The Prior Art Conundrum: Knock It or Not?

Dear Rich: I've just finished Patent It Yourself. The one point I'm struggling with comes in Chapter 8 (How to Draft the Specification).  I'm finding it difficult to reconcile two ideas: (1) "Don't say what the prior art can't do because this can make your invention or the problem you discovered seem obvious." (2) "...you should, as much as possible, try to "knock the prior art" here in order to make your invention look as good as possible." I appreciate the underlying conundrum (and that it's not of the author's making!), but this is currently my biggest concern. Since the 'non-obvious' requirement is likely to be a bigger hurdle than the 'usefulness' requirement, I'd appreciate an opinion on whether the greater risk here is in calling too much attention to the shortcomings of the prior art (making solutions seem more obvious) or not enough attention to them (making solutions seem less useful). PIY author David Pressman responds, "This is a good question. Here’s a better explanation that I had recently written in response to another person who had seen this dichotomy:

In the PRIOR-ART section don’t mention your invention or suggest any solutions. Never state that any prior art reference doesn’t teach a specific feature of your invention. When you knock the prior art, just state what's generally wrong with it from a novice's standpoint and don’t state that it doesn't have any specific novel feature of your invention. This is because (a) most people would not realize or consider that the lack of a specific feature is a disadvantage, and (b) it can make it seem like the lack of a specific feature was already known in the art and was not discovered by you, the inventor. For example, the right way to indicate this would be to write, "Smith’s system operates slowly." The wrong way would be to state, "Because Smith doesn’t eliminate the reaction waste products, this delays the refining process."