Ok, we have some movement here on the topic. Fat Knowledge and mr.snrub have posted some actual results in the comments section. Both of them plugged a Kill a Watt meter into their monitors, and then took measurements using the white Google page, and then a black Google page. Both pages were full screen when measured. Here are the results:
For an Acer 24" AL2423 monitor, the white Google was 65 watts. The black Google was 64 watts. White is more expensive by 1 watt.
For a 19 inch CRT, the white Google was 83 watts. The black Google was 60 watts. White is more expensive by 23 watts.
Finally, for a 19 inch LCD, the white Google was 35 watts. The black Google was also 35 watts. They are the same.
So stay with me here. The white numbers are similiar to what was in the original article; the spread is a little bigger between black and white, probably because the CRT is also a little bigger as compared to the one used in the Energy Star test. I don't think anyone is in disagreement that CRTs will use more energy showing an all white page as compared to an all black page. If so, say so.
Now for the LCD monitors. From our very tiny sample set of two, we see that it either (a) doesn't make a difference whether the page is black or white, or (b) that white is 1 watt more expensive to show. That pretty much agrees with the comment gallery, except that most thought that black should be a little more expensive. I'm going to make the grand assumption that they are very close to equal for right now.
Now I'm going to explain why I think the magic, energy saving background color that every global site should use is very close to black. Based on our tiny sample, and I understand it is tiny, LCD monitors use the same amount of energy regardless of the color of the page. True, we don't have any data yet for any other colors besides white or black (like red, green, blue...), but based on the data we have it makes no difference energy wise what is on the monitor. If that is the case, we can remove them from the equation. They don't matter. So forget about every LCD monitor out there, they don't matter in terms of energy and what is being displayed.
So what's left? All the CRT monitors. These are the ones where everyone agrees that showing a white page uses more energy than showing a black page. We can conclude then, just for the sake of the CRT monitors out there, that every background page should be black. For LCD monitors, it doesn't matter, but for CRT monitors, it does.
Finally, to calculate actual savings, we need a little bit more data. We need to know what the worldwide distribution of the monitors is in use, by type. We need to know how many CRT monitors are out there, how many LCD monitors are out there, and if there are other monitors (like plasma ,OLED, etc.) we need to find out what they do as well. If we had these numbers, then we could calculate some real savings. I'll tell you right now that that there are CRT monitors out there, and I think it's a pretty good size number. I'll also state that if we expand this out, and stop just looking at Google, and look at say, every single web page out there, we are talking about a lot of energy being saved.