Too many times. Of course there are patent trolls, but these (actual patents used by actual companies that make products) cases are not few. I suspect Apple thinks like car mfgs as told in Fight Club: if the cost of patent licensing exceeds what they think will be an eventual settlement or compensation awarded by a court - they don't buy the license.
I think it is a bit more complicated. Apple or others may decide they have patents that also cover the innovation and provide protection in the event of a court case and decide not to license the patents. Or, a company over values their patents and Apple decides they’d rather risk a lawsuit, which is what you argue.
Part of the problem is patent law that allows functionality, not implementation, to be patented. Koss, if I understand correctly, patented touch controls via a capacitive interface and wireless transmission of audio. I’d argue that patent law should protect Koss from someone using capacitive touch controls but not another method of implementing touch control; and wireless audio transmission is not exactly an innovative idea. You could patent a new lossless codec for wireless, but someone should be allowed to develop and patent a different lossless codec without violating your patent.
Koss also claims a patent on internal microphones for voice control; yet voice control is not a new idea. Is using an internal wireless make instead of and external one innovative enough to warrant a patent?
Apple may have decided Koss’ patents would not have held up in a court fight.
Another issue is the USPTO; I suspect the workload for examiners is so high they have to push applications through as fast as possible; but that means less time to thoroughly review an application. Approve it and let companies fight it out later or deny it and have to re-review it upon when appealed? Even so, time delays and backlogs mean two different examiners could issue patents for the same idea. That leads to patent fights as well.
The patent holders are the ones severely disadvantaged in either case - Apple f***s them by releasing a products with patent violations and makes tons of cash while years pass by and actual court deadline gets close. By that time they most likely already out-compete the competitor with their market share, and end up (worst case) paying some penny on the dollar type settlement to a company that's long since left in the dust.
Which is why striking a deal is probably the best solution. In Koss’ case, we don’t know what patents Apple has that could be used to invalidate Koss’; and Koss likely could not afford a court fight. Apple made an offer that spared Koss a costly fight they may not win, and ammunition to settle with Bose et. al.
Settling is not an admission of guilt but simply a way to end the uncertainty and avoid spending money on a court fight.
As I said earlier, a company I am involved with settled a lawsuit we’d probably win because it was a lot cheaper than a drawn out court fight. While winning would be nice spending the money was not in our best interest; and we wound up settling for our original offer anyway.
The only way such things would be fixed is extremely expedient court cases that don't get stalled for years, and where actual market share losses are added on top of other compensation to even things out. But that's never happening though.
A special patent court, not just for appeals, with judges with patent law experience would be a good start. That would avoid circuit shopping and non-technical juries deciding extremely complicated technical issues.