Sorry sentence was messed up. I corrected it.

I include neither torture nor death penalty in my definition of a civilized society.

Let's say you are a cop with a gun. There a criminal who threatens to cut of someone's finger and then hand itself in.

Should the cop be allowed to kill the criminal by shooting him? Our laws say, yes. It's a defensive act. Killing the criminal to prevent the finger from being cut of is right.

With torture it's a different matter. If a criminal did hid a hostage somewhere, we don't allow the cop to torture the criminal to give up the location of the hostage.

We value the good of not torturing higher than the good of not killing. If you look at the US constitution you will find a list of values. Those are all important. You won't find "don't kill" in that list, just "nor be deprived of life, liberty, or property, without due process of law".

So, if I understand you correctly:

• Death is so bad that you would never accept one extra death, whatever the compensating gains.
• Torture is so bad that you would never accept one extra instance of torture, whatever the compensating gains.

What do you do when you're in the unfortunate position of having to choose between deaths and tortures? E.g., some crazed criminal has set up an infernal machine that will either torture M people or kill N people, it's boobytrapped so that if you try to break it or otherwise stop it doing one of those it will torture M+N people and then kill them, but you do have the option to flip the switch from "torture" to "death" or vice versa.

Your comment above suggests you wouldn't accept any extra torture even to save multiple lives; since you say that preventing torture is (only) almost as important to you as preventing death, I guess you also wouldn't accept any extra deaths even to prevent multiple tortures. But that leaves you in a situation where, e.g., you wouldn't switch from 1000 tortures to 1 death, nor from 1000 deaths to 1 torture. That's pretty counterintuitive, to say the least. Here's one striking consequence: suppose you have a room with two such machines (operating on completely disjoint sets of people). One is currently set to "1000 tortures" (other option: 1 death) and the other to "1000 deaths" (other option: 1 torture). It seems like you have to either (1) endorse leaving them set that way even though switching both switches takes you from 1000D+1000T to 1D+1T, or (2) endorse at least one of the individual switchings even though it trades off torture against death, or (3) say that it's wrong to switch either switch alone but right to switch both, even though the two switches affect completely different people. All of those seem to me like very painful bullets to bite.

Oh, but in real life you would never have to make such an artificial choice between torture and death! Really? If permitting versus prohibiting euthanasia comes to a vote, what will you do? For that matter, as I remarked in a reply to you elsewhere, permitting (e.g.) driving means accepting some extra deaths for the sake of mere convenience. (But an awful lot of convenience.) Will you cross the road to buy something for me from the shop, if I pay you $10000? You will? Then you're accepting an extra risk of death for the sake of mere money. Should dentists be allowed to X-ray people's teeth? I bet doing so incurs a (tiny) extra risk of death from cancers of the mouth. Is it OK to trade off better teeth against death?

[EDITED to add the following two remarks:]

In practice we very rarely face explicit tradeoffs against either torture or death, and having a policy of never making such a tradeoff in favour of torture or death is probably a very good idea: most of the time it will lead you the way you find it better to go. But once you're dealing in politics -- which, you'll recall, is where we started -- you're inevitably looking at things that affect the lives of thousands or millions of people in subtle ways, and this is exactly the sort of situation in which heuristics like "never let anyone die" are liable to let you down and lead you in directions that end up doing more harm overall.

As it happens, in the present instance I find the article you linked to as odious as I expect you do. I do not think the alleged benefits the author is weighing against extra deaths are anywhere near good enough. But my objection isn't, and couldn't be, simply that the author is prepared sometimes to weigh other things against death. A serious attempt to set political policy on the basis of minimizing deaths at all costs would, I think, rapidly lead to disaster[1].

[1] Caveat: It's possible that minimizing deaths on long enough timescales ends up giving good short-term policies. (Maybe almost any objective that isn't entirely insane works reasonably well in the short term if you apply it in the long term.) But in that case, you can't just assume that minimizing deaths in the short term is a good policy; it could be e.g. that trading deaths against freedom in the short term ends up better for everyone in the long term.