Note: Eliezer Yudkowsky of the Singularity Institute for Artificial Intelligence asked me to say that he "points out that if the apparent order of our universe is something more than a huge, ten-to-the-something-ridiculous-to-one coincidence, our choices can probably affect our futures."
I do think he's making a mistake with anthropic reasoning, but anyway with that cheerful note in mind, here's my take on the matter :)
So, what's this all about?
Life, the Universe and Everything.
What about them?
Well, what the Universe seems to be like, what's the meaning of life, and stuff like that.
So you think you've got the Universe all figured out, huh?
Not exactly. No one can be really sure. But I have a pretty good idea, mostly pieced together from various ideas of other brilliant people while trying to adhere as much as possible to the philosophical principle that we shouldn't make a model more complex than it needs to be without any real reason.
What's your take on the matter, then?
The Universe is simply a multitude of self-aware information patterns. Non-self-aware patterns don't really exist in any meaningful (non-abstract) way in themselves, but only as perceived by self-aware patterns. (Here "to exist" therefore means effectively "to be perceived".) Think Max Tegmark's Level IV multiverse (though my view is based more on subjective worlds than Tegmark's) and/or the Theory of Dust in Greg Egan's Permutation City. (Addendum: Apparently also Hans Moravec has a similar view of existence.)
The gist of it is that any sentient subject — any self-aware mind, if you will — that can be (algorithmically) modelled exists in the same subjective sense that we do.
How'd you get to that?
Well, you know the "I think, therefore I am" thing? That's kind of a cliche, but it's a good start. Now, I know I am, but how do I know others I can perceive exist as I do, as beings having subjective selves? I don't, but it'd be sort of egomaniacial to assume that I'm the only one there really is. I therefore do not make the extraneous assumption that I'd be somehow special.
But why stop there? Sure, I can perceive other sentient beings and I can conclude that it's more plausible than not that they do, indeed, subjectively think that they exist also. However, I have only accounted the people that I can, at least in principle, perceive. Why should I make the extraneous assumption that I can, in fact, perceive everything that there is for anyone else to perceive — notably, every other sentient being?
I shouldn't. So, I'm left with the conclusion that as there is no reason to make the assumption that I could in principle perceive all sentient beings, it is philosophically more sound to hold, as a null hypothesis, that there may be beings that I cannot directly perceive.
Now, what exactly should that null hypothesis contain? Why, the most simple model possible would be nice. Now, there are actually two models that are comparably simple: first, we have the model that nothing exists. However, this model is trivially false, so we revert to the other simple model: everything exists, every possible world in the (metaphorical) eyes of every possible sentient being. Carve out some part of that multitude, and you're making the model more complex than it needs to be. So I don't.
Wait, so every possible sentient being exists?
Yep, that's right. You, me, countless other versions of us, and infinitely many other beings as well.
Other versions of us?
Yeah. You know, for example this you, who is reading this right now is a bit different than the you last year, last month, yesterday, or heck, even five seconds ago (in subjective time — objective time is an unnecessary assumption). There are, in this world model, of course a multitude of subjectively concurrent versions of us also. For example, there are versions of you that didn't even bother to start reading this web page, but from their subjective point of view the same amount of time has passed from that decision as has passed for you now.
Is that a bit like quantum uncertainty?
Not really, though I suppose it is a workable analogue to start with. However, in quantum mechanics just anything doesn't really need to be possible, but only a range of things. This is a limited view, as, quite separate from quantum uncertainty, it is possible to model a mind that thinks that it is a continuation of you in however strange circumstances. And because it's a possible mind, it also exists subjectively, as per our null hypothesis.
Are you saying that I am who I am only because I think so?
Yeah, what else is there?
What about physical continuity?
That continuity is just a part of your subjective experience. (Well, you could assume otherwise, but you'd just add complexity while bringing nothing new to the mix again.)
Well, how about algorithmic continuity and internal consistency?
Yes, well, you can have that basically with any two mind states. You just need to make a special case in the algorithm to cover it. Since it is trivial to have algorithmic continuity by defining the algorithm suitably, it's not a very good indicator of meaningful continuity.
