Far too many examiners equate a high level of "keyword hits" with "anticipation" or "obviousness," with no evidence they understand what they are talking about. Just string some keywords together, and hey presto! It's obvious!
My apologies for being so blunt, but this is a real problem. The Office should be embarrassed by the quality of many office actions.
The "keyword hit" attitude of many examiners is often impervious to reasoned rebuttal. If the examiner does not understand the invention or the references, the solution in many cases is not to better educate oneself, but instead to dig in deeper and stick to the original rejection.
Of course, many examiners do a better job than this, but there are far too many who do not.
Whether the solution is to hire better qualified examiners, or to provide adequate incentives to existing examiners to do a better job, I do not know.
Perhaps an examiner should be required to defend the rationale for a rejection internally within the USPTO before a skeptical review panel of his/her peers, before an office action is ever mailed in the first place. This would need to be structured with incentives to make the exercise meaningful, and not just a rubber stamp. Although this approach would be more time-consuming and expensive in the short run, in the long run it could improve examination efficiency by providing a quality check on office actions.