Has Montefiore and Formosa resisted the Gamer’s Dilemma?

In their paper, Resisting the Gamer’s Dilemma, Montefiore and Formosa (2022) provide a useful way of narrowing the Gamer’s Dilemma (Luck, 2009a). This narrowing helps those of us interested in finding a solution to the dilemma focus on cases where the dilemma seems to hold. Montefiore and Formosa then present a theory that, if true, would remove an underlying presumption of the Gamer’s Dilemma—that the intuitions grounding it are moral. In this paper, I first introduce the dilemma and then turn to how Montefiore and Formosa propose we narrow it. I then outline the theory they provide to resist the dilemma, before considering whether this theory is the most likely one available to us.

The Gamer’s Dilemma is a puzzle concerning the moral permissibility of two different acts you might perform when playing computer games. The first is virtual murder; where you direct your character to murder another in a computer game.¹ Many consider virtual murder to be permissible as  “no one actually dies when one character murders another” (Luck 2009b, p.281)—it is “just a game”.² However, this defence also applies to virtual child molestation; where you direct your adult character to molest a child character in a computer game.³ And although this is also “just a game” many consider this act to be impermissible. So, what is going on? Why do we feel so differently about these two acts? This puzzle can be presented in the form of a paradox.

Where permissibility here is understood as moral permissibility. Each of the above propositions seems plausible (or a least not implausible), but the set is inconsistent—they can’t all be true. This is the Gamer’s Dilemma.

Many seek to resolve the dilemma by demonstrating that 2 is false.⁴ Others seek to dissolve the dilemma by demonstrating that 1 or 3 is false (or both).⁵ Montefiore and Formosa (2022) seek to resist the dilemma by demonstrating that the intuitions supporting 1 are not moral (a line we shall examine further in "Resisting the dilemma").

Dissolving the dilemma is quite easy if we interpret its propositions as universal claims; like so,

For example, 3 seems false in light of the following case,

Bob works for a government classification board. One of his duties is to play new computer games and recommend appropriate ratings. Because (and only because) it is his duty to play such games he performs the act of virtual child molestation. (Luck, 2022, p. 1293)

In this case it seems Bob is doing something permissible (he is trying to protect the public from potentially immoral and/or harmful content), in which case 3 is false, and so the dilemma is dissolved.

Cases such as this one demonstrates that the universal interpretation of the Gamer’s Dilemma is pretty weak. So, a narrower version of the dilemma is required. In other words, the dilemma should be interpreted as having the following schema,

Where X and Y refer to the types of cases the dilemma should be limited to.⁶ The challenge is pinning down what X and Y actually are.

Because it is hard to come up with limitations that only include cases where the dilemma’s propositions are plausible, without also excluding some of these same cases. That is, a limitation that only captures all the cases where the dilemma’s propositions are plausible. (Luck, 2018, p. 61)

Montefiore and Formosa provide us with a very useful way of limiting the dilemma. But importantly these limits aren’t an attempt to pin down X and Y. Instead, they are a way of helping us to focus on purer cases of the dilemma—that is, cases that have fewer distracting features.

Montefiore and Formosa narrow the dilemma in a way that helps us focus on cases where the paradox holds; that is, on those cases where “the propositions of the dilemma seem plausible, and the set seems inconsistent” (Luck, 2022, p. 1296). They do this by suggesting that the cases of virtual murder and virtual child molestation in comparison should be appropriately analogous. Where cases are appropriately analogous when the virtual acts occur in maximumly similar contexts. An example might help here.

Importantly these games should not just be analogous in terms of their type or aims, but they should also be “equally positioned contextually, or as closely as possible, to the other (i.e., various contextual features, including degrees of player agency, perspective fidelity, context-realism and so on, are all kept the same for both modes)” (Montefiore & Formosa, 2022, p. 13). For example, if the cases of virtual murder under consideration take place in a game that contains a high degree of gruesome realism, then the cases of virtual child molestation in comparison should also contain the same degree of realism.

By establishing appropriately analogous cases we are now better placed to focus on cases that are in the Goldilocks zone, the zone where the conditions are just right for the dilemma to hold. For example, it may be that when the cases in comparison are both sandbox simulations games, which contain a high degree of gruesome realism, gamers are more likely to intuit that both virtual murder and virtual child molestation are impermissible (Ali, 2015, p. 273); in which case, the dilemma doesn’t hold. Likewise, if the comparison cases are story-telling games, where “neither the representations nor the viewpoint of the game are objectionable” (Ali, 2015, p. 271), gamers may be more likely to intuit that these acts are permissible; in which case, the dilemma will not hold here either. However, when considering competitive games, where the acts are depicted with a low level of realism, the conditions may be just right for gamers to intuit that virtual murder is permissible, but not virtual child molestation—in which case, the dilemma holds.⁷

To be clear, Montefiore and Formosa aren’t claiming that the Gamer’s Dilemma only holds in such analogous cases. For example, there may be cases where virtual murder is depicted more graphically (to some degree) than virtual molestation, and yet gamers still intuit virtual molestation to be impermissible, but not virtual murder. So, the Goldilocks zone will be broader than the analogous cases Montefiore and Formosa are pointing to. However, given how difficult it has been to find the edges of this zone,⁸ this focus on analogous cases is enormously helpful. For analogous cases are less likely to contain features that distract, and in turn make it more difficult for us to see the heart of the dilemma.

