(BTW this may be another point at which I’m not making myself clear:
I’m not arguing that the world must be lawful for prediction to be reliable. I believe that to be a false statement.
Instead, I’m arguing that one must assume the world is lawful in order to have any confidence in any prediction. By thinking one’s predictions can be reliable, one has already assumed one’s world is lawful.)
Because some things, properties, what have you, emerge that are more likely than others, independently of any concrete microscopic realization. There is no rule in the bit string universe that patterns with equal amounts of 1s and 0s must occur more often; they just do. This is inevitably the case whenever some macroscopic view that in some sense ‘forgets’ some of the detailed information of the microscopic level exists, I think.
So the world need not be lawful in order for prediction to be reliable, but I need to believe it is lawful in order to believe that prediction is reliable? I don’t see how that works. You concede that it may be the case that a) the world is not lawful, and b) prediction is reliable. So it may well be the case that I live in a lawless world and validly believe that prediction is reliable, because it actually is. But then why would I need to assume the world to be lawful, in this case? It’d be wrong! I don’t think you want to argue that believing in the possibility of prediction forces me to have a false belief about the fundamental nature of the world, do you? It would open up the paradoxical situation that belief in the reliability of prediction predicts the fundamental lawfulness of the world, which would be wrong, meaning that prediction wouldn’t be reliable…
What makes them more likely than others?
It sounds to me as though you’re describing a law–“patterns with equal amounts of 1s and 0s occur more often.” Inhabitants of this universe must assume there are laws in effect in order to make predictions–and lucky for them, they’re right. There is at least one law in effect in their universe: patterns with equal amounts of 1s and 0s occur more often. This is true in their universe, and it supports reliable prediction.
Your predictions aren’t valid–if the world is actually lawless, then your predictions are merely lucky. Makes no difference to you, of course–but that’s just part of how lucky you are.
You are consistently lucky. But it’s just luck. And it would make no sense for you to say “I predict that I will continue to be lucky.” Such a prediction is nonsense. (I feel like it’s self-contradictory but a few minutes poking around failed to yield a demonstration of this so I’m not sure.) Instead, you would have to assume (though you’d be wrong) that your predictions are catching on to some laws that are in effect in your world. If you didn’t assume this, you’d have no reason to go on making predictions.
Of course I do. It’s basically the very thing I’m arguing for.
As I already said, it’s possible for all we know that all of our coherent-seeming observations and accurate predictions are just a massive coincidence–but we can’t believe this if we intend to go on making predictions. If the world isn’t lawful, then in order to believe in the in-theory reliability of prediction, we must have a false belief that the world is lawful.
The last inference (“meaning that prediction wouldn’t be reliable”) doesn’t follow. As I’ve said, lawfulness is not required for prediction to be reliable. The right kind of massive random coincidence could yield a situation in which predictions are (massively coincidentally) reliable.
Lawfulness is not required for prediction to be reliable. But the practice of making predictions and expecting them to be reliable requires a presupposition that the world is lawful. If the world isn’t lawful, does this mean then that we must have a false belief in order to engage in the practice of prediction? Yes, it does.
(Lucky for us, though, the world is lawful. 
)
No, the bit-string universe is completely described by: the state of the universe at any given instant (‘whenever you look’) is a random string of bits. There is no law that 1s and 0s must occur with equal frequency; that just happens to be the case, on average. It’s something we get out, not something we must put in. It’s a result of self-organization. The reason is simply that there are more bit strings for which this equidistribution holds than there are bit strings for which it doesn’t; thus, if you are to draw a bit string out of a hat, one can predict with some confidence that it will be one with roughly equally many 0s and 1s. But again, there is no law that mandates this – such a law would be completely superfluous; things come out that way by themselves. It would require postulating a law in order to make things come out any other way, something which ‘tilts’ the probabilities in favour of, say, bit strings composed predominantly of 1s. But absent any law, what you’ll get is equidistributed bit strings.
