If you are invested in defending torture on the basis of its potential utility, it would be prudent to frame the issue by claiming that the torture of captured extremists has led to useful intelligence. That way, your detractors will respond either by arguing that torture did not lead to useful actionable intelligence, or that torture is ethically unjustifiable even if it is a useful method of information gathering. Either way, the pro-torture argument comes out ahead — because given this way of framing the issue, the worst-case scenario is that torture is ineffective and unethical. But, in fact, that is not the worst-case scenario. There is a scenario far worse even from the most utilitarian point-of-view — a scenario involving bad intelligence.