The recent announcement by Linux kernel developers that open source hardware drivers are the way to go got me thinking. You can make a business case for open source device drivers (you'll sell more hardware) or you can make an ethical case (it's the right thing to do). I believe in both, but for me the second view takes precedence over the first.
Let's start with the business case, which is easy and obvious: if you have a device with open source device drivers, you'll sell that many more of them to folks running Linux. My feeling is you'll sell that many more of them, period, since open source device drivers benefit every platform in the long run. Linux has long been the platform most associated with demand for OSDDs, but there's no reason why such things also can't benefit folks running Windows or the Mac or what have you. Rising tide, all boats, you know the drill.
The hard part has been convincing some hardware makers to offer OSDDs, or at least facilitate the development of same, when they don't feel there's a tangible reward in it for them. Why spend the money and effort when they'll get maybe only a fractional increase in sales? I suspect they may not have any real idea how big their potential Linux marketshare is, if only because solid marketshare information about Linux is tough to come by. (Yet another reason why the Open Source Census or Fedora's Smolt are valuable: they give us usage metrics that can be applied in other contexts.)
But potential sales alone shouldn't be the only reason. In fact, I don't think it should even be a primary reason. OSDDs make more sense in the abstract as a generic business practice. A company that uses an open driver architecture seems like a less treacherous outfit to do any kind of business with -- as a consumer, a manufacturer, or anything. Doesn't it make more sense to partner with, and buy from, a company that is less likely to play bait-and-switch games with its products? Furthermore, if you know other people are likely to warm up to that sort of behavior if you offer it first, isn't it in your best interest to do so?
Fans of logic puzzles may recall the Prisoner's Dilemma -- there, you get the best overall results if you cooperate rather than sell out your partners, even if you can't predict what they're going to do. Why see things as competition vs. cooperation, when you can compete (with those not using open architectures) by cooperating (with those who do)?