Edwards on the Incompatibility of Divine Foreknowledge and Human Free Will

In the book “Freedom of the Will”, Jonathan Edwards (1703-1758) put forward a strong ar-gument for theological fatalism. This argument, I suppose, can be considered as the universal basis for discussion between Fatalists and Anti-Fatalists in the 20th century, especially in the context of the most powerful argument for fatalism, introduced by Nelson Pike. The argument of Edwards rests upon the following principles: (a) if something has been the case in the past, it has been the case necessarily (Necessity of the past); (b) if God knows something (say A), it is not the case that ~A is possible (Infallibility of God`s knowledge). Hence, Edwards infers that if God had foreknowledge that A, then A is necessary, and it is not the case that someone could voluntarily choose ~A. The article argues that (i) the Edwards` inference Kgp → □p rests upon the modal fallacy; (ii) the inference „God had a knowledge that p will happen, therefore „God had a knowledge that p will happen” is the proposition about the past, and hence, the necessarily true proposition“ is ambiguous; thus, it is not the case that this proposition necessarily entails the impossibility of ~p; (iii) it is not the case that p, being known by God, turns out to be necessary. Thus, we can avoid the inference of Edwards that if Kgp is a fact of the past, then we cannot freely choose ~p. It has also been shown that the main provisions of the argument of Edwards remain significant in the context of contemporary debates about free will and foreknowledge (Theories of soft facts, Anti-Ockhamism, theories of temporal modal asymmetry, „Timeless solution”). Additionally, I introduce a new challenge for fatalism – argument from Brouwerian axiom.

The argument of Edwards runs therefore as follows: (1) Whatever has been the case in the past, has been the case necessarily. We know, for instance, that Barack Obama was born in the year 1961. It is not within our power to change (in the year 2020) the fact of the past that Obama was born in the year 1961. As a result, every true proposition about the past is a necessary truth.
(2) Suppose that the foreknowledge of God that I will write an article about God`s foreknowledge existed in the past. Hence, by (1), the foreknowledge of God is a necessary truth.
(3) If God`s foreknowledge that I will write today an article about God`s foreknowledge is a necessary truth, then it follows from God`s foreknowledge that I will write the article necessarily (by the principle that what follows from the necessarily true proposition is itself a necessarily true proposition). Call this principle Necessity Entailment (4) The divine foreknowledge exists. Hence, the truth of the statement that it is necessarily the case that I am writing an article about God`s foreknowledge in the year 2020, by (3), is true not in virtue of empirical observation (i.e., in virtue of the fact that I am actually writing this article), but because this proposition is logically entailed by the necessary proposition 2 that God had a foreknowledge 2000 years ago. Hence, it is not the case that I can refrain from writing the article.
Schematically, the argument of Edwards can be summarized as follows: (a) God had foreknowledge that A (let A be the fact that today I will write an article) in the past (Assumption) (b) It is not the case that God might not have a foreknowledge that A (from the definition of God) (c) If God had a foreknowledge that A, then necessarily, God had a foreknowledge that A (from the principle of necessity of the past) (d) Necessarily, "God has a knowledge that A" entails "A is true" (from the definition of omniscience; call this principle Truth-Entailment) (e) "It has been necessarily true 2000 years ago that "A" entails "It is necessarily true (now) that it has been necessarily true 2000 years ago that A" (Closure of the past and Necessity-entailment) (f) Hence, it is necessarily the case that A (now) is a necessary truth (j) If it is necessarily the case that A is a necessary truth then, necessarily, it is not the case that ~A (definition of necessity) (h) If, necessarily, it is not the case that ~A, then ~◊ ~A. Hence, it is not the case that I can voluntarily choose ~A. As a result, if the foreknowledge of God exists, I have no free will.
The fatalistic argument of Edwards has been developed further by Nelson Pike [Pike 1965]. Pike`s argument, very briefly, can be described as follows: (I) Suppose that A (now) (II) If, at t2 (now), A, then God at t1 (2000 years ago) had a knowledge that A (III) If God has knowledge that A, then ~◊ ~A. If God has a knowledge that A, then A is true.
