Baseball commissioning must have seemed so easy at first. Rob Manfred took over in January 2015, announced his intention to speed up the game and immediately got results. A nine-inning game shrank by six minutes from 2014 to 2015, matching the largest decrease ever. At that pace, baseball would be gone entirely within 30 years and Manfred could retire.
But as with two-homer games on Opening Day, this pace would not hold. Baseball added back four minutes per contest last year, and this year's games will almost certainly set a new record. Despite a few small changes for the 2017 season -- requiring managers to decide more promptly whether to challenge an umpire's call and making intentional walks automatic -- the average nine-inning contest is now 3 hours, 5 minutes -- the longest average in baseball history and up five minutes from last year. That's also already the second-largest year-over-year increase since integration. And since games get slower with September call-ups, the final month will quite possibly add a minute or two to that average.
Manfred, who remains insistent that the league will do something about this, has drawn a distinction between time of game and pace of game. As he told reporters in June:
Time of game is often what happens on the field competitively. How many runs get scored, how many guys on base, how many times you change pitchers. Those are things I'm not looking to control. Because that's about the competition. That's up to the clubs.
So, how much of the nine minutes added since 2015 fall under "what happens on the field competitively?"
There are more plate appearances per game.
On-base percentage bottomed out in 2014 and 2015, with the lowest leaguewide OBPs since the early 1970s. While strikeouts have continued to go up since then -- by more than a strikeout per game -- everything else has either held steady or benefited offenses: There are two-thirds of a walk more per game, 25 percent more home runs, slightly more hit batsmen and slightly fewer sacrifice bunts, sacrifice flies and runners caught stealing.
The cumulative result is that each game will have almost one extra plate appearance -- 0.78, to be exact. This partial plate appearance is responsible for about 1 minute, 20 seconds per game.
There are more pitches thrown per plate appearance.
This has been going on for decades, as the average matchup had gone from 3.57 pitches in 1988 to 3.82 per PA in 2009. Batters have become more patient as OBP has supplanted batting average in prestige. Pitchers' higher velocity and nastier stuff have led to more swinging strikes (which generally elongate at-bats). And the strikeout paradox -- in which batters and pitchers have simultaneously pursued strategies leading to more strikeouts -- has led to deeper counts overall.
After 2009, that figure of 3.82 pitches per PA held steady. We'd reached the peak, it seemed. The rate per at-bat didn't shift by more than one-hundredth of a pitch in either direction for six years, until, suddenly, it jumped to 3.87 in 2016 -- the highest one-year jump since pitch data started being recorded in 1988 -- and to 3.90 this year. An extra eight-hundredths of a pitch per plate appearance, applied to 76 PAs per game, adds six pitches. Each pitch lengthens an at-bat by about 24 seconds, so we've now added almost 2½ minutes more.
Three minutes, 45 seconds are now accounted for.
There are more plate appearances (and, thus, more pitches thrown) with runners on base.
Pitchers universally take more time to throw the ball when there is a runner on base -- about 7 extra seconds, on average -- and the aforementioned higher on-base percentage would tell us that a higher percentage of baseball is now happening with men on base.
In fact, about 43.2 percent of all plate appearances now come with men on. In 2015, that figure was 42.9 percent. This doesn't actually do much to our clock: It means about one extra pitch per game will be thrown with men on, slowing us down by only about 7 seconds on average.
There are more pitching changes.
Pitching changes ramp up as the season gets late. But if we just look at the first halves of the past three years, we can see 2017 is on track to set a new record for pitchers per game per side:
2015: 3.97 (4.11 for full season)
2016: 4.02 (4.15)
Not all pitching changes add time to the game. Most new pitchers come in to start a new inning, but about a third are made in the middle of an inning, which means that since 2015 there has been about one extra mid-inning pitching change for every 12 games. At about 2 minutes, 20 seconds per pitching change, we've added 12 seconds per game.
So we've accounted for a little more than four extra minutes. Where's the rest of the time coming from?
Defenses are taking more time between pitches.
So, this one is complicated. FanGraphs uses PITCHf/x time stamps to track how long pitchers spend between offerings within an at-bat. They call this stat "pace," and it has been going up hard:
2015: 22.8 seconds
2016: 23.2 seconds
2017: 24.3 seconds
Multiplied across nearly 400,000 pitches, this alone accounts for 5½ minutes. So, easy: Make the pitchers speed up by putting in a pitch clock, as Triple-A and Double-A did -- with great success -- and as Manfred seems prepared to do next year with or without the blessing of the players' association.
Except here's where it's a little complicated. Baseball Prospectus also uses PITCHf/x time stamps to track how long pitchers spend between offerings within an at-bat. They also call this stat "pace," and it has ... actually gone down:
2015: 21.7 seconds
2016: 21.8 seconds
2017: 21.1 seconds
The difference is in which intervals each method excludes as abnormal. FanGraphs tosses out any interval more than 50 seconds, and Baseball Prospectus throws out any more than 60 seconds. This is based on the shared assumption that something other than "pitcher stares in for a sign" happened: a visit from the catcher, a batter writhing in pain after fouling one off his shin, a shortstop coming over to talk about who will cover second on a comebacker, whatever.
But BP also throws out any intervals that include pickoff attempts. And it throws out any intervals following a foul ball. Foul balls generally take longer: The pitcher gets a new ball to rub up, the defense might have to get back in position after a chase, the batter is allowed to step out and adjust his batting gloves, etc.
Whether one of these approaches is "right" is less important than the fact they give us two different data sets with two different trajectories, and our answer might be in the middle.
In one set of a data -- the pitcher throws a pitch, then gets the ball back, then throws the pitch, then gets the ball back, and so on -- pitchers aren't working any more slowly this year than they did before. In fact, they're slightly faster!
But in the other, the game has slowed and keeps slowing. This suggests it's the slowness caused by mild interruptions -- even the rubbing up of a new ball, or the flip of a pickoff throw -- doing most of the damage to our schedules. It suggests pitchers, fielders and batters are taking more time to get back into rhythm after these interruptions. Where there's even the smallest pinprick in baseball's rhythm, wasted time rushes in to fill it.
Manfred will keep trying, and he'll have successes and failures along the way. The bad news is "what happens on the field competitively" has naturally led to somewhat longer games -- more baserunners, more pitches, more pitching changes. The good news, though, is there is a lot of stuff happening between pitches none of us would miss all that much. Baseball's natural pitch-catch-pitch-catch rhythm is fairly slow, but that's not what has been making it slower. To fix that, Manfred will need to seal the loopholes that let players interrupt the rhythm.