T Clark has rejected a nonexistent form of determinism, fdrake is banging away about his pet worldview, nobody wants to talk to anybody else. This thread was doomed to end this way since the big bang. — frank

In spite of your snotty criticism, I have found this thread very helpful in clarifying my ideas about determinism and predictability. Randomness has now been added which, although perhaps a bit outside the OP, I am also finding helpful and interesting. I hope it will continue. I'm pretty happy with the way things are going.

If you don't find it interesting or helpful, I can see why it wouldn't be satisfying, in which case maybe you should try something else.

Without wading too much into this, I deliberately avoided questions of 'in/determination' - indeed avoided the word(s) altogether - insofar as I think one can treat randomness - in the sense I outlined - without at all engaging in questions of determination and cause. I'll only say that I'm not convinced that one can make sense of the idea of indetermination or randomness ('ontological randomness'), and that what we need instead is a far richer conception of 'determination' than is usually presented, which is usually just fatalism evacuated of any causality whatsoever. — StreetlightX

As I indicated in the OP, I see the determinism/predictability question as just one in a larger set of issues. I would love to broaden that discussion or focus on a different question as we have started to do by bringing randomness into it. Would it make sense to do that in a separate thread. I had been thinking about suggesting that anyway, but I didn't want to disrupt the current discussion, which I am enjoying very much.

Alternatively, we could revise the title to read "Determinism vs. Predictability - Now - New and Improved with added Randomness!"

This is the salient distinction I was trying to tease out with fdrake. Putting it another way is to say that randomness is indeterminability. Ontological randomness would be ontological indeterminism, which is defined as microphysical events being not merely epistemically random, meaning they are not determined by anything at all, they simply happen without cause. — Janus

Are you using "indeterminability" as a synonym for "indeterminism?" I don't think that's correct. It seems to me it is closer to being one for "unpredictability."

I'm not sure I agree with "...they are not determined by anything at all, they simply happen without cause." Maybe it's in agreement with my position, but with all the new terms flying around, I'll need to think about it.

[Edit - changed "predictability" to "unpredictability" in first paragraph.

Thanks. Really helpful post, although it's helpful in making me think about broader issues, not necessarily about coming to conclusions about my original questions. That's fine with me. Some thoughts:

a system can be deterministic but not predictable; the light switch with external random source, predictable but not deterministic; any system with little random variation. — fdrake

Just to be clear, in my formulation, which I've labeled "pragmatic," if an event isn't predictable, it isn't deterministic. Billiard balls yes, multiple coin flips no. Let's work a little on definitions, please. What you are calling "ontological determinism" is what I called "determinism" in the OP, i.e. if someone knows the position and motion of everything at a given time, they can predict the state of the universe at any time in the future. What I think we are now calling "epistemic predictability" is what I am calling "predictability" in the OP, i.e. a system is sufficiently simple that it is practical for us to keep track of all the causal factors in order to predict future states. Is that correct? If so, I will be happy to use those terms in the future.

We can be in a state of great uncertainty with regard to the future of a deterministic system, like a chaotic one, purely due to our epistemic uncertainty concerning it; measurement precision of input variables and initial conditions. Allegedly there cannot be a state of ontological uncertainty with regard to the future of deterministic systems because (their future is not random because {their future states are completely specified by any input state}). So the chain of entailment goes: — fdrake

Maybe I don't understand or maybe I disagree. It is my understanding that chaotic systems are completely unpredictable given passage of sufficient time. Sufficient time is determined by a time scale which varies based on the system.

I want to say more and I will, but I have to go now for a few hours.

But when we come to flip the coin, it does form a distribution of heads and tails; this must therefore arise from variation in our set up; in which initial conditions we propagate forward along their trajectories. Where those initial conditions vary is due to the variability in the behaviour of our body material in a process held as equivalent (coin flipping, "fixed background"), not in states of knowledge regarding the coin. — fdrake

The fixed background is a formal system we analyze. We express our expectations (derived from rules of logical and physical possibility) as probability.

If testing agrees with our expectation, we then feel confident that the real system matches the formal system. Probability does not extruded through the system. It's a proposition about the system.

