I have often thought about what's *really* wrong with sorting by "Time left" and possible remedies to fix the issue. I think the main problem people have with "slow players" is not that they play slow, but more that games with long time limits have to wait most of that limit to bubble up the list of games to be played.

For example, say I have 5 games, with time limits of 1 day (game A), 3 days (game B), 5 days (game C), 7 days (game D), and 30 days (game E). And now, for the sake of the discussion, let's say that I play all games that have <3 days left to play on them (They're very mind-intensive games and I can only play once per day!). We will also assume that my opponents are *very* fast and move within an hour of me playing.
Day Games Played Time remaining
---------------------------------
1 A 1, 2, 4, 6, 29
2 A, B 1, 3, 3, 5, 28
3 A 1, 2, 2, 4, 27
4 A, B, C 1, 3, 5, 3, 26
5 A 1, 2, 4, 2, 25
6 A, B, D 1, 3, 3, 7, 24
7 A 1, 2, 2, 6, 23
8 A, B, C 1, 3, 5, 5, 22
9 A 1, 2, 4, 4, 21
10 A, B 1, 3, 3, 3, 20
11 A 1, 2, 2, 2, 19
12 A, B, C, D 1, 3, 5, 7, 18

etc,
I think most people can follow this. From my point of view, I'm keeping all of my games up-to-date and playing, on average, 1.92 moves per day. Out of 5 games, that's not horrible, I guess. Now, increase the number of A games to 10, B games to 20, C games to 30, D games to 100, and E games to 50, and the problem becomes much bigger.
The problem is that even though I'm making almost 2 moves per day, I have yet to make a move on game E, even after 12 days! Game D is going 6 full days per move (I only play if there are less than 3 days left, remember).
The days-per-move for each of the games is:
A: 1
B: 2
C: 4
D: 6
E: 29

One potential solution to this problem would be a two-tiered sorting/filtering mechanism. The first "group" of games would be sorted by games that have < X time left (1 day? user-defined setting?) The remaining "group" would be sorted by "percent of time left" or some other similar metric.

For an example of the "percent of time left" sorting, let's take our original example and also play games where "Percent time left" is <= 70% and a first "group" cutoff of 1 day (for the sake of brevity. Perhaps 50% is a better value). The Games Played ordering is by the ordering that they would be sorted by. Let's also change the play style so that a maximum of only 1 non-critical (> 1 day left) game is played per day:

Day Games Played Time remaining Percent Time Left
-------------------------------------------------------
1 A 1, 2, 4, 6, 29 100, 66, 80, 86, 97
2 A, B 1, 3, 3, 5, 28 100,100, 60, 71, 93
3 A, C 1, 2, 5, 4, 27 100, 66,100, 57, 90
4 A, D 1, 1, 4, 7, 26 100, 33, 80,100, 87
5 A, B 1, 3, 3, 6, 25 100,100, 60, 86, 83
6 A, C 1, 2, 5, 5, 24 100, 66,100, 71, 80
7 A, B 1, 3, 4, 4, 23 100,100, 80, 57, 77
8 A, D 1, 2, 3, 7, 22 100, 66, 60,100, 73
9 A, C 1, 1, 5, 6, 21, 100, 33,100, 86, 70
10 A, B, E 1, 3, 4, 5, 30, 100,100, 80, 71,100
11 A 1, 2, 3, 4, 29 100, 66, 60, 57, 97
12 A, D 1, 1, 2, 7, 28 100, 33, 40,100, 93


In this case, my moves-per-day is the same, 1.92.
The days-per-move for each of these games is (the numbers don't reflect the tables above, I've extended the tables for more accurate results):
A: 1
B: ~2.6
C: ~3.8
D: 4
E: ~12

The fast-moving games aren't really impacted at all, but the slower games have sped up considerably, at the expense of the medium-speed games. So even though it's a "slow" game, the 30-move game is being played, on average, once every 12 days. In essence, what sorting by "percentage of time left" does is bubble the slow games up to the top much faster, but still allows them to continue not being played if desired.