Localization and Censorship

May 23 · 17 min read

There are often differences between video games as they are published in different locales. Until fairly recently (and I am not sure of the current status of things), it was illegal in Germany to have Nazi symbology in media, so Nazis had to be replaced with Illuminati cultists or the like. Myth has it that you’re not allowed to have skeletons in media published in China. This doesn’t seem to be an official policy, but it is policy that you’re not allowed to mention Tibet or Taiwan. But most importantly, anime tiddies seem to consistently be censored in American publishing. (While I will continue to use this phrasing, this article will be SFW and otherwise perfectly academic.)

In the wake of news that Alabama Public Television is refusing to air an episode from Arthur that features a gay couple, there have been some comparisons to the claimed censorship of anime tiddies. The logic goes that you can’t be angry at Arthur being censored if you’re OK with anime tiddies being censored. This argument seems to hold intuitive weight to most people (even those who rightly reject it as nonsense), and my goal in this article is to expose how the misuse of the term “censorship” obscures the very significant differences between Taiwan imagery, Arthur, and anime tiddies. I hope to show that our common assumptions of defining censorship in terms of market restrictions are insufficient, but that a different sort of framework explains our intuition. I will finish with the hot take that censorship is not actually bad.

What is Censorship?

The first step to solving a problem is to define your terms. Let’s try to define “censorship” meaningfully, as it pertains to the localization process. Here are three ways that I see the word being used in games discourse, by various sides of arguments.

  1. Censorship is when localization removes content from the original media, for any reason. Let’s call this “Nintendo localization”.
  2. Censorship is when localization removes content, because of government policy. Let’s call this “Chinese localization”.
  3. Censorship is when localization removes content, because of the policy of another organization. Let’s call this “Sony localization”.

Of these, 2 is the most “legal” notion of censorship, and describes the regulations imposed in Germany or China. 3 is a reasonable definition at first glance and seems to apply to the case where Sony, as a platform holder, forces PS4 games published in the US to cut down on sexual content. These are often offered as counter-definitions to deflect accusations of censorship, and I used to hold to 2 until recently.

1, however, seems to be the definition of censorship as it is used by the more “anti-censorship” contingent of video game discourse. Consider the case of the localization of Fire Emblem Fates. Nintendo had full control over this localization, with no other company breathing down its back. The US government was also not involved. Yet this didn’t prevent gamers from decrying censorship in the localization in response to the modification and removal of voice lines, character personalities, and swimsuit costumes. This contingent organized under the hashtag #TorrentialDownpour, and a large collection of their complaints and comparisons can be found in this imgur album. At least some of their complaints are reasonable, and some of the text comparisons are incredibly interesting as a testament to how destructive — or transcreative — localization can be. Regardless, the notion that it is even possible for Nintendo to censor its own work is strange.

Our first task is to problematize all of these definitions, using the simple model of rational agents in a capitalist market.

Nintendo localization

Suppose that I am interested in selling cakes. I set up a cake shop in my town. However, it appears that people in my town do not like cakes, and prefer pies instead. As a result, I rebrand to a pie shop, and my new pie-selling business is successful.

When we talk about localization, we must remember that it is ultimately an act embedded within a market. Nintendo’s goal, both when creating the original game and when localizing it, is market success. What exactly this amounts to depends on the company, but some critical factors are sales and popular reception. In the case of Nintendo, we must observe that Nintendo has a well-established image of a family-friendly brand outside of Japan (perhaps in Japan as well). Nintendo games are the kind that you can play with your kids, or let them play on their own. Part of “market success” for Nintendo specifically is maintaining this brand image. (A relevant video on why Nintendo games never go on sale.)

Any act embedded within a market is obviously dependent on the conditions of the market. As I gave up on cake-selling because nobody wanted to buy cakes, we should expect Nintendo’s localization decisions to be controlled by the conditions of the target locale. Thus, the only dynamic at play is the basic dictum of capitalism: if you want to be successful in the market, you have to deal with the demands that people have.

Now, let’s put our observations together. Nintendo wants to maintain a family-friendly brand image. In the United States, even vaguely sexual content is generally not considered family-friendly. Therefore, by basic market logic, Nintendo should not publish games with even vaguely sexual content in the United States. Nintendo-style localization is in fact simply the self-interested action of a rational actor engaging in capitalism.

