@Luis:
Is the heat uptake by the oceans an ad hoc explanation? Not at all. Atmospheric and climate scientists have known about the importance of the ocean for quite a while now.1, 2 It's also observable and quantifiable. Quite frankly I don't think you have any good argument against it, and so far it seems that you're simply regurgitating the criticisms from skeptical websites without actually giving them much critical thought.
That's somewhat facetious. I could say the same about you: you are only regurgitating what the editorials of the IPCC tell you.
While it is true that everyone "knew" that the oceans were important, the argument is quite different this time, and you should not ignore this point. Of course you can always delude yourself and tell yourself that eurasia was always at war with eastasia and lesser thought crimes

, but what is true is that this argument of "the oceans ate the warming" is quite recent and new, coming after some discussions in 2009 by Trenberth et al:
"Global warming is still happening - our planet is still accumulating heat. But our observation systems aren't able to comprehensively keep track of where all the energy is going. Consequently, we can't definitively explain why surface temperatures have gone down in the last few years. That's a travesty!"
This is clearly an indication of the usual troubles that science must go through to eventually get to the best answers. I have zero problems with it (unlike many deniers / skeptics / whatever). What I have more problems with is this whitewashing after the facts and for the reports for politicians that "we always knew this, this was always taken account for, skeptics' lies, etc". What is true is that this explanation is recent and it still lacks any predictive power. The models still depend on many prejudices that may well turn out to be as false as the travesties that were plaguing climatology in 2009.
*head scratch* what you are describing sounds exactly like science. Propose model for complex system. => run model, find results slightly different than observed phenomena => re-evaluate model, find phenomena that were not counted on in the model => refine model, incorporating new bits => model now closer to observed results. Sounds like running a chemistry experiment. Sounds like honing a theory.
That's only the first step. If you only do what you are referring to you aren't getting a true theory, you are only capturing fine tuned and fudged models that represent the past exceedingly well. The real thing only comes when you actually predict data that is not in your present bucket of data and you get it right. Ensuring your models capture past data is trivial. Ensuring your model will be able to capture future unseen data is the Real ****.