L’Affaire le Laissez-Faire

A little over a month ago Valve issued warnings to the makers of adult visual novels and others which became publicized resulting in controversy and uncertainty regarding the status of games with sexual content on Steam. A few weeks later Valve reversed this implied decision and announced that they would open the store to anything that they “…decide [is] illegal, or straight up trolling.” A lot of commentary seemed to react to, or at least occupy the same space as, Leaf Corcoran’s (of Itch.io) response that the new hands off approach was “ridiculous” and to keep “malicious, derogatory, discriminatory, bullying, harassing, demeaning content” off of Itch. Gamasutra summarized a number of reasonably high profile comments here.

Given that this news is already a month old, the best I can offer here is a room temperature take, but I’m struck by how the focus has been on free expression when I don’t think this is really what is at stake here. For all the effort to form two poles between the Itch and Steam approaches, the main differences are technological. Before weighing in on any of the commentary, I’d like to begin by examining possible reasons for Valve’s actions.

The economics of keeping games on Steam

Valve’s announced laissez-faire approach to Steam should be viewed in light of the original news that there was a coming crackdown on sexual content in games, which invites us to ask why Valve might want to take a game off of its platform. Valve has a pretty good deal going with the Steam where games pay $100 to be listed (recoupable after $1,000 in revenues), and they take 30% of revenues, meaning their variable costs are matched by increases in revenues and their slotting fee ensures the developer is on the hook for at least some of the on boarding/vetting costs of a game that does not sell.

Since the on boarding costs are sunk, why would Valve ever want to stop selling a game? There are likely two sets of explanations: Costs directly relating to the title itself, and knock on effects from a certain class of product. Direct costs are fairly intuitive to understand. For instance, Grand Theft Auto (GTA) seems to be a perennial favourite when it comes to identifying entertainment products bringing about the decline of civilization and so carrying that title also means assuming the liability of angry people complaining that you are carrying a product that brings about the decline of civilization. In physical stores these encounters are diffused among different locations and staff. For a digital storefront they are taking on the sum of complaints which will be directed to their staff who could be tending to other requests from other games, and are doing so on the internet which tends to alter the character of the interaction. Nobody would seriously consider removing GTA from Steam, but dealing with the associated complaints do represent a real cost for carrying that particular game over a less controversial title.

Indirect costs are a bit easier to understand when considering a class of games. A common complaint about Steam now is the discoverability problem: Getting a game in front of its audience when more and more titles are being released. Some genres take less time to create games for and so it is to be expected that there will be a proliferation of low-cost titles entering the market on a daily basis. The indirect costs can be summarized as follows: Valve may take 30% of sales, but if the problem is severe enough that customers are shown games that do not result in sales over games that would have, everyone loses. These costs can be compounded by discouraging developers to release on steam or even create the game in the first place, though it’s doubtful we’ve reached this point (this will be covered later). Forgone sales are one of the more quantifiable indirect costs, but the main point in considering both costs is to show that some titles will be more costly for Valve to carry than others and that it is likely there is a class of product that can be identified as such.

In light of these added costs we can now consider Visual Novels. Visual Novels are relatively low-cost products with a straightforward production model. So long as the game can sell more than the costs of obtaining art, words, and someone to glue it altogether, it makes lots of sense to put it on Steam, and the tools built to create the first game (say a script that links certain phrases with displayed character emotions or a set of templates to produce characters faster) can be reused, lowering the costs of future games. None of this is to make a quality judgement, as there are clearly people who  enjoy these games, and some of them (Dream Daddy, Hatoful Boyfriend, Doki Doki Literature Club) seem to have appealed to players who do not normally engage with this genre, but this is a recipe for overproduction and the potential crowding out of other titles.