Other outcry about Nintendo’s localization practice ultimately reduces to this: Nintendo is engaging in capitalism and nothing more. If Nintendo changes character dialogue or personality (see the aforementioned imgur album), it’s because they think it would sell better as a result.

*This is not to say that these decisions are always correct. It may occur that, in fact, the American market is highly ripe for sexual content now that the Rockstar moms have grown too old to see them. (More on Rockstar moms later.) One famous example of such an error is the long-standing assumption that Americans were not tough enough for Japanese games, which led to the easy-ification of English localizations (Carlson and Corliss 2011). In fact, returning to Fire Emblem, long-time American fans have accused Nintendo of dumbing down the localizations too much in recent years (it is true that Fire Emblem localizations have added features like easy modes and disabled permadeath), and one persistent claim has been: If you made the localization as hardcore as the original, then you would get more worldwide sales. This is in fact a (possibly valid) claim as to market strategies in general. Do you go for a niche (“hardcore”), or do you make a widely appealing product? The fans seem to think that Fire Emblem would perform best as a niche game in all locales. Nintendo seems to think that Fire Emblem is best as a niche game in Japan and as a popular game outside Japan. It is important to recognize that these questions are empirical — that is, we’d have to collect data and perform science to come to a justified conclusion. It is an empirical assertion that publishing games with anime tiddies in the US would hurt Nintendo’s family-friendly brand image (but a seemingly obvious one); it is like wise an empirical assertion that Americans are not tough enough for Japanese games (but a seemingly silly one).

Sony localization

It is commonly thought that Sony forces PS4 games published in the US to restrict the quantity of anime tiddies. This was brought up recently in discourse on the modification of CG for localizations of Dragon Star Varnir, announced in advance on the localizer’s website. (I had the honor of citing this so-to-speak debacle in an actual academic paper.) To my knowledge Sony’s policy has not been publicly announced, but it seems reasonable to assume it exists, and here I will assume that it does.

Our question is now: does this addition of “force” — the explicit statement “You will not be permitted to publish unless you do X” — change the dynamics of Nintendo localization? Ultimately, a key part of the basic market dynamics in Nintendo localization is that nobody is forced to do anything. Nobody was preventing Nintendo from publishing anime tiddies, but Nintendo decided that it would not be a good look.

At a basic level, Sony localization still falls under the basic model of market capitalism. The game developer must select a target market for which to produce as in the Nintendo model, but at the same time must select from a publisher market, the most dominant of which are Sony-PS4, Microsoft-XBox, and Valve-Steam.

However, this act of selecting a publisher is in principle not different from selecting any other part of the development pipeline. What computers will you buy for your employees? Where will you rent office space? What food will you cater for lunch? Which publisher will you partner with? These questions are to be asked within capitalist markets where each product has its own tradeoffs. Maybe you rent office space in New York, but this prevents you from accessing the tech-savvy environment of California. Maybe you partner with Sony, but this prevents you from putting certain sexual content in your game.

There are similar questions that are not asked within capitalist markets per se, but are still subject to rational decision-making. For example, deciding on a programming language is a key starting point for software development. There is no market (programming languages are generally free — except in game development!), but there are still tradeoffs. Python allows fast development but is slow; Go is slow to code but assures safety; C is incredibly efficient but will blow up if you make a single error. We tend to think of these situations in terms of rational choice.

Thus, while Sony localization does involve force, the decision by the developer to partner with Sony in the first place was in principle a tradeoff they willingly made, with no force involved. And according to standard market logic, if they don’t like Sony’s anti-anime-tiddy policy, they could have simply selected a different publisher. Thus, there is in principle no difference between Nintendo and Sony localization, as both fundamentally reduce to rational actions made within markets, and thus neither are actually censorship.

Hopefully my abuse of “in principle” has ticked you off — this analysis is incomplete. We will need to call upon the ghost of Foucault to complete it, and a short discussion of Chinese localization will show why.

Chinese localization

Here, the question is: does the fact that China explicitly disallows certain political topics in media count as censorship for the purposes of localization?

