Ethics

[PWrong]

I want to try approaching general codes of ethics in the same way as utilitarianism. Here's what I've come up with so far:

A complete code of ethics should take any situation, and give you the most ethical action you can take. That is, it's a map from the set of possible universes, to the set of actions you can take. An action is simply something that changes the universe.

We can define "values" in a few different ways. A consequentialist set of values simply takes a universe, and gives you a number representing how "good" that universe is. Obviously you can't measure the value of an entire universe, so you just compare things. For instance, you could consider the difference between a universe in which someone is dead, and where he is alive. So a value system is a map from a universe to the real numbers.

You could say that sometimes an action is inherently bad, no matter what consequences it has. I think this is called deontological, as opposed to consequentialist. Under consequentialism, murder is wrong because it causes people to die, and death is bad. But some people believe that murder is inherently wrong, regardless of consequences. In this case, the value system is a map from the set of actions to the real numbers.

You could also have some combination of these. That is, a value system is a map that takes a universe and an action, and gives you a number.

Now it makes sense to say that, given a value system, we can find a unique complete code of ethics that maximises whatever we value.

So here's the rules we've found so far:
E: U -> A
A: U -> U
V: U x A -> R
and for every V, there exists a unique E.

where

E = code of ethics
U = set of universes
A = actions
V = value system
R = the real numbers

Here's some examples:

Moral nihilism:
Every universe is equally valuable
Every action is equally good.

Self interest (with respect to person X).
The value of a universe = the happiness of X, if X exists, and 0 otherwise
The best action is the one that maximises the happiness of X

Utilitarianism:
The value of a universe = total or average utility (happiness)
The best action is the one that maximises utility.

Consequentialist Christian
Value of a universe = happiness of God
Best action is the one that pleases God the most e.g. if God values life, then saving lives (including preventing abortions) is a good action.

Deontological Christian:
Value is based only on which actions God approves of e.g. praying

Self-interested Christian:
If you go to hell, value = -infinity
If you go to heaven, value = infinity
It doesn't matter what you do, as long as you're good enough to get into heaven.

What does everyone think of this?

[papernuke]

it makes sense to me a little, but what does it mean? the code of ethics?

[bo198214]

The problem I always have with this utiliarism is that neither is specified which populations utility should be maximized, but still more bad, it does not specify in which time range. Something can be increase utility for the next days (robbing a bank) but terrible in the next weeks and years (prison). Something can be good for the own generation but terrible for the next generations (exhausting all resources). And so on.
So you should specify a weight function first for the creatures (usually decreasing along the arrows)
me -> my family -> my region -> my country -> earth -> universe
humans -> mammals -> reptiles -> insects -> plants -> microbes
and second for the time
current minutes -> current hours -> current days -> current years -> current generation -> further generations

But if you do so and then integrate over the weighted utilities, its no more objective, i.e. different persons would do different things in the same situation, I think this is exactly what utilirianism wanted to avoid.

Of course you can say everything is weighted equally. But thats quite unrealistic (what do a care about 4 billion years in advance) if not even impossible because the integral becomes then infinite, at least what regards time. There is also the problem of unpredictability the more we look in advance.

[PWrong]

You integrate over the universe, from now to the end of the universe.

So you should specify a weight function first for the creatures (usually decreasing along the arrows)
me -> my family -> my region -> my country -> earth -> universe
humans -> mammals -> reptiles -> insects -> plants -> microbes

You can sum over all species. You can even take the utility of rocks into account. However, you will find that the utility of insects and plants is always zero, no matter what you do, because they don't have a nervous system and can't feel anything. As for animals, it depends on their preferences. Animals have some of the feelings and abilities that humans have, but not all of them. So for instance, everyone has the right not to be kept in cages, but only humans have the right to vote.

Of course you can say everything is weighted equally. But thats quite unrealistic (what do a care about 4 billion years in advance) if not even impossible because the integral becomes then infinite, at least what regards time. There is also the problem of unpredictability the more we look in advance.

