16 replies to this topic

[octopuppy]

I've seen an awful lot of rubbish on the boards about multiple parallel universes (or "multiverses"), string theory, and other such drivel since I've been on here. I suppose some people are bound to buy that sort of bunk, just like there are bound to be people who believe in ghosts and fairies and any hope for the Steelers to win it all this year. But the mere fact that people go around holding such delusions isn't the disturbing part. What's really bothersome is that some people think this almost classifies as science. Absurdity! Multiverses are inherently unprovable. Could I go outside right now and run an experiment that could either definitively prove or disprove the existence of even a single parallel universe? Could you even theoretically come up with an experiment that could ever prove or disprove the existence of multiverses? No, I say!

So, since the multiverse "hypothesis" is an inherently unobservable and unprovable speculation that as far as I know or care to look up does not even need to be invoked to explain any natural phenomena, then off to Occam's guillotine with it! Henceforth, anyone who posts anything about these mythical multiverses must acknowledge the fact that there is no scientific value in what they're saying, and must use quotes around it to help emphasize the fact that it shouldn't be taken seriously, and must use air quotes whenever talking about it in real life, and must give equal time to every other postulate like the Flying Spaghetti Monster.

Better late than never, I'll wade in with a defence of discussing multiverses since I've done a fair bit of that (this passed me by when you first posted it). Multiverses are a very relevant topic because it affects the kind of universe in which we would expect to find ourselves. In a recent topic I'd brought up the subject of whether we might expect physics to be simple or complicated, which is a very worthwhile issue, and the distribution of existing universes would have a bearing on that. Of course you cannot prove that there are many universes, but neither can you prove that there is only one. Which hypothesis is the more speculative? IMO it's the latter, and Occam's razor cuts the other way.
What precisely is the problem with an untestable hypothesis anyway? Your criticism seems to be based on the fact that untestable hypotheses are... untestable. But you have not ascertained that they are irrelevant. Take another untestable hypothesis, the existence of a god (for convenience I'll bundle the afterlife in with that). If there is such a thing, we would be wise to spend our lives securing an eternity in heaven, if not, we would be wise to live our lives to the full. Although untestable, it's hardly inconsequential. Incidentally, another reason why discussion of multiverses is relevant is in debunking the god hypothesis. One of the reasons for believing in a god is that our physics is fine tuned for life in a very improbable way. But it's only improbable if you make the assumption that this universe is all that there is; our version of reality has been selected to exist and all other possible worlds rejected. Why think that? Once again we have a choice between two untestable points of view which probabilistically have a great effect on unknown but important aspects of the world in which we live.
Untestable hypotheses may be very relevant in the choices we make. As Harry Callahan put it, "I know what you're thinking. 'Did he fire six shots or only five?' Well, to tell you the truth, in all this excitement I kind of lost track myself. But being as this is a .44 Magnum, the most powerful handgun in the world, and would blow your head clean off, you've got to ask yourself one question: Do I feel lucky? Well, do ya, punk?"

[dawh]

Better late than never, I'll wade in with a defence of discussing multiverses since I've done a fair bit of that (this passed me by when you first posted it). Multiverses are a very relevant topic because it affects the kind of universe in which we would expect to find ourselves. In a recent topic I'd brought up the subject of whether we might expect physics to be simple or complicated, which is a very worthwhile issue, and the distribution of existing universes would have a bearing on that. Of course you cannot prove that there are many universes, but neither can you prove that there is only one. Which hypothesis is the more speculative? IMO it's the latter, and Occam's razor cuts the other way.
What precisely is the problem with an untestable hypothesis anyway? Your criticism seems to be based on the fact that untestable hypotheses are... untestable. But you have not ascertained that they are irrelevant. Take another untestable hypothesis, the existence of a god (for convenience I'll bundle the afterlife in with that). If there is such a thing, we would be wise to spend our lives securing an eternity in heaven, if not, we would be wise to live our lives to the full. Although untestable, it's hardly inconsequential. Incidentally, another reason why discussion of multiverses is relevant is in debunking the god hypothesis. One of the reasons for believing in a god is that our physics is fine tuned for life in a very improbable way. But it's only improbable if you make the assumption that this universe is all that there is; our version of reality has been selected to exist and all other possible worlds rejected. Why think that? Once again we have a choice between two untestable points of view which probabilistically have a great effect on unknown but important aspects of the world in which we live.
Untestable hypotheses may be very relevant in the choices we make. As Harry Callahan put it, "I know what you're thinking. 'Did he fire six shots or only five?' Well, to tell you the truth, in all this excitement I kind of lost track myself. But being as this is a .44 Magnum, the most powerful handgun in the world, and would blow your head clean off, you've got to ask yourself one question: Do I feel lucky? Well, do ya, punk?"