(IV) If God exists in t1, it follows that if, at t2, A, God at t1 had knowledge that, at t2, A (V) It is not the case that the contradiction is possible. Hence, it is not the case that ◊ (A&~A), it is not the case that it is possible that God had and hadn't a knowledge that A, and it is not the case that it is possible that God existed and not existed at t1.
(VI) Hence, if God had an existence at t1, and, at t1, God had a knowledge that, at t2, A, then it follows from the assumption that I could have had a free will (i.e., to choose between A and the refraining from A) that: (A) If at t2, I can freely choose ~A, then it would follow from my voluntary decision to choose ~A that God had a false foreknowledge at t1 (B) If, at t2, I can freely choose ~A, then God who had a knowledge that A at t1, would not have knowledge at t1 that A, or God who had a knowledge that A would have had a knowledge that, at t2, (A&~A) (C) If, at t2, I can freely choose ~A, then God who existed at t1 and had, at t1, knowledge that, at t2, A, would have been nonexistent at t1.
Since none of these alternatives is acceptable, the argument seems to establish that free will is an illusion, and there is no way to avoid fatalism. Fatalism is not possible if there are contingent things and events; however, according to Fisk, Edwards denies the ontological status of contingency [Fisk 2016: 340]. Fisk correctly notes that Edwards deduces the necessity of certain things (for example, future events) from consequential necessity -if two things are in an "infallible connection", then the necessity of the second thing follows logically from the necessity of the first one. If A implies B, then B follows from A and therefore B is necessary. Let A be a fact of the past, and therefore a necessary fact (as Edwards believes). Now, if A and B are related in such a way that A is the cause of B or B follows logically from A, then B is necessary; B is necessary not by virtue of its own nature, but because B is ISSN 2075-6461. Sententiae, Volume XXXІХ, Issue 2, 2020. entailed by A. However, according to Fisk, in fact Edwards deduces the necessity of B not from consequential necessity, but from the necessity of a consequent, i.e. "philosophical necessity". Let me give a little example. Suppose I am married to Mary. If I am married to Mary, then I am Mary's husband. The proposition "I am Mary's husband" logically follows from the fact that I am married to Mary. But if I am married to Mary, then it is not true that I am not married to her; therefore it is impossible that the proposition I am not Mary's husband would follow from the fact that I am married to Mary. Thus, as Fisk notes, a consequential necessity (the impossibility for me not to be Mary's husband if I am married to Mary) would be legitimate in the structure of Edwards's argument only if it were impossible for me not to be Mary's husband (i.e. the necessity to be her husband) logically followed from the fact that I am actually Mary's husband [Fisk 2016: 340]. But in fact, the necessity for me to be Mary's husband does not follow from the fact of my marriage to Mary. Likewise, the fact of my marriage to Mary is not a consequential necessity, so Edwards cannot prove that the necessity of B follows from the infallible logical connection between A and B, and is not in fact a property of B. Fisk is certainly right that Edwards`s argument is flawed and the (necessary) truth of B does not follow from the consequential necessity. However, Fisk paid little attention to the question of why we should assume that A is not a necessary truth. The observation of Fisk gives us a little reason to disprove the (Necessity of Past). Finally, Fisk`s argument works successfully against naturalistic fatalism (if A is understood as a natural event, thing, or property), but has significant argumentative difficulties against theological fatalism (if B is caused or entailed by God. Barone depicted this issue in [Barone 2020]). We will demonstrate (in the last section of this article) that Edwards`s fatalistic argument is not only formally incorrect (as Fisk correctly observed), but even if it were correct, Edwards`s fatalism would be self-contradictory. Fisk provided no argument that Edwards`s fatalism is inconsistent and self-contradictory.
