I'll agree that those samples, compared to today's 18MP tech, don't look that great. As for what it could do vs what was available in 2000, or what 2MP could do today, i'll let the arguments continue.
But think about the other side of that article. RGB Bayer Arrays vs CMY Bayer Arrays. Pretty much as written in the article, CMY arrays just let more light through than RGB arrays. Punkt. So for the same sensor, with a CMY array, you're going to let in more light and need less gain, have less noise, than the same sensor with an RGB array.
And that's going to hold true for 2000, and 2012. I'm sure the RGB arrays in 2012 are better, let more light through, than RGB arrays in 2000. But what if we design a CMY array today? Would it not be also better than an RGB array today?
Sensors are always going to improve, interwiring will get thinner, wells deeper, whatever. But there's more to a sensor than just silicon and the bits of metal therein, playing with that will only gain so much. Besides this article, only Sigma (Foveon) and Fuji (with their hexagonal whatsit array) seems to even be thinking much past the boring old "can we pack more in?", and "can we make them bigger too?".
I'm curious as to why this CMY array never came about to mainstream use in the last 12 years? Patents? (ie, Kodak patented it and promtly went belly-up, ergo noone else could make one?) Processing? (ie, Adobe and DxO said they weren't going to write another raw-converter, and no camera maker had the balls to stand up to them?) Maybe it just didn't actually fill the right colour space and was really lacking in a certain colour, despite its much better ISO performance? (Maybe it had something to do with Magenta and Yellow being totally overlapping, RGB on the graph looks a lot more spread out and clearly defined. Or maybe because of that increased Transmission, the pixels saturate more easily = less DR?)