The typical pulseless collapse on land is cardiac or neurologic. Up until the moment of collapse, the victim was breathing, and the lungs are working normally. That is quite different from any cardiac arrest suffered in or under the water, and the new guidelines for compression-only resuscitation exclude drowning.
Gene's committee paper concluded, on the basis of a study of lifeguards, that rescue breaths were reasonable. Although I respect and applaud the work done by the committee, I really have to be dubious about this conclusion. Lay rescuers, with no training in artificial respiration, who are in scuba gear to boot, I think are highly unlikely to provide any kind of effective rescue breathing. I maintain my own position, which is that several rescue breaths are reasonable upon surfacing the victim, in case stimulation of the airway can induce spontaneous respiration. If two or three rescue breaths doesn't get the job done, you are most likely dealing with someone in cardiac arrest as well, or seriously neurologically impaired, and getting that person to where secondary care is available becomes a critical objective. And I don't believe that most lay rescuers in scuba gear would be as little delayed as the study suggests the lifeguards were.