A sign that we rely heavily on logical possibility is that if we test a system and statistical analysis of the outcome shows that the system isn't performing as expected, we don't update logical possibility. We start looking for the discrepancy in the real set up.

We can be in a state of great uncertainty with regard to the future of a deterministic system ..... purely due to our epistemic uncertainty concerning it; measurement precision of input variables and initial conditions. — fdrake

Such randomness isn't just a result of epistemic uncertainty; our knowledge of the coin and our bodies helps us little to change how coin flipping works; but nor is it a-causal ontological indeterminism - the system is fully deterministic; once a trajectory is fixed, the coin will land as it would land from the start. But when we come to flip the coin, it does form a distribution of heads and tails; this must therefore arise from variation in our set up; in which initial conditions we propagate forward along their trajectories. Where those initial conditions vary is due to the variability in the behaviour of our body material in a process held as equivalent (coin flipping, "fixed background"), not in states of knowledge regarding the coin. — fdrake

It has been my position in this thread that I don't think this make sense from a pragmatic point of view.

As I said in the OP:

It feels intuitively to me that in some, many, most? cases unraveling cause is not possible even in theory. It's not just a case of being ignorant. Part of that feeling is a conviction that sufficiently complex systems, even those that are theoretically "caused," could not be unraveled with the fastest supercomputer operating for the life of the universe. There is a point, isn't there, where "completely outside the scope of human possibility" turns into "not possible even in theory." Seems to me there is. — T Clark

Also, it is my understanding that some magicians (and cheaters I guess) can control their coin flips so that they can control whether a flip comes up heads or tails. I'm thinking through what that means for our discussion.

The equations that update climate models are deterministic, nevertheless they're run lots of times to produce "probability of rain tomorrow" and so on. — fdrake

Sure, I can see that the equations may be strictly deterministic. but that doesn't mean the system in the real world is. I don't think. Maybe. Kind of, sort of.

Are you talking about weather or climate models? I read somewhere that the appropriate time scale for weather forecasting is about a week. After that, predictions become very imprecise quickly. I don't know what the time scale is for climate models. I assume much longer. Also, climate models are greatly simplified as compared to actual climate systems. It is my understanding they can, accurately we hope, predict trends and tendencies reasonably far into the future, but not detailed specifics.

Are you using "indeterminability" as a synonym for "indeterminism?" I don't think that's correct. It seems to me it is closer to being one for "predictability."

I'm not sure I agree with "...they are not determined by anything at all, they simply happen without cause." Maybe it's in agreement with my position, but with all the new terms flying around, I'll need to think about it. — T Clark

My approach in this thread has been to try to get clear about the logic involved in notions of indeterminability and indeterminism and the contexts in which positing them or their antitheses makes sense. So, I am not at all trying to arrive at any metaphysical conclusions.

So, it seems to me that 'indeterminability' is a posit which belongs in the context of epistemology. A complex system could be strictly deterministic (i.e. without any actual random or chance events whatsoever) and yet future outcomes of that system could (in fact arguably would) be indeterminable. Saying that a system is strictly deterministic is an ontic posit. Saying that the same system is indeterminable, or unpredictable, is an epistemic posit. The weather system is a good example. Of course saying that the weather system is indeterminable or unpredictable does not mean that we cannot model the system and make more or less accurate predictions about it; it means that determinations or predictions are subject to degrees of uncertainty which become vastly amplified as the time-frames for the predictions are increased.

The next point I want to clarify is about randomness or chance, which I see as the being the same notions ( in this context at least), and I think positing them about a system constitutes an ontological claim, not merely an epistemological claim. You quoted "...they are not determined by anything at all, they simply happen without cause". It seems to be well-accepted in QM that (at least some) microphysical events are acausal, they simply happen, and that is what I was referring to in the sentence you quoted part of there. The claim that these microphysical events are acausal is not merely an epistemic, but an ontic claim.

But again, I am not making any metaphysical or ontological claims here, I'm merely trying to get clear about what these terms are being used to posit, and in what context, epistemic or ontic, such posits are apt.

