The original poster's concern related to the MTF plots, not the IS and weather sealing. If you check the MTF plots that he linked to (which you did didn't you), then you will see that the new lens is supposedly better. But, they are calculated ones, not measured, and we do not know how closely they can realise the plots. Secondly, as mentioned, sample variation exists.
Yes, I did check the plots, and long before this post existed actually. Yet, even then, my main interest was not really in the MTF plots but in the improved real-world functionality versus the elder model. I likewise realize there could have been sample variations, but really this kind of "unseen equation" can possibly exist in any face-off, so that just is part of the deal.
Still, I do not believe this has anything to do with the point. The point is, the image quality and sharpness were never
issues with the elder lens ... thus to offer comparisons here really isn't the central way to appreciate what are going to be the biggest potential advantages of the newer model.
Assuming the MTF performance was identical (whether it's supposed to be or not), what you still would have in the newer model is a super-sharp lens that delivers excellent image quality ... with some added advantages that the elder model does not.
Photozone has tested one sample of the new lens and the results at the centre of the frame are consistent with high resolving power. Not so stellar off axis though still very good. By all accounts the old lens was optically excellent. Frankly I would not worry unless I was using a >12MP sensor. Oh, that is what the OP is using.
Thanks for letting me know of the new Photozone tests. What's interesting is Photozone has rated the new 'L' lens higher on the 50D (which is what I have) than on the 5DMkII, both of which have > than 12mp sensors. The test also showed that the MTF resolutions were in fact superior across the board to the elder model, through a broader usable f/stop range, although in their conclusion they say the new lens is "probably not" technically superior to the elder. (Jeeze, guys, how about showing a little more decisiveness, eh?)
I find other aspects of their conclusions to be confusing also. On the one hand, of the upgraded build quality of the new 'L' model, Photozone said "Combined with the seals against dust and moisture (the build quality) is as good as it gets to withstand harsh environmental conditions,"
and yet in their conclusions section Photozone once again contradict themselves by saying the 'L' lens build quality is only "slightly" better than the old (which elder model they rate as only "decent"), and when the old model had absolutely so sealing at all. This makes no sense.
What also made no sense was that the review stated the new lens has Canon's "proven" IS system, when in fact the upgraded lens has a brand new
IS system, which is not yet proven, and which I therefore feel was THE most interesting potential aspect of the brand new 'L' model. Yet the subject of the IS was hardly even touched and what was said was more theoretical than actual field-tested. Overall, I felt that Photozone's 'test' was likewise rather ambiguous (and even self-contradictory) on the most critical elements of the new upgrade.
That being said, the new lens' AF capability was described as "very fast" and the accuracy was "excellent." And, of critical importance to any macrophotography, is the bokeh ... which the reviewer stated was "one of the best lenses in this respect that we've seen so far."
It is of course possible that the OP has picked a cherry from the old lenses, and/or a not so good one from the new.
What I think is the most probable of all is that judging this lens at this point, based on the woefully-inadequate reviews thus far, is a bit premature. Nobody has really analyzed what's new. I would like to see a really competent long-term review be done on this lens, being used in actual field conditions, where the added enhancements can really be put to the test and then compared to the limitations of the elder model.