Of course, if you wish, you can define yourself in terms of a particular algorithm. However, this is separate from subjective continuity, so you might have to redefine yourself if you notice that your chosen algorithm-self doesn't apply in your subjective reality anymore.
Wouldn't that special casing make the algorithm unnecessarily complex?
Yes. However, the overall model's complexity would still increase if we tried to include only "simple" algorithms as qualifying for subjective continuity. Also drawing the line would be difficult at best, and probably arbitrary. We don't want that now, do we?
Ok, but why am I this version of myself, and not some other version?
Because that's what makes you you. If you weren't asking that question, referring to the particular instance of you that you are actually referring to, you wouldn't be you. You would be somebody else — some other version of you.
Yes, everything can be explained by sufficient application of the Anthropic Principle.
Do you have any actual proof of all this?
No, there's no proof. I'll say that again: There's no proof. This is not science, and I make no scientific claims. Philosophically, the model should still be given due consideration simply on the basis of it being simple and explaining everything. These properties make it a suitable null hypothesis in my book.
That said, there are things that can happen that should increase one's subjective likelihood of the model being true. Spesifically, if something really aberrant happens, something that really shouldn't be happening if we have any clue about physics, perhaps something that's irreproducible but has solid evidence of having occured nevertheless, that's evidence for this model's correctness.
How's that?
This is because anything really out there — supernatural powers, miracles, all that jazz — would require a damn complex explanation. Every time this world is found to be inherently and significantly more complex than it was thought to be, one's eyes should wonder a bit to the direction of this simple model that explains, without much total complexity at all, also the existence of worlds of insane complexity. Alone, these insanely complex worlds would be astonishingly unlikely. In the aggregate, they're inevitable.
So what should we expect to see if the model is correct
Well, anything and everything, including a simple, boring world.
In the extreme end, there is a simple experiment that everyone can do to increase their subjective probability that this model is correct: Since, in the set of all possible sentient beings, there are also those beings who remember being you at every point in your subjective time, there are versions of you who are 200, 300, 400 subjective years old and so on. The possibilities to make mind-models who recall being you — and therefore are you to all intents and purposes — are endless, and therefore the subjective experiences are also endless. Of course, a random several thousand year old mind-model of "you" may bear little to no resemblance to your current self, but that doesn't erase subjective continuity.
So to test this hypothesis just see if you'll live forever. You don't even have to try to live forever, just do. It'll happen automatically if this model is correct, even if you are in a situation where you positively should have died. This, of course, includes all heaven and hell scenarios, being resurrected by apparently magic technology without being preserved while you were still relatively intact, etc.
If you don't quite get that, think about what subjective death really is: the complete lack of a mind who remembers that it was you just a fraction of a subjective second ago. This simply does not happen in this model; there are always possible minds with your personality and memories up until any given moment (give or take rounding errors, as we really don't need to remember everything or otherwise stay exactly the same to retain subjective continuity).
What, you claim that we are immortal already? But I know people who have died.
If this model is correct, this is not actually the case. This particular instance of you has just sort of lost contact with them; they are no longer a part of your mind-model's subjective experience (or, in other words, you do not share an intersubjective world with them anymore). They do continue in their own mind-models, in their own subjective worlds.
Now that's just wishful thinking.
Actually, no, since all the possible minds, among them all the possible future versions of us, contain also versions that are going through various forms of hell, both literal and figurative. I don't personally like the prospect (even though there would also be uncountable beings in heavenly bliss, again both literal and figurative).
My actual wishful thinking entails being able to actually make a difference; minimize the hells and maximize the heavens.
I see. So basically there are versions of me doing all sorts of crazy things, among them living forever in all kinds of heavens and hells, no matter what I do?
Well, yeah.
So if everything happens anyway, nothing I think I do really matters, does it?
Well, you're kind of right. If we assume that this world model is correct, no action you think you make has any real effect on anything. But there's a but.
We can't be sure that this model is, in fact, correct. This gives us a possibility of causes and consequences beyond our minds, and that is all we really need to act in a subjectively meaningful manner.
