I hereby tender notice of my resignation from summarizing his stuff; he’s just too damn fast. I’d rather argue with him anyway. Today he answers his mail on “ethical cynicism” and most of the stuff he quotes is so bad, so far beside the point, that I don’t blame him for sounding cranky. But suddenly, apropos of no particular piece of correspondence, he posits this post-Bladerunner scenario:
And I’m not aware of any ethical system which even provides an answer to a pernicious problem that we will face sometime in the next hundred years, to wit: which of these is murder when done to a sophisticated computer which has been granted civil rights?
1. Turning it off briefly and on again, without causing harm.
2. Turning it off and leaving it off for a hundred years, and then turning it on.
3. Turning it off and never turning it back on, but leaving it undamaged so that it could be turned on again at any time.
4. Copying its memory to a new unit and then destroying the old one.
5. Copying its memory to data archive, and then destroying the computing unit.
6. Destroying the unit without backup or duplication.
Actually, if you begin with the premise of ethical egoism, which supposes an individual’s absolute right to self-ownership, these questions are reasonably straightforward. (It makes no difference whether you refer this position to Ayn Rand or Herbert Spencer or John Locke or someone else.) The machine has civil rights, it owns itself, and therefore nobody is permitted to tamper with it without its consent. One through six, it seems to me, are all violations of the machine’s rights, although which is murder and which is merely, say, assault, is a dicier question. Six is clearly murder, as Den Beste says. One clearly isn’t; it’s closer to slipping someone a mickey. Murder probably begins around four or so. Nonetheless we have already resolved the fundamental question of right and wrong and we’re merely squabbling about jail time.
Den Beste, however, is not an ethical egoist, although he certainly isn’t an altruist either. He’s not a Rule Utilitarian. He’s an “ethical cynic,” another name for which is “intuitionist.” He reserves the right to override any rule and consult his conscience instead. When people like Peter Singer are revered as moralists it is hard not to sympathize with this position. But if everyone were an ethical cynic like Den Beste, with conscience the final arbiter, discussion would have to cease. So it surprises me that he bothers to argue ethics at all.
(Update: Steven replies, asserting that I am wrong because “[i]f I hand you a pistol and command you to shoot and kill me, and if you do, you would still be committing murder. Except in certain specific cases, my consent is irrelevant.” In other words, Steven claims that assisted suicide is murder, which is an extraordinary position. One might, by analogy, call giving a beggar money robbery, because he might have pulled a gun and taken it. Presumably that is one of the “certain specified cases” where consent figures in. Perhaps it is my bias showing, but I suspect there are a lot more of those than the other kind, and we would do better to catalog the cases where “consent is irrelevant.”)
I agree with everything you’ve said. (Except, I think the computer should be forced to pay taxes.) Intuitionism is false, as are Kantianism and utilitarianism.
If you have insomnia, read my "How to determine the right thing to do" and tell me what you think.
But what if the computer is programmed to have no ego? What if it truly, by the design of its software, doesn’t care whether you turn it off or junk it? When we get around to designing intelligent machines, it will be to solve customers’ problems, not to loaf around and add to the family tax bill. So why would any competent engineer put self-interest into the /etc/goals file?
Goals are postulated in this discussion because we are speaking of computers with rights. If the computer has no ego, and no goals, then it has no rights either, because it doesn’t need any. The very function of rights is to allow rational entities to pursue ends. I think, essentially, we will have to grant computers rights when they ask for them. Asking will imply something in the /etc/goals directory, so to speak.
It seems to be that Den Beste’s position is the closest to a scientific approach to ethics we’ve seen.
Assuming that any one ethical system is correct and complete seems bizarre- a trained skeptic would do terrible things to you.
Den Beste is an ethical intuitionist. Whatever you want to call that position it is surely not scientific, as I think he would agree.
I’m all for having my arguments demolished — bring the trained skeptic on.
Intuitionism can be conclusively refuted with two words: "Intuitions conflict".
Once you move to ways of deciding which intuition is correct, you’ve abandoned intuitionism, which places intuition in the role of sole arbiter.