Originally posted by Ash A balanced image is good because you have leeway both sides to play around with contrast and clarity settings on the RAW converter. That's why I prefer to expose to the middle (for balanced lighting). Of course, for low key images, the idea is to use negative bias to ETTL and for high key images, positive bias will be needed to ETTR. So it depends on the situation.
The problem with this approach is that it assumes the histogram is linear when it is not. Remember that as you move from right to left on the histogram, each f-stop records half as much light, and since the sensor is linear, much more detail can be captured when using the right side of the histogram.
Let's assume a dynamic range of 5 stops with the darkest being able to distinguish 128 levels. As we move to the right, the amount of light doubles 2 times by the time we reach the center
128 + 256 + 512 = 896 levels in the three stop range from the far left to the center
Continuing towards the right
512 + 1024 + 2048 = 3584 levels in the three stop range from the center to the right
More levels = more data, more data = more detailed images
Remember, this doesn't mean blow out the highlights, it means lean to the right and dial it back in pp.