Just going to put it out there that I got the impression that plasmid's post was stated with tongue firmly planted in cheek, since it doesn't fit the tone of voice of his usual posting. Though I guess it would be up to him to clarify on that specific point.

"Six. Definitely Six."

[octopuppy]

Just going to put it out there that I got the impression that plasmid's post was stated with tongue firmly planted in cheek, since it doesn't fit the tone of voice of his usual posting.

Well, he's playing devils advocate. I may have joined the game a bit late, but I wonder if there were more substantial arguments in store.

"Six. Definitely Six."

[plasmid]

Indeed, the main thrust of the initial argument is to criticize the stance toward religion that the scientific community seems to have taken: saying that religion is outside of the realm of science, and so cannot be proven or disproven. Scientists NEVER seem to take that kind of attitude with any other question they feel like tackling. To give up so easily on addressing a question seems shameful for any self-respecting scientist.

But I still feel like playing devil's advocate and arguing against the side that I actually believe in, because it’s both fun and helps keep me honest when I do it habitually. So here goes.



The problem with untestable hypothesis is that it seems to be trivially easy to prove that they are useless. If a theory is truly untestable, then no matter what happens, its truth or falsehood would not affect any outcomes. To use a concrete example, suppose we postulate that gravity is exerted because anything with mass emits "gravitons" in all directions that are massless, fast (say speed of light or so), and produce no electromagnetic field; but which draw other objects toward their point of origin almost like a wave. For simplicity, I'll say that the "fast" bit is untestable because even if you accept a competing theory – that mass exerts gravity by distorting space-time – then those distortions in space-time cannot travel faster than light due to relativistic considerations.

Would it be worthwhile at all to even address the question of whether gravity is exerted by gravitons versus warping space-time? And would the answer even be meaningful? In this case, I would lean toward no. I would also lean toward binning parallel universes in the same category of "perhaps enticing the imagination, but overall irrelevant."

Octopuppy's Dirty Harry counterexample does have its merit. Even if you aren't rainman and have no way of testing whether there's a bullet left in his gun (aside from the one potentially lethal way), it would be very useful if you had that information. And it would be nice to know for sure whether contributing tithe will pay dividends in an afterlife. And for that matter, it would even be neat to have a way of telling which way stock prices will go next week. The problem is that, although the solution would be valuable, if it can be proven that the solution is unobtainable then it seems logical to say that the doomed quest for such a solution should be abandoned.

[octopuppy]

Indeed, the main thrust of the initial argument is to criticize the stance toward religion that the scientific community seems to have taken: saying that religion is outside of the realm of science, and so cannot be proven or disproven. Scientists NEVER seem to take that kind of attitude with any other question they feel like tackling. To give up so easily on addressing a question seems shameful for any self-respecting scientist.

