Comparison of color temperature versus cholorphyll sensitivity graphs indicate that the general Earth plant prefers red light in a narrow band that corresponds closely with the color temperature of 2300K. 2100K bulbs will provide more energy to the plant then most 2700K HPS bulbs, both can be purchased, but if you could dial in the 2300K (the 650 to 660 nm range) that'd be best for inducing flowering hormones.
The blue range of light offers a wider swath for energy to be absorbed. Basically everything from UV up to about 500 nm (5000K) will be effective. Offering a lot of energy in this range will generally provide the highest values of PAR.
A combination of 7000K, 6500K, 6000K, 5500K, 5000K, 2700K, 2300K, and 2100K would be a very superior setup. LED offers the possibility for this configuration when you're talking the scale of hundreds of tiny little bulbs. The problem with LED is the casting distance, and you're not going to get the penetration power needed to grow big plants. This is an issue for the grower to come to terms with, but could be fine for the casual home gardener.
I'm probably wrong, but I don't think the Kelvin ratings have much to do with the issue here. Kelvin ratings are an average of the light spectrum produced by a light, and it's not consistent at all from one bulb with the same K-rating as another. That's why you have color spectrum graphs for 1,000's of SPECIFIC lights published all over the web and with some of the lights themselves, when you buy them. Sure, if you wanted a bulb with more small-wavelength blues in it for vegging, you would certainly look toward the higher Kelvin bulbs (4,000k and up). But unless the manufacturer has provided the spectral graph, you have little idea whether the light hits the peak points that you want it to hit.
Now, let's say you said, "Well, I found a 5,000k CFL that has a GREAT full-spectrum rating (CRI) of 95+ . . . well, that just means that the light being emitted is more broadly distributed across the visible light spectrum. So, it would probably be a poor source for growers, since it doesn't peak specifically in any wavelength, but uses a blend of all of them.
Colored LED's emit light at a specific wavelength, so that's why you don't often see Kelvin ratings, you see nanometer ratings. And that's why this thread was started. It doesn't appear that much research has been done on which EXACT wavelengths are most absorbed by cannabis. You can consult the GENERIC photosynthetic response graphs, that show the peak areas of light absorption . . . but the graph isn't specific to cannabis.
So the question still stands: when will we finally have studies specific to cannabis regarding which EXACT light wavelengths are best.
I suspect eventually we'll find that it varies by strains, depending on factors such as leaf color. And we'll eventually find that certain spectrums (such as 280nm UV light) might have benefits that we are only guessing at now. For example, the 280nm range might spark a "tanning"-type stress response from the plants, causing more resin production for protection from that light.
We're still in the infancy of testing specific wavelengths on cannabis. Hopefully things will move along quickly . . . and since LED are becoming more efficient (I think they double their efficiency every three years), it'll soon be a no-brainer.
LED's can target exact light wavelengths. Once we find the optimal wavelengths specific to cannabis, we'll soon be able to get significantly more grams of product per watt of light.