Ok... here it goes: if you record at a lower level the difference between the pianissimo (very low) and the fortissimo (very high) is reduced and this means a lack of dynamics. It's true that you can later normalize your recording, but this means increasing also the lowest levels. so your overall dynamics will always be reduced.
Just think of it in terms of decibels: in a quiet section the level is -say- 50 dB; then comes a peak and the level goes up to 90 dB; this means that you have a dynamic of 40 dB. If you reduce the recording level, you will never reach the 90 dB; it's true that you will never clip, but you will never attain those 40 dB of difference, and the same applies after normalization: the normalization process will give you a peak of 90 dB, but will increase the low levels as well, and so the difference between highest and lowest points will never be 40 dB.
Maybe professional studios have devices (like the old dbx) that can increase the peaks leaving the lows more or less unchanged, but the result will be less than natural and of course they cost an arm and a leg. This is the reason why all the manuals advice to record close to the clipping level, but without clipping. It is possibile, however, that with the hard disk recording process you have less control over the volumes.
I love to record at the highest possible level, but in the past had to redo more than one song because, after hearing the result, I realized that -at least in one point- I had gone into the "red zone".

Hope this all makes sense.

[This message has been edited by Dreamer (edited 06-03-2006).]