Muller agrees with Fisk: "Edwards appears, therefore, to confuse necessity of the consequence with necessity of the consequent, assuming that a necessity of the consequence entails an "infallible connection with [some] Thing foregoing" [Muller 2014: 273]. As a result, Muller does not accept Edwards`s understanding of contingency; for Muller, Edwards`s concept of contingency is not "genuine" [Muller 2017] -if Q follows from P then, by Edwards, it is not the case that Q could be not necessary or P could not be nonexistent. Barone [Barone 2020], however, objects to Muller. One of the most important Barone's arguments is as follows. Edwards argues that necessity is a connection of things and not an intrinsic property of the thing itself. If B follows from A, B is necessary because the truth of A implies the truth of B, but B is not necessary by itself. Take, for example, a certain fact of future C. It is necessarily true that the occurrence of C is entailed by the fact of the present (say A), but C is not necessary. If C were necessary, then it would be actual one. But according to our assumption, C is a fact of the future; therefore, the existence of C refers to a fact of the future. Thus, although C is necessary due to the connection between A and C, the existence of C is contingent, and therefore it is not true that Edwards denies the ontological role of contingency. Barone`s argument doesn`t look convincing. C would be contingent if and only if C could lack its intrinsic truth-value (that is, if C could have been true not necessarily and thus could have been nonexistent). Hence if C is necessarily true because C is connected with A, then C necessarily will be existent in future; and if C were contingent, it might not exist at all (regardless of a specific time). But C, of course, will be existent, if necessary, and thus ~C will be necessarily nonexistent, and the contingency of C definitely depends on whether ~C could have occurred. Barone tries to summarize his argument as follows: "…to say that, for Edwards, this infallible connection corresponds to an absolute and intrinsic necessity, then the claim is mistaken. We have clearly seen that Edwards does not mean that such an infallible connection of things and future events has its grounds and principles in the nature of the things and future events in themselves, as when, for instance, adding four to three is followed by seven because of the very intrinsic nature of four and three" [Barone 2020: 16] I do not think that this remark is helpful for the purposes of clarifying whether Edwards`s argument contains a logical mistake (that is, confusion between consequential necessity and necessity of the consequent). Barone argues that it would be mistaken to think that if C follows from the necessity of A, then C is intrinsically necessary. He says that, by Edwards, it is not the case that A could have had such an intrinsic value as necessity, only God has it. Thus, A is not necessary, but A is necessary being entailed by God, and so we can conclude that C is necessarily the case not in virtue of A`s necessity, but because it follows from God`s necessity 3 . I think that Barone`s argument is formally true; but even if this argument is true, it is not helpful for the purposes of understanding Edwards`s consequential necessity. Instead of this we could say that everything is necessary, because everything is entailed by God, but this entailment, of course, is not a source of the consequential necessity. Even if the argument of Barone is correct, and B is itself a kind of contingency, it would lead to some unacceptable consequences (as we will see in the last section of this article).
"in the remote past", "now" P(t1), P(t2) P is the case at t1 / t2 G God Kgp God knows that P Kg(t1)P(t2) at t1, God knows that P will be the case at t2

Divine Foreknowledge and Modal Fallacy
From the principle of divine omniscience advocated by Edwards follows that God knows every true proposition. This principle can be read as "If God knows the proposition p, p is true" (∀p (Kgp → p)), as well as "If the proposition p is true, God knows p" (∀p (p → Kgp)). Since God`s knowledge is infallible, we can necessitate the first reading of the principle of foreknowledge (∀p □ (Kgp → p)). Let p be any contingent proposition (for instance, "Obama exists"). Then, by (Necessity Entailment), Kgp → □p. By the principle of Divine Omniscience, we can guarantee that Kgp → □ Kgp; additionally, this inference is guaranteed by Edwards`s assumptions (1) and (2), since God`s foreknowledge about p is a fact of the past. As a result, we get an inference: (Inf) Necessarily, if God knows that Obama exists, then Obama necessarily exists.
Is it however the case that Obama exists necessarily? No. Obama is a contingent, but not a necessary being. Hence, (Inf) contains a modal fallacy. Consider, again, the revised argument for (1) and (2): (5) If God knows that p, then God necessarily knows that p. (6) If God necessarily knows that p, then p. (7) Hence, if God knows that p, then p.
Hence, P is true at t1. Assume now the (Necessity Entailment). From the (Necessity of the past) we have that P(t1) → □ P(t1), and thus P is necessarily true at t1. Thus, at t1, God necessarily knows that P will be the case at t2, and now we have from (Iterability) that P (such that P will be the case at t2) is true at t1. Thus P (from Iterability) could be true at t2 only if P were true at t1. Thus necessarily, the truth of P at t2 is entailed by the truth of P at t1. But P at t1, by (Necessity of the past), is necessarily true. So, by the (Necessity Entailment) P(t2) is also a necessary truth, and thus it is not the case that ~P could be true at t2.
