After listening to and commenting on the serving topic, I tried to do some research to educate myself. I'm no expert by any stretch but I perhaps this is a way to think about it.
Somewhere in a presser/interview/podcast I heard Morrison describe a good serve as "Fleen" (Flean?). Google says that means flat (hard with no arc) and clean (no spin.) That obviously applies to float serves, not topspins like Meuth occasionally uses. So if Fleen is the standard, a good serve would have pace, very little arc, and no spin so that the ball just clears the net and knuckles as it drops approaching the passer. I assume the ability to serve Fleen to a specific zone or seam is also part of the puzzle.
At this level with the pace good servers put on the ball, there is very little margin for error vertically. If you stretch a rope across the net from antenna to antenna, there is probably an 18 inch window that a Fleen serve can pass through between the tape and the rope to land in bounds. (If, as our home announcers sometimes argue, a server never missed short in the tape because that is a "bad miss", they would adjust their aim point up causing them to miss many more long - the optimal target through that net/rope channel to center the landing zone in the court is going to have some serves catch the tape given the natural variance of the athletic movement, the float, etc. It is true that it doesn't pressure the defense but the alternative would be lower expected points won due to more misses in total, which I'm sure is greater than the occasional out ball that the receiving team chooses to pass, which I think is their argument.)
In terms of aggression of serves, I'm sure this math is wrong specifically but hopefully directionally correct. The best teams in the country, those that we would hope to face in the tournament, side out somewhere between 60% and 70%. I take that to mean that every time we serve, our expected point value before hitting the ball is something like .35. If we serve a soft arcing ball fearing the service error and the team passes it as if it were a free ball running perfectly in system, my guess is their side out percentage jumps into the 70% range thus having an expected value for the server maybe around .25 On the other hand, if we serve tough and they pass a 1 ball or a low 2 ball, their sideout percentage could drop below 50%, giving us the advantage.
Using those figures, on a normal serve we therefore expect to lose .65ish points. An ace would gain .65 points versus expectation; an error loses .35 points. Doesn't that imply that an ace is almost twice as valuable as an error versus a very good team? Even if .65/.35 isn't correct, that's still a big difference from .50/.50 where an error and ace have the same offsetting value. My point is that even though the best servers in the country have more aces than errors, as a team you can have more errors than aces and still have the optimal strategy be to serve aggressively to maximize expected points.
I think that is the rationale. It's an altogether different question about whether we do that consistently or not and why. And to quantify what everyone is probably seeing with their eyes, this chart just hit the interwebs. I've heard Morrison say he is comfortable with a 10% error rate - my guess is that he'd like a higher ace rate and a little lower error rate since we are over double errors to aces today.
