For one thing, it uses the language of cause and effect. Bertrand Russell was the first to point this out: the language of cause and effect is not the language which physicists use to describe reality. Consider F=ma. You might be tempted to say that the force is causing the mass m to accelerate at a rate of a; after all, if I were to vary the force, I will vary the acceleration. But if you vary the mass you'll vary the acceleration too: is the mass the cause of the acceleration?
And notice what was just said: "if I were to vary the force..." i.e. you have introduced an agent. And you have spoken counter-factually about this agent: the agent could have chosen to vary or not to vary the force. I.e. using the language of cause and effect presupposes that an agent is the cause, and has the choice to cause or not to cause. Using the language of cause and effect to describe a physical situation is ultimately an anthropomorphisation.
Cause and effect are part of what Dennet calls "the intensional stance" i.e. the language and vocabulary which we are forced to use when describing the behavior of intensional agents. In order to at all make an intensional agent's behavior understandable, we have to treat them as uncaused causers. When courting a spouse, or raising children, you quickly learn that people cannot be treated as deterministic, stimulus-response mechanism. You have to presuppose that they can freely choose how to react to you.
The question of whether or not we have free will is therefore, not a question of whether we have to presuppose that we have free will; we do. Its just a question of whether the language of the intentional stance is merely a useful heuristic.
But arguing against free will by invoking the language of cause and effect, as the above definition does, is a category mistake, akin to arguing that 1 + 1 can't equal 2, because green + green equals green.
