Over-applying rationality: Indefinite lifespans

post by Bart119 · 2012-05-25T17:25:41.763Z · LW · GW · Legacy · 14 comments

 

UPDATE: One commenter said that arguments against the desirability of indefinite lifespans and their rebuttal had appeared before on LW and elsewhere. I am very interested in links to the best such discussions. If I'm going over old ground, a kind soul who can point me to the prior art would be much appreciated.

 

-----------

 

I am very impressed with this site in its goal of outlining cognitive biases and seeing how they apply in everyday situations. When you're trying to decide how to spend money to alleviate human misery, it works. Yeah, it's better to save 50,000 people than 5,000. The two alternatives concern the same moral intuitions. When faced with a specific choice among alternatives, you may find that the tools of rationality will apply and tell you what to do, which might be contrary to what you would have done without such analysis.

 

But when I see people trying to use Bayesian analysis for bigger questions beyond this, I think there is a substantial danger of being led astray by the method. When you can't find a clear way to analyze the situation and you are making low-confidence probability estimates of alternative futures and their utility, you'd do better to just put your rationality toolbox back on the shelf and decide the way you've always decided: gut feelings, intuition, doing what everyone else does, etc.

 

Let's take as a case study the popular view on LW that living as long as possible is a good thing. First, within the range of currently common lifespans, it's a good thing to live a longer, healthier life; that is uncontroversial.

 

But judging from the LW posts I've read, the prospect that science could reach a point where people could live indefinitely long is hailed as a great and noble goal. I think it would be terrible.

 

First, let me distinguish an indefinite lifespan with true immortality. Is there anyone here who thinks true immortality is within reach? The sun will go red giant, making earth uninhabitable. If we hop from star to star, we get a little longer. But there's stuff like heat death and entropy and all. Not to mention the accumulation of small, mundane risks over a very long time. Eternity is one friggin' long time.

 

If you don't have true immortality, you have a longer lifespan, and then you die. You still have to face the same profoundly settling issue. One wry formulation might be: whenever you do die, you're saving yourself the trouble of dying later. Different lifespans all end with the same unsettling matter of personal extinction. (Other thought: mortality is the most salient and immediate roadblock to finding a more satisfying meaning in life, but it's just the first one; if it was removed we'd find others beyond.) If you live 500 years instead of 100, you haven't achieved anything special. You haven't cheated death. You've just got an extra 400 years of living. The mundane stuff of eating, sleeping, thinking, seeing beautiful sunsets, chatting with friends, etc., and of course the less pleasant parts too.

 

The ecological integrity of the world is already under severe strain. Perhaps with technological and political improvements we could improve how many people could live sustainably on earth by some constant factor, but it doesn’t affect the current argument. Our population is limited. (You may think we're going to personally take off to colonize the stars. Let's assume for now it can't be done.)

 

Given a population limit, the effect of people living 500 or 50,000 years is that the available slots will soon be filled, and reproduction would have to be seriously curtailed. Children would be very rare.

 

I think that no matter how healthy they are, a world full of people who are over 100 or 10,000 years old with very few children would be a place that 'just isn't right'.

 

First, I estimate that the human mind isn't designed to live beyond 100 years (if that) and will tend to become unhappy. Such people think the same thoughts over and over. They get bored (a lot of people today get bored at 50). They still know they're going to die someday.

 

Second, they live in a world without children. (One thing I've never seen in a LessWrong survey is the proportion who are parents -- given a highly educated group predominantly in their 20s, I would estimate it is very low).

 

And aside from their own personal boredom and personal lack of children in their lives, they know they live in a world where everyone else is in the same position. It's an ossifying world.

 

Now let's put rationality into it. I imagine a Bayesian feeling comfortable and at home constructing an equation with two key constants being the number of people and the number of happy, productive years they get to live. Multiplication is in order. I'm not sure how the argument goes after that: Potential future lives that don't happen don't get to add to the utility (do they? Or at a discounted rate?) Even if they do, the utility of a new life has to be weighed against the lost utility of an existing person dying. We can see the equation coming out in favor of extending life as long as possible.

 

The argument on the other side can also be framed in Bayesian terms. My estimation is that the utility of a large majority of these people who are over 200 years old is going to be very small. We can multiply their utility and conclude that the world will be a happier place with a mix of children, young people, and people croaking after 90 years of happy, productive life.

 

I imagine a Bayesian frowning at this analysis. It seems imprecise. I could I suppose assign some sort of utility-reduction weight to those various factors and multiply them out, but it isn't really going to make the Bayesian very happy. It’s not going to make me very happy either. I would rather just consider the situation as a whole and assign a low utility to the bulk of the population that's hundreds of years old rather than break it into parts.

 

At one level, my argument with the pro-indefinite-lifespan faction is just a difference in what kind of a future world would be a happier place. We've plugged in our different assumptions and reached different conclusions.

 

But to what extent does framing the problem as one of Bayesian analysis bias people to prefer the indefinite extension of individual lives? If your favorite tool is a hammer, things tend to look like nails. My conclusion feels more naturally framed if we ignore individual utilities and just say: a world full of people living indefinitely long would suck. Spelling it out in terms of utility just doesn't add anything.

 

The practical implications are a separate question. Killing people when they get to be 90 is of course highly repugnant, as is asking them to kill themselves. But it might affect what sort of scientific research we fund and what drugs we approve, for starters.

 

 

14 comments

Comments sorted by top scores.