Great, but why? What it is about B that justifies the different ethical/legal criteria? Surely it's not the fact that eridius applied the label "experiment" to one and not the other?
How can it make sense to say that doing X is ethical "as long as you're just trying to make money, not publish scientific results"? Presumably, certain things are just wrong to do for whatever reason, and that is why scientific bodies user their publishing standards to prevent them from happening. Those reasons are surely just as applicable in other contexts; if scientific bodies don't have rules for those cases, its because they have no influence there, not because they think it's hunky-dory outside the lab!
"Well, yeah, MegaCorp burned people alive, but it's okay because they were just trying to make money, not publish results on human flammability."
How many more different ways can commenters make the above point, so that people will stop replying that "what advertisers do is not a Real Experiment, so they deserve a lesser standard"? That's a restatement of the inconsistency, not a resolution of it.
This is very different from the colloquial use of 'experiment' to talk about an A/B test or similar. It's a very specific context where we have very specific rules in order to prevent people from committing horrendous acts in the name of obtaining greater understanding.
Companies rigorously test and refine ads to manipulate emotions and behavior on a scale two or three orders of magnitude greater than Facebook's experiment.
Is Facebook's experiment different because they published the results in an academic journal?
Ad companies run experiments on us all the time. When they A/B test across different groups of consumers to see the effectiveness of an ad, they are running a psychological experiment on us without our consent (asking, "how will exposure to this ad affect a person's likelihood of buying our product?").
I think the rules that you're referring to are for federally funded research only. But there's a large body of commercial research that is constantly experimenting on us without our consent.
You should be at least as much worried about "manipulating human behavior for profit" as you are about "manipulating human behavior in order to publish the most effective ways of doing so"; otherwise, you're just letting yourself get tripped up by the labels.