation as representational phenomena. But it also covers murder and paedophelia representations in TV shows, books, and movies insofar as these media represent murder and paedophelia. On this view Luck’s dilemma is only new in using virtual representations rather than more traditional forms of representation which we see in books and movies. An appropriate response to this dilemma focuses on the representational aspect, for instance, by arguing that virtual representations of murder and paedophelia are permissible because engaging with representations does not commit us to endorsing their represented content.Footnote 22

By contrast, the simulation dilemma is about partially reproducing the essential properties of murder and paedophelia, not merely representing them. Virtual simulations of paedophelia may reproduce the visual appearance of a child molestation, and virtual murders may reproduce the psychological effects of murder on the murderer (assuming these to be essential features of these acts). In robust simulations, like Nozick’s experience machine, one may reproduce most essential features of these acts while still falling short of reproducing them. Unlike the representation dilemma, the simulation dilemma is about partial enactments of murder and paedophelia. As a result, the dilemma is narrower in some ways but wider in others. The simulation dilemma does not include all representational cases, since not every representation reproduces essential properties of what is represented e.g. virtual calculators that only look like calculators. But it includes non-virtual and potentially non-representational cases that reproduce essential properties. For instance, it includes murder and paedophelia performed in Westworld,Footnote 23 where Westworld is a nonvirtual but simulated world (since it partially reproduces essential features of the real world).Footnote 24 Finally, resolving the simulation dilemma involves offering an account of partially reproducing impermissible acts, and so offers no guidance on consuming virtual or nonvirtual representations of murder and paedophelia more generally.Footnote 25

One may be hesitant to divide the virtual dilemma into the representation and simulation dilemmas. Couldn’t the virtual dilemma cover both representations and simulations? Two points are worth making about this question: first, each of the two dilemmas may include the other, if we think all simulations are representations, or that all representations are simulations.Footnote 26 In these cases, answering the broader dilemma would answer the narrower one. The problem is that this requires further commitments. Neither view is immediately obvious. It is not clear that representing something requires reproducing at least some of its essential properties e.g. a painted house may represent a real house, but doesn’t seem to reproduce a house’s essential features. And it is not clear that reproducing essential properties of an item amounts to representing the item e.g. e-mail reproduces essential properties of mail, but it is not clear that it represents mail rather than simply being a type of mail. Second, even if we decide to focus on the virtual dilemma more generally, we still need to distinguish virtually representing acts and virtually reproducing them partially. Both issues matter, but they differ morally.

The three dilemmas

To summarize: the gamer’s dilemma can be understood as one of three distinct dilemmas depending on how we understand two key ideas in Luck’s (2009) original formulation. Depending on the weight we put on the gaming context, the dilemma can be about in-game or virtual acts. And depending on the type of virtual act we focus on, the virtual dilemma is about virtual representations or virtual simulations. This gives us three dilemmas: one about game acts, another about represented acts, and a third about simulated acts. We can see the three dilemmas at play in the literature around the gamer’s dilemma. Different participants have interpreted the dilemma in different ways, and have offered responses to different dilemmas, sometimes presenting one dilemma and responding to another. Here I consider a sample of accounts that roughly divide across the three dilemmas.

Participants like Rami Ali (2015) and Nader (2020) have focused on the gaming dilemma primarily. I treat the dilemma as one about acts in which the gamer is appropriately engaged with the video game. But I highlight that when we constrain the dilemma to gaming, there are at least three different types of games that demand different types of ludic engagement from the player, and where this mode of engagement is relevant to evaluating in-game acts. The three types are what I call sporting, storytelling, and simulation games. The first two types engage the gamer by offering a competitive event in a virtual space and telling a story in a virtual space respectively. By contrast, simulation games make no explicit demands from the gamer e.g. virtual actions need not move the story forward, or amount to winning the game. Instead, I maintain that “what characterizes these games is their focus on enjoying or exercising a virtual freedom in a given domain. In providing the player with a virtual freedom, the freedom to perform certain acts, or partake in specific events in a virtual world, such games simulate our natural freedom, and in this sense are simulations.” (p.270).

By including simulation games I cross from the gaming to the simulation dilemma. Games do sometimes offer a sandbox to experiment in. But when they do this, what they offer is not exclusive to or distinctive of gaming contexts. Running a scene in the Unity or Unreal engines can afford such a sandbox. As can a physics simulation. Unsurprisingly, these types of games make no reference to the game when being evaluated, and so elicit different intuitions from other cases. While storytelling and sporting games are evaluated by the in-game context, simulations are judged by the player’s choice to engage in certain virtual actions rather than with an in-game demand. Video games rarely offer pure simulations. When they do, this is usually part of a non-simulation game e.g. Grand Theft Auto offers a sandbox, but this is part of a broader storytelling game. It is also unclear that a game could be exclusively a simulation game in Ali’s (2015) sense. Such games would be unlikely to count as games rather than simulations. They would not meet e.g. Grant Tavinor’s (2009) definition of a video game, since they involve no objective oriented gameplay (there is no objective) or an interactive fiction (simulations need not be fictions),Footnote 27 nor C Thi Nguyen’s (2019, 2020) definition since simulations lack a ludic goal.Footnote 28

Instead, simulation games are games that allow one to enjoy one’s virtual freedom. The focus here is on the virtual, not the game. More specifically, the focus is on virtual simulation, through (at least) partly reproducing the subject’s non-virtual freedom. This breaks down Ali’s (2015) solution into two responses. One response focuses on game acts and maintains that because evaluations turn on the context a given game offers for performing murder or molestation, both acts can be morally permissible or impermissible. Another response focuses on simulated acts, maintaining that both simulated murder and molestation are morally impermissible. This also means that Nader’s (2020) supplement to my response more aptly responds to the gaming dilemma I am interested in. Rather than focusing on simulation games, Nader focuses on sporting games. And the response he offers explicitly includes the gaming component by appealing to the idea of consent in competitive gameplay.

