« Fluctuating inflation field | Main | John Rawls, anti-capitalist »
Philosophical implications of inflationary cosmology
Recent developments in cosmology indicate that every history having a nonzero probability is realized in infinitely many distinct regions of spacetime. Thus, it appears that the universe contains infinitely many civilizations exactly like our own, as well as infinitely many civilizations that differ from our own in any way permitted by physical laws. We explore the implications of this conclusion for ethical theory and for the doomsday argument. In the infinite universe, we find that the doomsday argument applies only to effects which change the average lifetime of all civilizations, and not those which affect our civilization alone.
Got that? Here is the paper. Here is brief background.
It seems if you count all possible universes (or call them parts of our multiverse, whatever) as normatively relevant, none of your actions matter in consequentialist terms.
As to how our world, and our decisions, matter at the margin, we delve into the murky waters of infinite expected values. With an infinity of alternatives out there, our little add-on doesn't seem to make any difference for the grand total. Why should even you raise the average outcome across universes? (TC yesterday: "No, Bryan, we are not leaping up Cantorian levels of infinity, it is just one version of you getting another Klondike bar.")
One option is that only our universe, or some other "in-group," matters. The other universes cannot count for less, rather they must count for nothing. I recoil at such a thought, but it does avoid the mess of infinities. Alternatively, we might embrace some version of Buddhism.
On the bright side, philosophic talk about modality is no longer so problematic but rather refers to facts about other existing universes. Since that problem threatened to bring morality to its knees anyway ("what do you mean, you "could" have done something different? You did what you had to do."), maybe I don't feel so bad after all. And who should care if I do feel bad? The other me feels fine. Infinity has its benefits, and there are many worse problems.
You should lower your probability that God exists, since the Anthropic Argument will dispense with the Argument from Design. Only the ordered pockets of the multiverse can wonder about why we are here and why things seem to run so smoothly.
That's a lot to swallow in one day, but it seems the probability of all those propositions just went up.
Addendum: Have I mentioned that inflationary cosmology and its implications fit my crude, pathetic intuitions? Since we have a universe, I feel it must somehow be a kind of cosmic "free lunch." And once you open the door for free lunches, why stop at just one? There is no good reason to rely on our locally-evolved common sense intuitions when doing philosophic cosmology.
Posted by Tyler Cowen on March 18, 2006 at 06:55 AM in Philosophy, Science | Permalink
TrackBack
TrackBack URL for this entry:
http://www.typepad.com/t/trackback/3576/4475530
Listed below are links to weblogs that reference Philosophical implications of inflationary cosmology:
» Ethics in a Multiverse from Parableman
Some people have suggested (usually to avoid the conclusion of intelligent design arguments) that our universe is just one universe among many, and in fact there's a universe for every possible way things could have gone. Whole TV shows have been based... [Read More]
Tracked on Apr 20, 2006 7:49:55 PM
» Ethics in a Multiverse from Parableman
Some people have suggested (usually to avoid the conclusion of intelligent design arguments) that our universe is just one universe among many, and in fact there's a universe for every possible way things could have gone. Whole TV shows have been based... [Read More]
Tracked on Apr 20, 2006 7:54:24 PM
Comments
Two concepts that have consistently receded over the centuries are "the infinite" and "god". Lots of things we thought were infinite turned out, on closer inspection, to be finite. The size of the world, the size of the universe, and the age of the universe for three. So let me just predict here that the "infinitely many distinct regions of spacetime" will sooner or later become "large but finite", and the whole consequentialist dilemma goes away, no matter how large the number is.
Plus, I'll take these kind of discussions more seriously when someone can tell me *reliably* what the weather will be tomorrow.
Posted by: Tom at Mar 18, 2006 9:15:50 AM
Tyler,
From an ethical perspective there is a simple way out of this,
positive discount rates, both temporally and spatially. Perhaps
at the cosmic level Frank Ramsey was right and discount rates
should be zero. But if one allows even a smidgen of a risk
premium or something else in there, such as I should care more
about that which I know and interact with (I know, Adam Smith
will lecture me about people in China, but he agreed that this
is how people certainly feel and behave), then poof, all those
infinite sums end up adding to finite amounts. The same occurs
if we add over all these other "universes" if we apply a discount
rate to them for their "distance," which, given that we are never
going to see or experience them, and do not even know they are
there for sure (definitely a risk premium on this baby), again,
back to finite sums out of infinitely summed sequences.
