-1

I'm looking for something like a generally applicable algorithm for determining which of several possible choices is 'most good'.

Some qualifications for that algorithm should be that: (a) it is applicable to a general intelligent entity. (b) should be, assuming equivalent information, invariant between intelligent entities. (c) should not result in an obviously dystopian future. [eg, utilitarianism, as far as I can tell, falls prey to the 'addict everyone to super-heroin' issue.] [others fall prey to the 'then, just wipe out humanity' issue] (d) it should be applicable to all choices in this universe (e) should not be easily breakable by hypotheticals (f) should be usable in the context of incomplete information

One possiblity is to assign every particle in the universe a quantum state. And to estimate the uncertainty in the resulting quantum states of all particles in the universe from t=now to t=end for each possible choice. (this is easiest to visualize assuming that the universe is inside a potential well and with particle number invariance - thereby generating a finite number of possible states) And to assign the choice with the largest estimated uncertainty to the 'maximum good' side.

Is there a generally accepted algorithm along those lines? Or is there reason to believe that such an algorithm is impossible? References would be useful.

  • First comment, it may be worth leaving out "secular" because if one doesn't exist for a secular moral code it's not at all clear that the same standards of evaluation wouldn't simultaneously rule out all such theistic descriptions. Second point, there should be many trivial such descriptions: "Round things are good," or "The choice to extend your leg is good." Of course these lack another important property of moral descriptions: Having motivation to accept them. – Addem Dec 27 '14 at 21:30
2

I don't quite get what your suggestion about quantum physical states is supposed to mean. Maybe (I'm speculating) what you have in mind there is the concept that intelligent life is "good", and you're looking for a quasi-objective way to measure and compare the existence of intelligent life?

Anyway, if I understand the algorithm you're after correctly, it needs first a definition of "good" by objective criteria, and then secondly a method to measure those objective criteria.

I assume the definition of good is supposed to be reasonably close to the usual ethical values. To put it differently, it should more or less agree with the intuitive judgement that suffering and brutality are "not good".

Does that mean we need an objective method to define and gauge suffering? For physical suffering that might be doable, but what about emotional suffering? That gets very complicated due to different personalities and different cultural values. (For example, if a vengeful person posted naked pictures of his ex-lover on the internet, some of the victims might just shrug it off, others would feel deeply humiliated and shamed. In an extremely prudish cultural environment, the victim might even be driven to suicide.)

Nevertheless, we might get a reasonably close approximation to those intuitive ethical frame conditions if we assume that emotional suffering would be in most cases a consequence of injustice, and injustice could be measured objectively by things like distribution of wealth and occurence of physical violence. I guess along lines like that an algorithm could be constructed.

But it would only apply to humans. You want an algorithm that applies to all intelligent entities. However, a non-human intelligence could have totally different values and ethics. For example, in a science fiction story Stanley Weinbaum imagined an intelligent plant that has neither the ability nor the wish to influence its environment; it even honestly does not mind at all if it's killed and eaten.

With that in mind, I consider it unlikely that a satisfying "one size fits all" algorithm could be found. Of course, relativistic worries about emotional suffering could be ignored and some objective values dictated, like "the more intelligent entities exits, the better. And the more they are allowed to act intelligently and creatively, i.e. create ordered but unpredictable things, the better."

With an algorithm like that, there remains your condition (c) to satisfy. You don't want an "obviously" dystopian future. What is "obviously"? I assume that leads back to the assumption I already made, that the results of the algorithm should not be to different from "intuitive" ethics. But it also hints at the problem that if an unwise algorithm were in power and inadvertently constructed a dystopian future, one should be able to stop it with some sort of emergency break, i.e. return to intuitive ethics.

That problem has some similarities with the classical question if the ends justify the means. I suppose a lot of people are uncomfortable with that concept because, like your algorithm, it might lead to conflicts with intuitive ethics that are hard to swallow (like large-scale political murders). With that in mind, it might seem safer to decide ethical problems on a case-by-case basis and accept that from time to time bad judgements happen and "retrials" should remain possible.

| improve this answer | |
1

Short Answer...

Depending on how we interpret your question, your best bets seem to be a choice between...

Long Answer...

If you look at this from a "Gods-eye" view of making decisions for the future of intelligence, then this at best leads to a vacuous solution, and at worst is impossible (under any reasonable definition of good). First, different entities regard different things as good, which means (a) and (b) are impossible. Furthermore, by (c) you seem to have an objective idea of "good" that is independent of the entities involved. So the only way to solve this is to arbitrarily define "good" as that which satisfies a-f, but since we already have a sense of what "good" is (even if we can't precisely define it), we'd be playing word games at this point.

You also don't define dystopian; for instance, if everyone were addicted to super-heroin, but the supply was endless and there were no ill effects, would that be so bad? I can think of many people who'd say it would be much better than what we have now... What about mind-controlled slaves, programmed to be happy? Is our culture of unhappy workers any better? Is freedom such an absolute good that it trumps the happiness of the individual?

Also, what do you mean by "breakable by hypotheticals?"

Additionally, I think the quantum state argument is irrelevant; we don't even know how matter gives rise to consciousness and "good" (in any non-vacuous definition) is a property of the perceiving (value-laden) consciousness.

So, the criteria are...

(a) Is applicable to a general intelligent entity.
(b) Is, assuming equivalent information, invariant between intelligent entities. 
(c) Should not result in an obviously dystopian future.
(d) Should be applicable to all choices in this universe 
(e) Should not be easily breakable by hypotheticals 
(f) Should be usable in the context of incomplete information

Again, having shown that this cannot be done in any universal sense, it may be time to look at this from an individual's perspective.

Well, some things pop out...

(a) is a subset of (b)
(c) can't be guaranteed thanks to (f)
(e) is too vague to use

As a result, I'm dropping (a), (c) and (e). This leaves us with...

(b) Is, assuming equivalent information, invariant between intelligent entities. 
(d) Should be applicable to all choices in this universe 
(f) Should be usable in the context of incomplete information

Essentially, the question becomes a choice function for any individual. Now, the closest way to preserve your original question is to ask how individuals should behave to maximize the universal good. For that, you could try the first formulation of Kant's Categorical Imperative as your choice function:

Act only according to that maxim whereby you can at the same time will that it should become a universal law without contradiction.

Now this isn't quantified, but it is an algorithm.

Alternatively, if you want to know how any individual can maximize his or her own good, then the closest thing to an answer to that is utility. I believe utility functions are flawed, but they seem to be the best starting point. Here's a passage from the link above (assume "pay" is any exchange of values, not just money):

Utility is taken to be correlative to Desire or Want. It has been already argued that desires cannot be measured directly, but only indirectly, by the outward phenomena to which they give rise: and that in those cases with which economics is chiefly concerned the measure is found in the price which a person is willing to pay for the fulfilment or satisfaction of his desire

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.