A noticeable number of Visual Novels also follow in the fine tradition of independent and European films trying to sell into the US market. Hollywood has an edge on production value and spectacle, and so rival offerings need to offer something to entice audiences away. While I’m sure there are defensible aesthetic reasons for doing so, it is not a coincidence that these films have a fairly high incidence of the leading lady, sometimes the leading man, and sometimes everyone else taking their clothes off. Sex doesn’t just sell, it’s high margin. As with film there are plenty of Visual Novels that do not fit this description, but there are many that do and by their very nature are the most attention getting. Whether this is because the Visual Novel field is crowded or it is just one of the conventions of the genre, a product in this space is more likely to contain sexual content than a product in another genre.

Returning to our evaluation of costs, it’s pretty clear to see how both the direct and indirect costs can be higher in the case of Visual Novels, specifically adult ones. If discoverability is a problem, the economics of Visual Novels are such that we’d expect them to be a larger contributor. More importantly, the higher incidence of sexual content means that Valve will need to spend more time vetting entries to ensure compliance, and likely committing to additional costs as patches will likely command greater scrutiny than a game like Opus Magnum. Finally, it is likely there will be a higher incidence of reporting/complaints with this kind of game, either due to hostility at the lack of quality or with regards to content. Whether any of this seems fair or not, it is understandable for a company like Valve to turn around and say that these games are more trouble than they are worth and to reevaluate their place on Steam. The focus on games with sexual content suggests that Valve was aiming to curtail simple games whose primary draw was titillation, regardless of their profitability. The subsequent focus on the content of these games (as opposed to their volume) is unavoidable, as it seems to have been the only condition for reevaluation, but the motivation here is not personal dislike on the part of decision makers at Valve but the fact that this content was associated with other costs that Valve no longer wanted to bear. This distinction is important as it explains why The Witcher 3 or GTA V did not face the same scrutiny.

The economics of opening up Steam

Just as games have indirect costs, so do policy decisions. A vocal subset of gamers were upset, but this doesn’t carry very much information nowadays. Likely more concerning to Valve were the expressions of uncertainty on the part of game developers. As Amazon, Netflix, and Google will tell you, content is king, and so Valve has a vested interest in being the first place developers think of when it comes to distributing their game. There are strong network effects at play with Steam, and while they can be difficult to unravel, they can do so swiftly. If the speculation in the preceding section was at least somewhat correct, then the uncertainty of the developers who have already been chosen for removal is no big loss, and it’s debatable if this can be read as “First they came for the adult Visual Novels and I said nothing…”, but the publicity resulting from the reevaluations serve as a reminder for something that has always been true: A ban (de facto or de jure) from Valve can destroy your company. A large number of developers are not just big unhedged bets on the PC gaming market, they are bets on Steam itself. If your business model is built around releasing lots of small, low-cost DLC, Valve is one submission pricing decision away from making it unaffordable, and the announcement can come right in the middle of production. If you made small games that can be completed in under two hours, guess what the refund policy did to your revenues. I’m not sure if Valve has ever done anything quite so heavy handed as what was implied by the reevaluations for sexual content, but introducing this level of uncertainty will leave firms wondering if their next title should maybe be a mobile game instead.

The uncertainty created by Valve’s decisions is something of their own making given that they are so tight lipped, but it also made the announcement of the new plans all the more noteworthy. A statement from Valve is also a useful point for a summary since everything above has only been speculation on news about Valve. We know firms were contacted regarding the reevaluation of their games for content and that this became a fairly prominent news story. From Valve we know they were responding to that public discussion, that they are changing course, they now prefer the laissez-faire approach and will collect their 30% while only intervening in the case of illegality and trolling. We’re left asking if this is a good decision and what the consequences will be. Following our reasoning above, this essentially means that the uncertainty felt by developers after the news about reevaluation was more costly than the decision to leave things on the platform. This seems plausible seeing as developer uncertainty can be an existential threat to Steam in a way discoverability is not, as well as the fact that Valve privately contacted the developers for reevaluation but publically posted when responding to the media coverage of that decision. Valve’s post does also spend a lot of time talking about discoverability which inclines me to think that this has always been the focus.

