If you are rated under 2000, then you are better off working on your concrete knowledge of chess (tactics, mates, endgames, even opening repertoire since those are more "factual" than a 1900-level players "educated guess").
Higher rated players require "higher-rated" strategy. They also use non-tactical move choice mechanisms. (In the game I gave as an example, both players are about 2000 rating, also one of them is a FM, but they use strategical reasoning: http://www.chess.com/forum/view/game-showcase/my-first-win-against-a-titled-player-fm?ncc=1) I haven't read the book, but if it's targeted at <2000 rating players, so be it; this doesn't mean there are no good strategy books for stronger players. Maybe someone could suggest one.
That's true. But a plan's main reason is saving time, because you can't afford to calculate all possible variations. Therefore you need to skip some of them. If you have some principles which ones to explore and which ones to omit, it will be much easier to you (and you will have at least slightly higher chance that you'll keep the better ones than if you don't apply any choice criteria). Strategy is about applying educated guesses.
I can agree with you on that, but then you are conceding the point and have essentially answered the original question asked by the thread-starter: "Would you recommend Silman's book?"...
If you are rated over 2000 FIDE, it is pointless to read it (Silman's intended audience is 1400-2100 USCF).
If you are rated under 2000, then you are better off working on your concrete knowledge of chess (tactics, mates, endgames, even opening repertoire since those are more "factual" than a 1900-level players "educated guess").
I may read Silman's books out of curiosity, but now I know that learning the concrete of tactics and endgames is the only way I will get better. An analysis of my games (and of others in the <2000 range) would reveal such.
Addendum: The mistakes made by those of us in Silman's intended range are not mistakes of "principles" but rather concrete (tactics, endgames).