PDA

View Full Version : Possibility vs Actuality



tashirosgt
2010-Jul-24, 01:35 AM
The typical scenario for a probability problem is that there are several "possible" outcomes and one of them becomes the "actual" outcome. We usually think of the possibilities existing at some time and the actuality happening at a later time. Is the concept of time a necessary ingredient in distinguishing "possible" from "actual"?

I suppose one advantage of a "many worlds" viewpoint is that "possible" and "actual" are the same thing.

danscope
2010-Jul-24, 04:45 AM
We can rely on the existance of time. It is sure and consistent. When you read this, you will be a little bit older. Anyone who denies this is
simply fooling himself that he exists beyond this realm. But if you want to exist in proper society, you might get a watch and be on time.
It may be to your advantage. And now it's time for a snack. :) It's about time.

Best regards,
Dan

tony1967
2010-Jul-24, 11:44 AM
I'm not really sure what you're asking here and what it has to do with life? I am a micro electronics and quantum mechanics engineer who’s done a lot of work on probability, so might be able to give an answer if I understood the question. Are you asking if the many world interpretation is correct does it mean there will be more worlds with life? Or is it something to do with the life span of the universe being so vast that there must be probabillity of life?

tashirosgt
2010-Jul-24, 02:54 PM
I'm asking a rather abstract question about definitions. (There was a long discussion in the thread "What is the meaning of the claim 'time is an illusion'?". I think it's just as reasonable to ask about the distinction between "possible" events and "actual" events.)

In a deterministic world, all things that are "possible" will actually happen or already did happen. There is no essential distinction between "possibility" and "actuality". To completely describe an event, you need to put a time stamp on it. Of course, you can adopt the convention that "possible" refers to events that are in your future, but (assuming determinism) all "possible" events are events that do happen at some time. There is no notion of "it could have happened, but didn't".

I'm not talking about the man-in-the-street's definition of "possible". To him, a "possible" event would be one that his knowledge does not allow him to rule out. I'm talking about events that are "possible" according to a theory of physics.

It seems to me that the concept of an event that "could have happened but didn't" must be tied up with the notion of probability. I don't see how to form a useful definition of probability without having events of that type. And I don't see how to have a physical theory of events of that type without resorting to probability.

tony1967
2010-Jul-24, 03:54 PM
Sorry can't help the question seems to be more of a metaphysical or philosophical one than a physical one, with our present understanding of reality. There maybe a chance an event will happen, but it is so small that it is greater than the lifetime of a universe. Therefore the major chance is it will not happen. If the universe is cyclical or is a bubble in the cosmic foam, then it always reverts to actual, however from the frame of the observer it can be a possible. In this cycle of the universe will life happen? Will it be common? In our universe we might be the only life, in the past universe the law of gravity could stop stars forming, etc...

Ken G
2010-Jul-25, 06:28 AM
I'm not talking about the man-in-the-street's definition of "possible". To him, a "possible" event would be one that his knowledge does not allow him to rule out. I'm talking about events that are "possible" according to a theory of physics.I'm not sure you can really draw the distinction you describe. There isn't such a fundamental difference between the incomplete information that a "man on the street" uses to make decisions, and the incomplete information that a theory of physics must use to make predictions. Determinism doesn't change that-- determinism just means that if you had all the information that is possible to have, you would know what is going to happen. The dirty underside of determinism is more relevant here, and is often overlooked: determinism also means that if you do not have all the information that is possible to have, then you can not know what is going to happen! Interesting when you put it that way, no? Because, how could anyone ever have all the information that is possible to have? The entire structure of science is based around not having all the information, it's about figuring out what tiny subset of all the possible information is the important information to have. Science is very much the art of making progress using limited information, and there is nothing about deterministic scientific theories that change this fundamental truth-- all deterministic theories do in practice is help you tell the difference between what you can know and what you can't, and everywhere in between, you will have to use the concept of probabilities.

In other words, in a normal view of reality, we would say that no matter how many things "could" happen, only one "will" happen. This is true whether or not it is possible to know in advance what will happen. What if we had a time machine for looking into the future, but the laws of physics were not deterministic-- say we could see the outcome of a quantum experiment before it happened, even though the wave function only gives us statistical probabilities. So what? Does that suddenly mean that quantum mechanics doesn't work any more, on the grounds that quantum mechanics speaks of events that "could" happen, whereas we know (from our time machine) that most of those things could not happen, since they don't? No-- even in the situation I just described, the non-deterministic predictions of quantum mechanics would still work just fine. If quantum mechanics says that 100 experiments should average 50 to come out "spin up" and 50 to come out "spin down", we can look into our time machine, and sure enough, that's just what we'd see. Quantum mechanics would work just fine, and the use of the concept of probability would work just fine too, even if we had such a time machine, and whether or not we actually used it.

Therefore, there is no connection between the use of probabilistic concepts, and the requirement that there "could be" more than one thing happening in the actual reality. That's because what "could" happen is all about what we know, and not about what "reality knows about itself." Physics is not something reality does to itself, it's something we do to reality, and we need to be able to use incomplete information. In other words, we can use probability concepts when our own information is incomplete, or we can use them when the fundamental information appears to be incomplete-- probabilities function for us just the same either way.

tashirosgt
2010-Jul-25, 04:14 PM
I Determinism doesn't change that-- determinism just means that if you had all the information that is possible to have, you would know what is going to happen. The dirty underside of determinism is more relevant here, and is often overlooked: determinism also means that if you do not have all the information that is possible to have, then you can not know what is going to happen!