By contrast to Ali (2015) and Nader (2020), Stephanie Patridge (2010), Sebastian Ostritch (2017), and Christopher Bartel (2020) focus on the representation dilemma primarily. Patridge and Ostritch initially seem to focus on the gaming dilemma since both frame their discussion in the context of an amoralist challenge that maintains that video games are “just games”, and so not subject to moral evaluation.Footnote 29 But though the problem they set up focuses on games, the solutions they offer focus on the representational element in games. Patridge argues that imagery in video games has social meaning, and some imagery has socially incorrigible meaning, so there is a difference between virtual murder and paedophelia. This response is indifferent to whether acts are performed in-game or not, what matters is that certain images play a representational role with an undesirable meaning in culture. Similarly, Ostritch (2017) highlights that in-game acts function representationally. He writes.

[T]he amoralist is right in claiming that the activity of playing a game cannot in itself be morally wrong, […] he is wrong in claiming that games are not subject to moral evaluation because they are “just games”. Games are not “just games” in a twofold sense. First of all, they may contain representations that require certain emotional reactions or attitudes on the side of the player. […] Second of all, […] games themselves sometimes transcend the boundaries of the fictional by endorsing a morally problematic worldview. (p. 125)

Unsurprisingly, Ostritch sees no difference between the representations we find in video games and in other representational works like movies and literature, and explicitly addresses this aspect of his solution.

Bartel’s (2020) response also focuses on representation. He notes that video gamers interact with video game imagery, and that these play a representational role. He argues that attitudes towards this imagery can be morally evaluated. For instance, he compares playing video games to bearing relations towards photographs and effigies, which can be experienced as morally significant.Footnote 30 Bartel’s (2020) position is somewhat continuous with his (2012) position. In Bartel (2012) it is less clear that the focus is entirely on representation because there Bartel treats virtual paedophelia as an instance of child pornography. But to be an instance of child pornography, virtual paedophelia has to reproduce essential properties of child pornography, properties like the visual appearance of a child. This suggests that Bartel’s (2012) focus may be on simulations as well as representations of paedophelia.

Finally, participants like John Tillson (2018) and Erick Jose Ramirez (2020) focus on simulations. Tillson’s (2018) main aim is to argue that there is a moral difference between simulating wrongdoing and consuming non-simulatory representations of wrongdoing. He argues that the former, but not the latter, are pro tanto wrong. Although simulations are a type of representation on Tillson’s view, a simulation is a specific sort of representation. Tillson writes “A representation is a simulation in the sense I mean, iff it is agential in the sense that a player selects actions to be undertaken by the player’s character(s) (i.e. it comprises agential affordances).” (p. 207) So, on Tillson’s view the problem is with simulating murder and paedophelia, since these (unlike non-simulatory representations) reproduce an essential property of these acts i.e. the agential role of the subject in engaging with them.

Ramirez (2020) also focuses on simulations. He describes the dilemma as being about the value of virtual acts, omitting any reference to games (as Luck (2018) does), and focuses on the idea that these acts are simulated. What Ramirez thinks is significant about simulated murder and paedophelia is their ability to reproduce responses that the original acts elicit. The resultant ‘virtually real experiences’ are treated as if they are real experiences. This imposes requirements on these acts. First, that they manifest ‘perspectival fidelity’ (to higher degrees), where perspectival fidelity amounts to the extent a given simulation offers a perspective that coheres with the perspective of the subject using the simulation. Second, that they manifest ‘context realism’ to higher degrees, where context realism “refers to the degree to which the rules of a simulated universe cohere with the rules that a subject believes the actual world is governed by” (p. 150). These features need not be shared with all representations, or all games. Instead, they are features that allow some virtual acts to reproduce essential properties of the original, properties like evoking the subject’s real responses to murder and paedophelia.

We may wonder, is one of these dilemmas the right gamer’s dilemma? We can say a few things about this question. First, if we stick with Luck’s name for the dilemma, the gamer’s dilemma sounds like a dilemma focused on games. But Luck’s (2018) discussion and the literature around the dilemma suggest that the answer is less clear. Second, in-game, represented, and simulated acts all potentially raise Luck’s worries, since acts of each type occur in video games. So it is not clear that one of the act types raises the correct dilemma. Third, the dilemmas are all independently interesting, so it is not clear that any one of them should be given priority. It is important that we evaluate our moral attitudes towards in-game acts given the prevalence of video games today and the contemporary use of gamification. It is equally important that we evaluate our moral attitudes towards representations given their prevalence in culture. And that we evaluate our moral attitudes towards simulations, particularly as extended reality technologies become more available and widely used for simulation. Finally, even if we think one dilemma is more significant, or that only one offers a genuine problem, it is important that we keep the different dilemmas distinct. Not doing so can obscure our view of whichever dilemma has significance.

Where does this leave Luck’s original dilemma? My suggestion is that we relabel Luck’s original dilemma as the video gamer’s dilemmas. Video gamers confront all three dilemmas since video games include game acts, represented acts, and simulated acts. The video gamer’s dilemmas are about committing virtual murder and paedophelia in games, about engaging with virtual representations of murder and paedophelia, and about engaging with virtual simulations of murder and paedophelia. In the end these dilemmas may be combinable into a larger dilemma e.g. focusing on acts that are not real, and this may be susceptible to a single broader resolution. But on the face of it each of these three dilemmas is different.Footnote 31 As such, the video gamer confronts three problems and not one in confronting Luck’s original dilemma.