What’s Wrong with the World

The men signed of the cross of Christ go gaily in the dark.

About

What’s Wrong with the World is dedicated to the defense of what remains of Christendom, the civilization made by the men of the Cross of Christ. Athwart two hostile Powers we stand: the Jihad and Liberalism...read more

Philosophy of time post

I have been working quite a bit recently on the philosophy of time, writing an article on God and time. I had some extra material that didn't quite fit into that article. If you want to engage in a little philosophy geekery, with perhaps a dash of philosophy of religion geekery, feel free to head on over to this (long) post at Extra Thoughts on the drawbacks of the position known as presentism.

Commenting is fine either here or there.

Comments (19)

I am putting comments here solely because it is easier for my browser to work with it here.

So what would God know sans creation? Presumably, for God to be omniscient about Napoleon at the "stage" (for want of a better word) of God's life in which God "was" (for want of a better word) not in time, God would have to know tenseless facts about Napoleon and the Battle of Waterloo. So a further epicycle would have to be added to the effect that tenseless propositions became meaningless at the first moment of creation.

Come, come. As long as both theories have difficulties with explaining God's interaction with temporal being, both theories should be willing to admit that "God is a special case" and not assume that "it isn't that way for God" means it isn't that way with respect to created reality. It is not incoherent to suggest that God can speak tenselessly but we humans cannot. A hint that this might be valid: the principle of non-contradiction is often stated as "X cannot be true and not-true at the same time and in the same respect. As long as we humans are going to attempt to understand tenseless statements (if we can understand them, that is) within a framework in which we are UNABLE to formulate the principle of non-contradiction without reference to time, we have to admit the possibility of a limitation there: Maybe our attempts at tenseless statements are not perfectly valid. Maybe we are allowing tense to creep in through the back door. Maybe our use of a supposed "unhinged" present tense verb for such constructions is really a pretense (hah!) at tenselessness that is carrying more temporal weight than we want. (Which would, rightly, put the coffin lid on Craig's odd notion that God's relation to time forces us into A-theory rather than B-theory: God's relation to time is his relation to a created thing, which is a bass-ackwards way to understand the thing "in itself", given God is transcendant to creation.)

Here is another indicator of the problem: we know that our minds work discursively - from thought to thought to thought - even about things that are not temporally ordered. Like the truth of the conclusion of Euclid's first theorem from the premises: the conclusion doesn't BECOME true at a later moment than the premises, the logical sequence is not a temporal sequence. We can even conceive of creatures whose minds do NOT work discursively, for whom all of the proper conclusions of a set of premises are immediately present all at one time. What we cannot do, however, is imagine clearly how they would present this to themselves (or each other). Even less can we imagine how such beings could speak about TEMPORALLY RELATED events in tenseless language. (We can think of a being not limited in a way we are limited, but we can think it by negation, not by positive knowledge.) If our very language (and our thought patterns) are so limited, we simply may not be able to express validly a non-temporally embedded way of speaking about things that are temporally related. We can logically admit that there COULD BE a tenseless language, but we cannot construct one ourselves, any more than we can carry out language non-discursively just because we can logically acknowledge discursiveness is a human limitation of thought.

Probably it would be better for neither side to try to allege such theological problems,

Oh. Well, right, that's probably better. [Of course, you only say that after you submit the theological problems with presentism... :-) Good rhetorical style!]

The connection to the philosophy of time, briefly, is this: We can apply certain mathematical theorems about sampling and representativeness to solve the problem of induction, but only if we take ourselves to be sampling out of a set of entities with particular statistical properties. So, for example, if I have eaten watermelon many times and found it sweet, this can be rationally connected to the proposition that the next piece of watermelon I eat will be sweet. But the connection runs through a proposition about the proportion of sweet watermelons to all the watermelons.

The "problem" doesn't even exist in hylemorphic theory such as Aristotelian realism. This completely solves induction (without Bayesian probability theory) in an epistemological approach without reference to any future at all. I know you are aware of this solution, I am just making it clear to others as well: there are other ways of solving the problem. And this way also happens to solve the biggest difficulty of strict presentism as well: an acceptance of the past as real in a sense but not real in exactly the same sense as the present is a perfectly valid option. It is just the way we can accept that a rabbit "becoming" a lion (because it is eaten) holds the situation where the thing that was a rabbit isn't wholly a rabbit in every sense, while it isn't yet wholly lion in every sense, but the reality is something imperfectly one and imperfectly the other.

