Miguel wrote:

It seems that those who deny PSR really have a big problem when it comes to justifying our explanatory practices, and explaining the apparent order in the universe.

On the other hand, there are aspects of our explanatory practices that seem to pull in the opposite direction. Two examples that come up occasionally in the literature: a plane crashes, and those investigating it sense that there must be an explanation (Rescher, cited by Pruss); a pressure cooker explodes, we say, because its valve got stuck (and not because it didn't have a hole in it somewhere else) (Putnam, cited by Feser).

In large part, "our explanatory practices" are to look for explanations like pilot failure (for example) in the former case and the valve's getting stuck in the latter case. Those are examples of improper functioning. To find such an explanation of an occurrence is really to be enlightened about the case. But explanations of that sort simply do not have to be available. The same is true in scientific inquiry. The unity of an explanation is part of what we take to recommend it. But a set of phenomena simply does not need a unified explanation of that sort. It's not unnatural English for one scientist to say to another "These data may look like they have an explanation, but so far as I can tell, there is none."

The defender of PSR will want to jump in and say, "Well, what he means is that there's no succinct, simple theory which predicts all of the data; but there is at least the explanation of each data point, the conjunction of which is the explanation of the data." I think this is in part to step away from "our explanatory practices." For even when there is a unified explanation of the data, there would also be the conjunction of the explanations of each data point, which will paint a different, less enlightening picture: is it obvious that these are explanations in the same sense? The former one is far more familiar ("I'm trying to figure out why he was so rude to me at the party last night"), but it is the latter in terms of which PSR will be formulated. So there seems to need to be some work to move from the credibility of the former sort of explanatory practice to PSR.

I don't intend to claim that it is impossible to fill this gap. It's been a long time since I've read Pruss's book, and I don't intend this as an objection to any of his arguments. It just seems to me that "our explanatory practices" form a rather heterogeneous, frequently interest-relative bunch, and characterizing them, let alone roping them together for a defense of PSR, is not a straightforward task.

Miguel wrote:

I don't see any issue. I think we come to expect simple explanations of improper functioning in the cases of a plane crash and a pressure cooker because of induction and our background knowledge about pressure cookers and planes. Without that, however, we're just looking for anything that can make the situation at least a little bit lucid and orderly; a reason why something was the case, and the same is true for the explanation of each data point even in the absence of a succint unified explanation of all the data. Our explanatory practices might follow different patterns in different cases, but there's never any clear occasion of a brute fact and we always seek to privilege explanations.

But the problem is that speaking physically and, it seems, metaphysically, the explanation which appeals to the broken valve is no better than an explanation which appeals to the absence of a hole drilled through any given location on the pressure cooker. The sense of 'explanation' in which the broken valve is the explanation of the explosion seems to be ineluctably interest relative. I mean, in the sense of 'explanation' that we need for a principle like PSR, neither the broken valve nor the absence of a hole is an explanation at all. One would presumably explain the explosion by appealing to the heating on the inside, the material of the pressure cooker, the absence of anywhere for the gas inside to escape, etc. Nevertheless, to cite such facts in answer to the question "Why did the pressure cooker explode?" would usually be some kind of joke. The answer is that the valve was broken.

I am not so much concerned with the possibility of brute facts. The greater problem for PSR has always seemed to me that the principle lacks content until one sorts out the question of what kind of explanation it is talking about. Like principles of sufficient reason, notions of brute facts could be indexed to senses of 'explanation'. For some senses of 'explanation', there are brute facts. What is called a coincidence, for instance, is a brute fact. It's an event that cries out for explanation but lacks one.

Again, I don't see that one does this part of our explanatory practices justice if one just insists that coincidences have conjunctive explanations which are explanations in the same sense as they seem to call for. Is there any explanation of the fact that A is in Pepe's Pizzeria at the same time that B is? Well, there would be if A and B went together, or if B is stalking A, or if they were both recipients of a coupon that must be used during a certain time. But if nothing like this is true, should we then say that the event is nevertheless explained by A's walking there and B's walking there? I don't see that we can do so without changing our topic.

One concern with saying so is the following. A fact which has an explanation is explicable; it makes sense to those who know the explanation and that the explanation explains the fact. It does not seem that explicability comes in degrees. But then it would seem that the fact that A and B are in Pepe's Pizzeria at the same time is equally explicable in the event that a) B is stalking A and b) there is no unified explanation of their joint presence, and they are no more related to each other than any customers of Pepe's ever have been.

I think the fact that A and B are there is, in some sense, a brute fact. Thus if there is any kind of brute fact the admission of the possibility of which is absolutely intolerable, this is not shown by mere consideration of its bruteness.

Miguel wrote:

The explanation for why both A and B are in Pepe's Pizzeria at the same time can be that A and B happened to decide to go to the Pizzeria at the same time, even though they didn't plan that or didn't know each other. So I wouldn't say the coincidence is a "brute fact"; actually we understand coincidences quite nicely, and it is perfectly intelligible to say that A and B individually happened to go to the same pizzeria because they each individually were hungry, wanted pizza, etc. The coincidence is not a brute fact, actually the coincidence is an explanation. We compare and consider coincidences as alternative explanations for a contingent fact: did they conspire to go to the pizzeria together, or did they just coincidentally happen to individually go to the same pizzeria?(emphasis added)

To call something a coincidence is not to explain it. The language you (naturally) use to characterize it shows that: if it is a coincidence, then they "just coincidentally happened" to be there. We speak that way because we do not think there was an explanation of their being there; they "just" were.