Sure, I can see that the equations may be strictly deterministic. but that doesn't mean the system in the real world is. — T Clark

That's right, there is no such thing as a completely deterministic system. That's a fantasy.

(1) No ontological uncertainty in deterministic systems because... — fdrake

There is a very real problem with this assumption, and that is that such a system is not real. A system cannot be completely deterministic, because it is always subject to outside influence. A completely deterministic system would be a completely closed system, which is impossible to construct, and even if it does exist somewhere naturally, it couldn't be observed. There is no such thing as an absolutely "fixed", or determined system, so it makes no sense to talk about what does or does not exist within such a system.

But again, I am not making any metaphysical or ontological claims here, I'm merely trying to get clear about what these terms are being used to posit, and in what context, epistemic or ontic, such posits are apt. — Janus

Sorry. I'm lost. I could keep track of some of what you wrote, but in the end, it spun off.

That's right, there is no such thing as a completely deterministic system. — Metaphysician Undercover

I agree, but from what you've said, I think you and I have different reasons for thinking so.

I do find it surprising that you don't understand what I wrote, since it is crystal clear to me, and I tried my best to express my thoughts clearly. If you are interested enough to want to understand, then indicate the parts of what I wrote you are having difficulty understanding and I will try to explain further, and hopefully clear it up.

One point where we may disagree, or be talking at cross-purposes, is this: from what I understand you think that it is incoherent to say that a system could be deterministic, if it is not epistemically deterministic. For example, the internal combustion engine is epistemically deterministic. That just means it is a simple system whose function is reliably predictable. Since the weather is not epistemically deterministic, being a complex system that cannot be reliably predicted, I take it that you would say that it would be incoherent to think that the weather system could be ontically deterministic. Did I get that right?

I do find it surprising that you don't understand what I wrote, since it is crystal clear to me, and I tried my best to express my thoughts clearly. — Janus

I have found it's very common that I write something that I think is clear but other people don't understand what I'm trying to say.

If you are interested enough to want to understand, then indicate the parts of what I wrote you are having difficulty understanding and I will try to explain further, and hopefully clear it up. — Janus

Sorry, I've gone back and reread the second and third paragraphs of your previous post twice and I just can't figure out what they mean.

I agree, but from what you've said, I think you and I have different reasons for thinking so. — T Clark

That we come to the same conclusion from different approaches is good support for the conclusion.

For example, the internal combustion engine is epistemically deterministic. That just means it is a simple system whose function is reliably predictable. — Janus

It isn't though, because the car breaks down when you least expect it. You're over simplifying "deterministic", and "predictable", in order to say that if you can predict something there is a deterministic system involved.

A completely deterministic system would be a completely closed system, which is impossible to construct, and even if it does exist somewhere naturally, it couldn't be observed. There is no such thing as an absolutely "fixed", or determined system, so it makes no sense to talk about what does or does not exist within such a system. — Metaphysician Undercover

The universe would be the closed system. If there are multiple universes, then the Multiverse would be the closed system. In other words reality itself is the closed system. Determining the motion and position of every particle within the universe would allow you to predict the future of the Universe and everything inside of it - something that may be beyond the ability of the human brain but maybe within the power of a computer.

Now, if the universe is infinite in space and time then that would make it impossible for the universe to be deterministic on large scales of space and time. We might be able to make predictions on small local scales, but our predictions become less reliable the larger and further we try to reach.

"Determining the motion and position of every particle within the universe would allow you to predict the future of the Universe and everything inside of it"

- I don't think omniscience will ever be possible. The physical world is not self-aware so the particles themselves don't know what they're doing and are just passively responding to the various forces.

"power of a computer"

- But such a massive supercomputer would itself exist inside the universe and so in order to make predictions it would have to be simultaneously aware of every particle that constitutes this computer and every other particle in the universe. I don't think this would be possible at the same time.

If clairvoyance could give actual, verifiable predictions of future events, that would be good evidence for determinism — T Clark

Why do you think that?

This is a statistical/probabilistic argument so bear with me.

