Baseball is in crisis. And major contributing factors are the data and internet revolutions. Analytics and data analysis have completely changed how the game is managed and played. It has meant fewer balls in play and longer game times. And a crisis of both demographics and viewership is already underway as a result.
Politics has undergone similar changes. The creation of new metrics for demographics and voter analysis has enabled changes to how politics is played and managed as well. Advancements in data analytics have allowed baseball to create new metrics like launch angle and situational spread to focus on offensive production. In politics, micro-targeting and data analysis have enabled politicians to calibrate their messaging to narrower and narrower groups of voters using wedge issues and culture calls.
A lot of this change started in the mid-1980s and, believe it or not, Steve Jobs was in the middle of it. And like so many changes that were celebrated and are now worrying us, big tech features as well.
Excel was released in 1985 and followed the successful Lotus 1–2–3 spreadsheet program that came out in 1982. Column organized data and the ability to conduct nearly 500 different functions to transform and analyze data made data crunching easier than it had ever been before. In 1991, things changed again when Lotus released Improv on Steve Jobs’s NeXt computer. Though NeXt didn’t work out, Improv left behind one of the most important tools on the tech era: the pivot table. It was released in 1993 and Excel adopted the pivot table function in 1994. The pivot table and functions based on it gave birth to modern data analysis.
In 1994, the major league season ended early in a strike. The prior year, the Toronto Blue Jays won their second straight World Series. A quick look at some statistics from that year show that six teams scored more than 800 runs that season, the Detroit Tigers had the most at 899. Six teams struck out more than 1,000 times that season. The free-swinging Tigers led in that category as well with 1,122 strikeouts. In 2019, the last season before COVID disruptions, the changes to the game leap off the page when you look at the statistics. Every single major league team struck out more than 1,100 times. In 1993, only two teams had combined batting averages under .250; in 2019, that number was 15. In 2019, 12 teams scored more than 800 runs. (A small note: there were only 28 teams in 1993 compared to 30 in 2019.) The data on home runs is still more jarring: in 1993, no team hit more than 181 home runs. In 2019, only five teams DID NOT hit that many. With the advent of data analysis, digital video and photography in coaching, as well as other new technologies, teams have learned how to score way more. They do it by swinging way more, caring less about hitting anything other than homers and just not caring about strikeouts. Indeed, in 1993 the team that struck out the most were those same Tigers with 1,122 . In 2019, the team that struck out the least were the Houston Astros with 1,166. You read that correctly. In 1993, the team that struck out the most times in the season still did so fewer times than any of the thirty teams in 2019.
The subcategorization, summary, and carving up of data has enabled teams to do all kinds of things. They can now easily identify how likely a hitter is to ball to the left or right against a right handed pitcher that throws mostly off speed pitches, for example. They can then use digital photography and analysis and then plot those data to focus that player’s attention on the optimal launch angle to hit a home run in any situation. That kind of data tabulation and analysis would have been almost impossible before computers, but now the data model can be built and analytics ran on it in mere minutes. The results can be used to position the infield and outfield where a hitter is most likely to hit the ball in that specific situation. The results are more runs, way more strikeouts and home runs. All of that together also means more pitching changes and fewer balls in play.
In the time between 1994 and today, data modelling and analysis have done similar things to voters as they did to baseballs: they have enabled teams of analysts to undertake highly specific subset analyses (micro-targeting) enabling predictive or preventative action. Political consultants can, using a series of data related to demographics, income, and geographic data, micro-target individual voters or identify the specific wedge issues that will prevent a small subset of their electoral coalition from defecting to the other party or parties. In short, political consultants now no longer care about batting average either, only home runs. Or, more to the point, preventing the other team from getting homeruns. They no longer care to try and triangulate their party’s focus to attract the greatest number of voters. They just care about the specific issues that get them the exact number they need: 50+1%.
Money ball has come to politics, just as it came to baseball and the systems of both have adapted poorly. In politics, across the western world, political systems are based on the premise that parties will seek fairly broad electoral coalitions, particularly in first-past-the-post systems. Party systems, as they developed in both the U.S. and U.K., grew around appeals to the working and middle classes in the mid-nineteenth century. The modern American party system emerged as a result of Andrew Jackson’s mass politics and his appeal to middle-class Americans. In the U.K. the expansion of the franchise led to the growth of the Liberal Party and the reactions and adaptation of the Tories along similar lines. Either way, the focus was on building parties to attract broad coalitions of voters around major issues.
These motivations to go big are no longer present now that the voters can be sliced and diced in databases, sorted into optimal groupings so that politicians now know how to get to fifty percent-plus-one only. That’s all they need after all. You don’t get extra power by winning 70–30 rather than 50.1–49.9. The incentive to go after broader coalitions just doesn’t exist when data can be used this way.
In baseball, this has led to a near complete breakdown in the pace and flow of play. Game times are increasingly long and fewer and fewer balls are put in play. In ’93 the average game lasted 2 hours and 52 minutes; in 2019, it was 3 hours and 10. That’s almost twenty more minutes to try and keep viewers interested or attendees in their seats.
In short, the game is getting more boring.
In politics, the stakes are considerably higher and the consequences of this data analytics revolution affect citizens everywhere. Many important thinkers like Ezra Klein in the US and Matthew Goodwin in the UK have pointed out the impact of micro targeting on political parties. There is also excellent political science literature available. In short, data analysis and micro-targeting has decoupled political parties from their need to seek broad support. They now only need the right configurations of small groups to win.