All theories with an Absolute Now are forced to take a view of the meaning of physics and relativity theory that is anti-Einsteinian. If you have one Absolute Now, you must deny relativity theory as usually construed and hold that there is absolute time which yields a preferred frame of reference.
Second, he rejects the view that time can be quantized into smallest possible units--temporal atoms known in the philosophy of time as chronons. Craig explains that the chronon view has hugely problematic consequences. One such consequence would be that motion itself would be discontinuous! The idea that things move in a smooth and continuous fashion through space would have to be an illusion. Everything would have to jump.

As far as I understand it, strict "standard" quantum interpretations probably require that reality "jumps" both with respect to chronons and with respect to spatial changes - "planck" units. So "motion" would be discontinuous in both respects, not just one. (Indeed, some interpretations deny any "fundamental" difference between the temporal and spatial dimensions anyway, so whatever is said of one must be said of the other.) The notion is abhorrent to most interpretations of B-theory with real human causality (with free will) as it is to A-theory with human causality, so it isn't really a special problem for A-theory. I suspect that many solutions to the problem will be compatible with A-theories.

He points out that it seems quite impossible to "make up" temporal duration out of strictly durationless moments--the temporal equivalent of geometric points (which take up no space)....Any larger, technically divisible, unit of time chosen for the Real Now would be chosen arbitrarily.

And the answer (well, one answer) is a principled Aristotelian moderate realist response: the "now" isn't a temporal interval, because the now doesn't "construct" periods of time any more than points "construct" lines as parts thereof. Motion isn't the sort of reality that is "wholly real X" during this block 2, and then "wholly real X-1" during the next block 3, etc on to "wholly real Y". It is imperfectly real as X and imperfectly real as Y during the transition, with no moment constituting a proper PART of the change. Aristotle explicitly rejects the now being an interval in his discussion of time, and he does it with a perfectly consistent account of change. He may be wrong, but not because of a philosophical problem with the NOW.

It is telling, though (as you show) that Craig has to resort to nominalism to "solve" the problem, and I agree with you that it isn't really a solution. It would probably take a much more committed nominalist than Craig is to make a go of that.

Is there any sense in which a soft moving block theory can be taken as virtually the same as a weak B-theory? Or, to put it another way, does all B-theory require an absolutely complete resistance to any special frames of reference (outside of human temporal consciousness)?

I think what you're asking in the last paragraph is about something like a growing block view or a moving illuminated block view. Depending on whom you ask, some will tell you those are varieties of B theories and some will tell you those are a variety of an A theory. Craig criticizes them (in a section I did not discuss) as an uneasy hybrid. McTaggart has a famous dilemma (meant to show that time is just unreal, period) in which he criticizes such views on the grounds that they can give no account of the _speed_ with which the Now moves through the block. McTaggart suggests that this question of speed creates a vicious regress: The speed, he suggests, must be relative to some higher-level time. But then that higher-level time would have its own "now," which must move within some yet higher-level time, and so forth, ad infinitum, so that the speed of the moving-now-in-the-block at the initial level has never really been defined at all. Craig takes this to be deadly to all such hybrid views. I'm ambivalent as to whether some sort of "intrinsic speed" solution is available, but since I see no good motivation for such views in any event (any Real Now seems to me otiose), the question hasn't engaged much of my attention.

The thing is, there isn't as much of a problem with the Now on a growing block or illuminated block view (or it seems to me that there isn't), because it doesn't literally *constitute reality*, as it does with presentism. The problem really is acute for presentism among the A theories. I'm pretty sure that if you acknowledge that all parts of the block are in some sense real, then the moving Now really could be dimensionless, because you aren't building reality out of it at any moment. It could be, so to speak, the edge of the knife. Though I admit that I'm not quite sure what that moving Now _means_, either, which is why I'm reluctant to postulate it unless forced by strong philosophical considerations.

Come, come. As long as both theories have difficulties with explaining God's interaction with temporal being, both theories should be willing to admit that "God is a special case" and not assume that "it isn't that way for God" means it isn't that way with respect to created reality. It is not incoherent to suggest that God can speak tenselessly but we humans cannot.