Imagine a coin flipping experiment. No two flips are causally connected i.e. each event is independent of the next. It is extremely improbable that you'll get a 1000 heads in a row but it isn't impossible. A clairvoyant person could be just one very lucky dude/gal if you prefer.

The universe would be the closed system. If there are multiple universes, then the Multiverse would be the closed system. In other words reality itself is the closed system. Determining the motion and position of every particle within the universe would allow you to predict the future of the Universe and everything inside of it - something that may be beyond the ability of the human brain but maybe within the power of a computer. — Harry Hindu

Complete prediction is not possible from within the closed system. See Determinism and the Paradox of Predictability.

Complete prediction is not possible from within the closed system. See Determinism and the Paradox of Predictability. — ChrisH

Right. So, complete prediction of the closed system in not possible, but that isn't to say that the universe isn't deterministic in that the states-of-affairs in local areas aren't predictable, and that is all we really need. Do we really need a complete prediction of the closed system to accomplish what we want at any given moment? NASA can still get spacecraft to Pluto without knowing where every atom in the solar system is. And if we could acquire the motion and position of most of the particles in the universe would that allow us to narrow down the possible futures of the universe so that we can at least eliminate contradictory predictions?

Maybe I don't understand or maybe I disagree. It is my understanding that chaotic systems are completely unpredictable given passage of sufficient time. Sufficient time is determined by a time scale which varies based on the system. — T Clark

I think you're right, there are systemic reasons why chaotic systems are chaotic, even though (AFAIK) there isn't just 'one thing' which is chaos. Even if the system is sensitive to initial conditions, there has to be a reason for why it's sensitive to them.

One of the things that makes a chaotic system chaotic is how it acts to disperse points away from themselves (called topological mixing); that there exist (sets of) states in the system which travel so far and so fast away from themselves (under the evolution of the system) that their trajectories never return to where they came from after an amount of time. This occurs when (and only when) there exists a state that can evolve arbitrarily close to any other state in the system. There are related notions for this that rely upon probability; if trajectories return in the above sense with probability zero, or if there exist points which go everywhere except collections of states with probability zero, the system will still be chaotic in some sense.

Not all the points of a chaotic system have to have this property for the system to be chaotic. Only some of them do. The trajectory might get stuck somewhere in the state space, like falling down to the bottom of a hill and being unable to get back up its slopes again, and these 'somewheres' are called attractors. Attractors come equipped with sets of initial points that will eventually end up in them, and these are called basins.

Generic points of chaotic systems usually do not belong to basins of attraction, most places in the state space don't lead to being stuck in a rut, so those points never end up getting stuck in a stable repeating pattern of behaviour. This most is why chaotic systems usually have divergent trajectories from small changes in initial conditions; introduced by measurement/instrumental error or limitations of computer precision in representing numbers; though precisely how quick nearby trajectories diverge from one another depends on the system and on the trajectory itself (discussed in the mathematics of the Lyapunov Exponent). The presence of chaos does not depend on the divergence rates, but how much it effects predictability does.

In this regard, a chaotic system can be said to be more predictable (relative to others) when its trajectories diverge slower (than them). How quickly they diverge quantifies the predictability of a chaotic system without an appeal to uncertainty of the initial conditions (like measurement error), the uncertainty of the initial conditions is amplified over time into divergent patterns of behaviour within the measurement precision of the input.

It is extremely improbable that you'll get a 1000 heads in a row but it isn't impossible. A clairvoyant person could be just one very lucky dude/gal if you prefer. — TheMadFool

Keeping in mind that flipping 1,000 heads in a row is no less likely than any other specific series of heads and tails, there are 2 ^1,000 possible combinations of heads and tails. Of course 1,000 heads could come up on your first flip. It's much more likely you will flip coins until the end of the universe before it happens. That, to me, is a fine definition of impossible, which is the case I've been trying to make since the OP.

I think you're right, there are systemic reasons why chaotic systems are chaotic, even though (AFAIK) there isn't just 'one thing' which is chaos. Even if the system is sensitive to initial conditions, there has to be a reason for why it's sensitive to them. — fdrake

As usual, I think you know a lot more about this than I do. Reading more about chaos and complexity are high on my reading wish list. Any particular recommendations?