Not as big a deal as you and this Bostrum guy are making it.
Posted by: Barkley Rosser at Mar 18, 2006 11:35:02 AM
Tyrone is posting as Tyler now!
Posted by: Macneil at Mar 18, 2006 1:30:32 PM
There is not actually any argument for such a rate of discount across time or space, especially since we are talking about well-being here, not real resources to be invested at real rates of return. Positive discounting, in this context, would imply that a very very large number of people count only for epsilon. We should be suspicious of constructs that dismiss the interests of many people simply to make the answer tractable. That can be done in positive economics, but is not justified in normative economics. A risk premium won't get rid of them either.
Posted by: Tyler Cowen at Mar 18, 2006 3:09:02 PM
"And who should care if I do feel bad?"
Why should you feel bad if the universe doesn't?
If everything that can possibly happen happens in some universe, then low probability universes must vastly
outnumber reasonable probability universes. And since the combination of positions and energy levels of
the basic units of my fingers are fairly limited if they stay attached to my body, universes in which my
fingers do not spontaneously become an expanding ball of plasma must be incredibly rare. So while you
could cut off your own fingers to save you in another universe from having to cut off your own fingers,
any universe in which Tyler has fingers at all is exceptional. Indeed, any universe in which Tyler isn't
killed by someone else's fingers turning into a random expanding ball of plasma is also incredibly rare.
So I say if you are one of those incredibly rare Tylers who has fingers, do your best to enjoy them while
you have them. You owe it to the infinite number of Tylers whose fingers randomized.
If all possible things that can happen happen, then anyone who is capable of thinking at all is likely to
be obliterated the next instant. This would suggest that we should all increase our discount rates to
massively high levels. However, because our chance of obliteration is so high, it simply is not worth
the effort to change our discount rate. My advice is just keep on enjoying your life as best you can for
if you are capable of enjoyment at all you are an incredibly rare gem in an infinity of dross.
Posted by: Ronald Brak at Mar 18, 2006 9:52:39 PM
Tyler,
But, these infinite other universes remain hypothetical.
The theory says they could be there, but they also might
not be. The risk premium is there in the very stochastic
nature of the quantum formulation.
And as for our infinite future, well, the sun will blow
up in a few billion years. If some of our descendents
manage to get out of the system before then, more power to
them.
Posted by: Barkley Rosser at Mar 18, 2006 10:37:20 PM
Since nothing we do can possibly affect people in other universes, whether spatially, temporally or dimensionally distinct it makes perfect sense to count utility in those universes as zero.
Alex
Posted by: Alex Tabarrok at Mar 18, 2006 11:00:51 PM
I will slash them to ribbons with my razor.
--Occam
Posted by: Robert Schwartz at Mar 19, 2006 12:21:12 AM
Remember, this whole inflation stuff is a hypothesis, essentially supported only by the microwave background data and a load of wishful thinking.
Posted by: A Tykhyy at Mar 19, 2006 12:21:41 AM
Forgot a qualifier -- //messy// microwave background data. The interpretation of this sort of data is very complicated and inevitably involves drawing on theory for pointers -- a fertile ground for seeing what you want to see.
Robert Schwartz: +1.
Posted by: A Tykhyy at Mar 19, 2006 2:59:41 AM
It seems that there a lot of very speculative claims being thrown around recently.
Most recent example: http://news.bbc.co.uk/2/hi/science/nature/4819370.stm
It must have something to do with a wish to attract some of the attention and funding that string theory (another highly speculative theory) has received.
Astronomers and physicists must think that if they can make fantastical claims and attract enough public attention they will be able to obtain funding for these risky projects that otherwise would be left unfunded.
Posted by: Nick Borst at Mar 19, 2006 3:34:27 AM
How does one falsify this hypothesis?
Posted by: Matthew Cromer at Mar 19, 2006 8:24:04 AM
See the Bostrom piece at the link, pp.16-17 for a good response to Alex and some of the other comments.
Posted by: Tyler Cowen at Mar 19, 2006 8:56:52 AM
If everything that can possibly happen happens in some universe, then low probability universes must vastly
outnumber reasonable probability universes.
Not necessarily. High probability universes might have more copies of themselves than low probability universes. (Of course, if every universe exists an infinite number of times, how is one to define "more" and "less?")