Following the market logic of our previous two sections, we should actually answer: no. The logic is the same as for the Nintendo model. The market in China does not respond kindly to discussions of Tibet and Taiwan, so the decision of a localizer to omit references to them in a Chinese localization is simply self-interested rational action within a capitalist market.

This example should strike you as severely problematic. How is it possible that selling pies to people who don’t like cakes is described in the same way as omitting references to Taiwan in your media under threat by the Chinese government? This is because models of rational capitalist action do not take into account the power dynamics that underlie the generation of market conditions. In other words, the difference between “people don’t like cake in this town” and “the Chinese government won’t accept references to Taiwan in our game” is a matter of power, to which we now turn.

It’s All About Power

I hope to have shown two things in the previous discussion. First, I analyzed the common definitions of censorship in a model of economic rationality. Second, I showed that this model and the question of “force” are insufficient, and that our intuitions on censorship really return to the question of power relations.

Intuitively, you might consider that Chinese localization is wrong because it is top-down, or “imposed” in some respect, whereas selling pies is perfectly normal because it’s simply reacting to a natural state of affairs. This is the correct intuition. To clarify it, we should ask: who is determining the market conditions, and how and why are they doing it?

In the case of Chinese localization, the market conditions in question are determined exclusively by the Chinese government. They are enforced via coercive methods, and their purpose is to maintain the cultural stability of the PRC.

In the case of selling pies, the market conditions are generated by the random organization of preferences in a town. Perhaps the organization of this town was determined by a certain city planner, but the fact that my cake business failed is a side-effect and does not implicate the city planner. In other words, nobody knows and nobody cares.

A short examination of power dynamics thus reveals why Chinese localization is objectionable: we have some opposition to exclusive government control of media, coercive enforcement, and propagandistic orientation. The reason why governments are special targets is because they seem to be the only agents capable of coercive enforcement.

Our analysis of Nintendo localization also becomes more interesting. Consider the specific problem of Nintendo censoring sexual content for American markets because it is not “family-friendly”. American puritanism, which is what makes sexual content “problematic”, was historically developed and is maintained by a contingent that uses it for sociopolitical power (on a related note, ask yourself sometime why the GOP is so intent on propagandizing abortion). This makes it surprisingly comparable to Chinese localization. However, we should identify two major differences: it is not coercive, and the relation between puritanism and anime tiddy is indifference. That is, political puritans do not really care about sexual content in video games (like how the city planner does not care about my business), whereas China is exceedingly concerned with references to Taiwan.

(I claim that American puritanism is not concerned with controlling anime tiddy primarily because the sexual objectification of women is generally permitted in American culture. This is the point on Rockstar moms: the people who were most involved in opposing sexual content in video games were conservative parents, pundits, and local church officials, who are products, not sources, of political power. Of course, their outcry can lead to actual political change, but rarely anything of substance. But when there are gay people on television — then judiciaries, state legislatures, and Congress take action without prior public outcry. It is the action of government in silence that reveals the powers at play.

Another reason to believe that the relation is “indifference” is that the sociopolitical contingents which drive American puritanism are largely state apparati in the central and southern US, the GOP, and Murdoch-owned media. However, none of these contingents take part in localization, which is something that occurs largely on the coastal US (coasts offer maximal proximity to other countries, and California for this reason is a centerpoint of Pacific Rim flows), where the GOP and Murdoch have little direct control. Thus, if puritanism affects the localization of sexual content, it is unlikely that puritanism is anything more than a background force.)

I think that this point on indifference is the most important. It’s only meaningful to talk about censorship when the “source of power” exerts active control, rather than indifference, over the market agents. Thus, I would say that American puritanism effects censorship with regards to discourse on sexuality, which it is intent on controlling, but not with regards to anime tiddies, to which it seems to be indifferent. In both cases it is non-coercive. This makes the Arthur episode doubly censored: it is first censored by the Alabama government (in full coercive Chinese style), and second censored more generally by American puritanism, which forbids discussing The Big Gay.

So What About Sony?

Sony localization is the most interesting case because the power dynamics are unclear. As we stated, Sony localization is the result of developers making a choice about their development pipeline “in principle” in the same way that they decide on a programming language. But there is clearly a difference between Python’s low speed and Sony’s content restrictions, and it comes back to the question of intention and indifference.