However, we are suspicious of (10). In order to derive (15) we should presuppose that, by Edwards` principle (1), every proposition about the fact necessarily has a modal profile. If it is so, then the consequent of the inference P → □P 5 should be read as de re modality. Does Edwards`s inference satisfy this requirement? No. Presupposing the principle that the fact of the past is a kind of necessity (because it is not witin our power to change this fact), we presuppose, equivalently, the following principle: (NP) Necessarily, if something has been the case, then it is impossible that something has been not the case.
By substitution (NPA) Necessarily, if p is a fact of the past, then it is not the case that ~p could be the fact of the past. Hence, it is impossible that ~p.
Hence, we can reconstruct the argument of Edwards for the (Necessity of the Past) as follows: (17) P is a fact of the past (18) If p, then impossible that ~p (19) If it is impossible that ~p, then necessarily, p (from the principle ~◊~p↔ □p) (20) Hence, if p, then p is necessary.
But the inference from (17) and (18) to (19) is dubious. If p is true (by the principle of fixity of the past), then, of course, ~p is untrue. However, (17) says nothing about the necessity of p. For if p is true in virtue of (17), the correct inference from (17) should be as follows: (18*) If p, then impossible that p&~p 5 One of the most relevant arguments against the possibility of inference like KgA → □A is to argue that KgA is so-called "soft fact" about the past, contrary to "hard facts" (see an example from (1)). This argument is known as "Ockhamist Solution" (OS). According to (OS), some facts of the past are soft, and thus not necessary. See Plantinga [Plantinga 1998] and Widerker [Widerker 2015] for detailed development of this line of reasoning. See also Hoffman and Rosenkrantz [Hoffman & Rosenkrantz 1984], Hasker [Hasker 1988], and Adams [Adams 1967]. Fischer [Fischer 1983] gives an argument against (OS), and Zemach and Widerker reply in "Fact, Freedom, and Foreknowledge" [Zemach & Widerker 1987].
By the distribution of necessity: (22) If it is necessarily the case that if p, then p, then if necessarily p, then necessarily p.
Are we able now to derive from (22) that "if necessarily p, then necessarily p" entails "p is necessary"? We can build up this inference as follows. Firstly, (18*) and (22) entail that if p, then necessarily, p → p. By (19), secondly, from the impossibility of ~p follows the necessity of p. Hence, if it is necessarily the case, that p → p, and □p, then p →□p. However, we cannot make this inference since the inference from the second step of this argument to the conclusion contains modal fallacy: the necessity of p does not follow from the impossibility of ~p, because the impossibility of ~p in the second premise of the argument is to be read as de dicto modality 6 . What follows from the second step of the argument, it is the conclusion that, necessarily, p, but not the inference that p is necessary. Compare: (23) Necessarily, if something has been the case, it has been the case.
(24) If something has been the case, it has been the case necessarily.
At t1, A will happen at t2. Hence, at t1, God essentially knows that A will happen at t2. Is it derivable from the previous sentence that A will happen necessarily? No. We can conclude that, necessarily, if God had, at t1, a knowledge that A will happen at t2, A will happen at t2. Here the necessity is a necessity of the sentence telling us that A will happen, but not the necessity of A itself, i.e., it is not the case that A will happen necessarily. God had knowledge (at t1) that A. We can therefore legitimately conclude that if God had knowledge that A, then it is not possible that God didn`t have a knowledge that A (in the same way as in (18)). But the main idea of the usage of (18) in the context of divine foreknowledge, as it was shown above, is ambiguous. Suppose that we know that KgA. Hence, we can conclude that KgA → A. Then, it follows from KgA that ~◊ (KgA & ~A). But (by 18*), (impossible that ~A) does not follow from KgA. If KgA, and respectively, A, then it follows that A & ~A can`t both be true, but not that A is necessarily true, and the impossibility of ~A follows from KgA only if A is necessarily true. To see why it is so, consider the following argument: (P1) Necessarily, Lionel Messi is either a man or a woman (P2) Lionel Messi is not a woman (P3) Thus, Lionel Messi is necessarily a man (P1) is true in virtue of the Law of Excluded Middle (let us accept for the simplicity of the argument that there are only 2 genders). (P2) is true because Messi is not a woman (he is a man). (P3) however is false; (P1 -P3) is an example of a logical mistake, known as Sleigh`s Fallacy. (P3) would follow from (P1 -P2) only if (P2) had the meaning that Lionel Messi is not a woman necessarily. However, Lionel Messi could have been born as a woman. The fact that Messi is a man is a contingent fact about Messi. Thus fatalist can argue that in virtue of the fact that Messi, necessarily, is either a man or a woman, and Messi actually is not a woman (the fact that Messi is not a woman is logically equivalent to the fact that God at t1 had foreknowledge that Messi is not a woman at t2), then such a state of affairs as Messi is a woman is impossible at t2. However, the fact that Messi actually is a man is compatible with the property of Messi possibly be a woman. Actually, in the world in which Messi is a man Messi, of course, has this property. He is possibly a woman and actually a man in the world in which he is a man.