The uncertainty actually solves the problem. When you make a decision, you care about all of eternity, but the effects of your decision won't last that long. At most, it will last until the end of life on earth. If you're a politician, most of your decisions will only affect the people in your country until your laws are repealed. It might affect other countries, or your own country later on, but the effects won't be as great. You also don't know whether they will be +ve or -ve effects. As a good approximation, you can treat them as negligable.

[Hugh]

However, you will find that the utility of insects and plants is always zero, no matter what you do, because they don't have a nervous system and can't feel anything.

How do you know that insects don't feel pain? Insects have the same neurotransmitters that are associated with pain reception in vertebrates. What are they there for? Ever see an injured insect look like it is in pain?

Can you really say that insects definitely do not feel pain?

Pain is meant to warn of damage to the body. It seems unlikely that there is no mechanism for this for insects. We have discovered how nociceptor cells work in our body, but that doesn't mean that there is no equivalent in the insect world that we have yet to understand.

[PWrong]

We have discovered how nociceptor cells work in our body, but that doesn't mean that there is no equivalent in the insect world that we have yet to understand.

To be honest, I have no idea. But I trust biologists to understand this stuff. If there is no known mechanism by which insects can feel pain, they probably can't.

Anyway, this is off-topic. I want to talk about ethics in general. All of Bo's complaints about utilitarianism apply to the other codes as well.

[Keiji]

This is a very interesting idea.

Now uncertainty is the problem. You could give money to a charity helping sufferers of tsunamis, for instance, but there's a very slim chance that doing so would, via chaos theory, cause very subtle changes in the air pressure and actually cause a tsunami, going against what you wanted to do.

So, I suggest that the value of an action be not related to what it WILL do, but what it is INTENDED to do.

[PWrong]

You can take probabilities into account and make a lot of basic approximations. One very simple assumption that you can sometimes make, is that the result of an action will be close to its intended result. In your example, you intend to do a good thing, but there is a small probability that something bad will happen. Overall, the action will be good (especially if you remember that the same action could actually prevent a tsunami somewhere else, or that not donating money could also cause a tsunami).

Let's consider a situation where you intend to do a good thing, but there is a large probability of something bad happening as a result. Suppose you shoot a random person, on the off chance that he might be the next Hitler. If you're right, you may have done a good thing. But it's far more likely that you're wrong. Overall the action is bad, despite your good intentions.

Nevertheless, your idea could be a consistent code of ethics itself. Start with a value system based on universes e.g. utility. Then your modified value = the value of the intended outcome of your action.

A slightly better code would be:
Modified value = value of the expected outcome of your action

[Keiji]

A slightly better code would be:
Modified value = value of the expected outcome of your action

That's a much better idea.

[jinydu]

Sorry to be nitpicky, but I think you meant "extended real numbers" rather than "real numbers", PWrong. The set of all extended real numbers, unlike the set of all real numbers, contains positive infinity and negative infinity.

Also, I could imagine the range of the function being something other than the real numbers. For instance, you could have a moral system where every action is either "must do", "must not do" or "may or may not do".

I'm curious about one point though: You say that a consequentialist would assign values to different possible states of the universe, while a deontologist (which may not be a word) would assign values to actions. How would you distinguish between states of the universe and actions?

[PWrong]

Sorry to be nitpicky, but I think you meant "extended real numbers" rather than "real numbers", PWrong. The set of all extended real numbers, unlike the set of all real numbers, contains positive infinity and negative infinity.

Ok, I'll give you that. Although some codes of ethics might not need infinity. There's some evidence of an upper limit to pain and pleasure, so utility might only go from 0 to 1.

Also, I could imagine the range of the function being something other than the real numbers. For instance, you could have a moral system where every action is either "must do", "must not do" or "may or may not do".

That's true. The only requirement is the range be an ordered set. But I think the more general the better. It would be nice for some codes if it was a group too, so we can add stuff. Consider a universe containing two closed systems, A and B. We could have (as an axiom) that the value of the universe is the sum of the values of A and B. That is: |A U B| = |A|+|B|

I'm curious about one point though: You say that a consequentialist would assign values to different possible states of the universe, while a deontologist (which may not be a word) would assign values to actions. How would you distinguish between states of the universe and actions?

