Kind of like how software security most often fails in the implementation rather than the theory. :)
Take the Rockstar suit against Android. We know who the real parties in interest are (Apple, Microsoft, etc.). Rockstar isn't going after small businesses who are more likely to settle than go court. They're going after Google (which will almost certainly fight back). I don't know about the validity of the claims, but they're at least plausible enough that Google felt compelled to bid for the patents.
Nevertheless, the Rockstar suit feels wrong. And it's because Google created Android independently of anything Rockstar (or the previous holders of the Rockstar patents did.
You may be right, but a similar claim could be made about any patent case. (i.e., it's a weak argument.)
Consider you invent and patent a time machine. I've never met you or done business with you, but I read your patent and "independently" start producing a very similar time machine. You might have a problem with that. Obviously, Google claims they've never read the Rockstar patents or that they're irrelevant, but equally obviously, I'd make exactly the same claim regarding the time machine.
You would be correct that this wouldn't save you in court -- under current patent law, independent inventions still counts as infringement. But my argument is that in terms of what feels "right" or "just", independence is the key test. If it turns out later that Google did in fact read Rockstar's patent and copy them, the outrage at Rockstar would probably be mitigated.
> Obviously, Google claims they've never read the Rockstar patents or that they're irrelevant, but equally obviously, I'd make exactly the same claim regarding the time machine.
You could claim that, but I could also prove it false. If you read the patent by downloading it online, there's an electronic paper trail. Unless you're Doc Brown, building a time machine sounds complicated enough to require multiple people, so there are now witnesses to your copying. I'm not a professional investigator, but if I were, I'm sure I could find some other ways of proving or disproving copying -- especially if there were bajillions of dollars on the line.
That said, if you had a time machine, I suppose you could just go back in time and patent it first and avoid the issue altogether. But temporal mechanics mechanics aside, if you're curious on how to make an independent invention defense work, I've written a much longer post on it at http://www.techdirt.com/articles/20121011/14171220681/yes-in...
Also under current patent law, "obvious" inventions are not patentable. There's a tension involved in saying an invention isn't obvious, despite its independent development in various places.