Had the story not been public then eliminating the biggest problem spots (cheap titillating content) was seemingly the lowest cost way of clearing the channels and helping discoverability. The uncertainty created through the subsequent publicity was more costly than the gains from improved discoverability and so the decision was reversed. However, reversing the decision doesn’t solve the original problem and so the second best option is to invest in technology to help mitigate the discoverability problem. The opening of the platform adds another dimension to the problem, but I suspect in part it lessens the burden on the people handing reports and curation as the only real decision rule will be the Valve defined category of “Trolling” which may not reduce the volume of content complaints but can reduce responses to boilerplate “Valve does not curate…” messages.  The case for Valve’s reversal being caused by developer uncertainty is also strengthened by the fact that Valve merely announced the new policy with later implementation, while their usual pattern has been to implement then announce (even if it’s meant letting a negative story fester for longer than we’re accustomed to from other companies).

What does it mean to have an open Steam? The economist Joel Waldfogel has a good and readable summary on the effects of digitization on music, movies, books and television. At the risk of oversimplifying, Waldfogel’s conclusion is that digitization has created a golden age for these media and this is due to projects that otherwise would not have been realized but for the fact that digitization has lowered costs. Conclusions like this normally inspire a dismissive “Well that’s obvious” (if it helps, you’re very smart and I’ll be sure to ask you next time), but analysis like this is valuable because it helps us understand the mechanisms that brought about the outcome. I’ll take a point from Waldfogel and present it as a puzzle: Should the effect of lowered costs of production translate into marginal gains in niche areas, or will it produce unexpected hits that produce substantial gains? Most readers are likely going to know of some indie successes and so say that it will produce hits, but this does not harmonize with the ‘obvious’ conclusion that digitization produces a golden age. To see why consider that under the old model there would be gatekeepers who evaluate a given project and if the returns exceed the cost of production then they would approve it. Lowering costs allows more projects to get made, but these are projects with returns lower than the previous threshold, meaning we should see a proliferation of niche projects, not hits (which would have been produced under the old system). Where do the hits come from? The hits come from the fact that gatekeepers aren’t that good at predicting future outcomes for present day production and so lowered costs are allowing more hits to be realized outside of traditional channels. This is observed in the data through the growing share of self-published/independent/non-traditional projects in revenues or best seller lists. Since the quality or success of a given product is something of a random draw, the gains from allowing more products on Steam have an outsized return even when some of the ex-ante losers turn out to be ex-post losers.

If we think Waldfogel’s analysis has applications to gaming as well, then distribution platforms need to be better at predicting final quality, or the platform should be open as possible to projects that previously would not have passed gatekeepers such as Greenlight or a large publisher. But Steam already seems to have achieved this through the Steam Direct program. Morgan Jaffit’s (Defiant Studios head, makers of Hand of Fate 2) analysis indicating more winners and more losers seems to be exactly in line with the experience of other creative industries as reported in the Waldfogel article (though the an increased share of ex-ante losers in revenues is something of a matter of speculation given the notorious difficulty of estimating sales on Steam). This is why it is important to recognize the mechanisms through which the gains in other media were made, since these gains rest on the unpredictability of outcomes and the failure of traditional gatekeepers to manage it. The most relevant question stemming from the new lassiez-faire approach is whether or not quality guidelines are analogous to gatekeepers in terms of evaluating the value of products. If you feel that they are an impediment to high value products being realized, then Valve’s decision is a good one, will likely be profitable, and we’ll all benefit. If you feel they are largely irrelevant (possibly even orthogonal) to a project’s value, then the best case scenario from this decision would only be the incremental gains that will come from incorporating niche products.

