This article started life as a Usenet discussion. Thank you everyone who contributed. Perhaps mercifully, my system crashed shortly afterwards and I lost my cached store of posts! So, rather than try to reproduce all the threads, I've written this self-contained article.
While I was conducting a complicated argument about consciousness in the Usenet group alt.atheism, another thread caught my eye concerning the possible motivation God may have had for letting mankind fall into evil. A theist poster had suggested that God allowed man to make a free choice in order to see what he would actually do. One might ask why God, who knows everything, should need to do experiments to find out how his creation works, but the idea is that a being with free-will makes his own choices that even God cannot predict. With free will, a "gedanken" experiment would be inconclusive, a real choice needs to be exercised in the real world. This led to some discussion as to whether free will is a coherent concept at all which terminated, in true Usenet fashion, in a morass of arguments about time travel.
This scenario of free choices seems to clash with God being able to know what the choice would be in advance. However, the idea of a God outside time goes back at least as far as Augustine, so there is no problem with having Him "observe" the events from His vantage point. Free-will would be questionable if God could "predict" what we will do simply by knowing what we are like. This would make us deterministic machines. Watching us make our own decisions in the future has no such implications.
The argument moved on and another poster said that if God knew what man was going to do, then, when man came to choose, he must have been "constrained" to make the particular choice that he did. Thus his will cannot be said to be free. I believe this to be a muddled argument, and I am in good company here with the atheist philosopher A. J. Ayer. It is, of course, correct that if God knows something then it is so and cannot be anything else. Thus in a trivial sense, the fact that it will be so imposes a logical constraint on whether it will be so. However, this is mere tautology. If X then X
The problem is that the word "constrains" was being used too loosely. For a choice to be free, it need only be free from causal constraints. The fact that it will be X does not cause it to be.
Some posters recognized this but still maintained that any kind of constraint means that the will is no longer free. I do not agree. The freest being in the world can still only make the choices that he makes, he cannot defy logic by choosing something he doesn't choose. Nothing is free to be what it is not. If this kind of "constraint" mattered then the word "free" would already be meaningless before any ideas of foreknowledge were brought in. It is certainly hard to define free-will exactly - if we say things like "the ability to make decisions which are really our own" then we probably come close to what most people feel about it, but then we have a problem in saying what "really our own" actually means. However, the aspect of free-will that is relevant to foreknowledge is much simpler - a free choice is not predictable. There's nothing incoherent about that idea: modern physics has several phenomena which are inherently unpredictable.
The reply to this was that knowing things in advance is impossible. The contention was not that it happens to be impossible owing to the laws of physics but that it is impossible in principle, simply because they "haven't happened yet". I believe this glib dismissal of foreknowledge covers a serious logical mistake.
It's an obvious fact about the physical world that we can remember the past and reconstruct it from evidence but cannot do so for the future. However, such an observation cannot be extrapolated to beings outside time or even to new-fangled gadgets of the physics laboratories which may be able to steal information fleetingly from the future. Unfortunately, "not happened yet" is not just used in this simple sense. People use it to suggest that events in the future are "fluid" whereas events in the past are "fixed".
This is a curious idea. The future seems fluid to us because we don't know what it will be. However, what people actually mean by "fluid" goes much further than this. Conventional logic requires that all propositions be either true or false. Sometimes it is necessary to label them as "unknown" and you can concoct a calculus which has three logic values: true, false and unknown. However, "unknown" is always secondary to the other two values and the rules for handling it are simply the rules that cover the two possibilities.
The idea of a fluid future is different. We now have a fourth logic value: fluid. A proposition that is fluid is neither true nor false. It may become true or it may become false. We do not know which it will become. Meanwhile it is neither, it is fluid. Of course, fluid futures rule out foreknowledge in one easy move. Nothing in any possible world can know whether something is true or false if, in fact, it is neither.
Now this might seem pretty harmless but the problems arise when you say "Ok, we don't know whether it is true or false. However, we do know that it is eventually going to become either true or false." Why doesn't knowing that a proposition is going to become true or false mean that it is already true or false? Why does it have to be something different again, this thing they call fluid? And the answer comes back: "Because it hasn't happened yet". Let's look at a concrete example to see how bizarre the idea of "fluidity" really is. The proposition "Potter will be king of England in the year 2020" is deemed to be neither true nor false but fluid. The proposition "In 2020 it will be true that Potter is king of England" is, however, deemed to be a normal proposition, either true or false, because it does not speak of fluid things, it speaks of the resolution of this fluidity at the appropriate time.
