Just an average player in the high minors today is a fairly dangerous hitter by historical standards, particularly with whippy fast bats. For most pitches and deliveries, the pressure for perfect execution, pitch after pitch after pitch, means considerable strain on the arm.
Arbitrary pitch counts are a crude and lazy means of addressing this, but they are an effort to meet a real threat to pitchers’ health. A better approach would be for managers and pitching coaches to really learn their pitchers, and understand their particular signs of tiring. A given pitch count may be just fine one day and too long another, because every game is different. But adhering to a simple number gives managers something of an ass-cover after both lost games and arm injuries.
Having so many relievers is itself an effect of the fetish for pitching changes. And it’s not as if there is no field-level cost to teams for devoting half their roster to the bullpen. I’ve seen plenty of games lost for want of a decent pinch-hitter or positional substitution.
There’s certainly a legitimate need for more relievers today than even thirty years ago. But there’s also a managerial over-reliance on pitching roles (closer, set-up) and tiny platoon advantages.
Most of the time, IMO, if one of your better starters is in control of the game and not showing physical signs of tiring, changing him out just because you can is the wrong move.