The amount of money generated for universities by patents on CS/engineering research is quite small, though. The total for all fields in 2009 (according to an unfortunately paywalled report: http://chronicle.com/article/Table-Licensing-Revenue-and/125...) was $1.8 billion, of which the vast majority went to a handful of schools for blockbuster biotech patents (e.g., NYU has been pulling in $100m+ annually from the Remicade patents). I would be surprised if UW non-bio patents are bringing in more than low-single-digit millions, i.e. not much compared to what they get in corporate donations, not to mention NSF and DARPA grants, and even the state subsidy and tuition.

The former Dean of Georgia Tech's CS dept quotes John Hurt from the NSF as claiming: "Of 3,200 universities, perhaps six have made significant amounts of money from their intellectual property rights." I don't think there have been any blockbuster university (non-bio) tech patents since the Akamai (MIT) / PageRank (Stanford) pair ten years ago.

This is really more of an issue with how universities value the research, I think. Most of the time, there isn't a very clear immediate business benefit, so they're willing to part with the IP for very little money.

Prime case is the Mosaic browser IP and the University of Illinois Urbana-Champaign. All the research was done by the National Center for Supercomputing Applications for the university, and when it was completed and released, didn't really bother to try to get anything from it. The university basically sold the rights to an offshoot company for a very small fee-I believe it was under $1 million, and didn't see any money from it after that.

Part of that is a deliberate strategy on the part of some universities, based on a guess that they'll get more in donations from corporations, spinoffs, and successful alumni voluntarily, than they would get by being tight-fisted with IP. Plus, it increases the university's prestige to have their stuff involved in more products, and their alumni involved in more successful ventures, so there's a cost to doing anything that would complicate stuff up-front. The goal is to get it on the back-end instead: don't hassle the not-yet-successful startup, but if they IPO, nag to see if they're willing to kick in a $50m building or write you into their will.

I'm not sure which version of the calculus is right, but it's at least plausible to me that universities stand to lose more in donations/prestige than they'd gain in royalties if they took stricter approaches to patent licensing.

I don't have a problem with a university owning a patent unless the work was somehow funded by the Federal Government, in which case it should be disallowed.