So we should act as if our actions matter not because it's likely, but because we can't be sure they don't?
You hit the nail on the head.
So, what is this meaning of life you were talking about — on the off chance that it matters?
To make the world a better place for everyone.
Why not just for me? I like me better.
That's an overly complex value decision, and ultimately counterproductive. It's understandable, of course; our evolutionary programming tends to make us favor ourselves and our kin over others. However, everybody doing this decreases the sum total of happiness in the world, and consequently, on average, decreases also the happiness available to you. Giving the same weight to the happiness of every being in the world makes the value system simpler and cleaner and in the long run benefits most everyone, as the amount of negative or zero-sum gaming goes down.
I do realize that that argument is not going to beat everyone's evolutionary programming; egoism is rather deeply embedded in humanity. Still, happily there is a basis for altruism as well in said programming, you just need to extend it further than it's really evolved to. Also, have a look at The Meaning of Life FAQ for more arguments on this subject. (Note, however, that the author considers the FAQ rather obsolete, and it's only on-line because there's nothing better to replace it. I agree.)
Well, I'll think about it, but what could I do to make the world a better place, anyway?
As the TMoL FAQ explains, there is this thing called the technological Singularity (see also "What is the Singularity?" by the Singularity Institute for Artificial Intelligence). If we can reach it in a proper and Friendly fashion, most of the world's problems, material, social and psychological, could be solved. As people who get the concept and actually actively work towards it are currently few and far between, one more person can make a noticable relative difference.
This Singularity stuff sounds too good to be true; what's the catch?
Well, it is exceedingly difficult to do this, and it may very well fail. However, a positive Singularity would have such an enormous impact on the wellbeing of all sentient beings, current and future, that even a small chance of success makes efforts towards it worthwhile.
Also, a positive Singularity would make an efficient shield against any Existential Risks — scenarios that could "either annihilate Earth-originating intelligent life or permanently and drastically curtail its potential".
Myself, I do believe that — assuming causality — it's most likely that one or the other of the Existential Risk scenarios will get here before we can launch a Friendly Singularity. Still, as I think it's our only reasonable hope to survive this century, it's worth a shot.
Aren't we apocalyptic, but there's nothing new here; people have been proclaiming that the end is near for ages and it still hasn't come.
There is something new here. We haven't had weapons capable of destroying the world for very long now, so that's pretty new. Also, it's pretty new that a guy like George W. Bush has his hand on the Button. Still, I don't think that Bush is going to end the world, but the point is that already some select individuals do have that power. When technological progress continues, it will get easier to destroy the world before it gets harder; new technologies can generally be much more easily applied to destruction than to constructive uses. E.g. bio- and nanotechnological advancements carry with them the danger that at some point, anyone knowledgeable in the fields may, with modest resources, create self-replicating virii, bacteria or nanobots that will eat the biosphere, or at least the humans. Sure, it'd take a kind of craziness to do so, but the humanity isn't exactly lacking in ambitious sociopaths.
And that's not even taking into account the accidents that could happen with those technologies.
Ok, but why don't we just ban the dangerous technologies?
Now that'd be suicide — in a very real sense of the word, since without e.g. advanced bio- and/or nanotechnology, it seems like we're eventually all going to die whether we want to or not. But there are other reasons too: We need advanced technology also to avert any possible Existential Risk scenarios, all of which do not depend on advanced technology.
And then there's the practical argument that even if some country bans or severely limits the development of certain powerful technologies, that just means that the development will move underground and/or to other countries. Even if you somehow managed to enact the bans globally, enforcing them would be another matter entirely. So that'd still leave the underground, where the people doing the research would likely on average be a bit more lax about safety and a bit less altruistically motivated than we'd like.
To make a long story short, bans and strict limitations would only really limit the development done by the good guys, and those are the ones that we need to get there first.
I see your point, but I'll have to think about all of this.
By all means. This is not easy to absorb for most people. Be sure to look at the links mentioned earlier, they provide more comprehensive introductory material than I can explain in one simple page.