I think it does, for two reasons.I believe he is suggesting that one has to look at a scene in terms of luminosity, rather than colours in order to visualize what a scene can look like as a B&W image. That is a learned skill, but has nothing to do with the mechanics of B&W conversion from a colour image.
Look at the thumbnails in the article linked in the document that Ted linked. You'll see 7 B&W renderings of each color image. Which would you previsualize? Note that in the first row, the two that were on average rated both most pleasing and most accurate across images--decolorize and Smith 08--yield dramatically different results in that one instance. The methods apparently differ on several dimensions, but one of them is what you wrote about in post 17, that is, how much they darken specific colors:
That is, a final B&W conversion is generally not simply a desaturation, or extracting the L channel data. Even the defaults in Photoshop and Lightroom aren't a simple desaturation. So, there are lots of conversions. Whatever you visualize, and whatever your camera does in creating a B&W thumbnail, may correspond to one of them, but only one. And if people could previsualize accurately, they wouldn't need to play around with underlying colors. They would know what adjustments to make--like adding a red filter in the days of B&W film.Most people that do B&W conversion will also play around with the underlying colours to enhance the look, just like they did in the B&W film days; enhancing certain colours and downplaying others.
It's for that reason that I wrote that I can sometimes approximately previsualize, but only approximately.
My second reason is a hunch. That is, I am very skeptical that the human visual system can accurately estimate luminosity differences independent of hue and saturation. I actually started constructing a test panel that would have a series of rectangles, mixing differences in luminance with differences in hue, but I got bogged down. For example, I would be willing to bet that when the luminosity difference is constant, people perceive it as larger when the adjacent colors are close to complimentary or highly saturated. However, I have no evidence to back this up. I'm pretty sure I have made this error myself, but I don't keep track of all of my mistakes. There is no library large enough.
Ted, do you know of any research that addresses this?
In any case, back to the earlier point: my advice to novices to try to previsualize, regardless of whether you aim for color or B&W. However, after the fact, feel free to make use of every tool you have to make the image into what you want. If you decide after the fact that the image is more pleasing or more effective in a different form than you planned, then be glad that you discovered that. The point of the effort is the final result, not the process.
Dan