Ah, but don't forget: I am responding here to Craig's view, and he emphatically *does not* hold that God is just different, once creation happens, or that God is a special case with regard to time, once creation happens. Quite to the contrary: Craig takes our theory of time within creation as telling us what things must be like *for God*. If A theory is correct, then, he holds, it must be true for God. God, since creation, travels through time with us, on Craig's view. So if presentism is correct, and if presentism should be understood as involving no tenseless truths (which is my suggestion, not Craig's), then that would be true for God as well since creation, given Craig's notion of God's relationship to creation and time.

Oh. Well, right, that's probably better. [Of course, you only say that after you submit the theological problems with presentism... :-) Good rhetorical style!]

That little stylistic move probably reflects my mild frustration with the particular, published, theological criticism I'm answering ("Jesus is still hanging on the cross on the B theory"), as well as my surprise at, a couple of times, having had charming and intelligent young men sit across from me at a table and bring it up as if it is unanswerable. It doesn't hurt to point out the possibilities in the other direction ("Jesus' death is unreal on presentism.").

The "problem" doesn't even exist in hylemorphic theory such as Aristotelian realism.

Hmm, to confess my ignorance: Perhaps by choosing watermelons as an example, I gave the impression that the issue arises only for things which a hylemorpist would say have natures or essences. That's fine in itself, but it really has little to say for the problem of induction when it comes to things that are just aggregate patterns of past events, like traffic patterns. (I myself would argue that postulating an essence doesn't entirely solve the issue for watermelons, either, but I'm willing to waive that.) There comes a point, I would say, where the most robust realist and essentialist has to be able to do a garden variety induction in order to have some idea about what is going to happen in the future, and his essentialism just isn't much to the point. An Aristotelian who has to commute still needs to know when the traffic jams are going to happen. For such purposes, I think he is tacitly relying on the kind of solution we give. It's Bernoulli in that case, by the way, not Bayes. Much as I love Bayes's Theorem, I think induction is almost certainly its "own thing" rather than reducible to Bayesian terms.

That's fine in itself, but it really has little to say for the problem of induction when it comes to things that are just aggregate patterns of past events, like traffic patterns.

I don't see why we need the model of induction to be on the population of "all traffic jams on Interstate 90" (which would be past and future) in order to carry out the useful kind of induction that we actually do. It is sufficient to induce on "all past traffic jams on I-90" and apply the premise: I don't know of any reason this population will differ from those in the near future, though they may be a separate population with their own rules that I just don't know about yet." That is to say, I think induction on non-essence features of the universe, used to predict future events, to be based on a negation of knowledge of distinction, not knowledge of sameness. And without knowledge of sameness, there isn't any intelligible "population" that includes the past events as well as the future ones that is part of the process of induction.

Many people would say that induction that is not in reference to things that have essences is, inherently, only a probabilistic guess, based on assumption that future events are "in the same category", without being able right now to actually specify all the criteria that constitutes "the category." Or, "of course, I only meant sweet watermelons - eating watermelons - not ethanol watermelons developed for bio-fuel, of which we had no inkling at the time."

That is to say, I think induction on non-essence features of the universe, used to predict future events, to be based on a negation of knowledge of distinction, not knowledge of sameness.

I couldn't agree more. Spot-on. That's what we say, too. But as it happens, the mathematics that best models the inference that uses this insight does make reference to sampling and to the fact that most large samples are representative of the population.

But actual computations can only take into account current and past samples, and there is no certainty whether these are either "large" samples or representative samples of "the population", not without assumptions. I'm an actuary, and one of the constant things we have to include in our analyses is the assumption that our mortality table being used (i.e. a table of past experience) is similar to future activity. It is always an assumption. We always accept that, as an assumption, there is no way of knowing that the samples we have are "representative" enough to cover the future predicted events, until after the fact. The assumption is always carefully tailored so that we exclude ANY reasonable basis that the future we are predicting would be too different, but that's negative knowledge - we don't know of any reason our past experience won't be a good predictor - not positive knowledge that our experience sample is a good sample for the future events. Once you include future events, we simply don't know whether any specific sample we are working with is either large enough or representative enough.

Maybe I am just not getting what you mean. Or that I am not familiar with the math you are talking about.

there is no certainty whether these are either "large" samples or representative samples of "the population", not without assumptions.

Actually, "large" is pretty straightforward, and the margin of error decreases as size of sample increases. The genius of the math is that representativeness does not have to be assumed but rather can be inferred, because most large samples _are_ representative, so it's possible to get a rough idea of how likely it is that one's sample is representative from its size. Even more beautiful, "largeness" is not relative to overall population size. I'll see if I can scare up a pre-print version of the chapter or the predecessor article.

