On Baseball, Game Design, and Output Randomness

Anthony Rendon had a sub-.400 slugging percentage on low fastballs, yet he still homered off Clayton Kershaw in Game Five of the NLDS. (via Keith Allison)

The competition between a pitcher and hitter is often described as a chess match, so often in fact that in 2013, Miles Wray at Vice wrote, “It’s a shame that describing the interplay between pitchers and hitters as a chess match has achieved Unpardonable Cliché status.” For such an overused expression, it’s not an apt comparison. The analogy contends that while baseball is an athletic game of speed, power, and grace, above all else, baseball is a thinking person’s sport. Like chess, it’s mostly a game of deception, preparation, and strategy. However, through the lens of modern game design, if baseball is a strategy game, it’s a terrible one.

The key difference between chess and baseball is that in chess a decision will always reap the expected return. The right move in chess will always yield a good result. As Clayton Kershaw demonstrated in Game Five of the NLDS, sometimes the right pitch will have the worst possible outcome.

On a 1-0 pitch, Kershaw threw Anthony Rendon a fastball on the inner half of the plate and well below the strike zone. Rendon generally doesn’t hit low fastballs as well; he had slugged just .395 against four-seamers at the lower border of the strike zone throughout his career. With Kershaw’s velocity below 90 mph, perhaps it wasn’t the optimal pitch, but it was a good pitch. Rendon, of course, homered despite never having done so on any pitch quite so low. That cut the Dodgers’ lead to just one run, and the next batter, Juan Soto, would erase the deficit, and the Dodgers would never have a lead for the brief remainder of their otherwise excellent 2019 campaign.

If pitching and hitting were truly like chess, Kershaw would have been rewarded for making a good pitch. Maybe Rendon would have swung and missed to bring the count even or maybe he would have grounded out to third. Conversely, if baseball were really like chess, any time Kershaw throws a center-cut fastball, it would be hit 450 feet. The outcome of a pitch is so far removed from the process of selecting it that it can occasionally render any strategy moot.

The problems a pitcher, a manager, or a front office face are mostly solvable with pitching tracking data and run expectancy tables. The only reason Kershaw’s Dodgers haven’t won a World Series in the seven consecutive years they’ve won the NL West is that baseball is a highly variable game. A manager isn’t moving pawns on a board when he makes make a pitching change; he’s rolling dice. No baseball decision is guaranteed success or failure. Baseball, ultimately, isn’t a strategy game. It wasn’t designed to be. It’s designed to be random chaos, a push-your-luck gamble where the best team doesn’t always win, and baseball is so much better for it.

When designing a game, this question must be answered: Who is this game for? The answer determines how the game is constructed and how players (and perhaps spectators) will interact with it. Chess is a game for people who want a mind-crunching and fair game. One  reason chess persists for lovers of strategy games is that there is no randomness in the game. The board is set up in exactly the same way every single game, and actions have fixed outcomes. Chess is the quintessential perfect information game, a game where there’s no hidden information, no random events, no unintended outcomes. A knight will always take a pawn when moved into that pawn’s square. Compare that to a game like Risk where an invading army could vastly outnumber the defenders but still lose the battle. A less skilled player could conceivably win a game of Risk with a series of lucky rolls, but the better player will always win a chess match.

The absence of randomness in chess makes it fair, but it can also make it unexciting and even predictable. There are no stories of plucky underdogs working their way up to grandmaster. Chess rewards only skill. Randomness is a common part in games because while it makes things less fair, it can also invigorate a stale system, present new challenges, and create stories. It’s all dependent on what form that randomness takes. In game design, there are two major forms of randomness: input and output.

Input randomness is any indeterminate element generated prior to a player’s decision. Settlers of Catan utilizes input randomness at the beginning of the game when players shuffle and draw hexagonal tiles to construct the board. Each different board state will yield certain resources more readily than others and it’s up to players to analyze which hexes will be most fruitful when placing their first settlements.

In baseball, there’s very little input randomness at the game level, but this affects how teams construct their big-league rosters and farm systems. Negotiating with free agents and working out trades during the offseason isn’t so dissimilar from drawing and bartering storefronts and factories at the beginning of Chinatown. Drafting amateur players isn’t so far removed from drafting dice in Sagrada or selecting bonus cards in Terra Mystica. 

Output randomness is variance in results from performing the same activity. Common examples of this include rolling dice and drawing cards. In Settlers of Catan, players roll dice at the beginning of their turn to see which hexes yield resources that round. You may have made good decisions in where to place your first structures, but a few bad rolls at the beginning of the game can put you in a hole. In baseball, throwing a slider (or any pitch) is similar to rolling dice to resolve an attack roll in Dungeons and Dragons.

When a pitcher throws a slider, he could be said to roll a d20, add his breaking ball modifier to it, and if it beats the hitter’s defense (how well he hits sliders) the pitcher will get a swing and a miss or a ground out. An entire season of baseball could be simulated in this way assuming there is sufficient data on each pitcher, hitter, and fielder.

