Not enough (official) torture in the world?

Torture is permissible where the evidence suggests that this is the only means, due to the immediacy of the situation, to save the life of an innocent person.

Given the choice between inflicting a relatively small level of harm on a wrongdoer and saving an innocent person, it is verging on moral indecency to prefer the interests of the wrongdoer.

The analogy with self-defence is sharpened by considering the hostage-taking scenario, where a wrongdoer takes a hostage and points a gun to the hostage's head, threatening to kill the hostage unless a certain (unreasonable) demand is met. In such a case it is not only permissible, but desirable for police to shoot (and kill) the wrongdoer if they get a "clear shot". This is especially true if it's known that the wrongdoer has a history of serious violence, and hence is more likely to carry out the threat.

I had a feeling you might bring up that one . You will remember, of course, that I didn't predict a decrease in drug use after legalization -- only that the increase would turn out to be modest, especially in hard drugs. Still, I am confident that on Mr. Bagaric's reasoning, you and I will eventually find a way to settle our agreements and reach a common misunderstanding.

So, Bagaric has already made a judgement. The suspect and candidate for torture has been already sentenced by him. Without a trial.

How small is that level of harm? If the "wrongdoer" is able to resist the first stages of torture, then should we continue until he is physically or mentally destroyed?

The problem with Bagaric perspective is not the torture in itself. It is to define the situations when torture must be applied.
Someone is suspect of having put a bomb in a public place. But the police doesn't know where. So, according to Bagaric, we must torture the suspect.
But how does he know that the suspect is more than that? Well, following Bagaric ideas, there is a way to know. Torture. If he is guilty he will say, sooner or later, the place where the bomb is. But, if he is not guilty?
Well, we will only know that when he dies under torture. "After all, he didn't really know. Bad luck. Now, let's pick the next suspect. Maybe this one is the right one".

There does seem to be a certain amount of bootstrapping in Bagaric's argument.

Thomas wrote:

By contrast, I would contend that it is foolish to pass laws without considering the probability that people will sometimes break them, lie about breaking them, and get away with it.

I agree with you, Thomas (that, in itself, is worth noting). But, to be fair to Prof. Bagaric, I think he takes this possibility into account. As he states:

The first is the slippery slope argument: if you start allowing torture in a limited context, the situations in which it will be used will increase. This argument is not sound in the context of torture. First, the floodgates are already open - torture is used widely, despite the absolute legal prohibition against it. Amnesty International has recently reported that it had received, during 2003, reports of torture and ill-treatment from 132 countries, including the United States, Japan and France. It is, in fact, arguable that it is the existence of an unrealistic absolute ban that has driven torture beneath the radar of accountability, and that legalisation in very rare circumstances would in fact reduce instances of it.

In other words, according to Bagaric it is not a valid objection to argue that legalizing torture would increase its usage, since torture is already widely used. Legalization would, if anything, militate against the worst abuses by formalizing torture and regulating it. Now, frankly, I find that argument less than persuasive, but it is at least a perspective that deserves to be considered.

In fact, since we are ready to torture someone because we suspect him, that means we considerer that person already guilty, without trial. We are depriving him of any rights. He cannot present his own defense.
Then, if after torturing him, we discover he knows nothing about the bomb, that fact doesn't prevent us, according to Bagaric's argument, of trying again the same methods in the next suspect. Assuming we have 100 suspects, we torture all of them, and, if the last one knows where the bomb is, we have succeeded.
That reminds me the methods used in middle-age to decide if someone was guilty.

I think Bagaric is very conscious of the extreme danger of his argument. That is why he uses the word "wrongdoer". Not suspect, but wrongdoer. But that leads us to the question: what method has Mr. Bagaric do decide if someone suspected of some crime is a wrongdoer? Since we are going to torture him, there is only one answer: we will only know after torture.
A vicious circle.

The analogy with self-defence is sharpened by considering the hostage-taking scenario, where a wrongdoer takes a hostage and points a gun to the hostage's head, threatening to kill the hostage unless a certain (unreasonable) demand is met. In such a case it is not only permissible, but desirable for police to shoot (and kill) the wrongdoer if they get a "clear shot". This is especially true if it's known that the wrongdoer has a history of serious violence, and hence is more likely to carry out the threat.

There is no logical or moral difference between this scenario and one where there is overwhelming evidence that a wrongdoer has kidnapped an innocent person and informs police that the victim will be killed by a co-offender if certain demands are not met.

I do see a very large logical difference between the 2. It is simple risk/reward. With a single hostage taker holding a gun to the head of a hostage it is easy to calculate that risk/reward. The reward is great since it stops the threat. The risk is based solely on the attempt not succeeding which is small because the intent is to take a "clear shot." So you have a high reward to low risk factor.