Also, for those of you who understand real analysis, are we talking about countably many infinite universes, or uncoutably many infinite universes? If uncountably many, then you can have "more" or "fewer" infinitely, any universes depending on how much of the number line these universes take up if you arrange them adjacently. If countably many, you wind up with the measure of any set of universes being zero, and any two sets of countably many universes map onto one another in a one-to-one fashion unless you consider in some sense for their to be twice as many whole numbers as their are whole even numbers.
Posted by: Glaivester at Mar 19, 2006 9:28:12 AM
Tyler,
I think you already answered Glaivester in implying that
we are dealing with countable universes (this is not a story
of going to Cantorian higher levels of infinity), although we
are taking your word for that.
Regarding the relevant pages you just emphasized. OK. I have
not read the underlying lit on inflation theory, so maybe it is
true that this latest observation is consistent with a theory of
infinite universes with "all possible futures" happening "within
our own spacetime" (and hence immune to agnosticism about modal
realism).
Nevertheless, I would say that from the standpoint of ethics,
intentions connected to reasonable probabilities of cause and
effect regarding one's own actions (I defended free will a while
back on one of your earlier threads) should be the basis for morality
and ethics. It may be that some day we will understand our causal
links with these other subsections of our universe. But until that
time we have no idea what connection anything we do has to them, if
any. Hence, we should pay attention ethically to the links that we
do understand and worry about them. Certainly the possibility of these
more cosmic links does not undo our responsibility ("here on earth in
this piece of spacetime") for those we do know about and maybe can do
something about.
Posted by: Barkley Rosser at Mar 19, 2006 9:39:34 AM
Question:
Why does the existence of alternate universes dilute the meaningfulness of our universe? Does the fact that we have 6 billion people on Earth today make your life and your happiness any less important? The only way this works is if you assume a constant meningfulness that is spread throughtout however many universes there is.
Posted by: Glaivester at Mar 19, 2006 9:43:46 AM
You do not have to think much of the fine-tuning argument to see that the multiverse response to it is a bit sophomoric. The usual, and more or less sound insofar as it goes, response to the fine-tuning argument is that one cannot infer design just because the probability of the universal constants is low. To infer that, one needs to know the prior probability of design versus non-design. Just as one can’t from winning a lottery that it was rigged in your favor (because of the high prior probability that it was NOT rigged), one can’t infer design from low probability constants.
The multiverse argument, however, seems to concede just this point, and then attempts a very bad alternative refutation of the fine-tuning argument. It argues that one can infer non-design due to the high likelihood of the distribution of intelligent, seemingly-designed beings (i.e. beings like us) in the universe. This is bad, first of all, because it commits the same logical error, turned on its head, as the regular fine-tuning argument: it infers non-design from a high likelihood on non-design without reference to any prior probability. And secondly, because it doesn’t realize that the fine-tuner can simply move his probability arguments up a notch and argue that a machine capable of producing any distribution of finely-tuned being would have to be finely-tuned itself. Thus, I see no good reason, based on these or any other current cosmological findings, why anyone should alter their estimations of the design argument or any other argument about God.
Posted by: Jthaddeus at Mar 19, 2006 3:41:59 PM
Have you considered that if the infinite variety of universes weakens the telelogical argument for God (because of the anthropic principle), it strengthens the ontological? With an infinite number of universe encompassing every possible universe, then there must be a universe in which exists a (locally) infinite being. Now the question is if the Russell/Whitehead descriptive principle or something similar would prevent every version of said infinite being from being so infinite as to be able to open windows into the monads that are all the other universes, or at least into universes that are otherwise indistinguishable from those universes that cannot have such windows.
Posted by: deRien at Mar 20, 2006 12:56:18 AM
Bostrom's work fails on so many levels that it is hard to manage. In order for a being to reason about infinite consequences, he must have information about those consequences. While our hubris is great, we really cannot discuss ethics of those who can process infinite data.
I have a positive discount rate on a dollar in the future not because it makes my series converge, but because _I don't know_ if I'll really get the dollar or not. My discount on Aid to Africa (tm) is high firstly because _I don't know_ if my aid will get there or rot on the docs.
Any reasoning about the effects of my actions outside my light cone is superstition. And reasoning which attempts to place significanct value on causality within my light cone but removed more than a very limited degree is rank speculation.
The claim that the universe is truly infinite is highly suspect, and in any event cannot be observed. Plenty of manifolds are closed, a fact which we should not dismiss. All the claims relating to the likelihood of the existence of multiple or infinte like observerses rely on this claim. But they also rely upon a presumption of uniformity of energy distribution in this setup. Again, a difficult presumtion.