Therefore despite the fact that (P1) is true, and thus it is necessarily true that Messi is a man only if Messi is not a woman, it is not the case that Messi necessarily is not a woman in the world in which he is a man. Thus, fatalist is unable to argue directly for the impossibility of ~A from the truth of A. The problem is that Messi is not a man at t2 necessarily, even if God at t1 knows that Messi is a man at t2. Suppose that God at t1 knows that Messi is a man at t2. Thus at t2, Messi is actually a man and possibly a women, and so the proposition Messi is a woman at t2 is possibly true, contradicting the statement of fatalism according to which it is impossible for Messi to be born as a woman at t2, if God at t1 had knowledge that Messi will be born as a man at t2. Suppose again that Messi was born in the year 1987 as a man. The fact that Messi was born in the year 1987 as a man is the fact of the past, and thus, according to the Edwards`s (Necessity of the Past), the proponent of fatalism concludes that if Messi was born in the year 1987 as a man, then Messi was born in the year 1987 as a man necessarily.
However, as demonstrated above, the fact that it is true that Messi was born in the year 1987 as a man does not express a de re truth. The truth of this fact is not a truth about Messi. Given that Messi was born in the year 1987 as a man, we can conclude that it is a necessary truth that in the year 1987 Messi was born as a man (de dicto truth), but it is not true that in the year 1987 Messi was born as a man necessarily. Regarding this issue, Fisk asserts that "Edwards`s view of the necessity of the consequence a priori rules out contingency" [Fisk 2016: 340]. According to Fisk, the necessity inferred by Edwards as a consequence of his argument is not a consequential necessity, but rather a necessity of the consequent, contrary to Edwards`s statements that the necessity he describes is the necessity of consequence and not a necessity of the consequent [Fisk 2016: 340]. Thus, Edwards confuses a de re and de dicto types of truth, so he is unable to infer □A from KgA, and also he is unable to infer □A from A. If it is so, then the argument from theological fatalism fails in virtue of the fact that it rests on a modal fallacy.
Another relevant proposal showing how to reject (1) to avoid the fatalistic consequences is the argument of Arthur Prior [Prior 1968]. The logic of this argument is as follows. Suppose that KgA is a fact of the past. Hence, it follows from the Law of Excluded Middle that □ (KgA v Kg~A). From the distribution of the operator, we get □KgA v □ Kg~A. Assuming that KgA (at t1) entails A (at t2), and Kg~A (at t1) entails ~A (at t2), then it follows from □KgA v □ Kg~A that necessarily, at t2, □A v □~A. The last sentence is fatalism. Consider however the proposition "It is now the case that it will be the case". This sentence is equivalent to "It is now (say at t2) the case that it has been the case that it will be the case". Hence, if God had foreknowledge about the fact of future, then the proposition "It has been the case that God had knowledge that it will be the case" is true. But, as it was pointed out by Edwards, if the fact about the past is true, it is true necessarily (compare with the fact of Obama`s birthday from (1)).
