A Preliminary Analysis of the Decline in Average Shot Length in American Cinema From 1940-2009 as Evidence of Shortening Attention Spans.
“What ideas are convenient to express inevitably becomes the important content of a culture.” –Neil Postman, Amusing Ourselves to Death
In 2015, there were a flurry of news stories arguing that average attention span had decreased to 8.25 seconds, allegedly less than that of a goldfish at nine seconds. These claims were met with both a alarm and skepticism. The problem of course is that the precise definition of attention span is hard to nail down, much less how we should measure. Still, it is hard to fight the impression that our cultural ability to focus on a singular thing has declined.
Concerns about decreasing spans are nothing new and writings on subject can be found back into the 1960s. The best known in the broader culture being Neil Postman’s Amusing Ourselves to Death published in 1985. Here Postman proposes that the proliferation the Television in the 1950s began to fundamentally transform numerous things about our culture, but specifically relevant here that it decreased our ability to focus on a single thing for an extended period of time, an ability Postman argues that we’d gained as the result of the printed word and broad literacy.
Supporting this with hard data is more challenging. Ideally, we would have repeated tests of some sort that had been performed since at least the 1940s or 1950s where we’d at least be able to make apples to apples comparisons between both periods of time and age cohorts over time. No such data exists. Instead, we are left to look for other, indirect indicators. Here, I propose using the Average Shot Length of commercial films. Before we go into the reasons why let me clarify what average shot length is and what it isn’t. Average shot length in not the entirety of the run time of the film. Instead, it is a measure of the time between “cuts” in a film.
The following reasons suggest that the increase in rapidity of cuts and the resulting lower average shot length could be a valid lagging indicator of reducing attention spans:
-
Viewing films is a periodic rather than daily, even continuous practice. (Note, it can be argued the introduction of widespread HDTV in the late 2000’s altered this, hence ending the study with 2009.) This makes them unlikely to create changes in cognition rather than responding to them.
-
Commercial films are expensive to make and so must, on the whole, respond rapidly to changes in consumer taste.
-
Films produced by US studios were primarily concerned with domestic revenue during this period and so those changing tastes can be localized to the US.
Using a sample of over six thousand films from the period we see the following:
-ASL Chart –
This chart makes it plainly clear that following a decade of increasing in the 1940s, average shot length begins a steady drop until by 2009 it’s approximately 1/3 of it’s peak. So the decline certainly correlates with TV being introduced, becoming widespread, and it’s consumption increasing. This means the hypothesis is viable and should be explored further, however this does not yet constitute proof of causation.
One additional indicator of how films respond to commercial pressures is the very different set of changes to the average running times of films during the same period:
-Running Time Chart-
The commercial pressures I propose we see in effect here are slightly different. Running time actually continues to go up and then levels off. In distinction to average shot length, which is shortened to keep consumer eyes on the screens, run time is lengthened for two reasons:
-
The consumer must feel like they are getting a fair quantity of product for their money. Shorten the running time too much, and it will produce dissatisfaction with ticket prices.
-
The movie theater makes its money primarily on concessions and a longer film means more consumption. Related to that there as been a consistent decline in the percentage of the ticket price that is actually retained by the theater. This means that, up to a certain point, holding the same people in the theater for longer periods of time becomes more profitable than additional showings, especially in the late 2000s as the percentage returning to the studio for major films near 90% and in a few specific cases even reached 100%.
Limitations
Average shot length is not a commonly reported statistic of films and the backbone of this study is Barry Salt’s database of shots and average shot lengths which is widely available on the internet and Cinemetrics. This database was then merged with information exported from IMDb by year and title in order to gain additional data on each film. The sample is limited, once merged, cleaned, and reduced to only US films, to about 5000 films from the time period. This is a visual representation of the data set used:
-Data Set Combined Chart-
The database contains a good number of samples from each year, but it is weakest during the 1970s and 1980s where additional data would likely reduce the standard deviation we see in those periods. This data set is large but not truly random. It does not appear to be purposefully biased in any way, but the exact selection criterion for which movies are present is unknown.
Further Study
Further cleaning of the existing data should allow a future attempt to correlate the decreased average shot length with box office success. Additionally, Cinemetrics has a much a larger set of data for additional films. This dataset however is full of irregularities and would require extensive cleaning and verification to be remotely usable. Cinemetrics is also beta testing an VLC plugin that would calculate shot length as the film as displayed. A tool similar to this capable of running in an automated fashion over an entire archive of films would be the method to produce the best possible dataset for the strongest analysis.
Finally, access to the Nielsen total family data for the respective years would enable a strong case to be made for the correlation between TV viewing habits and ASL data.