--
Oh, the odds of Alex's fingers exploding are quite low in every observerse in which physics is coherant enough for Alex to exist in the first place. Rearrangements can make the external probability whatever you like, but that doesn't change the internal probability.
Posted by: Nathan Zook at Mar 20, 2006 12:33:10 PM
deRien,
Interesting points, but I’m not convinced that the actuality of other “possible” worlds does much for the ontological argument. The more or less mainstream, as I understand it, take on that argument is that it at best shows only that if God or some necessary being is possible, then God exists. If it’s true that all possibilities are realized, that still begs the question as to whether or not God is one of those possibilities.
Posted by: Jthaddeus at Mar 20, 2006 11:53:24 PM
"[T]he odds of Alex's fingers exploding are quite low..."
But if everything that can happen does happen in some universe, then surely the protons in his fingers must spontaneouly decay in many of them? If Alex's fingers remain as conventional fingers over say the next second, then the number of different universes branching off from his fingers will be quite limited compared to the number of possible universes that will result if they happen to under go an extremely improbable explosion. If everything that can happen happens, no matter how improbable, then wouldn't universes where everything explodes for no adequately explained reason be the most common sort?
Now some people might say that I don't know what I'm talking about, and there's a very good chance that they'd be right. But I guess there is a very small chance that I'm currently in a universe where I'm making some kind of sense. (Ooh look! Flying pigs!)
My idea may sound kind of silly,
Posted by: Ronald Brak at Mar 20, 2006 11:59:20 PM
Ronald,
You're not being silly, just confused. There are two different ways to approach your error. They might be complimentary.
First, don't confound the a class of many cases with a class of high (total) probability. A class that has a million cases each with one-in-a-billion odds is still a .1% class.
Second, don't let your eyes deceive you. While it may appear that there are more potential histories in which one's fingers explode than not, this is not true. In almost all potential histories, not only do Alex's fingers remain intact, but so do those of everyone on earth! This has to do with the distribution of the probabilities of various quantum fluxuations.
Posted by: Nathan Zook at Mar 21, 2006 9:15:06 AM
Interesting. Looks like I'll have to hit the books, Nathan. (Now where did I put my club?)
Posted by: Ronald Brak at Mar 22, 2006 12:19:08 AM
It's worth keeping in mind that what physicists are postulating in order to avoid the cosmological fine-tuning intelligent design argument is not the view that David Lewis calls modal realism, i.e. the view that the possible worlds that modal talk requires are concrete entities just like the actual world. In Lewis' lingo, the hypothesis of many universes is really just a hypothesis that the actual world is a multiverse. His concrete possible worlds are not connected to the actual world in any spatiotemporal or causal way, and these postulated worlds are connected at least in a causal sense and perhaps in a spatiotemporal way if the scifi lingo about their being other dimensions is to be taken literally (though perhaps it shouldn't). I'm not sure that anyone is making the mistake here that Lewis warns against, but I had the sense that we might be on the verge of confusing the two.
Posted by: Jeremy Pierce at Mar 31, 2006 7:55:41 AM
Now that I've had a chance to come back to this and read your argument carefully, I have one comment on the ethics. I think you're assuming something that I wouldn't grant. The argument is that the net result wouldn't be any different if I did an action usually considered better worse, and therefore the action isn't really better or worse because the consequence of either action would be the same.
Now anyone who denies consequentialism isn't going to buy this. Just because two actions lead to the same consequences doesn't mean they're morally equivalent. Your argument rests on the consequence being the same, as if that's the only morally relevant issue. Non-consequentialists won't accept that.
But the other problem is one I think a consequentialist might even accept. Consequentialists don't always believe that my action's moral value depends just on whatever future state of the world occurs afterward. It depends, for many consequentialists, on which aspects of the future state of the world have something to do with my action. If my action causes the bad, then I am to blame. If it causes the good, I'm to be evaluated positively. So if I'm the one who does the bad thing, and some duplicate of me in another part of the multiverse does the good thing, then I'm to blame and he is to be congratulated. So my action is bad even if my not doing it would logically (but not causally) entail someone else doing it and lead to an exactly similar result somewhere else. That means we should blame the one who does the bad thing, even if the overall multiverse isn't going to be different just because one particular person in one of those worlds did it this way rather than another way.
Posted by: Jeremy Pierce at Apr 16, 2006 10:59:24 PM