In the US, data targeting is the logic of the Republican Party. Roughly thirty-five percent of Americans support Republican politics. This 35 percent is made up of several constituent groups: the very wealthy, the South, and evangelicals. That’s the great appeal of Trump. He brought white working class voters to that coalition, giving Republicans a chance at 50% + 1. Voter analytics let the GOP target these coalition members in specific states, allowing them to capture power without a broad majority in 2016 (and nearly in 2020.) Elections have become about micro-targeting voters in specific areas rather than appealing to potentially new voters that don’t fit into a party’s data model.
And this is just the beginning. With revolutions in cloud computing and the powerful processing chips available in the last decade, we are at the cusp of a revolution in AI and machine-learning enhanced data analysis. And politicians will not only be able to analyze voter behavior, they will become more adept at predicting it as well. Think of a situation where a more disciplined, but equally dangerous Trump-like politician can leverage predictive analytics to understand exactly how he (or she) can get just enough votes in Wisconsin and Pennsylvania to win the presidency. They don’t need to care if they lose the popular vote by 10 million, or 20 million for that matter.
The technology is here or very nearly here already. Algorithms and machine-learning enhanced models that take advantage of this new processing power allow for the crunching of vast data sets to predict the behaviours of small groups of voters.
So what do we do about this? Is this really even a bad thing?
We have to do something, not because these changes are necessarily for the worse, but because the interactions with these new technologies for both systems — baseball and politics — is creating bad outcomes. Boring baseball games and a politics of paralysis are results of these technologies and their deployment. It’s an unintended consequence, but a consequence nonetheless. We need to reform systems before the technology disrupts these systems further.
In baseball, predictive analytics combined with the largely situation-driven nature of the game will make positioning fielders and determining pitches ideal fodder for smart algorithms to manage games and further reduce offence and excitement in the game. In politics, using predictive analytics and micro-targeting, politicians will be able to correlate their messaging precisely to get the 50+1% that they need. This move, combined with reductions in voting rights in the United States, is what makes a counter-majoritarian Republican politics possible.
Political deadlock has been the result so far, particularly in the United States. After all, analytics has enabled congressmen to stay in office by leveraging just a few issues to keep their seats rather than seeking a moderated position that attracts majorities across groups. Instead, he or she can just leverage a minor wedge issue to just turn out enough of one group to make the difference. It makes compromise and pragmatism — the only way that politics can really be functioning in a large and diverse democracy like America’s — difficult or even impossible. Instead, more and more, this is leading to people in both parties that are less and less democratic. The system isn’t responsive to this problem, so actions that undermine the system are a logical reaction. Dangers to American democracy are increasing.
At the same time baseball is rapidly becoming unwatchable. As a stats-head and a diehard baseball fan it kills me to admit: statistical analysis is bleeding the game of its entertainment value. The average game length has increased by 17 minutes (that’s more than half of a sit-com episode by more modern time measurements) since 2007. Young viewers (as in those under about fifty) are abandoning baseball in droves for shorter and more exciting screen sports like basketball. Or, you know, there’s one of the many streaming platforms that were created while you were reading this. Baseball is in danger too.
Thankfully, these things have been created that have disrupted both of America’s national pastimes can be un-created, or at least innovated against. Jayson Stark, a baseball columnist has come up with just one great idea to slightly alter the rules of baseball by forcing teams to give up their designated hitter once they take out their starting pitcher. It would force teams to make a material sacrifice when they try to game the system on the pitching side. Similar ideas are gradually being developed on the political side too.
One horrible idea is the elimination of the senate filibuster. I know, as a liberal, it initially seemed appealing to me too. But it’s got two big issues. First, it will obviously be used against the Democrats once Republicans get the senate back, which they inevitably will. More than anything though, it’s a bad idea because it doesn’t actually solve anything. It does nothing to push candidates to have to appeal to more than a narrow sliver of voters. It does nothing to promote pragmatism of broadly supported policy choices. The only way to do that is to stop campaigns from being able to use this data-driven technologies to slice-and-dice the electorate. Banning it isn’t possible; the Supreme Court would never allow it. A constitutional amendment is the only way to change the rules; unfortunately, that isn’t as simple as 30 owners getting together to change the rules like in baseball.
Instead, America should starve the data analytics systems in their political system of their constituent element: the data. Laws should be passed or executive orders issued to prevent anyone (including corporations) from using a person’s data in any type of data analytics project without paying them for it. Second, additional regulations over political advertisements on social media should be passed to prevent the appeals to the slivers over wedge issues which are so often cultural and racial.
The data explosion has had myriad good results for medical research and population analysis, among many other things. But there have also been tremendous costs, many of which we have been slow to realize. Big picture, baseball doesn’t matter. If Major League Baseball become a third-tier sport airing only on ESPN 8 (The Ocho) then who really cares? But the manipulation of the political system and the deepening divides between political groups is a serious thing. Indeed, there are few things more serious. It threatens liberal democracy everywhere. And data analytics and this exciting technology has done nothing but harm to political consensus making and good policy. We all need to step in and save America (and, if we have time, baseball) from the Excel nerds. And quick.