Why do so many people who claim humans have free will dismiss the deterministic-random dichotomy, and regard free will as a third option?
We recognize that a system is deterministic when we recognize the laws that determine its future state. We recognize that a system is random when we fail to recognize any laws that determine its state. So, what does it mean to recognize that a system is free?
We are familiar with the idea that a man can make a decision of his own free will. A decision made by free will is not made under duress or under "abnormal" external influence (e.g., the decision was not induced by mind-altering drugs). Using this same definition, we can see that it really doesn't matter whether the decision-making process was fully or partially deterministic. What matters is whether the decision was significantly determined by a certain class of external influences.
Thus, there is really no direct connection between "free will" on the one hand, and determinism-randomness on the other.
Where, then, does the mistaken link between free will and determinism come from? The problem arises when we disproportionately focus our attentions on the determinism required for coercion, and ignore the determinism required for free decision-making. For an external influence to coerce a decision, there has to be some causal link between the influence and the decider. However, for the decider to be able to make a free decision that weighs options against the decider's values, there must also be some determinism. Otherwise, a person could not be trusted to make any given decision freely with any reliability at all. In that case, we would not even need a word for "decision."
One clue that the link between free will and determinism is a false one comes from a linguistic analysis of the statement "humans make free decisions, i.e., decisions that are neither deterministic nor random, nor a simple combination of the two."
How would we recognize the difference between humans making free decisions and humans making non-free decisions? After all, if we are going to name a decision as free, we had better be able to recognize free ones versus non-free ones. Our inability to do this is a clear signal that we have stretched our definitions beyond recognizable limits.