As far as I understand it, strict "standard" quantum interpretations probably require that reality "jumps" both with respect to chronons and with respect to spatial changes - "planck" units. So "motion" would be discontinuous in both respects, not just one.

In traditional quantum theory, time is smooth. The chronon is part of several alternate formulations of quantum mechanics. The chronon is not a part of standard quantum mechanics as far as I know.

The Chicken

The first paragraph in my last quote was supposed to be blockquoted. Don't know what happened.

The Chicken

The genius of the math is that representativeness does not have to be assumed but rather can be inferred, because most large samples _are_ representative, so it's possible to get a rough idea of how likely it is that one's sample is representative from its size.

Again, maybe we are speaking in different tongues, but for actuaries this is what is meant by an "assumption", not an inference.

Even more beautiful, "largeness" is not relative to overall population size.

Only if you don't need to put a definite confidence level / reliability value in the process. Here is a portion of an online sampling size wizard:

Sample Planning Wizard Demo: Population and Sample Data

Specify population and sample properties.

If you don't know the exact values for these data, estimate. For practical suggestions on how to estimate missing values, click the Help button below.

Total population size

Standard deviation

So, yes, if you are willing to give up reliability, you can ignore the size of the population.

Actually, "large" is pretty straightforward, and the margin of error decreases as size of sample increases.

If I am sampling mathematicians for some characteristic, perhaps a sample size of 10,000 is "very large". If I am sampling pennies, 10,000 isn't all that large a sample. If I am sampling bacteria, 10,000 is a fairly small sample. If I am sampling atomic particles for some feature, 10,000 is probably a hopelessly small sample size. Likewise, if my population is "all traffic accidents", past and future, I don't know whether 10,000 is a big enough sample to be a reliable, useful sample, because I don't know what the size of the total population is. Maybe the total population is in the trillions, because people will go on having accidents for the next billion years. It is my impression that the theory of sampling relates the sample size to the total population. Without knowing the size of the whole, the theory doesn't provide an estimate of the reliability. Sure, you know enough to know that

and the margin of error decreases as size of sample increases

but you can't actually estimate WHAT SIZE the error margin is.

you can't actually estimate WHAT SIZE the error margin is.

I feel like Buzz Lightyear. "Can!"

Tony, meet Mr. Bernoulli. Have a look at the writing sample I recently sent you.

Likewise, if my population is "all traffic accidents", past and future, I don't know whether 10,000 is a big enough sample to be a reliable, useful sample, because I don't know what the size of the total population is.

Nope, that's an error. It really is. As long as you can define a relevant reference class, the needed sample size for a reliable estimate of population characteristics doesn't scale like that with the size of the total population. In fact, you can sample an urn *with replacement*, making the number of possible future draws indefinitely large, and still rationally estimate characteristics of the population of balls in the urn.

It is my impression that the theory of sampling relates the sample size to the total population.

If it helps, my layperson's understanding of sample size calculations is this. If you're surveying a truly randomly selected group of subjects, every subject that you add to the sample increases your accuracy. So, if you sample 11 people in a large population, your survey is slightly more accurate than a sample of 10. If you survey 12, it's slightly more accurate than a sample of 11. However, if you graph your accuracy as you add subjects, the line forms a curve. Going from a sample of 10 to 11 might not increase the accuracy in a very large population much, but the _rate_ of increased accuracy is much much greater than if you go from 100 to 101. Basically, the level of accuracy starts to approach a limit.

Depending on how tight your accuracy needs to be, that limit might be reached at an astonishingly low number.

The sample size definitely need not be related to the size of the total population, which might in fact be unknown. There is a way to calculate how much accuracy your increased sample size gives you, and adding more sampling does increase accuracy quite surprisingly quickly.

Depending on how tight your accuracy needs to be, that limit might be reached at an astonishingly low number.

Right. Or the opposite, a quite high number. Sample size needed to provide SUFFICIENT conclusiveness is a function of (a) population, (b) confidence level required, and (c) confidence interval that is tolerated. That's clear. Sample size as a function of population alone is not a linear function: If you graph required sample size as a function of population and hold the other 2 constant, the curve can flatten off to a nearly horizontal line.

