I'm reminded of an economics article that I read on cheating in sports. It turns out that no amount of punishment of cheaters would be enough to stop cheating in bicycling. What would be required is that if ANY bicyclist is caught cheating, EVERYONE who had worked with him who hadn't reported it would have to be punished. Don't just punish the cheaters, punish those in a position to have conceivably known who didn't report it.
The scientific equivalent would be to not punish bad research articles, but to punish everyone who went through the lab who could have raised the flag and didn't.
I'm pretty sure that they didn't try modeling that. It would have been interesting to know what that would have shown in their model.