Gallup's interpretation is that this isn't really about torture -- rather, it's about investigations. We Americans like to investigate!
While a slim majority favors an investigation, on a relative basis the percentage is quite low because Americans are generally quite supportive of government probes into potential misconduct by public officials. In recent years, for example, Americans were far more likely to favor investigations into the firing of eight U.S. attorneys (72%), government databases of telephone numbers dialed by Americans (62%), oil company profits (82%), and the government's response to Hurricane Katrina (70%).I don't necessarily debate this interpretation, but I think it's somewhat incomplete. Although many people regard torture as a moral absolute, for others (perhaps most others) it is more of a sliding scale: certain types of torture may be permissible against certain types of persons in certain -- presumably fairly extraordinary -- circumstances. A Pew poll released last week, for example, has 15 percent of Americans saying torture is "often justified" against terrorism suspects and 25 percent saying it is "never justified". The majority of 56 percent are somewhere in the middle, saying torture is "sometimes justified" (34 percent) or "rarely justified" (22 percent).
Thus, people may want an investigation into the torture so that they can see whether or not this was the "right" type of torture. They want the details, because they think the details matter.
This is perhaps compounded by the fact that Gallup used the deliberately ambiguous phrase "harsh interrogation techniques" rather than "torture". An ABC-Washington Post poll, which did use the phrase "torture", did not show as significant a number of people who were inclined to think the interrogations were OK but nevertheless wanted an investigation into them.