Well an action is a transformation that sends one universe to another, in some time. Only a small subset of the possible actions are available to any given person. This subset is the space we maximise over.

Let's consider a very simple universe, a point in R². Let's ignore time to some extent. Our value function is just a map f: R² -> R. For this example, let f(x,y) = x^2 + y^2. So the further from the origin we are, the better.

The set of actions is the set of all maps A: R²->R², but suppose we can only perform linear maps, for which the determinant is less than 5. We know our current situation is x. All we need to do is find the A that maximises f(Ax).

Unfortunately I didn't design this problem as well as I thought. I thought the answer would be just a scale factor of 5. But that won't work if we know what x is. I'll try to come up with a better question later.

EDIT: This was my 1000th post .

[Keiji]

Sorry to be nitpicky, but I think you meant "extended real numbers" rather than "real numbers", PWrong. The set of all extended real numbers, unlike the set of all real numbers, contains positive infinity and negative infinity.

As far as I am concerned, there is only one infinity, just as there is only one zero. Besides, the use of infinity in this situation is silly. No matter what happens, there will always be something better than it and there will always be something worse than it. Therefore either there should be no upper and lower limits on utility, or the limits should be asymptotes that are approached but never reached.

[PWrong]

No matter what happens, there will always be something better than it and there will always be something worse than it.

Not in every value system. In christian ethics, there's nothing worse than going to Hell, and nothing better than going to heaven. I think most modern christians believe that Heaven is infinitely pleasurable and Hell is infinitely painful.

Therefore either there should be no upper and lower limits on utility, or the limits should be asymptotes that are approached but never reached

I think the latter is true. If you're torturing someone and you apply twice the force, you won't cause twice the pain. Similarly, having 10 billion dollars is only slightly better than having 1 billion.

[Keiji]

Not necessarily. Having a bad life and going to heaven is not as good as having a good life and going to heaven, right?

[PWrong]

I guess it depends who you talk to. I think some christians would disagree. If the pleasure of heaven is infinite, any other pleasure is negligable, and the sum of the two is still infinite.

Also, as Jinydu pointed out, the value doesn't have to be real. Some christians might have a two-valued system: hell and heaven. Catholics would have purgatory in there too.

[bo198214]

I am still concerned by the sum over the time.
If you take it till infinity then you necessarily sum up to infinity,
unless you provide a valuation function like t^(-2).
So what do you propose to overcome this (at a first glance technical) difficulty?

Apart from that, to include infinities into the value scale might lead to paradoxes. What if one goes to hell and one to heaven then you have -∞+∞=? If you take the scale from 0 to ∞, then only one person need to go to heaven and the world is saved (maximum utilty).
So I would propose to only admit values that one can compute with, i.e. the real numbers (or maybe also the positive real numbers).

Fix: oo -> ∞ ~Rob

[PWrong]

If you take it till infinity then you necessarily sum up to infinity,
unless you provide a valuation function like t^(-2).

There are versions of utilitarianism that have these (although most versions are more philosophical than mathematical). The only problem is that it's arbitrary, but there's no inconsistency.

Here's a common example of the problem: If you kill one person, you're preventing his children, grandchildren, and so on, from living. There's a few ways of solving this problem, but I can't remember them exactly.

Apart from that, to include infinities into the value scale might lead to paradoxes. What if one goes to hell and one to heaven then you have -∞+∞=? If you take the scale from 0 to ∞, then only one person need to go to heaven and the world is saved (maximum utilty).

That's a good point. I suppose you could just set heaven & hell to 1 & -1 respectively, and set anything that happens on earth to 0. Then the value for christian ethics would be (no. of souls in heaven) - (no. of souls in hell). You could put some other numbers in there, if you believe that heaven is twice as good as hell is bad, or something. The bible doesn't have any specific information, so christians can believe whatever they want as long as they're consistent.

Fix: oo -> ∞ ~Rob

[bo198214]

There are versions of utilitarianism that have these (although most versions are more philosophical than mathematical). The only problem is that it's arbitrary, but there's no inconsistency.

Yes, thats what I meant. You need it, but its arbitrary. So utilitarianism is arbitrary.