Really? I was under the impression that this stance was more generally taken by proponents of religion when invited to come up with evidence for their claims, or any suggestions as to how such evidence might be obtained. There may be some people out there claiming that there is verifiable evidence of religious claims, but such claims have a habit of evaporating under close examination. The more common position is that religion is indeed outside the realm of science which leaves nothing for the scientific community to do. If I claim the existence of a being whose existence cannot be scientifically tested, why would scientists wish to test my claim? They are wasting their time if I am right, and wasting their time if I am wrong.

The problem with untestable hypothesis is that it seems to be trivially easy to prove that they are useless. If a theory is truly untestable, then no matter what happens, its truth or falsehood would not affect any outcomes.

If it is temporarily untestable, there may be outcomes. You can test the god hypothesis by dying, but then it's too late to act on it. Same with the Dirty Harry dilemma.
The multiverse hypothesis may be permanently untestable, but still change what we would expect to see from physics, and this effect could perhaps be verified. That wouldn't prove the hypothesis, because the effect is only a probabilistic one and we can't be sure it isn't caused differently. But the same goes for any theory of physics. A physics theory might be disproved, but cannot be proved correct. If it makes useful predictions and we verify those predictions to be true, then we have failed to ascertain whether that theory of physics is universally true or not. It is at best an untestable hypothesis.

To use a concrete example, suppose we postulate that gravity is exerted because anything with mass emits "gravitons" in all directions that are massless, fast (say speed of light or so), and produce no electromagnetic field; but which draw other objects toward their point of origin almost like a wave. For simplicity, I'll say that the "fast" bit is untestable because even if you accept a competing theory – that mass exerts gravity by distorting space-time – then those distortions in space-time cannot travel faster than light due to relativistic considerations.

Would it be worthwhile at all to even address the question of whether gravity is exerted by gravitons versus warping space-time? And would the answer even be meaningful? In this case, I would lean toward no.

I don't know whether both those theories would make precisely the same predictions about the behaviour of gravity. Maybe they would seem to at first but then on closer examination one of them would produce a slightly different result under certain circumstances. Then it may have proved worthwhile to consider them both. Alternatively, one of them might suggest other behaviours beyond gravity. Physics is about looking for descriptions of the workings of the universe which are as simple, elegant, and consistent as possible. In the quest for elegance, sometimes one hypothesis suggests another and a wider consistent picture of things is formed. That in turn could produce useful predictions which validate the process. But it still doesn't prove anything to be correct.

And for that matter, it would even be neat to have a way of telling which way stock prices will go next week. The problem is that, although the solution would be valuable, if it can be proven that the solution is unobtainable then it seems logical to say that the doomed quest for such a solution should be abandoned.

If you secretly find out that a top investor is putting all her money in a stock which she firmly believes will go through the roof next week, would you invest in it too? You will only know if the tip is a good one when it is too late to act on it. Until then it's an untestable hypothesis, but that's no reason to ignore it. The odds seem pretty good in that case, and probability is what a lot of this comes down to. Take the example of a universe fine tuned for life. Here's 3 explanations for why:
1) Dumb luck
2) God did it
3) This universe is one of many, so it's just what you would expect
The first explanation is ludicrously unlikely. The second is unlikely because there is no reason why a god should happen to exist. If it can be shown that a multiverse is exactly what you would expect (albeit by rather speculative and unprovable means) you have a strong probabilistic argument for that, but it can not be verified. The probability cannot be calculated and other explanations cannot be excluded. Occam's razor is vague like that. Is a simple hypothesis more likely to be true than a complex one? How much more likely? It's hard to be precise but it doesn't render Occam's razor worthless.

[plasmid]

Individual scientists when talking in private will likely say that religions are absurd, but the scientific community as a whole doesn't denounce religions such. To incontrovertibly prove my point, I'll refer to the Simpsons episode where Homer gets really smart and scientifically disproves the existence of religion and hands the proof to Ned (who promptly destroys it). It happened in parody because it hasn't yet happened in reality. While that generally has more to do with social reasons and not wanting to get your funding cut and such, the defense of "it's an untestable hypothesis so we don't have to address it" is invoked as the more purely scientific reason why the issue is ducked. The original post was a critique of that excuse.

