Open source, the US military, and fighting off open-source threats
Open source is more than a software strategy. It's a way of fighting military battles, and one that is highly disruptive to the US military.
I've long been a reader of The Atlantic, and found this article highly interesting. The article talks about how the US military can defeat an open-source insurgency, and actually has direct implications for proprietary software vendors who are trying to stave off threats from open-source software vendors.
After four painful years, the US military has stumbled upon...the only model for fighting a mature open source insurgency: a decentralized model of security that forgoes centralized defense/police forces in favor of a plethora of independent militias. The success of this model in reducing violence (at least in the short term) in Anbar province, has led to its replication in other provinces.
In the software context, you don't beat open source with a single, unified front (i.e., single point of failure against disparate hordes of "barbarians"). As John Robb points out, you do what IBM has done: you become part of the insurgency, joining and/or funding open-source communities to steer them in your direction.
In this model, the US military aligns itself with a plethora of militias (in this case hundreds) regardless of political/regilious/ethnic/tribal affiliation under the plausible promise of local autonomy. It is made fact as funds, weapons, and local control flows to these militias.
The IBM example almost to a "T." IBM has made itself a friend of open source while simultaneously holding it off its shores in many areas. Microsoft, on the other hand, has tried to preserve a central line against all open source, and is under siege as a result. You don't beat guerrilla warfare with a traditional army, unless you're prepared to drop nuclear bombs (patent suits, anyone?).
In short, the best way to beat open source is to join open source.