Consider again our Counter-Strike-esque game (where you play as a terrorist or child molester), but this time molestation is depicted with a very high level of graphic realism, whereas murder is depicted at a very low level. Gamers are more likely to consider such instances of virtual murder to be permissible, but not virtual child molestation. So, the dilemma seems to hold. But we can easily point to a possible relevant difference in this case—the level of realism. Perhaps, actively inflicting very realistic depictions of harm on computer game characters is wrong⁹; a justification which would explain our intuitions in this case. However, such a resolution doesn’t really help us figure out why those same gamers still find virtual child molestation to be impermissible, but not virtual murder, in cases where the realism remains the same. So, resolving this non-analogous version of the dilemma hasn’t got us very far; for the resolution is not scalable to all the cases that occupy the Goldilocks zone.

So, it would seem more fruitful to firstly focus our approaches on analogous versions of the dilemma (as Montefiore and Formosa do), and then determine whether such approaches can be expanded outwards to cover non-analogous versions. To do otherwise seems backward, or, at least, provincial. This is because, for many of us, the aim is to find a single solution to the dilemma that will cover all the cases in the Goldilocks zone (the X and Y cases).

And why look for a single resolution? Because, unless we have reason to do otherwise, we should try to maximize explanatory strength. That is, we want as few resolutions as possible to resolve as many instances of the dilemma as possible. (Luck, 2022, p. 1295)

Of course, it may be that in the final analysis a single solution doesn’t exist. But it seems too early to give up on this at the moment.

So this limitation on the dilemma is, for practical reasons, a good one to initially adopt. However, Montefiore and Formosa offer us more than a practical limitation to the Gamer’s dilemma – they also offer us a way to resist it.

Montefiore and Formosa give us reason to suppose that the intuitions that support proposition 1 of the dilemma may not be moral. That is, although gamers may intuit that X cases of virtual murder are morally permissible, such intuitions are only prima facie moral ones. This is because, according to Montefiore and Formosa, such intuitions are not morally grounded. If this is correct the dilemma is resisted insofar as the puzzle is no longer a moral one (in the sense that there is no longer a clash of morally grounded intuitions).¹⁰ This resistance to the dilemma stems from the interpretive flexibility of virtual killing.

Following Patridge (2011, 2013) Montefiore and Formosa suggest that virtual murder may be interpretively flexible. That is, virtual murder takes place in a context that allows it to be interpreted more flexibly—and so can be understood differently to what we might initially think. Virtual murder takes place in computer games, games which have a long history of virtual justified killing; they are places where, typically, gamers play “goodies” who are permissibly killing the “baddies”. As such, a conventional permissibility has arisen regarding most virtual killing.

This conventional permissibility has developed from fixed and multi-directional shooter games like Space Invaders (1978) and Asteroids (1979), to first-person and third person shooter games such as Doom (1993) and Max Payne (2001) and beyond. (Montefiore & Formosa, 2022, p. 11)

So, given that virtual killings typically occur in computer games where they represent morally permissible killings, Montefiore and Formosa suggest gamers are more likely to intuit that virtual killing in general is conventionally permissible.

As a result, the conventional permissibility of justified killing in games may bleed into, and cause, gamers to construe many, if not most, instances of unjustified killing as being (at least potentially) conventionally permissible. (Montefiore & Formosa, 2022, p. 10)

Put crudely, because we are so used to justified virtual killing, virtual killing has become conventionally permissible; consequently, many cases of virtual murder are intuited as conventionally permissible (rather than morally permissible) because virtual murder is a type of virtual killing.

So, on this understanding, the intuitions seemingly supporting 1 (that X cases of virtual murder are morally permissible) are grounded in convention, not morality.¹¹ And, if this is true, the dilemma has been resisted. This is an important result. So, it is worth considering if there are reasons to resist this line.

It is reasonable to suggest that the gamer’s intuitions are influenced by non-moral considerations. However, this doesn’t mean they are completely grounded by non-moral considerations; a possibility that may dilute this resistance to the Gamer’s Dilemma. For if the intuitions supporting proposition 1 are sufficiently morally grounded the dilemma would remain in place. An example might help flesh out this possibility.