Specifically, while the design of Python exerts restrictions on the types of artistic expression you can produce with it, the designers are indifferent to the product of these constraints. In fact, the relationship between choice of programming language and limits on artistic expression is so contorted that it’d be impossible for any language design to construct meaningful limits. On the other hand, Sony’s content restrictions are designed with the express, direct, short-range, intentional purpose of restricting the types of artistic expression that can be produced when partnering with Sony.

What complicates this picture is the fact that Sony is one of alternatives: you can always go to a different publisher. In the cases of Nintendo and Chinese localization, not dealing with power meant losing out on a market, but here, not dealing with Sony just means publishing on Steam instead. In other words: If you don’t like McDonalds, go to Burger King.

I will give a brief outline of how this problem might be solved. In addition to discussing the power dynamics of Sony’s content restrictions, we would want to discuss the power dynamics of content restrictions in the entire publisher market. We need to ask what it means that McDonalds and Burger King are the only two options, and recognize that even in a supposedly free market, there are still no good-quality burgers. Since there are only a small number of publishing platforms (Nintendo, Playstation Store, XBox Store, and Steam basically exhaust the market for non-mobile games), we might also consider oligopolistic analysis. Tentatively, I think that the oligopolistic aspect is key. The fact that there exist many people for whom the PS4 is their sole gaming system means that there is in fact no alternative to Sony for developers who wish to target this market. The exercise of why Windows or Linux are therefore optimal as gaming platforms is left to the reader.

Tying Threads

Under the logic of market rationality, all three of Nintendo localization, Chinese localization, and Sony localization are explicable in terms of rational agents making self-interested actions. If we want to understand why only some cases are objectionable, we need to analyze the power relations in question, which are more complex than the three basic categorizations.

The power dynamics generally associated with Chinese localization are government coercion and explicit control of culturally-relevant topics. The removal of Arthur from public broadcasting in Alabama is an example of this, but the systematic suppression of nonconforming media in China is a more classic example, and involves more thorough coercion.

The power dynamics of Nintendo localization are widely varied. There may be a political design, with either explicit control of culturally-relevant topics (eg. control of discourse on sexuality), or control that really turns out to be unintended side-effects (eg. control of anime tiddy). Generally it is not coercive.

The power dynamics of Sony localization are unclear and require market analysis. Specific providers like Sony generally have explicit intentions in controlling the artistic expression of their partners, but we need to analyze the market as a whole to determine whether developers have a meaningful “choice” in the matter. In the case of video game platform publishing, I expect that the oligopolistic structure would reveal that developers have little choice and in reality are constrained by the whims of Sony.

If we want to give a functional definition of censorship that can distinguish these cases, I propose that we consider censorship whenever market conditions are designed by a sociopolitical force with the goal of limiting a certain kind of speech. (This notion of censorship applies to market-based speech, such as localization of video games, but would need some tweaking to apply to personal expression, which is not constructed in a market.)

I do not think coercion is necessary, since methods like public education and media propaganda can end up blocking out certain viewpoints from public discourse. In the United States, both leftist discourse and Naziism have been (perhaps no longer) censored in this respect: the government won’t stop you, but the media has inculcated into everyone that VENEZUELA BAD, so for many years you had no other choice but to bow to the idols of Thatcher and Reagan while in public. Obviously, though, coercion makes censorship more severe and permits less dissent: we might talk about hard and soft censorship in this respect.

What does this ultimately mean for anime tiddy? It’s reasonable to say that the removal of sexual content in Japanese->English localization occurs due to the assumption of a certain American puritanism, which is a sociopolitical force. However, this is a consequence to which puritanism is largely indifferent, and I thus hesitate to call it censorship. Various other modifications may be more or less like censorship. Changing characters’ personalities is a response to basically impenetrable cultural stereotypes, and is akin to selling pie. The case I would yield is the modification of certain character dialogue. Consider this example from the Fire Emblem Fates imgur album:

Left: Japanese original. Middle: Literal translation of original. Right: Localized version released in the US.

