It's an important question and I wouldn't want to dismiss it with a quick general answer. First of all, in the 1920s and 1930s pitchers threw a much wider variety of pitches than they do now. In that era teams had no pitching coaches; pitching coaches started in the late 1940s and didn't become universal until about 1960. No pitching coaches.
In that game, the dominant assumption was that a major league pitcher should be expected to be able to throw any pitch. I remember reading a story about a pitcher who was called up sometime in the 1930s, and in his first game the manager signaled for him to drop down and throw side-arm. He couldn't do it; he couldn't throw side-arm instead of overhand, so he got sent back to the minors. A major league pitcher was expected to be able to throw a curve, a change, a fastball, and, in most cases, a knuckleball. Once the slider came into the game, almost everybody threw a slider, and the "curve" included a big curve and a hard curve and a drop curve. It was standard repertoire.
Once the pitching coaches came into the game, the pitching coaches generally believed that the pitcher should focus on the pitches that he threw BEST, rather than being expected to throw everything. Modern pitching coaches don't want you throwing your third- or fourth-best pitch. They want you to know what you can do best and to focus on that. Over time pitching repertoires became more and more limited.
There's a parallel process in basketball players moving from high school to college. A high school player may have 30 different shots that he tries to make. He may shoot 10-foot jumpers, 25-foot set shots from the corner, 6-foot hook shots, 15-foot floaters, may shoot off the drive, may shoot off the pass. The college coach in almost all cases will sort through the things that the player does in high school, and tell him which ones he can continue to do, and which ones he should stop doing.
Things go wrong, in that process. Sometimes the college coach tells the player that he doesn't want him shooting that high arcing jumper from the corner, but the player thinks he can hit that shot and does not accept that it's not part of his game anymore. The player will transfer to another school. Even very good coaches sometimes lose excellent players because they don't understand their players' skill sets. Bobby Knight lost Larry Bird because he didn't really understand Larry's game. It happens.
The same thing happens to baseball pitchers. When Charlie Leibrandt came up with the Reds, Johnny Bench didn't want him using his fastball as an out pitch. He wanted him to throw his fastball out of the strike zone as a setup pitch, then try to get the batter out with changeups and sliders. It didn't work. Leibrandt failed with the Reds, then became an outstanding pitcher when he got with another team and realized that Bench was totally wrong, and that he needed to use his fastball in the zone. It happens all the time; pitchers fail with one team, go to some other team and improve tremendously because the team lets them throw the stuff they need to throw.
A few years ago, probably ten years ago, I developed a method to measure the variety of pitches thrown by each pitchers, a "repertoire index". A pitcher like Tim Wakefield, who just threw knuckleballs on every pitch, had a repertoire index not much above 1.00, while most pitchers had repertoire indexes in the range of 2.20 to 2.60.
But there was one pitcher, Shaun Marcum, who had ridiculously high Pitch Repertoire Indexes, much, much higher than any other pitcher. They were around 4.00, meaning that he would throw any of four different pitches, one as much as the other. Nobody else was over 3.00, although maybe a couple of guys were at 3.15 or something; I don't exactly remember. I remember than Marcum was an outlier.
But Marcum was also dramatically more EFFECTIVE than he should have been, given his stuff. Not to disrespect him in any way, but the second half of his career, his arm really was shit. BUT HE STILL GOT PEOPLE OUT. He was pitching well with an 86-MPH fastball, and he was able to do that, it was clear to me, because the hitters just never knew what was coming. He would throw any pitch at any time.
So it was never clear, to me, that the conventional wisdom of pitching coaches--that the pitcher should focus on his best pitches and stay within his repertoire--was actually correct. Ideas like that take possession of the way a group of people think, often when there is nothing to them accept that everybody else says this so it must be true.
I THINK this is beginning to change or has changed, but I don't really know. It's complicated by the growth of these 1-inning pitchers, who mostly just throw 1 or 2 pitches, but I BELIEVE that starting pitchers now are more likely to throw something off-beat on a count.
El Duque is relevant here (Orlando Hernandez). He would just throw any goddamned thing on any count, like Marcum; you just never knew what he would throw. Greinke is like that. I believe that there is more tolerance for that now than there was 10 years ago. But again, I am not really certain.