What about the case of a computation randomly distributed over a random universe? According to your experience, the next step will follow logically from the previous one; but in the ‘actual’ universe out there, the previous step may not even have occurred yet, or might never occur, and again, everything is perfectly random at bottom. Yet still, your predictions are valid, and you have reason to expect them to be so.
But even so, if the belief in the reliability of prediction predicts a fundamentally lawful universe, yet the universe is fundamentally random, then we have a clear case of a false prediction – which means that prediction isn’t reliable (since in this case, it’s necessarily false). I don’t see why that doesn’t follow?
Perhaps we should get back to basics: the bit string universe, where macroscopic properties are built from elementary propositions, the truth value of which is determined randomly – i.e. where everything corresponds to a random bit string --, would you agree that it is lawless? If not, why not? And can you give an example of a universe you would call lawless?
The law I quoted isn’t “ones and zeroes must appear with approximately equal frequency.” The law is “Ones and zeroes appear with approximately equal frequency.”
The creatures in this universe, presumably, haven’t seen all the ones and zeroes in their world–just the ones in theird neighborhood. (If they’ve seen them all, then prediction is impossible because there’s nothing to predict.)
Having seen only some ones and zeroes, they predict that the ones they’ll see later will also have an approximately equal frequency of ones and zeroes.
To predict this, they have to presuppose their world is lawful.
And their presupposition and prediction, in this scenario, both turn out to be correct.
There is a law operant in their world–ones and zeroes occur with approximately equal frequency.*
And their prediction is correct–ones and zeroes they encounter in their future will tend to occur with equal frequency.
*Again, there’s no modal terminology here–the law isn’t that they “must,” it’s just that they “do.” Laws do not require modal terms in their formulation. “What goes up comes down.” That’s a law. You don’t need the “must.”
What reason do I have to expect them to be valid?
Presupposition isn’t prediction. Also, prediction doesn’t have to always be right in order to be reliable.
It may or may not be lawless. It depends on how the random determination turned out. If, purely by coincidence, the propositions turn out to all be compatible with the statement “what goes up comes down,” then the universe turned out to be lawful.
I ahem don’t think it’s actually possible–it’s logically impossible, in fact, is what I think–for a universe to be lawless. But that’s not what I’m arguing for here and I’m not nearly even close to feeling as sure about that as I am about the fact that we have to presuppose lawfulness to make predictions.
It occurs to me that I might think something even stronger than what I’ve been arguing for–but which you may be more inclined to agree with.
I think (if I understand this term correctly) that in order to engage in the practice of prediction, agents must presuppose that their world is compressible.
How does that hit you?
I’m not talking about creatures, and I’m not limiting the amount of 0s and 1s that ‘can be seen’ or anything. Here’s a complete state of the universe: 110001100111001001000100100111. Here’s another: 010100011110001110010101101010. And so on. If I gave you enough of these bit strings, you could determine that they follow no rule. From this, you could predict that most bit strings will be equidistributed.
This is just an observation – I don’t see how one could grant it the ontological status of a law. Laws, at least the way I understand them, have to be prescriptive, rather than descriptive – something turns out a certain way because of a certain law holding. Otherwise, everything is indeed lawful, as one always can resort to mere description. Then it’s vacuously true that something turning out a certain way implies the existence of a law to that effect, merely by virtue of it turning out a certain way.
Well, it’s a computation, which means there is a function that embodies the transition from one state to the next – knowing that function enables you to carry out predictions, i.e. calculate the next state, which always will come true, as you will subjectively experience the next state.
The propositions in themselves are wholly arbitrary; what matters is that the dynamics, i.e. the way one state follows the next, is completely random – if that isn’t lawless, then I don’t know what is. And if there’s nothing that’s lawless, then the term ‘lawful’ looses its meaning as well.
I think about lawfulness in terms of compressibility, yes. But the bit string universe is completely incompressible by construction – a bit string is random iff it is incompressible. Nevertheless, because macroscopic properties can be found such that distinct microscopic states correspond to indistinguishable macroscopic ones, a ‘lossy’ kind of compression is possible, though one looses the information about the detailed microscopic state – the description ‘as many 1s as 0s’ does not distinguish between ‘0101010101…’, ‘00000…11111…’, ‘00110011…’ and countless others; but with respect to these kinds of properties, the description of the world is indeed compressible. So if the ‘true’ microscopic nature is hidden, or does not play a role in some other way, a macroscopically lawful, compressible description emerges.