Now back to the issue of untestable hypotheses in general. One important distinction to make is the difference between something being testable and being provable. Proof is tough. Testing is much easier. With the stuff around my house, I can't prove that my microwave oven is emitting microwaves since I can't directly observe them. But I can tell that my coffee gets rewarmed if I put it in there, and infer that since it's unlikely to be getting heated by any other mechanism, then my microwave is probably actually emitting microwaves.

Testable hypotheses are fine. A "working model" of the universe that's correct a lot of the time can be a handy thing. It does involve subjectivity about how much evidence something needs before you can consider it testable enough to be pretty likely to be true, but that's the situation we're stuck with.

As for untestable hypotheses, like the existence of dimensions outside of the three accessible to us (plus time if you want to count it as a dimension) and wormholes and such, they have no role in anything except making for an entertaining science fiction story. It would be a completely different matter if the hypothesis would lead to testable conclusions – if gravitons were discrete particles that exerted force in quantum increments on discrete collision events as opposed to warped space-time that would have a uniform effect without such stochastic events, and if you could measure the difference between those two possibilities, then it would suddenly become a testable hypothesis that you could investigate. Or even if gravitons would, say, theoretically cause subtle influences in say the trajectories of nearby photons that warped space-time wouldn't, you could look for those influences to at least be evidence of gravitons. But unless you can come up with an experiment, the hypothesis is untestable.

So, is there any role for untestable hypotheses? Like most people, my gut feeling was that they can and should be around and are nothing to be afraid of, which is part of the reason why the OP comes across as such a criticism of the assertion that religion is an untestable hypothesis and should therefore not be addressed scientifically. (The real argument behind the OP is that religion is in fact a testable hypothesis, even if not a disprovable one – just look at all the brainden posts and you'll find tons of evidence attempting to support or refute it.) But the more I think about it rationally, the more I have to admit that truly untestable hypotheses would just skew the way we think about things in ways that are probably more likely to be wrong than right. While it's true that two different untestable hypotheses might lead to vastly different outcomes, such as the Dirty Harry example, I can't come up with any rational reason to support adopting one untestable hypothesis over another if you by definition can't accumulate any evidence to support one over the other.

As for the existence of multiverses, I'd say there's a fourth option for how we ended up in a universe that's so finely tuned to support life. It's that the universe doesn't really need that much fine tuning; life can arise under a lot of different conditions. While it's difficult for me to say how plausible that hypothesis is, it is perhaps a question that could be addressed by someone more knowledgeable about how many free parameters the universe has and how they would impact the ability to form complex polymers that could go on to become life as we know it. However, in the absence of such knowledge or of any discrepancies in how the observable universe would act under the two hypotheses, speculation on the existence of multiverses versus a wide range of universal parameters that could support life would fall under the category of untestable, and worth considering only until you're reasonably satisfied that they actually are untestable. (Again, I find myself compelled to entertain these untestable hypotheses, but can't rationally explain why I should!)

[octopuppy]

...It happened in parody because it hasn't yet happened in reality. While that generally has more to do with social reasons and not wanting to get your funding cut and such, the defense of "it's an untestable hypothesis so we don't have to address it" is invoked as the more purely scientific reason why the issue is ducked. The original post was a critique of that excuse.

I disagree. I think the point is much more that the hypothesis is constructed so as to be untestable. This has happened gradually by a process of natural selection. Religions which make testable claims, like "the gods live at the top of mount Olympus" have had a tendency not to survive. The testability of their claims may or may not be a factor in their extinction, but may have eventually proved to be a critical weakness if they didn't die out by other means. Early Christianity made various claims such as "The heavens move around the earth", or "The earth was constructed 6 thousand years ago", and these have been shown to be incompatible with the evidence, but the religion has survived by largely turning its back on these claims. Where a claim is testable, like "The power of prayer helps to cure the sick", there have been tests performed and the claim is duly discredited. For the most part religion has now retreated into the realm of the untestable. Scientists are used to it being that way and thus tend to ignore the topic. If you wanted to test the existence of God, how would you go about it (given that the postulated god occupies no physical space, has no predictable physical effects, is omniscient and omnipotent, so can easily fiddle the results, and doesn't like to be tested)?