If we now consider the scenario in which Potter becomes king we have the interesting situation in which the proposition "It will be true to say that Potter is king" is a normal proposition, either true or false, but the proposition "it is true to say that Potter will be king" is a fluid one. Unfortunately it is not clear what the difference is! In normal logic, to assert that a proposition is true is to assert the proposition itself. Thus "It will be that Potter is king" is a normal proposition but, according to the fluidity merchants, "Potter will be king" is not, it is "fluid". However, "It will be that Potter is king" is just a clumsy way of saying "Potter will be king". It is the same proposition. Thus it is said to be simultaneously [either true or false] and [neither true nor false].
To put it bluntly, the invention of a "fluid" truth value is incoherent. Its sole purpose appears to be to prevent anyone talking about foreknowledge. It's a very powerful weapon too. Since propositions are now allowed to be fluid rather than merely unknown, it is impossible to apply conventional logic to them. This is handy in a discussion about God's foreknowledge. He can't know the future because it's fluid. Blip blip blip.
I prefer to state the obvious and refer to future events as "unknown". Even free-will events are going to have a specific outcome which will be determined by a choice. It's just that the choice hasn't been made yet so it's not yet known. Furthermore, to be free, it can't be predictable - there must be an element in the choice that is not determined by the person's state and circumstances at a the time. That does not imply that God (or a machine or a crystal ball) can't "see" our choices from His vantage point.
The discussion continued and some interesting syllogisms were posted purporting to prove the incompatibility between foreknowledge and free will. These turned out to require a fluid future as a premise. Most of the arguments boiled down to saying that even if a man chooses X, /X is still "possible" because of free-will. Therefore it is possible for X and /X, which is clearly absurd. Thus the free-will premise must be false (reductio ad absurdum).
The compatibility of actual X with the possibility of /X gives rise to the term "compatibilist". The compatibilist believes that even if I'm going to do X tomorrow, I am "free" to do /X and that therefore /X is a "possibility". (/X is "compatible" with X.) The poster rightly pointed out the illogicality of compatibilism. To me, it illustrates the problems that the fluid future idea gives rise to. There is no need to invoke free will to find problems with fluid futures: anything that re-writes logic for its own convenience leads inevitably to nonsense.
It is always necessary to remember that modal logic requires that we state the conditions: "X is possible" means nothing on its own; "X is possible given Y" simply means that "Y does not imply /X". It does not grant possibility to X if it is otherwise impossible! However, to the compatibilist, the fact that I am going to choose to do X tomorrow does not rule out my doing /X. Unfortunately, many people who stoutly defend the idea of free will do so as compatibilists. Since a non-fluid future does not remove free will and does not require the use of mental gymnastics, I take the incompatibilist view. If I'm going to do X tomorrow then I'm "free" to do /X but will actually do X. As to what will cause me to choose X - I will choose freely, so nothing outside myself will be forcing me. I have not seen any argument against incompatibilist free-will, either formal or otherwise.
Whether free will really exists is another matter. If it does, it means you can make a choice which is not entirely constrained by your previous nature or by external circumstances. The unknowability of the future only impinges on the subject of free-will if we make entirely arbitrary assumptions about time that require re-writing conventional logic.
The subject moved on again and various scenarios were posted trying to prove that the very idea of foreknowledge leads to insurmountable logical difficulties even without the help of fluidity. For example: If God tips you off that you are going to make a big mistake tomorrow and get yourself killed, you can do something to avoid the problem. Or must we assume that events will conspire to force you to your fate after all? Whether from God, a crystal ball or by a quantum contraption, foreknowledge shouldn't organize your circumstances spontaneously.
This is actually much the same as the well-known "grandfather paradox". What happens if you go back in time and kill your own grandfather? It's something that fantasy writers have to deal with and the usual trick is to invoke special "rules" for time-travelers. This is ok for a story but it won't do for a proper argument. The answer must lie in logic and known physics. Of course the question presupposes that time travel (if only for information) is possible, but it's an interesting logical conundrum, whether the real world permits it or not. And, of course, the paradox can be created without sentient beings, indeed it works rather better with a robot, programmed to do whatever is necessary to prevent a future that some contrivance has revealed to it. I've analyzed this in some detail in Time Travel Paradoxes Resolved (coming "soon").