Another way of examining this question is whether or not there is such a thing as a pornographic masterpiece. That is, is there a work of pornography (realized or unrealized) of genuine artistic merit that we have suppressed through these guidelines that would achieve a level of appeal beyond its traditional audience? Most definitions of pornography deny any aesthetic value, but if this is the only argument then we are simply begging the question. The same might be asked of hate speech (is there a work of bigotry of such sublime beauty that failing to publish it would impoverish our culture?), though this would certainly be illegal in some jurisdictions and so still banned. While there have been independent successes outside of Steam (Minecraft or Dwarf Fortress for instance), none of them seem to be products that would have failed to pass the Steam Direct system. These exceptional cases are always dangerous ground to tread on, and so it is possible that the unavailability of traditional commercial channels really has prevented a pornographic masterpiece from being realized in gaming, but it doesn’t seem like a product has emerged that has achieved universal acclaim within its own niche, let alone as a contender for mainstream success.

The reason we care about whether or not the gains from removing content barriers are incremental (i.e. niches) or the realization of mainstream hits is that it is what we should measure against the cost of lowering these barriers. For this we return to the discoverability problem and see just how far apart Itch and Steam really are.

The costs of discoverability problem

If we believe Waldfogel’s analysis applies to Steam then Steam Direct has produced gains above and beyond the incremental benefits stemming from lower costs for games. However, we started with the speculating that the impetus for reevaluating adult visual novels and similar games was the discoverability problem brought about by implementing the more open system. Waldfogel briefly addresses this in the “more research is needed” section at the end, though the idea of information cascades (the Bikhchandani, Hirshleifer, and Welch 1998 reference) does seem to be in line with the “big open” strategy employed by recent successful releases. Waldfogel’s summary does suggest that the cost of search (which we’ve termed the discoverability problem) may not increase, but I am operating on the assumption that this is not the case for games given that developers (who are better informed than I am) describe it and Valve has dedicated resources into mitigating it (and they see almost the entire market).

All of this may seem a bit abstract, so here’s an illustration of what this means. Suppose everyone had an app like Tinder which contained true information about potential romantic matches, but was optional to use and had some search preferences. Let’s say the search preferences start with the default setting of people who are very similar demographically to you and if the love of your life is present in this group, you will find them within two or three swipes. The next setting will put you in a pool of likely matches which, if it contains the love of your life, will find them in 7 to 12 swipes. Finally you can include literally everyone but you will not find the love of your life until 400 to 328,922 swipes. The growing number of swipes represents the increasing search cost as we grow the pool in search of that special m’lady or m’sir. Our choice of search setting is going to depend a lot on our expectations of how likely it is we’ll find our match in each of the groups. The cost difference between the default and second option is more than double, but is broad enough to be pretty appealing to risk averse people who want to maximize their chances. The literally everyone option is so costly that even if we turn the chance into a certainty, we may prefer to reserve it as a last ditch option (or abandon the app entirely).

Switching to Steam Direct seems a bit like switching to the second option in the example above. It does increase the cost of search, but the reward is worth it, especially if we’re looking to expand beyond what’s familiar. But we are unlikely to want the pool to increase indefinitely since the search costs don’t merit the increased chances of finding the love of our lives (game or person). If the benefits of removing content restrictions are analogous to the ones obtained through the reduced costs of games, then we should be willing to entertain higher search costs, but if the benefits are limited to the incremental gains of niche products then we may not willing to accept any additional inconvenience.

A key difference for Steam is that at the moment we are collectively setting one search preference for everyone, and so we’re deciding if we want to go from the expanded category to the literally everything one. It’s true that some people are not able to find the love of their life from the current choice (maybe this explains the desire for pornography), but the inconvenience to this group is considered the lesser evil compared to the universal inconvenience faced by the entire platform and its attendant frustration and likely abandonment by some users. To return to games specifically we can characterize search costs on Steam as the time it takes a customer to find a game they like, the forgone opportunity of Steam to make a sale by serving up the wrong recommendation, and the equivalent lost sale to the developer.

Valve’s new approach involves investing in technology to help lower search costs, including allowing them to restrict the pool of potential games they are exposed to. I don’t think anyone really knows how successful they will be at it. If Valve succeeds, then we will all benefit through faster matches to games we enjoy the most. If the system doesn’t work as well as they plan, then I suspect we’ll see a few more applications of the “trolling” criterion to keep users from encountering offensive content they wouldn’t buy in the first place. When looking at the Steam Direct guidelines, it becomes apparent that Valve hasn’t really given up a lot since half of the rules are already legal matters, and the remaining can credibly fall under a the trolling criterion. In fact, Valve offered some clarity on the condition and used a banned school shooting game as an example of what would not be permitted.

