I read a summary on Wikipedia[1] and that helps. But why should the fact that someone didn't think (or document their thoughts) on something be evidence of non-obviousness?
I think of touch UIs and the patents there, for example Apple's rubber-band scrolling patent[2]. Sure, it's quite probable that no one thought of this before if they weren't designing touch UIs. And even if they were, they might have thought of that, plus several other ideas, while developing. Why should this make any difference?
For instance, no one has made scrolling that intentionally segfaults if you scroll fast enough (or insert other silly thing here). You won't find prior art on this. Should it be eligible for a patent just because it's novel? There should be some sort of criteria where the effort required to invent something is taken into account. If it's likely to come about merely as a result of playing in the space, then what does the public gain by issuing a patent?
What knowledge is contained in the "reverse PIN dials cops" that merits protection? Even if nobody had that idea before (or bothered to document it), what does that matter? If you patent can be constructed just by asking a simple question ("think of some ways to alarm when being robbed at an ATM"), well that should be grounds for it not being valid.
Edit: Another example. Things like algorithms. Look at a simple database indexing system: ISAM. Sort your data, sample every so often to form an index. Repeat if the index is too large. Ta-da. Going back far enough, this was very novel. But any worker that had to figure this problem out would arrive at the same solution. It's nearly as fundamental as binary search.
1: https://en.wikipedia.org/wiki/Inventive_step_and_non-obvious...
2: http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=H...