I'm not sure it makes any difference to my primary position - if predicting future states of a system is so difficult as to be practically impossible, I don't think it makes sense to consider the universe deterministic. I think that's what we have been calling "ontologically deterministic."

As usual, I think you know a lot more about this than I do. Reading more about chaos and complexity are high on my reading wish list. Any particular recommendations? — T Clark

This (and the whole channel) is excellent for visualisations and doesn't skimp on the math.

I'll take a look. Thanks.

but that isn't to say that the universe isn't deterministic — Harry Hindu

No.

Do we really need a complete prediction of the closed system to accomplish what we want at any given moment? — Harry Hindu

I can't think of an instance. It depends what you're attempting to predict. For most relatively closed subsystems I'd have thought that it's not a problem.

It isn't though, because the car breaks down when you least expect it. — Metaphysician Undercover

The car is epistemically deterministic in the sense that the problem can be identified and the car repaired or an irreparable part replaced. On the other hand the human body is not like this; many things can go wrong that we do not fully understand and repair is often impossible.

Sorry, I've gone back and reread the second and third paragraphs of your previous post twice and I just can't figure out what they mean. — T Clark

Again, if you're interested enough to identify which parts or words are giving you trouble, I will be happy to explain. If not, no problem.

Keeping in mind that flipping 1,000 heads in a row is no less likely than any other specific series of heads and tails, there are 2 ^1,000 possible combinations of heads and tails. Of course 1,000 heads could come up on your first flip. It's much more likely you will flip coins until the end of the universe before it happens. That, to me, is a fine definition of impossible, which is the case I've been trying to make since the OP. — T Clark

The universe would be the closed system. — Harry Hindu

The existence of free will demonstrates that the universe, as we know it, is not a deterministic system, nor closed system. To say that there are multiverses which comprise a closed system is nonsense, indicating that you do not know what a "system" is..

The car is epistemically deterministic in the sense that the problem can be identified and the car repaired or an irreparable part replaced. On the other hand the human body is not like this; many things can go wrong that we do not fully understand and repair is often impossible. — Janus

The fact that human beings can identify the problem after the fact, and repair it by replacing the worn parts, does not make the system deterministic. After all, human beings built the system in the first place, and it is the fact that the machinery will break down which makes it non-deterministic.

The existence of free will demonstrates that the universe, as we know it, is not a deterministic system, nor closed system. To say that there are multiverses which comprise a closed system is nonsense, indicating that you do not know what a "system" is.. — Metaphysician Undercover

Sure. A system is an assortment of interacting parts - like neurons, people or universes. Free will is an illusion.

The fact that human beings can identify the problem after the fact, and repair it by replacing the worn parts, does not make the system deterministic. After all, human beings built the system in the first place, and it is the fact that the machinery will break down which makes it non-deterministic. — Metaphysician Undercover

If you predicted that it would break down, and it eventually does, then that is deterministic. Deterministic means that the outcome of some system is capable of being predicted by some mind. It follows some logical pattern. It is logical.

If you predicted that it would break down, and it eventually does, then that is deterministic. — Harry Hindu

That's not true. When you predict that something will happen, and it does, this does not mean that the thing is deterministic. This conclusion would require a further premise which states that something can only be predicted if it's deterministic.

Deterministic means that the outcome of some system is capable of being predicted by some mind. It follows some logical pattern. It is logical. — Harry Hindu

Neither is this true. Minds can predict things which are not deterministic by many different means, like chance, by some system of statistics and probabilities, or through vagueness in terms . I can predict the outcome of a coin toss. If I am right, I've successfully made the prediction. I can also predict that if I flip the coin 100 times half will be heads and half tails. If the score is 51 to 49 I can employ vagueness to claim that it's close enough to count as half and half, therefore my prediction was correct. For a prediction to be correct, it is not required that the thing predicted is deterministic, nor that the thing follows any logical pattern, it only requires a successful strategy by the predictor.