Can we say that the sentence "It has been the case that God had a knowledge that it will be the case" is equivalent (in this context) with a statement "Obama was born in the year 1961", i.e., are these statements express the facts of the past in the same way? In accordance with Prior, no. At t1, there is no such fact as "It will be the case that A" (let A be this fact), since there are no "facts" about future contingents at all 7 . Hence, by Prior, we are within our rights to reject not only (at t1 it will be the case that at t2, A), but also (at t1 it will be the case that at t2, A v ~A). Hence, the inference from (1) to (4), following the argument of Prior, is unsound.

Free Will, Truth, and Contingency
Necessarily, if (at t2), A or ~A, God knows it. Suppose that I can voluntarily refrain from A. A, hence, is contingent. Moreover, by (5), if ~A, then God necessarily knows that (a) ~A and (b) ~A is contingent. But if God necessarily knows that ~A (or A) is contingent, whether it follows from KgA that ~A (or A) that the contingent event ~A (or A) turns out to be necessary? Suppose that, at t2, A. We have therefore a challenge coming from the Edwards` assumption (4): If at t1, God had a knowledge that, at t2, A (A is contingent) will be the case, and God`s knowledge about A is necessary, then A is itself necessary (not contingent). It would mean, however, that it follows from the premise that God necessarily knows about a contingent state of affairs that this state of affairs is necessary. Hence, in order to avoid this conclusion, we should provide an argument that God`s knowledge about future contingents does not change their ontological status.
Consider the argument of P. Weingartner [Weingartner 2008: 113]. Suppose firstly that, at t1, God has a knowledge that, at t2, A. A is itself contingent, i.e., A CON ↔ df. ◊A & ◊~A, and it follows from the definition of divine omniscience that whatever God knows, it is true. Thus, the argument of Weingartner [Weingartner 2008: 113] goes as follows ("T" abbreviates "truth") 8 : Hence, it is derivable from KgA and A CON that Kg (T(A CON )). Is this argument sound? Let us substitute A CON for ◊A & ◊~A. Consequently, if, at t1, KgA, then at t2, Kg (T(◊A & ◊~A). The last clause entails that it is possible that KgA → Kg (T(◊~A). But KgA, by (26) and the principle of Infallibility of divine foreknowledge (KgT(A) → T(A)), entails T(A). 7 Zagzebski [Zagzebski 2015: 192-193] indicates that the problem of future contingents rests upon the modal asymmetry of past and future. Her argument goes as follows (◊ t2 is a symbolization of "possible at t2") 1) At t2, A → at t1, KgA (at t2) 2) A CON 3) And by substitution of A CON for A:

25) KgA 26) KgA → Kg (T(A)) 27) T(A) & A CON → T(A CON
(33) A CON → □ KgA CON Thus, if (T (A CON )) follows from KgA CON (from 26 and 28; if God knows that A, God knows that A is true. A is contingent. Thus, God knows that A CON is true), and KgA CON is a kind of necessity (i.e., □ KgA CON ), then it is a necessary truth that A is contingent (not necessary) (from the principle that what follows from the necessary truth is itself a necessary truth): (34) □ KgA CON → □T (A CON ) Thus, it follows from the Weingartner argument that if A is necessarily contingent, then A does not change its ontological status even being known by God. If God knows that some contingent fact about future (A), then God definitely knows that A is contingent, and thus □KgA entails that A is necessarily possible, but not the statement that A is necessary (see Weingartner [2015: 108-109, 113]. Hence, given (32) and the fact that A is not necessary, we have: (35) □ KgA CON → □ Kg (~□A) (35) is the alternative reading of the Edwards` principle (3). By Edwards, necessarily, if God knows A (a contingent fact about future), then A is itself necessary. And by (35), necessarily, if God knows A, then God necessarily knows that A, and God necessarily knows that A itself is not necessary (only contingent). Thus, the consequent of (35), contrary to the consequent of (3), does not include fatalistic implications. The only way to preserve fatalism (in accordance with (3)) is to argue that if God knows that A, and A is contingent, then God necessarily knows that A and not-A is possible (i.e., God knows that A will happen voluntarily), and it follows from the knowledge of God that A will happen voluntarily that it is necessarily the case that ~A will not happen, i.e. it is necessarily the case that God, at t1, does not have knowledge that, at t2, ~A. Hence: (36) □ KgA CON → □ Kg (◊~A) (If, necessarily, God knows that A is contingent, then necessarily, God knows that it is possible that ~A).
