Ed Felten has an interesting post (“Freeing the Xbox”) about efforts to “hack” the Xbox to open it up so that it can run the Linux operating system. And Tim Lee at the Technology Liberation Front has a series of related posts on this issue. Ed notes that given the price/power combination of the Xbox, it is an attractive target. How we should we evaluate this? Here is what Ed says:
The real action here is in Microsoft’s strategy of selling the Xbox hardware as a loss leader, and the tendency of the Xbox Linux work to frustrate this strategy. Xbox Linux creates value for its users. Should public policy be willing to destroy this value in order to enable Microsoft’s pricing strategy? My instinct is that it should not, though there is a plausible argument on the other side.
This is a reasonably balanced statement, but I don’t think that this is still the right way to frame this issue, and I do want to make clear the other side. Ed puts it as users vs. Microsoft, but this is as least as much about user vs. user and we have to understand what the Linux hack means for people playing games on the Xbox.
I think that we have a number of moving parts here: (1) monopoly v. competition in the games market; (2) selling the game platform—the Xbox—for above or below marginal cost; and (3) Microsoft’s design alternatives to the Xbox. Take these one-by-one.
If Microsoft has monopoly power in the games platform, hacking the Xbox might be a good thing. This would create, as Ed puts its, value for Linux users and would, as an economist would put it, expand output. As reduced output is the central harm of monopoly, responsive steps that expand output often improve matters. I don’t have a professional opinion about whether Microsoft has monopoly power in games, but I doubt it, especially when we focus on competition in games platforms over time. We have seen a lot of market-leader turnover in that market, which is usually a sign of meaningful competition.
If Microsoft is pricing the Xbox above the marginal cost of making it, Microsoft should be delighted to see users find new uses. If I decide that the Xbox is the world’s finest ottoman, Steve Ballmer should be happy. But if Microsoft is pricing the Xbox below marginal cost—I assume that we think that—then we have an issue, and this gets us to Ed’s central point. Ottoman users won’t buy any games, and they will inflict costs on Microsoft through their purchases.
We are now in the world of systems—or platform—competition. Multiple pieces work together. This is very much in the news right now given the French threat to regulate iTunes and the iPod, which is exactly this sort of system. The Financial Times had an editorial on June 15th ($) supporting Apple’s effort to build a closed system; I have a prior post on this here.
Pricing is a tricky issue. Adobe gives away Acrobat Reader so that it can sell the Acrobat authoring program. Gillette gives away razors—or sells them below cost—to sell blades. Selling the platform—the razor or the Xbox—for below cost can be in the collective interest of the users of the platform.
Try a concrete example. Light game players value the Xbox with one game at $200. Heavy users value the Xbox with multiple games at $300. The marginal cost of making an Xbox is $200 and the marginal cost of producing each game is zero. If Microsoft has already created the Xbox platform and the games, what would we like to happen? The marginal costs of serving both types of users are covered by their values—especially if we add epsilon to the light user’s value—so we should want both types of consumers to be served.
How do we do that? Assume that we set separate prices for the Xbox and for games and that we have six different games created. If we set a price for the Xbox that covers marginal costs, we set the price of the Xbox at $200. That means that we need to set a price of zero for the games, otherwise the light user will not buy at all. If we do that and if we have one light user and one heavy user, total revenues are $400. Of course that means we have just covered marginal costs and have received nothing to cover the fixed costs of the system.
Try a different pricing strategy. Set the price of the Xbox below marginal cost—say $150—and raise the price of games to $25. We can sell one Xbox and two games to the light user and one Xbox and six games to the heavy user. That gives us a total marginal cost of $400 and total revenues of $500. We are making progress, as now we are recovering some of the fixed costs of creating the system. With game prices necessarily about marginal cost, game purchasers contribute to the overall cost of creating the platform. The more games that you buy, the more that you contribute. Whether we can pay for those costs depends on the number of light and heavy users.
Now introduce a third group to the system, Linux users and the ottoman set, who buy one Xbox and no games. For now, assume that they value the Xbox at $189, $11 less than the marginal cost of serving them.
Having them buy the Xbox is a social bad, not a social good. They value it at less than the marginal cost of making it, so we don’t want them to buy. But allowing them to buy limits how we can allocate prices across the Xbox and the games. We can’t set the Xbox price at $150 anymore, as we need to boost prices of the Xbox to get rid of the ottoman types.
So set a price of $190. The Linux/ottoman users drop out of the market. What price will be set for games? At a price of $10, the light user buys the Xbox and one game, the heavy user the Xbox and six games. Total revenues are $470 against marginal costs of $400, giving rise to $70 that we can use against the fixed costs of creating the platform and the games.
But our platform creator would charge a higher price for games. Charge $190 for the Xbox and $18.33 for each game. The light user drops out of the market. The heavy users buys the Xbox and all six games for a total price of $300. Marginal cost is $200, so the contribution to fixed cost recovery is $100.
Note what has happened here. The mere presence of the ottoman users has pushed light users out of the market. Our ability to sustain the platform is the same—with $150/$25 pricing and with $190/$18.33 pricing we get $100 towards fixed cost recovery—but we have lost users. We had to raise prices of the platform to get rid of the Linux/ottoman users, but then we would have had to set low prices of games to keep light users in the market. If those users create benefits for third parties—if light users value use at $200 but they don’t capture the full value of their use, so someone else gets $5 from their use (not implausible in a world of network externalities)—we have inflicted a loss on society.
A bunch of numbers already, so I won’t do too many more. If Linux/ottoman users actually value the Xbox for more the marginal cost of creating it, getting the Xbox into their hands would be a good thing. The $150/$30 pricing structure wouldn’t be sustainable in the face of Linux users who valued the Xbox at $211. Prices would rise at least until we hit marginal cost and that would push the light users out of the market. But we would just be trading one group of users for a second and, at least in the framework of the example, I have no basis for preferring one group of users to a second.
The point of this (extended) example is to try to make clear that our focus needs to be on the consequences of opening up the Xbox for users as a whole, conditional on how Microsoft will respond to the knowledge that the system can be hacked. That means that Microsoft will take that possibility into account in its pricing decisions, and may do so for its design decisions. We may start to see a preference for services over products if freestanding products are more permanently hackable than services which require repeated interaction.
This has been pretty complicated already, but consider other issues. I suspect the Linux crowd would say that Microsoft has closed the Xbox because Microsoft was concerned about creating a cheap computer that would run something other than Windows, and if that is right, we would need to take into account cross-platform competition. And both the Xbox gaming market and Linux market are characterized by network externalities, so we have spillovers to deal with.
Finally, on innovation, two points. First, innovation isn’t an end in itself; it is just a means to whatever metric we are using to gauge social welfare (utility, jollies, happiness). Second, as I hope the example makes clear, we need to focus on all users. In the first example, the possibility of opening the Xbox could lead to the loss of one set of users, without adding any other users, and the lost users could be an important source of innovation. Who knows where the next Red vs. Blue will come from?
[Disclosure: The Law School has received grants from Microsoft in the past, and, I believe, continues to do so.]