I really think it should be the opposite. Recent code should be clearly visible and older code should progressively interpolate into black. The current setup is extremely unintuitive for me.
Old code is mostly good code. It's new code that has varying quality.
http://thomasleecopeland.com/2014/09/18/olde-code-finder.htm...
Actually, the first version only found files where x% of the lines were written by a particular author. Then someone gently suggested checking the line age, which made perfect sense.
https://secure.phabricator.com/diffusion/ARC/browse/master/s...
More, I think context plays a big role and it is difficult to see what new is brought to the table at times. Worse, what old was accidentally discarded?