Statistical and technical tests? They should have someone look at them. I've been in a position to choose photos for publication, and the first (doctored) version of this one is not only would not have published, but I would have called the Reuters photo desk and asked them what they were smoking when they let it through.
Speaking of which, has anyone seen any media outlet that picked up this photo? Yahoo is fully automated, and picks up stories and photos straight off the wire. Did this slip past anyone other than Reuters that has actual human editors?
FWIW, I think I just figured out what Rooters was up to. In the undoctored pic, the source of the smoke is clearly coming from one one place/building. In the doctored version that's not as easily to see, especially if you're not examining the pic closely. A cursory glance might make you think that the IDF blew up an entire area of the city. Which was undountedly the intention.
I wasn't so sure they were actually fakes at first, but http://www.freerepublic.com/focus/news/1678623/replies?c=6
makes a very compelling case.
My point is, that these crude fakes could/should have been detected mathematically, by simple tests of internal validity, without recourse to external tests or human intervention. There is no substitute for a "human in the loop", but a cheap, simple statistical test would have been a bell ringer, alerting editors to the possibility of fakery.