In the variations sidebar, maybe you could allot some CPU to calculating and displaying the opponent's best pre-move so we can see the nature of their mistake that allowed our tactics. Usually they simply violate the LPDO rule ("loose pieces drop off") but sometimes they had a guard-and-simultaneously-threaten defense that would be rewarding to investigate.
Sometimes they were actually lost even at the pre-move (i.e. there was nothing they could do to save the position anyway), which perhaps lowers the objective training value of the problem but is still of interest to us "problem archaeologists".
And then there are those rare, precious positions where it *looks* like they were lost at the pre-move but there is a nifty defense just over the search horizon, at 18 ply or so.
For example, in
http://chesstempo.com/chess-problems/41825at the pre-move, Crafty 22.1 calculates 1. Rd5! (instead of the tempo-wasting 1. Rd7? as in the game) c2 and now:
a) 2. Rc5? Nb3+ 3. Kb2 Nxc5 4. a7 Kd2 5. a8=Q c1=Q+ 6. Ka2 Nb4#
but
b) 2. Kb2! Kd2 3. Rc5 Ne5 4. a7 Nd3+ 5. Ka2 Nxc5 6. a8=Q and suddenly we realize we can't play 6... c1=Q because white can force stalemate by throwing away the new white queen. So we would have to play 6... c1=R, which is still up material but is a serious grind to actually win. (Can anyone see a better method for black at move 3 or earlier and confirm it with their computer?)
Anyways, another one for the "wish list". :-)