These (1) and (2) techniques that you speak of, both of them are applicable to all strategy games, and to master any strategy game, one needs to master both of these techniques, and master combining them as well.
An example of (1) from chess would be to know that the queen is a powerful defender and that the pawns are good attackers. Or that two bishops can work together to make a knight into a very strong attacker along a certain line.
An example of (2) from GalCiv2, would be that I know that founding an Influence Base at a certain point will force the AI into declaring war on me if he wants to win, therefore I should be ready to defend it before building it.
People just tend to focus on (1), because it always stays the same, allowing one to rely on ones memory, rather than thinking. (You only have to think to figure it out, after that you practise it in and rely on memory.)
The power of (1) is only unbeatable when the game is unbalanced, meaning that (1) forces you to take all the right actions and the enemy cannot do anything about that.
In good games, like GalCiv2, StarCraft and chess, every action has a cost and therefore, if the enemy can detect or guess your action they can choose to counter it.
For example, say building only cavalry are the 'best' according to (1). The other player can then guess that you will do so and build much pikemen to counter you. Then cavalry in this instance are not the best choice, but you can anticipate this and plan accordingly. To really beat your opponent you need to spy on him, but then again, he can build a bunch of cavalry where you expect them to be while his second army is elsewhere and follows a different plan.
The same rule determines whether an AI is good or not. The AI needs to be able to calculate the cost of different actions (i.e. use (1)) and guess what the player is doing by examining his actions and counter it (i.e. use (2)).
For more information, check out Dave Sirlin's book (for free on his website), "Playing to Win".