An interesting newsletter by Adam Mastroianni pulls together data from movies, television, music, books, and video games to show that in each of these domains, the biggest hits dominate an increasing share of revenues, and are themselves increasingly dominated by franchises. Here are a few representative charts that Mastroianni put together:
Lines going up from Pop Culture Has Become and Oligopoly by Adam Mastroianni
A lot has been written about how franchises dominate the box office today, and some have used this to argue that progress in culture (or at least some forms of culture) has stalledout. Seeming to bear out that conclusion, Mastroianni’s post showed this problem extends to pop culture in general.
But in fact, this problem goes beyond pop culture too.
A 2021 paper by James Evans and Johan Chu titled Slowed Canonical Progress in Large Fields of Science shows that, in science, top papers garner more and more citations, turnover of top papers has slowed, and everyone cites the same papers (academic franchises?). In other words, science seems beset by similar problems as pop culture. Fortunately, Evans and Chu’s paper sketches out a model of why this happens, and I think their explanation can be extended to pop culture as well.
The heart of their explanation has to do with the total quantity of scientific papers, which has expanded enormously over the last 100 years. In the figure below, each point represents a field in a given year. In the left figure, on the vertical axis we have citation inequality. In the right figure, it’s correlation across years of the ranking of the top 50 most cited articles, by field. In each figure, on the horizontal axis is the size of the field (# of papers/yr).
These figures show bigger fields have more inequality and more ossification of top cited papers.
From Evans and Chu (2021)
Mastroianni’s newsletter plots trends related to franchises and concentration against time, but we know from the work of people like Joel Waldfogel that the number of books, movies, songs, and TV shows has gone way way up over time (I assume the same for video games, but Waldfogel doesn’t look at that).
From Digital Renaissance by Joel Waldfogel
I suspect that if you re-plotted Mastroianni’s figures so that the horizontal axis was the number of new works per year, you would get an even tighter relationship. In other words, in any field of content production, whether it’s scientific journal articles, movies, music, TV, books, or video games, as the number of new works per year increases, the most popular stuff garners an increasingly large share of the audience, and is increasingly composed of franchise-like content.
Why?
Dynamics of Attention
I think the key idea across these domains is network effects; the value of a work is higher when it has a larger audience. If you’re a scientist, you want to work on research related to what other people are working on; similarly, you want to communally experience art with other people. To take advantage of network effects, it is necessary for the audience to coordinate on what to give limited attention.
Evans and Chu, talking specifically about scientific papers, argue that when the flow of new papers is small, people can sample very broadly and form judgements which are shared and debated; over time, a consensus about which papers are best can gradually emerge.
But when scientific fields get too big, this dynamic no longer works. There are two problems. First, in a big field it takes longer for people to arrive at a consensus if they are each independently sampling over an increasingly small share of all the papers. It just becomes less and less likely two people have read the same paper and can build a consensus around it. Second, even while this process takes longer to play out, it is disrupted at a faster rate by the inflow of new original works. People lose interest in hashing out the merits of older papers and move onto the new thing. As evidence of this dynamic, note the increasing correlation of the rankings of the top 50 papers, as fields get bigger. It gets increasingly rare for new papers to dethrone older champions.
And when that does happen, it follows a different dynamic than this gradual process of consensus building via independent assessment and discussion. Instead, as fields get bigger, the hits achieve their success via some kind of viral channel that focuses a lot of attention quickly on the paper (a favorable writeup in the NYTimes? A big-name author? Twitter vitality?).
The figure below is the median time until a paper enters the top 0.1% most cited (vertical axis) against the size of the field (horizontal). Note small fields can slowly build up steam, and take a long time to reach top 0.1%. But in big fields, it’s more like instant fame or bust. You either make a splash instantly or people move on to the next thing.
From Chu and Evans (2021)
I suspect the same dynamics underlie the pathologies in pop culture. When a domain is small, high quality can be identified by a similar process of consensus building via independent sampling. But when a domain gets big, audiences become too fragmented for that approach to work well. Just as scientists lose interest in hashing out the merits of older papers and move onto the next thing, so too do pop culture audiences lose interest in disputed art of the past (which few of their friends have seen anyway) and move on to something new.
That also means, as a field grows, the best strategy for a commercial art producer changes. When a field is small, it may be a viable strategy to simply make the highest quality (mass appeal) art possible, and then count on enough people to see it and convince others to do likewise. A consensus about it’s quality will emerge.
But when a field is large and crowded, that strategy may no longer be viable. With few exceptions, the only art that can benefit from network effects is art that quickly captures a large share of audience attention. If that art happens to be high quality, then the audience will identify it and you’ll have a big hit. But even if it’s not, if the network effects are strong enough, people might settle for mediocre work that they can enjoy communally.
To capture audience attention, you need to finance art that is capable of sending powerful signals to the audience, to coordinate their attention prior to release. One of the best ways to do that is to build franchises; in that setting, the quality of prior art is a credible signal of the quality of new art.
If you believe franchises tend to be lower quality than the best original stuff, then in the past, this franchise-based strategy would have been worse than one that only focused on producing the highest quality stuff. That’s because, in the past, the smaller number of new products would have meant the highest quality stuff would eventually be identified and attract more attention than comparatively mediocre franchise offerings. But that’s no longer the world we live in.
So it’s not that we’re running out of ideas; we have more than ever! But that very fact is also a curse that means the most popular stuff looks less and less creative. It has to look like what’s come before, in order to win the war for attention in an environment of superabundant choice.