I retract the above. I’m using “compressible” incorrectly.
For us to talk about “prediction,” we have to be talking about an epistemological state which involves having “seen” some of what’s in the world, and having “not seen” some other things that are in the world. If information is available about every bit of the world, then prediction is impossible by definition. (Note that by “every bit” I mean not just every bit in the world’s “space” but every bit in its “time” as well.)
Then their equidistribution is a law that holds true in that universe.
I don’t know how else to say this?!
I predict that every future bitstring you show me will tend to be approximately equidistributed in terms of ones and zeroes. On what basis do I make this prediction? On the basis of the following law: ones and zeroes tend to be equidistributed.
You throw the apple up, I predict it will fall down. On what basis? On the basis of the following law: Things that go up come down.
(BTW it’s not directly relevant, but you’re wrong to say I could determine they follow no rule. No matter how many bitstrings you give me (assuming that number is finite) the series will be compatible with an infinite number of candidate rules.)
It’s controversial, I think (though I don’t actually know tbh) but in my view a law is just a fact about patterns that hold in a system. Nothing prescriptive is necessary in the formulation of a law. I know for sure that many philosophers in the early 20th century thought this as well. I used to disagree with it but lately I’ve come around to it.
To think I can rely on prediction, I have to think there are patterns that hold in the system I inhabit. If I don’t think there are such patterns, I’ve got no reason to think my predictions will tend to have any validity.
I don’t need to think there’s anything modal about the laws themselves. I do need to think that the truth of these laws supports certain (physical) necessities regarding my own experience.
Then that function is a law.
Lawful just means (as I’ve been using the term) containing patterns.
Some patterns are better than others, but you’re right that anything at all could be called a “pattern.” Hence my move to talk of “compressibility” in my most recent post–though I now think that’s not the right word to use.
The view is something like this:
For me to engage in prediction, I have to presuppose that the world I inhabit contains patterns of a type I can grasp and which hold not just in my local environment but everywhere.
That’s certainly not an empty presupposition–it may well be false. But without the presupposition, you’re left with no reason to think your predictions are likely to be true.
“Everywhere” might be relative here (I’m not sure). To make predictions about region X, I need to think that the aforementioned patterns hold in X, but not necessarily everywhere-everywhere. (So “everywhere” might be relative to particular domains of discourse or something.) But in physics at least we make predictions about the entire universe. So to do physics, we have to presuppose the patterns we’re familiar with hold throughout the universe. We may be wrong about this, but we have to presuppose it to get the practice going in the first place.
(And if the world can be reduced to physical truths, then this presupposition isn’t just necessary for physics but for every human activity.)
It was realizing that that led me to the subsequent retraction post.
The anthropic principle.
ETA: replied to a post I now realize may be outdated
According to this article, the “regularist/necessitarian” debate is a live one. I guess you and I are on different sides of it.
A cornucopia of views are described here as well.
It looks like I am for Supervenience, (described in the section titled, confusingly, “Systems,”) and you HMHWare for Antirealism (described, unconfusingly, in the section titled "Antirealism.)
Heh, don’t I recall correctly that I’m for compatibilism and you’re for strong determinism when it comes to free will? I feel like there’s a similar dynamic in our choice of positions on the present issue.
But then, everything that happens happens according to a law – the law that mandates that it happens in the way that it happens to happen. I don’t think I can see how this would be a valuable notion of law – it’s just saying that ‘things happen the way they happen’.
It’s also a useless notion if you want to be able to make predictions, as it can’t be abstracted – if the law does not give you insight at what the salient features are of something happening a certain way, i.e. why it happened that particular way, then you can’t use this law to predict the outcome in a related but similar situation, as there’s no way of ascribing causal force to any element of the situation.
