I generally managed to engineer my games in such a way that standard reactions are inadequate. (Comment to his 9th move in the game Kasparov-Kamsky, Linares 1993, cited in Igor Stohl, Garry Kasparov's Greatest Chess Games, vol. 1 (Gambit, 2005), p. 307.)
Many of us don't bother looking for novelties, but simply do the best we can to arrange our repertories in accordance with the most optimistic evaluation we can find in modern theory. Others of us, more ambitious, do try to find new moves, either by means of switching on Fritz or his cousins, or by looking at different ways of implementing our preferred plan.
Kasparov's idea is different, though it doesn't exclude either of the means given in the previous sentence. It's akin to a prophylactic approach, but with a twist. The standard sort of preventive approach stops the opponent's idea before it can take shape, but Kasparov's tweak is to allow the opponent to execute his idea (e.g. a routine development scheme or typical middlegame plan) and then show it leads to serious, hard-to-foresee trouble. (A classic example, as Stohl notes on that same page, is Kasparov's fantastic win over Karpov in game 16 of their 1985 match.)
We might not be Kasparovs (and I use that "might" rhetorically), but we can learn from his approach. Sometimes the first step to finding a great idea is knowing what questions to ask, and Kasparov has added a valuable question to our toolkit. Let's use it!
Meanwhile, click here to see the two games mentioned above, with brief notes highlighting the way they exemplify the technique we've discussed.