That the characterization of some fact as a coincidence does not explain it is also shown by the fact that facts you would agree lack explanations could still be coincidences. Suppose, per impossibile if you like, that brute facts (in some metaphysical sense), such as the popping into existence of A in Pepe's Pizzeria, are possible. Then it would be a coincidence if A were to spontaneously teleport into Pepe's Pizzeria while B were there (supposing A and B to be friends), but to characterize it as a coincidence would not explain it, because the fact would have no explanation.

If I say "It was a coincidence that A and B were in Pepe's Pizzeria at the same time," then I will be speaking falsely unless the fact that A and B were in Pepe's Pizzeria a) "cries out" for explanation and b) lacks explanation. It is not a coincidence that two strangers are in Pepe's Pizzeria, though it is if two friends are, and there is not some explanation of this. (Note that there is no particular explanation which some occurrence needs to cry out for, for it to be a coincidence.) In other words, it is part of the notion of a coincidence that a need for explanation goes unmet. If what we want to call a coincidence turns out to have an explanation, then we would call it an apparent coincidence. The thesis that all apparent coincidences have explanations is thus quite revisionary of our ordinary explanatory practices--unless here we are using 'explanation' otherwise than we did in the characterization of coincidences.

There's something interest relative about the notion of crying out, which is clearly indispensable to the notion of a coincidence. It is noteworthy, though, that it does seem to be a feature of the occurrence which is not modulated by the interests or beliefs of anyone presently existing. That is to say, it is not because an occurrence strikes me or anyone else as crying out for explanation that it is a coincidence; if something does strike me as crying out for explanation (say, the sum in someone's bank account) but does not (because that is just the amount of money he should have, based on his salary and expenses), then I'm wrong in characterizing it as an apparent coincidence (which might be a way of insinuating that the money was acquired through unseemly means).

(It is a red herring that A and B in my original case did not pop into existence, since my point in calling attention to coincidences is not to suggest that they lack explanations in the sense of interest to people formulating cosmological arguments but rather to point out that our ordinary explanatory practices involve denying that coincidences have explanations.)
 

What do you take a coincidence to be? "A coincidence might be an explanation" is most naturally taken as saying that the fact that is a coincidence explains something else, which is not anything I have denied, but which also does not seem to be relevant to our discussion. (That A and B were both in Pepe's Pizzeria, albeit coincidentally, might be an "explanation" (explanans) of the fact that they both witnessed the theft which took place there that night.)

One might take that sentence, alternatively, to mean that the identification of some fact as a coincidence suffices to explain it, or implies that it has an explanation. That would contradict what I have said, though it is a rather obscure and doubtful remark. But to assess its plausibility, we should first hear what you think is being claimed by someone who says that a fact is a coincidence. I have given my own analysis; the person is claiming that the fact cries out for explanation but lacks it.

It is unresponsive to what I have claimed to respond that "everyday coincidences are not brute facts"--for what I have explicitly argued is that, if 'explanation' is polysemous, then 'brute fact' is polysemous, so that coincidences might be brute facts in one sense even if they are not brute in another. In an exceedingly common usage of 'coincidence' and 'explanation', it seems to be the case--indeed, a truism--that coincidences are brute facts; to call something coincidental just is to deny that the apparent need for explanation is met, where this apparent need is something that not all facts have.

Now, one can argue that in fact what is being denied is not that the occurrence has an explanation (in some sense though not perhaps in others) but that it has an explanation of a certain sort (a unified explanation): in other words, one is just claiming that the explanation is nothing more than "the chancey and orderly individual activities of A and B".

I should first of all repeat that I have not denied that there is some sense of 'explanation' in which every contingent fact has an explanation. I have only argued that 'explanation' is polysemous, that 'coincidence' is correctly analyzed as I have analyzed it, and that the notion of explanation involved in the analysis of coincidences (a major part of "our ordinary explanatory practices") is one according to which some contingent facts do not have explanations.

More directly, I don't think that our descriptions of occurrences as coincidences typically carry any commitment to the regularity or orderliness of the 'component' occurrences, the composition of which yields the "crying out" for explanation which is characteristic of coincidences. I think this is borne out in the way people talk about occurrences. (I am not giving ordinary language a privileged place in philosophy; I am just scrutinizing the immediacy of the proposed relationship between "our explanatory practices" and PSR.) When it's suggested that there is some explanation of some conspicuous event, to describe it as a coincidence just is a way of denying that (one typically says, "A and B were just there--stop suspecting a reason"). It is no mystery why in our ordinary practices an additional clause--"but there is some more complex and less satisfying conjunctive explanation"--is not implied: whether or not this is true, it is not the point of the discourse. (For one thing, the situation in which a description of some occurrence as a coincidence is rejected because, again per impossibile if you like, the event lacks an explanation even in the PSR sense, is one hard to imagine; such a thing seems not to be before our minds when we affirm or deny that some occurrence was a coincidence.)