Aaugh rather than Arragh: Windows Piracy

Bill Gates once famously argued that if people are going to pirate software, he’d rather them pirate Microsoft’s. This is hardly surprising in that Microsoft products run on the majority of PCs in China (traditionally a poster-child for software piracy) and, by design or accident, has meant that Microsoft is a leader there with its network externalities base likely to keep others, including Google, at bay for some time yet. This is especially the case in Microsoft’s enterprise area where, as Chinese companies secure US and European export contracts, they must establish that they are compliant with international intellectual property agreements and come to Microsoft in pursuit of a paid license. When you have an export contract, it is likely that paying Microsoft is a minuscule fraction of your IT costs has, all of a sudden, become cheap. Indeed, put in these terms, this is almost a stunningly efficient arrangement. Use the software for free and then, when income comes in, you pay. It is like a corporate income-contingent loan that is somehow enforced by others!

Anyhow, the issue of piracy and who pirates and why continues to fascinate. This is why a new paper by Susan Athey and Scott Stern is so interesting. It was written while each was affiliated with Microsoft Research and so is the first of these papers I am aware of to be based on Microsoft’s own data. They examine the piracy of Windows 7. Specifically, because Microsoft continues to interact with pirated versions of its software (all the user sees is a desktop saying the copy was pirated but otherwise can use the product and even receive updates), their anonymized data allows them to look at different patterns of piracy across countries.

The first interesting finding is that reports that a copy zero hack was responsible for pirated copies was largely true. Before the release of the final version of Windows 7, one of the original equipment manufacturer master product keys issued to Lenovo was hacked and released online. Basically, users could then easily pirate Windows 7. The key was discontinued but Microsoft allowed users to register with the key. The paper finds that this seed was responsible for most of the pirated copies in their dataset. The point here is that, if it is the case that only a few instances of hacking occur, then the hacking is not arising at the level of individual consumers. That means that trying to dissuade individual consumers from investing time in hacking by offering free limited functionality products is likely to be fruitless. This also implies, as confirmed by the study, that when Windows is not pre-installed (as it is for leading OEMs), then piracy is likely to be much higher.

The second finding regards the characteristics of the countries with greater piracy rates. Here is a graph of piracy incidence.

So, for Georgia, 80 percent of machines have a pirated copy of Windows 7. So what drives piracy? It turns out that institutions matter. Countries with strong legal and political institutions have less piracy, and controlling for this, GDP per capita doesn’t have a significant impact on the piracy rate. In addition, if you have faster broadband there is more piracy (no surprise there) but the more innovation oriented the country is the less piracy there is. This is consistent with my ‘export-driven success’ theory sketched above.

A third finding is that most of the piracy was of Windows Ultimate. Basically, if you are going to pirate, pirate the best. This is good news for Microsoft in that, absent price considerations, users do actually want to install the best version on machines even if it is much larger as a download. But it is also bad news in the sense that it may be Microsoft’s use of price discrimination of different versions (something Apple doesn’t do — well, its free so it is hard to tell) that may drive some consumers to piracy.

The final interesting finding is that IP enforcement does not seem to have any impact on piracy rates. This is no surprise to anyone who understands how the internet works but may be a surprise to policy-makers and corporations lobbying for anti-piracy measures. So when the UK banned the Pirate Bay in 2012 here is what happened to piracy rates:

See that? Nothing. Zip. Nada. In other words, anti-piracy measures are possibly the most ineffective policy instrument ever. Why? Because the Internet.

The outstanding issue that the paper does not address is the question of how much is Microsoft losing as a result of piracy? If Microsoft were able to enforce DRM and cause machines with Windows 7 to become inoperable, how many would purchase a license? One argument has always been that it is ‘ability to pay’ rather than ‘willingness to pay’ that is a limiting factor in the potential losses Microsoft faces. In other words, a pirated copy removed does not translate into a paid copy because the users don’t have the money. The analysis here suggests that that isn’t the case. Instead, it is a permissive culture that is leading to increased piracy. That means that the losses to Microsoft could be potentially large.

That said, the institutional environment cannot be controlled by Microsoft and, as we have seen, it takes one talented hacker to release the beast. In that world, if you are running a platform, you don’t want to version in a way that isolates some users; to do so would reduce the value to all due to network effects (see this paper by Andre Veiga for a proof). In that sense, Microsoft, whether by design or not, has engaged in a perfect platform strategy in the face of those institutional constraints. In that respect, given constraints, one suspects that potential losses are possibly minimal.