This is essentially how Strat-O-Matic baseball is played, and the board game version of baseball has achieved cult classic status for lovers of baseball statistics, but the game mechanics haven’t brought it to a wider audience. Strat-O-Matic baseball currently ranks 744th among all strategy games on Board Game Geek but 275th among thematic games, which would indicate that people like it because it’s a baseball game and not because of the in-game decisions it forces a player to make. If all the baseball were removed from Strat-O-Matic and the game became an abstract contest to see who could move the most cubes around a circle before failing a roll three times, it wouldn’t have the same cachet.

Neither type of randomness is inherently better than the other though each has its advantages and disadvantages. Elements of input randomness are often consistent for every player and are generally viewed as being fairer. No one would argue that Catan’s randomly generated map benefits any single player, only the player with the most experience and the greatest capacity for probabilistic analysis. Input randomness rewards good play.

Output randomness is often unfair. Game designers like Keith Burgun have argued that output randomness does not belong in strategy games. A player should be rewarded for making the right move and be punished for screwing up. It’s unfair and frustrating to employ a clever strategy just to have it backfire. Good process should always yield a good result. 

A Hardball Times Update
Goodbye for now.

Output randomness, however, can sometimes generate the most memorable experiences a player might have. It’s no accident that role playing games like Dungeons and Dragons rely heavily on dice rolls; i.e., output randomness. The purpose of those sorts of games isn’t to concoct the best strategy, it’s to tell a story, and dice rolls create unpredictable moments and facilitate creativity.

The randomness and chaos of baseball might make it unfair and frustrating for Clayton Kershaw or for the Dodgers front office that built the strongest, most consistent team of the decade but has gotten unlucky in the postseason. Baseball, though, isn’t for any particular front office or individual player. Baseball is for everyone. The unpredictability is what makes baseball worth watching. If the purpose of baseball were to make it fair (i.e. to reward the best strategy) there wouldn’t be a postseason. The regular season would be a 145-game round-robin where every team plays every other team five times each, and the team with the most wins at the end would be declared the winner. An alternative would be to make every postseason series a “best-of-75.”

Those might be the surest way to crown the best team the champion at the end of the season, but it would be horrendously boring. The current system might be flawed, but it allows for emergent stories. We remember teams like the 2004 Red Sox and we’ll remember the 2019 Washington Nationals because those teams overcame the odds and ended droughts that were themselves the result of bad luck. We don’t have the same fondness for the teams with the best process.

Newest Most Voted
Inline Feedbacks
View all comments
2 years ago

Great piece.

Marc Schneider
2 years ago

It seems like, over the course of a long season, the output randomness is eliminated or at least reduced. The better team will usually win the division (although injuries, in-season acquisitions, etc. can change this). It’s the playoffs where randomness has the most effect, especially with the additional rounds. In effect, there really isn’t such a thing as a “favorite” to win the World Series because there is so much randomness in getting through the rounds.

2 years ago

I think the word ‘fair’ is too often mis-used when applied to output randomness. A ‘fair’ coin is one that lands heads or tails with equal probability. Betting against a scenario in which 10 consecutive flips of a fair coin land heads might be good strategy, but being wrong would not make the outcome ‘unfair’, just unlucky.

In baseball this comes up regarding blown calls. Too often I hear instant replay or robo-umps justified on the basis of fairness — that a blown call is ‘unfair’ to one side. This is only true if the umpire is biased such that bad calls are more likely to favor one team over another (i.e. not a fair coin). Absent such bias, bad calls are not at all unfair, just unlucky, and part of the output randomness that this article rightly celebrates.

2 years ago
Reply to  mgwalker

Blown calls are unfair, even if you’re assuming they are distributed randomly. Blown calls increase the variance of games. This increase in variance would give an underdog team a greater chance in winning.

2 years ago
Reply to  ryanredsox

I agree with the second and third of your sentences.

2 years ago
Reply to  ryanredsox

I duno, you have human officials and that’s just part of it. The strike zone simply cannot be called consistently by humans. That’s part of the game too.

2 years ago
Reply to  mgwalker

Blown calls are when your D20 roll beats whatever number it’s trying to beat and the DM says you die anyway.

2 years ago
Reply to  mateodh

I guess drawing a distinction depends on whether or not one is willing to consider the umpires and their impartial judgements to be ‘part of the game’.

2 years ago

amazing. calling to mind a lot of parallels in golf and golf course design. great read!

2 years ago

I stood with excitement when I saw that Kershaw pitch before Rendon swung, so maybe not the best example. Ryan Zimmerman’s homer on a very high pitch would probably be a better example.

2 years ago

so alright, given this randomness and chaos, will anyone ever “beat the streak” on MLB.com? In a world with a universe of possibilities, is it even close to likely?….20 years and counting, tick tock. Everything is all about probability, but the question is, are we accounting for all possibilities with those probabilities?

2 years ago

Very enjoyable read and you touched on a bunch of my favorite games.

I’m a believer in expanded playoff series, under the theory that talent will out over a long enough run. (That said, my love (or tolerance) for Cinderella stories depends on the sport. Would anyone want to see a best of 3 Super Bowl system? Probably not. All the ‘Any given Sunday’ clichés would be out the window.)

But I wonder if 7 games isn’t the sweet spot for series length. It is something of a magic number in memory research, with seven (plus or minus two) being the maximum number of pieces of information people can hold in their head at one time.