Nice post. I often do just what you have suggested ... after checking that the remaining time will allow me to execute the longer, slower, but more reliable path.
A trivial but classic example is mating with K+Q vs K. Although it might take 10 or fewer moves from the worst position, I usually use the queen-goes-a-knight's-move-from-the-opposing-king-whilst-cordoning-him-off technique, then move my king in for the mate while my opponent shuffles between his two remaining squares. Takes much longer (move-wise), but it's (nearly) error-proof (watch out to leave the opposing K two squares).
On the other hand most of the time I play it safe and crawl into a shell if I get up say one major piece. So like I said , it seems to depend on whats happening in the game. 
In tactics training. The best move is the one that leads to mate in the fewest moves. I have not problem with that for tactics training since it give a clear metric for determining the "correct" move. However, in a game, I would argue that the best move is the one that leads to mate most reliably. And for most of us, a mating sequence that is more than two or three moves contains uncertainty, especially if there are still a lot of pieces on the board. Many times, I thought I had it only to be disappointed that my opponent found some obscure move I had not considered.
Therefore in a game, I will gladly take a slow reliable path to victory as opposed to a quicker but uncertain path. Admittedly, the path I see as uncertain, may actually be certain, but calculating the certainty is beyond my ability.
So if I think I have a mating attack, but I am sure I can with the Queen, I will win the Queen because I have more confidence in my ability to win with an overwhelming material advantage than in my ability to correctly calculate a complex mating attack.
In addition, when I achieve a material advantage, I will simplify. I especially like to trade queens when I have an advantage. Often the computer analysis will say that trading Queens was a mistake, I am assuming that the computer says that because I can win faster with my Queen. However, I figure there is always hope as long as you still have your queen.
Recently, I achieved an overwhelming material advantage against a higher rated opponent who had sacrificed a lot of material in an attack that did not work out. I forced him to trade his Queen and a pawn for my Queen and a bishop. I suspect the computer will say that was wrong. But I think I still have enough of a material advantage to win and removing his Queen reduces the likelihood that he will find a winning or drawing tactic.
Obviously, you cannot comment on the wisdom of my exchange without seeing the position and I cannot show the position since the game is on-going. But in principle, do you think simplifying is the way to go with an advantage? And do you think the "best" move is the most reliable or the fastest?