As an American feminist, the original dialogue seems to me to be rapey and homophobic, and as a localizer, I would refuse a direct translation on the grounds of it being culturally unacceptable. Feminist and queer criticism in the US are oriented directly towards criticizing and reducing these kinds of media representations. We could say that feminist criticism censors a direct translation and leads to the incomparable recreation on the right. I would agree with this, and it leads into our final point on censorship.

Censorship Isn’t “Bad”

From my claim that Naziism is or was softly censored in American discourse, you may have considered that that’s a good thing — isn’t it beneficial that we don’t spend several hours each day discussing whether nonwhite races and the Jews need to be exterminated? Not censoring Naziism would make public discourse much less meaningful.

Giving a content-agnostic definition of censorship, propaganda, or similar terms reveals how seemingly innocuous organizations also engage in these supposedly evil practices. Consider scientific, medical, and legal discourse. In these discourses, whatever you say must accord with scientific standards, and these standards are coercively enforced (you may lose your grants, your license, etc if you break them). Scientific discourse censors non-scientific speech much in the same way that China censors Taiwan-related speech. If there is a moral difference, it is in the consequences; scientific censoring establishes scientific rigor and enables scientific progress (good things!), whereas Chinese policy creates artificial cultural stability and suppresses “free thought” — but also enables efficient capitalist growth in a global economy (we might call this Lee Kuan Yew’s Wager for its well-known application in Singapore).

In Foucauldian terms, power is productive. As we cannot say that power is “inherently bad” (this is a naive misreading of Foucault), it would also be wrong to say that censorship — under a content-agnostic power-based definition — is “inherently bad”. Rather, censorship, just like power, is necessary for meaningful discourse.

You can take this as a pointer for how a notion such as “freedom of speech” (ie. no censorship) is fundamentally contradictory with any sort of productive discourse (eg. scientific rationality), which requires the establishment of power structures and the censorship of “non-science” in order to function. From Aihwa Ong’s Flexible Citizenship:

But, as Barry Hindess has argued, there is an ambiguity in the doctrine of liberalism that poses both a political community of naturally autonomous subjects and a political community that promotes rationality, self-control, and responsibility.

While I consider the notion of censorship useful — as it exposes a common method of discourse control — I try not to use the word, as it is too emotionally charged with images of Evil China. The same goes for propaganda, which we in practice apply to Evil China but never to the American military.

So — is localization censorship? Sometimes. But that’s not morally meaningful. For censorship to be good or bad, we must return to its process and its consequences, rather than fixating on a fairly common power structure. The common error people make is assuming censorship is bad, and defining it post hoc to cover only the types of censorship they dislike. If you press on the definition and show how it encompasses eg. scientific discourse, then their position falls apart. Instead, we need to put the moral convictions front-and-center, and show that it is those moral convictions that really construct how various types of censorship are acceptable or unacceptable.

Other Reading

Mandiberg, Stephen. 2017. “Fallacies of game localization: Censorship and #TorrentialDownpour”. The Journal of Internationalization and Localization 4:2.

  • For more about Fire Emblem Fates and how localization involves the “original creator’s vision”. There is also a trenchant observation which identifies how even selling pies — that is, responding to completely natural states of affairs — may ultimately be problematized:

If games are localized to target the largest possible audience, they are necessarily translated to fit into dominant discourses. In so targeting the game these localizations reproduce normalized discourses and politics of cultural identity whether they are good or bad (Venuti 1998). By continually erasing foreign elements in the process of localizing games the localization industry runs the risk of never, to quote Venuti, “creat[ing] a[n] [audience of players] that is more open to linguistic and cultural differences” (Venuti 1998, 87). Such an eventuality would be just as much of a “scandal” as censorship.

Carlson, Rebecca and Corliss, Jonathan. 2011. “Imagined Commodities: Video Game Localization and Mythologies of Cultural Difference”. Games and Culture 6(1).

  • If you’re interested in the mistakes made in localizations, this article talks about the mythologies of assumptions localizers make about “American gamers” and “Japanese gamers”.

Ong, Aihwa. Flexible Citizenship. 1999.

  • This book has almost nothing to do with localization, but it inspired my power-based perspective, and my analysis would have been pretty garbage without it.


Written by


Software engineer, epic gamer, and student of Barthes and Skinner.