Adapted and extended from a twitter thread on this topic.
The Curse of Plenty
An interesting newsletter by Adam Mastroianni pulls together data from movies, television, music, books, and video games to show that in each of these domains, the biggest hits dominate an increasing share of revenues, and are themselves increasingly dominated by franchises. Here are a few representative charts that Mastroianni put together:
A lot has been written about how franchises dominate the box office today, and some have used this to argue that progress in culture (or at least some forms of culture) has stalled out. Seeming to bear out that conclusion, Mastroianni’s post showed this problem extends to pop culture in general.
But in fact, this problem goes beyond pop culture too.
A 2021 paper by James Evans and Johan Chu titled Slowed Canonical Progress in Large Fields of Science shows that, in science, top papers garner more and more citations, turnover of top papers has slowed, and everyone cites the same papers (academic franchises?). In other words, science seems beset by similar problems as pop culture. Fortunately, Evans and Chu’s paper sketches out a model of why this happens, and I think their explanation can be extended to pop culture as well.
The heart of their explanation has to do with the total quantity of scientific papers, which has expanded enormously over the last 100 years. In the figure below, each point represents a field in a given year. In the left figure, on the vertical axis we have citation inequality. In the right figure, it’s correlation across years of the ranking of the top 50 most cited articles, by field. In each figure, on the horizontal axis is the size of the field (# of papers/yr).
These figures show bigger fields have more inequality and more ossification of top cited papers.
Mastroianni’s newsletter plots trends related to franchises and concentration against time, but we know from the work of people like Joel Waldfogel that the number of books, movies, songs, and TV shows has gone way way up over time (I assume the same for video games, but Waldfogel doesn’t look at that).
I suspect that if you re-plotted Mastroianni’s figures so that the horizontal axis was the number of new works per year, you would get an even tighter relationship. In other words, in any field of content production, whether it’s scientific journal articles, movies, music, TV, books, or video games, as the number of new works per year increases, the most popular stuff garners an increasingly large share of the audience, and is increasingly composed of franchise-like content.
Why?
Dynamics of Attention
I think the key idea across these domains is network effects; the value of a work is higher when it has a larger audience. If you’re a scientist, you want to work on research related to what other people are working on; similarly, you want to communally experience art with other people. To take advantage of network effects, it is necessary for the audience to coordinate on what to give limited attention.
Evans and Chu, talking specifically about scientific papers, argue that when the flow of new papers is small, people can sample very broadly and form judgements which are shared and debated; over time, a consensus about which papers are best can gradually emerge.
But when scientific fields get too big, this dynamic no longer works. There are two problems. First, in a big field it takes longer for people to arrive at a consensus if they are each independently sampling over an increasingly small share of all the papers. It just becomes less and less likely two people have read the same paper and can build a consensus around it. Second, even while this process takes longer to play out, it is disrupted at a faster rate by the inflow of new original works. People lose interest in hashing out the merits of older papers and move onto the new thing. As evidence of this dynamic, note the increasing correlation of the rankings of the top 50 papers, as fields get bigger. It gets increasingly rare for new papers to dethrone older champions.
And when that does happen, it follows a different dynamic than this gradual process of consensus building via independent assessment and discussion. Instead, as fields get bigger, the hits achieve their success via some kind of viral channel that focuses a lot of attention quickly on the paper (a favorable writeup in the NYTimes? A big-name author? Twitter vitality?).
The figure below is the median time until a paper enters the top 0.1% most cited (vertical axis) against the size of the field (horizontal). Note small fields can slowly build up steam, and take a long time to reach top 0.1%. But in big fields, it’s more like instant fame or bust. You either make a splash instantly or people move on to the next thing.
I suspect the same dynamics underlie the pathologies in pop culture. When a domain is small, high quality can be identified by a similar process of consensus building via independent sampling. But when a domain gets big, audiences become too fragmented for that approach to work well. Just as scientists lose interest in hashing out the merits of older papers and move onto the next thing, so too do pop culture audiences lose interest in disputed art of the past (which few of their friends have seen anyway) and move on to something new.
That also means, as a field grows, the best strategy for a commercial art producer changes. When a field is small, it may be a viable strategy to simply make the highest quality (mass appeal) art possible, and then count on enough people to see it and convince others to do likewise. A consensus about it’s quality will emerge.
But when a field is large and crowded, that strategy may no longer be viable. With few exceptions, the only art that can benefit from network effects is art that quickly captures a large share of audience attention. If that art happens to be high quality, then the audience will identify it and you’ll have a big hit. But even if it’s not, if the network effects are strong enough, people might settle for mediocre work that they can enjoy communally.
To capture audience attention, you need to finance art that is capable of sending powerful signals to the audience, to coordinate their attention prior to release. One of the best ways to do that is to build franchises; in that setting, the quality of prior art is a credible signal of the quality of new art.
If you believe franchises tend to be lower quality than the best original stuff, then in the past, this franchise-based strategy would have been worse than one that only focused on producing the highest quality stuff. That’s because, in the past, the smaller number of new products would have meant the highest quality stuff would eventually be identified and attract more attention than comparatively mediocre franchise offerings. But that’s no longer the world we live in.
So it’s not that we’re running out of ideas; we have more than ever! But that very fact is also a curse that means the most popular stuff looks less and less creative. It has to look like what’s come before, in order to win the war for attention in an environment of superabundant choice.
Adapted and extended from a twitter thread on this topic.