Since it is still a function of the other 2 parameters, WHERE it flattens off, at a relatively modest sample size, or at a "large" sample size, is not a fixed value - it *depends*. It is easy enough to construct a combination that requires quite a large sample size - especially by constricting the confidence interval allowed. Here is the output for a "Sample Size Generator"

Confidence Level: 99%
Confidence Interval: 2
Population: 10,000
Sample size needed: 2938

If you increase the population size to 100,000, we get
Sample size needed: 3994.

Now, if we change the confidence interval to 0.5, it looks quite different.
Population: 10,000
Sample size needed: 8,694

Population: 100,000
Sample size needed: 39,963

Still tends to flatten out, but much more slowly at these population levels.

If you don't know what size your population is, AND you have high confidence level and narrow confidence interval requirements, yes, there is still a "large enough" sample size that will do, but it may be quite large.

But this is ALL VERY INSIGNIFICANT to the overall topic, so I don't think we should belabor it.

To go back to the "duration of NOW" problem of presentism, you say Craig says this:

He considers and rejects two views: First, he rejects the view that the present moment has zero measure, that it is durationless. He points out that it seems quite impossible to "make up" temporal duration out of strictly durationless moments--the temporal equivalent of geometric points (which take up no space).

I don't think we should just assume that you would be forced into a position of "making up" duration by bits of "the now" accumulated, strung end to end. Whether under strict presentism or other forms of A-theory, or under B-theory as regards the conscious "now", either way, it is certainly possible that the "now" simply doesn't measure time as intervals thereof, but MARKS time. Just as geometric points don't measure length, but mark it. The end-point of a line segment doesn't measure a portion of the line, doesn't extend over any length.

The Now has to be the constituent entity from which all of time is made, because only what is real in the present is real at all

I don't think that the Now has to be a constituent entity even under Presentism. For, time regards change. What is real in a change is the thing changing, and the attribute with respect to which it is changing is in a sense real and in a sense not-real (precisely to the extent that it is changing, to the extent it is becoming X and thus not-Y). But the CHANGE ITSELF , if it is real at all, is "real" in another sense, and it is not real in exactly the same sense at any moment as compared to any other moment, nor is it real in the same sense as the thing suffering the change. There is, in a change, the substrate, i.e. the subject of change, and the condition with respect to which it is changing. The subject WAS really not-X and BECAME really X (say, it was really not-here, and came to be here). Real is said first of the thing, which has the potency to be X, and not of the motion in which it became X. Real is said of the motion in a derived sense.

Given time regards change or motion, then it is derivative to that reality that underlies motion. If time is a measure of change, time cannot be a primary reality, it would have to be a derivative reality rooted in things that have primary reality. That the ball was there and became here doesn't dispel the reality of the ball all the way through the change - it isn't changing with respect to its ball-ness. So, while the ball "is real" only "at the moment of the now" is valid in a certain sense, it's not the critical sense for understanding the realness of the ball, that sense is more applicable to the realness of its condition "there" or "here" - but even for that, it is the reality of "being here" that is primary, the time of the change being due to the change, not the other way around. The Now doesn't lend reality to the thing. One might say, it is not "because we are standing at the Now that the ball IS real," rather, the realness with respect to becoming here is measurable with respect to before and after, and the Now is the terminus of the two segments of time measure before and after.

This is all straight Aristotle, and for the life of me I cannot see his account assuming a "presentist" stance. I think that the above does not require either A-theory nor B-theory as such.

The Now doesn't lend reality to the thing. One might say, it is not "because we are standing at the Now that the ball IS real,"

Which is precisely where the view you are espousing differs from presentism.

So, yes, the Now cannot be dimensionless in presentism. This problem really is not acute for other theories in the way that it is for presentism.

Which is precisely where the view you are espousing differs from presentism.

I didn't think presentism implied that the Now is causal to the reality being, I thought it only implied a necessary condition. Necessary conditions can be causal or consequent to the thing, can't they?

Post a comment


Bold Italic Underline Quote

Note: In order to limit duplicate comments, please submit a comment only once. A comment may take a few minutes to appear beneath the article.

Although this site does not actively hold comments for moderation, some comments are automatically held by the blog system. For best results, limit the number of links (including links in your signature line to your own website) to under 3 per comment as all comments with a large number of links will be automatically held. If your comment is held for any reason, please be patient and an author or administrator will approve it. Do not resubmit the same comment as subsequent submissions of the same comment will be held as well.