Thus, the defender of Pike-Edwards`s fatalism could argue that if (35) was true, it would be true that God could, at t1, have a knowledge that both A and that ~A is possible, and therefore there is a possible circumstance in which God knows (A & ~A). But according to argument (V), it is impossible (even for God) to have knowledge that (A & ~A). By assumption, God knows (at t1) that, at t2, A.
Thus, fatalist can argue that if it is the case that A, then it is not the case that ~A, and it follows from the fact that if God has a knowledge that A, God necessarily knows that A and thus God necessarily does not have knowledge that ~A. But if God does not know that ~A necessarily, then (from the principle that God`s knowledge is infallible, i.e. God have only true knowledge), it follows that ~A not only untrue but also can`t be true. Thus, we can reintroduce a fatalistic argument (as the objection to (35)) as follows: (37) □ KgA CON → □ Kg (~◊~A) Hence, it follows (from the principle of infallibility of God`s foreknowledge) that if God knows A, ~A is impossible. But if God knows that A is A CON , i.e., A CON is true (from 31), and A CON is necessarily true (from 34). But by the definition of A CON and (34), the fact that A CON is necessarily true, and thus ~A is necessarily possible, So it would follow from the assumption that God knows that A is contingent, and God necessarily knows that it is necessarily the case that ~A if it is the case that A: (38) ((□ (◊~A)) & (□ Kg (~◊~A)) → ~◊~A (38) seems to be contradictory. Finally, from (38), by the rule of (Exportation), we have (□ (◊~A)) → (□ Kg (~◊~A) → ~◊~A). Now we have (□ (◊~A)) → (□ Kg (~◊~A) → ~◊~A) → ((□ (◊~A)) → (□ Kg (~◊~A)) → ((□ (◊~A) → ~◊~A)) (from the rule of Propositional Logic (P → (Q → R) → ((P → Q) → (P→ R)). Thus we have ((□ (◊~A)) → (□Kg (~◊~A)) → ((□ (◊~A)) → (~◊~A)) (by Modus Ponens) and so we have ((□ (◊~A)) → (□ Kg (~◊~A)). From (Omniscience) we have that (□ (◊~A)) implies (□Kg(◊~A)) (that is, if it necessarily the case that ~A is possible, then it is necessarily the case that God knows that ~A is possible). And from (Infallibility) we have that (□ Kg (~◊~A)) implies (~Kg(◊~A)) (that is, if it is necessarily the case that God knows that ~A is impossible, it is not the case that God knows that ~A is possible).
So we have: (39) □Kg(◊~A) → ~Kg(◊~A) Since (39) is impossible, we must conclude that it is not the case that it is possible that □ KgA CON entails ◊~A. Hence, fatalism survives. Contrary to fatalism, the argument of Weingartner results in the inference: (40) KgA CON → □A CON However, the consequent of (40), as it has been proved above, does not follow from the antecedent. The correct consequence from KgA CON is (40*) KgA CON → □ KgA CON Or (40**) KgA CON → T (A CON ) Nevertheless, the consequent of (40**) is derivable from the consequent of (40). Hence, if A CON is true, then A CON . By Brouwerian axiom, A CON → □◊ A CON . Then, taking into account the definition of Divine Omniscience, we can construe the following fatalistic inference: The validity of (41) is quite obvious. It follows from (Omniscience) that it is necessarily the case, if Lionel Messi is a football player in the year 2020, God in the remote past had a knowledge that Lionel Messi will be a football player in the year 2020. And the (Necessity of Past) gives us that if (at t1, KgA), then (necessarily, at t1, KgA), and so (at t2, A) only if (necessarily, at t1, KgA). A sentence (42) is true too. By (42), if Lionel Messi is not a baseball player in the year 2020, it is impossible that God in the remote past had a knowledge that Lionel Messi will be a baseball player in the year 2020. Thus, the validity of (42) follows directly from (Infallibility). And by (43), it is necessarily the case that if Lionel Messi is not a baseball player in the year 2020, it is impossible for Lionel Messi to be a baseball player in the year 2020. The argument (41-43) has a form (□ (P → Q)) & (R → ~◊ Q), so (R → ~◊ P). Hence (41-43)  Thus if (by 45), A (i.e. A CON ) is necessarily possible, and God knows that A is necessarily contingent (by 33), thus by (Hypothetical Syllogism; hereafter (HS)) ((P → Q) & (Q → R)) → (P → R)) we have from (33) and (45) (43) is true. If at t2, ~A, then this fact trivially entails that ~KgA. If (43) is true, and it is necessarily the case that the fact that, at t2, ~A implies the impossibility of A at t2, it seems very plausible to think that it follows from the impossibility of A at t2 that KgA is impossible at t1. Our reason is as follows. If ~A at t2, then from the (Infallibility) we have that it is not the case that KgA at t1, and thus it is the case that Kg~A is the case at t1. But Kg~A at t1 is a fact of remote past and thus, by the Edwards`s principle (Necessity of the Past) is a necessary fact. But if Kg~A is a necessary fact, then God couldn`t have a knowledge that A at t1. Thus:  (48) and (49), by (HS), we have ~KgA. From (43) and (Omniscience) of the form A ↔ KgA, we infer □ (~KgA → ~◊KgA), and by (Axiom of Distribution) we have (□~KgA → □~◊KgA). Given the fact that, from (Infallibility), it is not the case that God could have a false knowledge, then ~A implies □~KgA. So we have (□~KgA → □~◊KgA) and □~KgA, and by (Modus Ponens) we have: (50) □ ~◊KgA And now we have from (50) and (KgA → □◊KgA) (by (Modus Tollens)) that (~KgA). From (Omniscience) we have that (~KgA → □ ~KgA). We have also (□~KgA → □~◊KgA). Finally, from another application of (HS) we obtain that ~KgA implies □~◊KgA (51) ~KgA → □~◊KgA But (51) is very implausible: in accordance with (51), if God does not have a knowledge that A, God couldn`t (at all) have a knowledge that A. ~A is contingent, and thus it is the case that if at t2, ~A, then at t2, (~A & ◊A). Thus, A is possible at t2, if ~A is actual at t2, but from (51), it is impossible for God to have a knowledge that A, if ~A is actual, and thus it would be impossible for God to have a knowledge that A is possible at t2 if A is possible at t2, contradicting (Omniscience). Hence, we can suspect that the metaphysical principles underpinning (41-43) are incorrect; we are suspicious about the idea that KgA necessarily entails ~◊~A, and the idea that ~A entails ~◊A. Consequently, we can doubt the validity and consistency of fatalism. Thus, we have a good reason to reject (4) and (h) of Edwards` argument (and thus to reject the whole argument), since this argument, as was shown above, has many false instances.

Conclusion
The argument of Edwards determines indirectly (via Pike`s argument) the content of discussions between fatalists and the advocates of free will in the 20 th and 21st centuries. The core of this argument is the principle of closure and necessity of the past. Historically, the first (and one of the most serious) objection to the argument of Edwards-Pike is a distinction between "hard" and "soft" facts (which can be traced back from W. Ockham). Current debates about the problem of foreknowledge and free will (inspired by "Ockhamist Response") center on the discussions between Ockhamists (Plantinga, Widerker) and Anti-Ockhamists (Fischer) on the logical and metaphysical validity of "soft fact conception", as well as between the supporters of the conception of temporal modal asymmetry (Zagzebski, Weingartner), and the proponents of "Timeless Solution" and Supervaluationism (Cobreros, De Florio, Frigerio). In the light of Fischer`s critique of Ockhamism, we can doubt that the "soft fact conception" is an adequate solution to the problem of necessity of the past, but we can nevertheless be sure that this principle (i.e., the necessity of the past), at least in the version of Edwards and Pike, rests upon the modal fallacy. Hence, without the additional argumentation, this principle is unable to disprove the logical possibility of free will. However, it is clear that this principle can be consistently modified, and every defender of free will theory necessarily faces variations of (1).
It has also been demonstrated that the argument of Edwards has many false instances and contradicts the Brouwerian principle (if something exists, it is necessarily possible). Of course, the question of whether we should accept the universal validity of BP is debatable.