I mean this is neither pathological nor academic. If you are the leader of a company you have also to look how you make your investments. Do you want to make more profit now, or invest everything to make later profit. Or even think in so large terms that your children can profit of your decisions.

Or if you are programmer in that company. Do you write the code cleanly and maintainable doubling your needed time (and leading to angry customers) but you can then fix later occuring problems in a wink. Or do you anyway assume that the company goes bankrupt or that you leave the company soon.

Or even more drastic in politics. If the president has to change every 4 years, the time valuation function (of the president) is 1 for the next 4 years and 0 after that. Of course in the honest case, otherwise the time valuation is 1 only for the months before the election.
But whether thats the best (in whatever sense) for the country is another question.

So my conclusion is that there is no general "best" if taking time into account. Which is mathematically supported by the impossibility to determine a best time valuation function. And by this arbitraryness you can not take it as a code of ethics (which should be regarded equal for all people).

[PWrong]

So my conclusion is that there is no general "best" if taking time into account. Which is mathematically supported by the impossibility to determine a best time valuation function. And by this arbitraryness you can not take it as a code of ethics (which should be regarded equal for all people).

The valuation function may not be equal for everyone, but the total utility is. Integrate your own utility using your own valuation function. Then integrate the next person's utility using his or her personal valuation function. Do this for everyone, then add them all up. For instance, one person may love anticipation, while the other wants everything now and hates waiting. So it's natural to take this into account. Psychologists can make up tests to find out what the average person's valuation function is.

Do you want to make more profit now, or invest everything to make later profit.

In this case, you just go by your own preferences. It depends what you want. But you should ask yourself "how will I feel about this later on" rather than going by what you feel now. I admit it gets more difficult when unborn children will be involved.

Or even more drastic in politics. If the president has to change every 4 years, the time valuation function (of the president) is 1 for the next 4 years and 0 after that. Of course in the honest case, otherwise the time valuation is 1 only for the months before the election.
But whether thats the best (in whatever sense) for the country is another question.

Well, it's clearly better for the country if the president cares about its future. So the president should take this into account. He uses his own "step function" only to integrate over his own utility. For everyone else, he should use the average valuation function.

[Keiji]

I can't see what's wrong with my idea: [for Christian ethics]

0 = go to hell
1 = go to heaven
0.5 = neutral

Someone's life utility is determined from a Schmitt threshold of their utility and past utility. The life utility can only be zero or one, and starts at one (e.g. if they died as a baby before doing anything consciously they would go to heaven).

So someone who sinned and repented would go to heaven but someone who sinned and then remained neutral would go to hell, and the same works the other way round, someone who was good and turned evil would go to hell but someone who was good and then neutral would go to heaven.

[bo198214]

The valuation function may not be equal for everyone, but the total utility is. Integrate your own utility using your own valuation function. Then integrate the next person's utility using his or her personal valuation function. Do this for everyone, then add them all up.

Thats an interesting solution. Everybody has to tell his current valuation function to the government, that they can make their decision based thereon. But really difficult computation becomes, when we have to take into account future changes of the valuation function

For everyone else, he should use the average valuation function.

Whats this?

[PWrong]

Thats an interesting solution. Everybody has to tell his current valuation function to the government, that they can make their decision based thereon. But really difficult computation becomes, when we have to take into account future changes of the valuation function

It doesn't have to be so complicated. The government doesn't need to know each indivual's preference. That would be an invasion of privacy. They just take a survey, and work out what the average person's valuation function is. You wouldn't just ask them, you'd give them tests to see how long they're willing to wait for something, e.t.c.

For everyone else, he should use the average valuation function.

Whats this?

It's just the valuation function of the average person. You've got a bunch of functions f_i(t), so you just take the average over i.

I can't see what's wrong with my idea: [for Christian ethics]

0 = go to hell
1 = go to heaven
0.5 = neutral

What does neutral correspond to? Purgatory?

The problem with hell being zero is that zero utility is equivalent to not being born. I'm pretty sure going to hell would be worse than not being born. Zero utility would be better for purgatory.

Someone's life utility is determined from a Schmitt threshold of their utility and past utility.

What's a Schmitt threshold?

