A few weeks ago, I wrote an article titled "The rise and fall of catcher framing," using the new advanced Baseball Prospectus data on catcher defense to show a recent narrowing of the range of framing ability. It appears that, in about 2008, a small number of teams began to aggressively pursue catchers who were good at receiving and teach their bad framers to be better. Since then, nearly every other team has followed suit, and the terrible framers have improved or fallen out of the league, making it much harder to gain a big advantage via catcher defense.
Since that article, I haven't been able to stop thinking about just how crazy this shift and counter-shift has been. In less than 10 years, a brand new method of winning games was "discovered," developed, and adopted wholesale into the game. It's possible this says something about framing specifically; it appears to be much more teachable than other baseball skills, which has perhaps enabled this narrowing of the range of talent. It's also relatively unique in that umpires can become aware of both the general concept of framing and who the standout framers are, and, as Jeff Sullivan has pointed out, start consciously attempting to counter their efforts. In any case, it's been a remarkable decade for catcher receiving.
However, framing isn't the only big advancement MLB has undergone. There have been numerous shifts in strategy or style that seemed seismic at the time and are now the norm. One of the areas with the most obvious and drastic changes is the bullpen. I see two major changes in the past. The first is the rise of the dedicated reliever and the ossification of clearly defined relief roles. The second, which is ongoing, is the move away from those roles. Looking at how those new strategies spread through baseball provides an interesting contrast to framing's adoption.
The creation of the modern reliever
In the early part of the 20th century, there was no position of "reliever" per se. Pitchers did occasionally enter games in relief, but usually only in blowouts. Those same pitchers would throw most of their innings as starters, however; no one was dedicated to relieving other pitchers (as @oldhossradbourn will undoubtedly tell you). As Rob Neyer writes in this great, in-depth documentation of the rise of the reliever from the National Pastime Museum:
[e]arly in the 20th Century, the best relief pitchers were . . . starting pitchers.
From 1908 through ’11, Mordecai "Three Finger" Brown ranked among the National League’s best pitchers, going 102-43 in that span. But in addition to his 123 starts, Brown also made seventy relief appearances, and led the National League in saves in all four of those seasons.
The first reliever by a modern definition didn't come until much later, when Clark Griffith was manager of the Washington Senators. Griffith himself had been a pitcher for 16 years (plus several more as player-manager), and in the last three years of his career, he started 20 games but relieved 38, perhaps planting the idea of a dedicated reliever in his mind. In 1923, he acquired Firpo Marberry, a pitcher with a hard fastball and not much else, exactly the kind of pitcher fans today identify as a relief candidate.
As sabermetrician Bill James wrote of Marberry in his book, The Bill James Guide to Baseball Managers,
Marberry was the first pitcher aggressively used to protect leads, rather than being brought in when the starter was knocked out. Thus Marberry is, in my opinion, the first true reliever in baseball history... He was a modern reliever–a hard-throwing young kid who worked strictly in relief, worked often, and was used to nail down victories.
Marberry went on to pitch 44.2 innings in 1923, most in relief, and in the following year, set a new major league record by saving 15 games and played a major role in the Senators' World Series victory with three appearances in relief. Saves might be basically useless from an analytic perspective, but they do indicate how a manager is using his bullpen, and I'm going to use them (and the Baseball Reference Play Index) to show how long it took the rest of baseball to adopt Griffith's innovation.
Marberry broke his own 1924 record twice more, in 1925 (16 saves) and 1926 (22 saves), showing that the Senators continued to use him in innovative fashion. It would be more than a decade, however, until another team approached the heights of dedicated bullpen usage set by the Senators, with White Sox pitcher Clint Brown making 53 relief appearances in 1937 for 18 saves, and Johnny Murphy of the Yankees making 38 relief appearances in 1939 for 19 saves.
With Brown repeating his 18 saves, 1939 marked the first time two different teams had players clear the 15-save threshold. It still took the league as awhole more than 20 years to adopt the strategy; it wasn't until 1963, four decades after Marberry's debut, that half the teams had a player with more than 15 saves, and most teams were employing some number of dedicated, full-time relievers.
The rise of the closer
Once relievers became an established part of baseball, the next change baseball underwent was the development of specified roles for different relievers and the solidification of those roles. Hall of Famer Dennis Eckersley, who is thought of as the first modern closer, was traded to Tony La Russa's Athletics in 1987 and became the first pitcher to be used almost exclusively in the ninth inning. His 45 saves in 1988 were one short of the single season record, and he (and La Russa) changed the bullpen forever.
I calculated the standard deviation of saves among all players with at least one inning pitched, by individual teams and leaguewide. A higher standard deviation indicates a more spread out sample, with greater extremes on either end. In this context, it indicates more players with very many saves and more with very few, as opposed to a number of players with some amount of saves. The A's figure in 1987 was 5.2; over the last decade, it had averaged 4.2. In 1988, it spiked to 11.5, the highest in MLB and the second-highest since 1960.
The change is also visible in the league-wide standard deviation. Over time, the number of pitchers with any playing time in a given season has increased, which would push the standard deviation down if all other trends were held equal. It's notable, therefore, that that hasn't been the case. The following graph shows the three-year rolling average of the standard deviation to remove some of the fluctuations. (I also removed the strike-shortened 1994 season.)
As you can see, the chart is fairly steady until the early '80s, showing the influence of La Russa and others on the game. By 1993, the year with the highest league-wide standard deviation, the league had entered the modern era of capital-C Closers, and it's remained there ever since. Compared to the four-decade development of the full-time reliever, dedicated closers arose at lightning speed, between 15 and 20 years in total.
Both of them, however, appear to be left in the dust by framing's wholesale integration into the game, which took about 10 years in total. I acknowledge that this isn't an apples-to-apples comparison, but it's useful to think about why the adoption of framing has proceeded so much faster. Framing requires only slightly different behavior from the catcher, while changes in bullpen usage required totally new routines from pitchers, and baseball players are not the most welcoming of change, as a rule.
I think, however, this has as much to do with the modern era and the tools of analysis being used by teams as it does with the inherent skills behind the shifts. The work put in to calculate framing's impact on wins and losses meant that teams knew relatively quickly that this wasn't something they should ignore. The changes in pitching had a less easily quantifiable impact on the game, which meant that adopting them was more of a stylistic choice than a strategic imperative.
Framing was not the first major shift in the strategy of baseball, and it certainly won't be the last. The speed with which the league adopts the next big insight will depend on what it is, but also on how we can measure it. Not only is baseball is changing, it's changing faster than it ever has before.
Henry Druschel is an editor at Beyond the Box Score. You can follow him on Twitter at @henrydruschel.