But if you only make the observation that things come down if thrown on a Sunday afternoon, wind coming from Northwest, precisely 13.7°C, then using your descriptive notion of law all you will have learned is that things come down if thrown on a Sunday afternoon, wind coming from… etc. If there is nothing that requires for things to come down if they are thrown, no prescriptive element (like, say, gravity), then all you can ever learn is the description of what happens.
Well, but for any amount of certainty you might require before you believe that the bit strings are random (as long as it’s strictly less than 100%), I can supply you with enough bit strings such that this amount is met.
But again, then lawfulness is empty, because everything is lawful, and prediction is impossible, since in the absence of a prescriptive element, there’s nothing to point to as being the reason of something happening a certain way, so you can never say, ‘because [some element] is present, this is going to happen’.
But a law of what? The universe, being random, does not conform to it; it exists only by virtue of your computation being possible, and embeddable in the underlying random dynamics.
Indeed. But the random does not contain any patterns – this is what makes it random.
There’s no pattern you could grasp in a series of random bit strings. Nevertheless, you can make predictions about these bit strings – such as them being predominantly composed of equal amounts of 1s and 0s. It’s precisely the lack of pattern that allows you to make this prediction.
I don’t know how you’re able to say that “ones and zeroes are equally distributed” isn’t a pattern. It’s a fact that tends to hold true through several iterations of a system. That’s a pattern, by definition. Isn’t it?
If you are going to engage in the practice of prediction, you must presuppose that there are patterns in your world that do hold in situations similar to but distinct from those you are directly familiar with. You may be wrong in that presupposition. Your predictions may only be lucky ones. But to make the prediction, you have to make that presupposition.
A pattern, to me, implies regularity, repetition. Random bit strings have neither; that’s what makes them random. There are no patterns to randomness. (And anybody who says differently is, to paraphrase von Neumann, living in a state of sin.:p)
A pattern in a string of bits minimizes the surprise of any new bit’s value; in a random string, any new bit’s value is as surprising as the one before.
If any fact that tends to hold true through several iterations of a system is a pattern, then ‘there is no pattern’ is a pattern – which doesn’t make sense to me.
This may just be a terminological issue then. What would you call a phenomenon whereby the same can be said of every part of something, once you’ve divided it up into parts the right way?
Well, the “pattern” (as I’ve called it) isn’t a pattern of bits, it’s a pattern of strings. My surprise at the equidistribution of the ones and zeroes is minimized because I already know to expect that pattern to hold.
Part of me wants to say “if ‘there is no pattern’ is a pattern, this just means there’s no such thing as a patternless thing. No contradiction there.” To which you’d probably reply that this makes “pattern” an empty concept. In response to this I’d emphasize that in my latest revision of the view I want to defend, I put something admittedly vague in there about the pattern needing to be “graspable”. (I now think what I want to say is–we have to presuppose that there is an exceptionless but possibly ungraspable pattern to our world, which is capable of supporting exception-ful but graspable patterns in that part of the world which is accessible to us cognitively and sensorily.) “Graspable pattern” certainly isn’t an empty concept.
Another part of me wants to say that there are heierarchies of patterns, such that we might say something like “‘there is no pattern-sub-1’ is itself a pattern-sub-2.” Then what do we have to assume to make predictions? That there’s some pattern-sub-something, I guess? I know when you’re doing this kind of heierarchy with truth predicates you’re not supposed to be able to talk about “truth-sub-anything” but I don’t know if the same logic would apply in this case.
That’s because philosophy is a mental analysis of existence, not a direct manipulation of symbols or discrete movements or objects.
The brain of a savant can react to a physical stimulus in much greater speed - in lack of a better term - than the average human. What it cannot do is comprehend ideas or abstract mental constructs that are philosophy’s basic building tools.
Cite? I am unaware of any rule or analysis implying that a savant can’t be extraordinary at comprehend ideas or abstract mental constructs. For example, brains have been described as analogy machines. Some analogies are more abstract than others, and some can be more or less adept at recognizing very abstract analogies.