As for untestable hypotheses, like the existence of dimensions outside of the three accessible to us (plus time if you want to count it as a dimension) and wormholes and such, they have no role in anything except making for an entertaining science fiction story.

Not at all! The goal of physics is not to tell us the truth about how the universe is, since there is no path to that goal. Rather, physics seeks to find a way of describing all that is known about the universe in one cohesive system, correct or not, and if that system can be described in a way which is elegant and simple, so much the better. Newton's laws of motion were good physics. They held consistently with all that was known at the time. They turned out not to be strictly correct, but that's not the point. Because they encapsulated all known mechanics in a simple mathematical system, they proved to be extremely useful. Any elegant mathematical model of the workings of the universe is worth looking at, and if it makes elaborate predictions, so much the better. The things you mentioned arose not because people made them up, but because people tried to encapsulate what is known about physics into as simple a model as possible, and arrived at a model which predicts physics as we know it but also has a few exotic side effects. Maybe the side effects provide a way of testing the model, but if not, they are not a basis for rejecting it.

As for the existence of multiverses, I'd say there's a fourth option for how we ended up in a universe that's so finely tuned to support life. It's that the universe doesn't really need that much fine tuning; life can arise under a lot of different conditions. While it's difficult for me to say how plausible that hypothesis is, it is perhaps a question that could be addressed by someone more knowledgeable about how many free parameters the universe has and how they would impact the ability to form complex polymers that could go on to become life as we know it.

This is essentially a physics matter, and while there may be a degree of speculation involved, it does appear that in order to allow complex chemical structures to occur, a very fine balance needs to be struck. Imagine if you were inventing physics from scratch, so the number of dimensions, relationships between dimensions, ratios between forces and so on could be absolutely anything. Let's say you were simulating a universe in a computer and these things were random. What are the chances that anything interesting would occur? Almost nil. You need a delicate balancing act between opposing forces in order to produce complexity. More likely everything would just implode/explode and that would be it. More likely still, you wouldn't even have something like space-time as a framework in which things could happen.

However, in the absence of such knowledge or of any discrepancies in how the observable universe would act under the two hypotheses, speculation on the existence of multiverses versus a wide range of universal parameters that could support life would fall under the category of untestable, and worth considering only until you're reasonably satisfied that they actually are untestable. (Again, I find myself compelled to entertain these untestable hypotheses, but can't rationally explain why I should!)

Let's consider an analogous untestable hypothesis. Let's say you're an early astronomer, around Galileo's time. You think that probably the Earth goes round the Sun, and you're contemplating the unlikeliness of life on Earth. You realise that life requires a fairly narrow temperature range and the presence of certain elements, and although you don't know all the specifics, it seems pretty unlikely that if Earth were the only planet in existence, that it would have life on it (and you're not buying all that God stuff). Since the Earth does have life on it, you reason that it is probably not the only planet. Maybe other planets orbit the Sun, but even so, if the Sun were unique, it would still be pretty unlikely that it would have a planet of the right size, distance and chemical composition to support life. So you reason that the Sun is probably not the only such body in existence. You hypothesise that the stars may be similar bodies, too numerous to count, and too far away to examine and test your hypothesis. That last part of the hypothesis is fairly speculative, but it's generally pretty good reasoning, and while it may be a bit of a stretch for someone of that time, I don't think it impossible that such an inference might have been made, had someone thought of it. From the point of view of such a person, this is an untestable hypothesis, and seems likely to remain so for all time. Would you say that it was unworthy of consideration?