Interesting... I had wondered if requiring a progressively increasing DAG file - like Ethash - would be one way to make ASICs less-effective (I no longer think they can be held at bay indefinitely). In fact, there is already a real-world example of this: the ASIC developed for Ethash only delivers a relatively modest increase in performance vs. cost vs. energy usage over a typical 6-8 GPU rig, mainly because it has enough memory on board to deal with another 2+ years worth of expected DAG growth.

It will be some sort of DAG and Cryptonote algo's alliance?