Sometimes you just get blindsided. Back in the mid-1960s, shortly after the laser was invented, a couple of smart guys named Norman Wild and Paul Leavy were working for a defense contractor in New Hampshire called Sanders Associates (now part of BAE Systems). They came up with a clever idea for spotting snipers on a battlefield, based on a laser and a special property of the eye. Most of us have seen the way a cat's eyes seem to glow in the dark: their retinas are retroreflectors much like a bicycle reflector or a safety vest, that send light back preferentially the way it came. Human retinas do that too, though not quite as obviously. (Before smart digital cameras, photos taken with direct flash tended to show people's eyes glowing bright red for this reason.)
Anyway, Wild and Leavy realized that a sniper's eye looking through a rifle scope was a very efficient retroreflector that could be detected from a long distance, at least if he was looking anywhere near you. Their scheme scanned a collimated (parallel) laser beam around and looked for anomalously-bright reflections. If you see one, shoot at it, and there'll be one less sniper to worry about. All being fair in love and war, no worries. They filed a patent, and since the field was so new, they were able to make very broad claims covering essentially every optical system that sent a collimated laser beam out and received a collimated beam back, such as a laser microscope or many kinds of optical disc. Sanders would have owned that till the mid-1980s.
But then the US government classified it, so the patent wasn't published. Sanders Associates lost the benefit of their patent for the time being, and the field of laser optics flourished, with many systems using the collimated-in, collimated-out principle because it's so convenient. Then in 2003, it was declassified, and the Patent Office reissued the patent. (There were a few of them in the same general field, actually.) This meant that this 1967 patent time-travelled 36 years into the future, and still had its 17 years to run--53 years after the filing date is a long time in technology.
That was the patent we were up against. I was working as a consulting expert for defendant Samsung in a case at the US International Trade Commission (ITC). They'd asked me to be the testifying expert, but ITC cases go so fast that I wouldn't have been able to keep my other commitments. Thus I built a reverse-engineering setup in my lab, tested the accused products, and assisted the other expert repeat the tests himself so that he could testify about them. I didn't read his report, but from my observations it looked like some of the accused optical disc products infringed, but some didn't because they didn't use collimated beams. (This was another case where I had to give the client the bad news.)
Samsung settled out of the case shortly thereafter, but in the end Optical Devices lost the case, since the ITC held that they didn't have "prudential standing", which basically means that they didn't completely control all the patents at issue. A close call for a lot of other outfits!