I saw that post in CM thread. It's certainly possible to get better returns than multipools, and rig rentals sites kind of prove it since there are people paying 20-50% premium for hashrate (or maybe they are all one big charity), but I think that CONSISTENTLY earning twice as much as WP/CM/etc is not that easy. I have a feeling that 0.004 number - if it's true - was taken from a short period (e.g. 1 day) or from a long period that includes "better times" just a few weeks ago. You also need to question why someone would step into a multipool's thread to brag about it. I would either silently mine the shit out of it with rented rigs, or I would sell it as a service.
It's certainly possible... since I switched to other algorithns, I have been averaging .005BTC/MH+ scrypt equivellent consistently (some days at .007BTC/MH). It takes a fair time commitment to manage this tho.
In interesting developments, nicehash released an auto-algorithm switching profit based system. But it's way too simple so far - no ability to consider power costs or user prefences, but their implementation is straight forward (port for each algo, but all ports dead except most profitable).
If WafflePool took this same idea, but instead waited for authentication, checked the user's pre-defined algo weights, and rejected/accepted auth based on the best algo, we would have a pretty excellent multi-algo-profit-switching pool =].