I agree if we are talking about predicting an exhaustive and complete outcome. However, I think predicting an event that is only a partial description of an outcome (like "the car turns over" or "the car does not turn over") doesn't require complete information. A theory might predict it from only the "relevant" information.



What if we had a time machine for looking into the future, but the laws of physics were not deterministic-- say we could see the outcome of a quantum experiment before it happened, even though the wave function only gives us statistical probabilities.

If such a time machines is possible then the future would be predictable provided there is only one future (as opposed to a many worlds view). In that case I agree that "possibility" , in the sense of something that "might happen, but might not" is not a fundamental aspect of nature and neither is "probability". However, "Possibility" and "Probability" would still be mutually dependent concepts, even though they are not fundamental aspects of nature ( i.e. not like mass, energy, distance).

Ken G
2010-Jul-25, 05:28 PM
I agree if we are talking about predicting an exhaustive and complete outcome. However, I think predicting an event that is only a partial description of an outcome (like "the car turns over" or "the car does not turn over") doesn't require complete information. A theory might predict it from only the "relevant" information.But even if you are missing the location of a single stick on the pavement, it could possibly make the difference between whether the car turns over, or even if the accident happens or not. Hence, in any situation where you lack complete information, you might need to employ a concept of probability. Not always-- if I let go a ball, I wouldn't say I'm employing probability that it will fall, I can be sure of that, based on experience. But for other questions, I'll need probabilities, whether or not there is some underlying determinism present.



If such a time machines is possible then the future would be predictable provided there is only one future (as opposed to a many worlds view). In that case I agree that "possibility" , in the sense of something that "might happen, but might not" is not a fundamental aspect of nature and neither is "probability". However, "Possibility" and "Probability" would still be mutually dependent concepts, even though they are not fundamental aspects of nature ( i.e. not like mass, energy, distance).I'm just saying that probability is a tool used by physics, and it is very useful to physics whether or not it is a "fundamental aspect" of nature. Indeed, I don't think there really is such a clear line between what is a fundamental tool of physics, and what is fundamental to nature.

tashirosgt
2010-Jul-27, 01:50 PM
But even if you are missing the location of a single stick on the pavement, it could possibly make the difference between whether the car turns over, or even if the accident happens or not.

In any practical statistical problem, I prefer the Bayesian approach, so I'm sympathetic to the view that probability might merely be a measure of the state of "the observer's" knowledge and not some fundamental property of nature. However, I don't completely trust that idea! You point out the possiblity that deterministic predictions might depend on very small details. Aren't probabilistic predictions subject to the same difficulties? If we doubt our ability to make a deterministic prediction for the outcome of an experiment whose result is "the car turns over" or "the car does not turn over", then why should we be confident that we can predict the probability that the car turns over? (I'm not implying that you asserted that we should be confident; it's just a question that arises.)

Ken G
2010-Jul-27, 03:00 PM
In any practical statistical problem, I prefer the Bayesian approach, so I'm sympathetic to the view that probability might merely be a measure of the state of "the observer's" knowledge and not some fundamental property of nature. However, I don't completely trust that idea! And I agree on both counts-- we use probability as a tool, and we should never completely trust any of our tools.


You point out the possiblity that deterministic predictions might depend on very small details. Aren't probabilistic predictions subject to the same difficulties? If we doubt our ability to make a deterministic prediction for the outcome of an experiment whose result is "the car turns over" or "the car does not turn over", then why should we be confident that we can predict the probability that the car turns over? We can't, because we don't know if there is any such thing as "the" probability the car turns over. The only question we can ask is if a large ensemble supports the probability we asserted. It's like weather prediction-- if the forecast is 30% chance of rain, we cannot know if that individual prediction represents the "correct probability" of rain, no matter whether it ends up raining or not. But we can take all the times that the forecast was 30% rain, as a large ensemble, and ask if it rained 30% of them. Since the goal of probability is to get the right statistical trend, to within our ability to classify together situations that involve similar knowledge, all we can ask is whether or not we met our goals.

I think you're asking a question I've also wondered about-- what if half the time a 30% rain forecast occured, a better forecast would have said 40% chance of rain, and the other half the time, a better forecast would have said 20% chance of rain? How would we ever know the 30% forecast could have been improved in that way given the same information that we had? This is a slippery point about probability-- there isn't any such thing as the "real probability" that something will happen, it's all a question of what you are trying to gain from the use of probabilities. You can be meeting your goals, yet improvement might still be possible if you tried a different scheme that better maximizes the potential of the information you have at your disposal. The best scheme gains the most advantage from the same input of information, it just uses that information better, but it's impossible to know if you have achieved that "best scheme."

An example of this is known to anyone who plays the card game "bridge." In bridge, there is a maneuver called a "finesse." All else being equal, a finesse has a 50% chance of succeeding. However, all else is never equal-- there is always a lot of additional information available to the player who knows how to use it. Would the opponent have bid differently, or played their cards up to that point differently, if the finesse will succeed vs. fail? Did they hesitate to make a lead, or do they seem concerned or anxious about something? Expert players can use all that information to change the probability of the finesse working, and make better decisions about whether or not to try it. That doesn't make anyone else wrong who imagines the probability is 50%, it just means they are not using a better probability, one that will help them win in the long run. Experts in bridge (or poker) do win in the long run, for just that reason-- they are playing with better probabilities, and the only way we know our probabilities are not the best is that we keep losing-- but only if we are playing against such an expert.