The life utility can only be zero or one, and starts at one (e.g. if they died as a baby before doing anything consciously they would go to heaven).

Actually I think in some versions, you start out with original sin. So you have to be baptised or you go to hell.

[Keiji]

I can't see what's wrong with my idea: [for Christian ethics]

0 = go to hell
1 = go to heaven
0.5 = neutral

What does neutral correspond to? Purgatory?

No, because as I said below the life utility determines whether they go to heaven or hell and it can only be one or zero. Those values were the limits for ethics values of individual actions.

The problem with hell being zero is that zero utility is equivalent to not being born. I'm pretty sure going to hell would be worse than not being born. Zero utility would be better for purgatory.

Someone who isn't born has no soul, obviously. So it doesn't matter what ethics value you have for not being born.

Someone's life utility is determined from a Schmitt threshold of their utility and past utility.

What's a Schmitt threshold?

A Schmitt threshold basically starts the output at one value (in this case, one) and when the input goes below a certain value (e.g. -0.5) the output changes to zero, and when the input goes above a certain value (e.g. 0.5) the output changes to one. So for between -0.5 and 0.5, the output value is determined by what was previously the input.

The life utility can only be zero or one, and starts at one (e.g. if they died as a baby before doing anything consciously they would go to heaven).

Actually I think in some versions, you start out with original sin. So you have to be baptised or you go to hell.

Well I'm no Christian, but I'm sure that God would take pity on those who died before they had a chance to do good?

[PWrong]

Well I'm no Christian, but I'm sure that God would take pity on those who died before they had a chance to do good?

Nope. Apparently God decided in His infinite fairness that they should be punished, because Adam and Eve ate the apple .

Someone who isn't born has no soul, obviously. So it doesn't matter what ethics value you have for not being born.

My point is that having one soul in hell should be worse than having no souls at all. Also, one soul in heaven and one soul in hell is equivalent to no souls anywhere.

So the 0,1 thing is a measure of how good you are (according to God), while the value of your life is either -0.5 or 0.5, for hell or heaven respectively?

[Keiji]

No...

Example: Some person, lets call them P, is born and baptized and whatever else they do, and grows up to whatever age your average person gains their moral conscience. Now let's imagine that during P's childhood they were a bully in primary school. Well, their ethics value would be something around the ethics value of a primary school bully. So that would be, lets say around -0.6. Their life utility says they will go to hell, unless they fix it. So suppose that P decides to right their wrongs and ethics value goes up to say, 0.4. Well, their life utility is STILL at going to hell due to the Schmitt threshold. However, as soon as their ethics value passes 0.5, they will be going to heaven. Suppose that later in their teenage years P gets in a bit of trouble with the police and ethics values go back down to say, -0.2, but their life utility is still at going to heaven.

Get it now?

[PWrong]

Ok, that makes sense. So it's a function transform, like fourier or laplace, except even uglier. I found some graphs of how it works at hyperphysics

I can't see the point of doing it that way. It doesn't seem fair, but then fairness was never a strong point of the heaven/hell idea.

[PWrong]

I just read something about the "four reminders" in Buddhism.

1. Human life is precious
2. Death is inevitable
3. The laws of Karma cannot be avoided
4. Suffering permeates all existence

I wondered about 1 and 2. Some might say they contradict each other. There's no point keeping someone alive, even though their life is precious, because they're going to die anyway. The 1st reminder is pretty vague anyway (I do like the emphasis on suffering in buddhism though. It sounds similar to utilitarianism).

Anyway, I'd like to look at the ways in which life and death can be treated in a code of ethics. Here's what I've come up with.

1. No inherent value. Life is only good because it increases something that does have inherent value, like utility. Killing is wrong because it usually hurts, and because it prevents future happiness. Have children only if you think they'll increase whatever it is you value. You have to be very careful with this.

2. Value = total number of lives. Have lots of babies. However, killing is ok, because everyone dies anyway.

3. Value = integral of number of lives over time. The longer you live, the better. Killing is bad, having babies is still good.

4. No inherent value in life, but killing is inherently wrong.

Can anyone think of one in which life is valuable, killing is bad, but you're not morally required to have kids?