The intuitive permissibility of wearing the hijab in Iran (which is currently compulsory in many public places) will probably be more pervasive than in France (where it is currently banned in many public places). This difference might be explained, at least partly, by non-moral considerations (such as the social conventions predominant in these countries). But it may also partly be explained by moral considerations (such as religious convictions and human rights concerns). And it may be hard to separate the two types of considerations; for moral considerations can influence social conventions and vice versa. However, just because these intuitions might be influenced by non-moral considerations, doesn’t mean that the moral considerations in play aren’t sufficient to explain the different intuitions held by the French and Iranians on this issue.

A similar possibility remains in play regarding the gamer’s intuitions. Consider the following point Montefiore and Formosa make regarding the gamer’s intuitions about virtual murder,

…we contend that without killing being a conventionally permissible dimension of videogame actions, virtual unjustified killing, or virtual murder, would not be intuited as morally permissible as readily as it is by gamers. (Montefiore & Formosa, 2022, p. 10)

This is plausible. However, it is compatible with the possibility that the intuitions supporting proposition 1 of the dilemma are sufficiently morally grounded to explain why gamers consider virtual murder to be permissible. Of course, it is also possible that they are sufficiently conventionally grounded.¹² The answer is ultimately an empirical one—however, we may at this juncture be able to consider which theory is more likely.

There are (at least) two competing theories regarding the intuitions supporting proposition 1 of the Gamer’s Dilemma:

I will leave it to the reader to make their own determination regarding which of the above theories is more likely. A determination that will turn on criteria such as simplicity (which theory presumes less), background fit (which coheres better with what we already know) and strength (which is able to explain more). However, I would like to make the following points to help the reader make this determination.

Theory A is stronger than B in the following way. A, by itself, resists the Gamer’s Dilemma (by explaining away the presumption that gamers intuit virtual murder to be morally permissible). While theory B, by itself, does not represent a solution to the dilemma. Of course, B is compatible with potential solutions to the dilemma (for 1 can remain in place if a relevant difference is found), but by itself the puzzle remains. This counts in theory A’s favour.

However, B is stronger than A in the following way. B can explain the intuitions that underpin particular variations of the Gamer’s Dilemma—intuitions that A may struggle to explain. This is because if we replace virtual murder and virtual child molestation with other types of virtual wrongdoings, we still get results akin to the original Gamer’s Dilemma. For example, in the game The Sims,

…players can trap innocent characters in door-less bare rooms for simulated years, in circumstances such, were the game world actual, it would be false imprisonment of a most tortuous kind. Yet we hardly blink an eye. However, if a game were introduced which enabled players to direct their Viking hordes to rape the women of a village, we would be less dismissive. (Luck, 2022, p. 1229)

Likewise, with the game Skyrim,

…players can take very valuable items from innocent characters in circumstances such, were the game world actual, it would be stealing of a particularly egregious kind. Again, few eyes blink. However, if a game were introduced that enabled players to shout homophobic insults at gay characters, we would be less dismissive. (Luck, 2022, p. 1229)

It is not clear that the intuitions supporting these related dilemmas can be easily explained by theory A (at least as it stands). Why? Because virtual false imprisonment and virtual stealing are not instances of virtual killing. There might be another virtual action (instead of virtual killing) that Montefiore and Formosa could point to render virtual false imprisonment and virtual stealing conventionally permissible, but it is not clear at this stage what that would be.¹³

However, theory B is able to explain why gamer’s intuit certain virtual actions to be permissible, but not others. It is because there is a moral difference between these actions¹⁴ and our intuitions can latch onto this difference. And although, as many trolley problems illustrate, our intuitions may be difficult to justify, and sometimes conflict with each other, they may still be morally grounded. In other words, it is because of moral considerations that gamers intuit virtual murder, virtual false imprisonment, and virtual stealing to be morally permissible. Note that my aim here is not to suggest that A is false, and B is true. Only that, in respect to these variations on the Gamer’s Dilemmas, theory B may be stronger than A.

The original Gamer’s Dilemma (where virtual murder is compared to virtual child molestation) seems to be an instance of a larger dilemma (where other virtual actions may be compared with similar results). In which case, many of us will want to develop what Davnall (2021) refers to as a deflationary approach to the dilemma, where we take “the gamer’s dilemma just as an instance of a more general problem” (p. 226).

Although it is not clear whether Montefiore and Formosa’s theory (as it currently stands) is able to resist this larger dilemma, it may not have to. It might be the case that there is one solution for the original version of the Gamer’s Dilemma, and different solutions for other variations. However, this would be surprising (and economically disappointing) given the similarities between such cases. So, some of us will want to (at least until we know better) aim for a single solution to this larger puzzle.

Yet, Montefiore and Formosa’s approach does provide those of us interested in a single solution with a helpful way forward. As it is advantageous to initially approach the dilemma with analogous versions in mind, and then if a solution is forthcoming, expand outwards to non-analogous versions. An approach that should help filter out some of the noise that obscures the heart of the dilemma.