This brings us to Itch. Itch’s submission guidelines are extremely permissive, a fact that tends to be lost when you see people who have never used the site complaining about censorship in response to its founder’s Tweet. Itch also does not have the resources that Valve does in terms of being able to automate the curation of a given user’s pages and so they are less likely to propose a technological solution to the problem. Itch faces the same issues of discoverability and desire to maximize revenues as Valve but needs to employ solutions that reflect its capabilities. It is difficult to find a more direct solution than saying ‘don’t put this kind of stuff on our site and if you do we’ll ban you.’ If anything this is likely more a timely reminder of Itch’s existing policy rather than any kind of policy shift (so far as I can tell, the rules have not changed substantially since Corcoran issued his tweet or, indeed, years before he did). In fact, based on the clarification Valve offered to the definition of trolling, there seems to be very little difference between Itch and Valve in terms of what they consider (or will consider) unacceptable on their platform.

If it turns out there are substantive differences between Itch and the future Steam, it will be an interesting experiment in terms of whether or not these content guidelines actually do prevent the realization of good games. At the moment Steam is not my go-to source for interesting and unknown games but Itch is. The free speech warriors reacting to Corcoran’s tweet seem not to have been interested in the relative freedom of Itch’s platform to Steam Direct and Greenlight, and Itch has done fine without them. Gaming as we know it has also not collapsed under Valve’s relatively more restrictive policies, and Itch’s ban buttons are not likely to steer the industry into some kind of Hays Code. In truth, most if not all of the percieved difference is likely explained by Corcoran’s directness and Valve’s vagueness.

It’s fun to wrap oneself up in the illusion of a principled stand on free speech or against intolerance, but there really isn’t a lot for either side beyond rhetorical posturing. People who want to see the Valve decision an unmitigated win for free speech need to reconcile that with Valve’s follow up comments which seem to be banning exactly the kind of titles we’d expect a reasonable set of content guidelines to ban. Likewise, Itch’s permissive submission policies do not seem to have reduced the storefront into the dumpster fire predicted for Steam, nor has it seemed to produce the controversial hit that has demonstrated the outsized welfare gains from removing content restrictions. This is why I think the difference is merely technological and if anything Steam is shifting closer to Itch. The small scale experiment of Itch also suggests that the gains from loosening content guidelines are incremental for their respective niches rather than an opportunity for previously censored hits to finally be realized.

Seeing the realization of Valve’s promises will make all the difference, but my own expectation is that this decision will result in much needed curation tools being made available to users and developers, pornographic content will be made available (though the most egregious examples will likely still be banned) but will need to be opted in, and there may be some improvements made to the recommendation engine. Because the new policy will likely mean the entry of niche products, a ‘first do no harm’ approach would mean that most of the changes should be imperceptible to users, since these are products they have no interest in buying in the first place. So long as Valve is able to ensure these niche entries aren’t a nuisance to the rest of us, it’s hard to see this move as a bad one, but it hinges on both the new entries not being as bad as the worst that people assume (the hate speech case) and Valve being able to segment the user base in a way that doesn’t cause games we would have otherwise enjoyed becoming invisible to us. Of course, if Valve finds this problem is much more difficult than they anticipated, they have left the details open enough to make the changes modest and we may find the store isn’t all that different from the way it is now.


Note on affiliate links: I have an affiliate status with GOG.com for which I am given a portion of sales for traffic I drive to the site. The inclusion of a given title is for illustrative purposes first, but when it is available on GOG I will provide such a link. Naturally I encourage you to do your own price comparison or buy on your preferred platform. I include, on occasion, affiliate links from other broadcasters to support people who helped me in a given post or the cast in general. For this article, Humble Links support the streamer JessyQuil.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s