I bet if I worked on analyzing and then memorizing 1000 selected games (many from world championships, famous tournaments, candidates matches, etc) that I'd improve a good deal.
The point of contention both here and with MdlM, I imagine, is how efficient it is compared to other methods. Not just for rating gain in the short term, but also how well rounded it is, allowing for smooth improvement in the future.
I wonder if anyone is going to claim they tried it and didn't improve.
I am waiting for this, as well.