A common complaint for some time has been that films are longer than they used to be, and longer than they have to be. People argue about it on blogs, on Twitter (just the other day I became involved in such a discussion), and journalists and statisticians write articles about it. There is however no consensus as to whether films actually are longer now than they used to be. Some claim that they are, others respond that the data does not support that belief and that films are on average as long as they have been for some time. It is common enough that our perception of things is wrong, that even if it feels like something is in a certain way it often is the case that things are quite the opposite. Books on statistics are filled with examples of such gaps between our beliefs about something, such as crime rates, and the reality, we tend to believe crime rates are much higher than they are, and that they are always increasing even when they are not. So it is possible that films are not getting longer, even though to many it feels like they are.
I am one of those who believe they are getting longer. I worked as a projectionist for about ten years, 1995 to 2005, which gave me a Rain Man-like relationship to the length of new releases, and I feel quite strongly that while it is rare for a mainstream film these days to be shorter than two hours, it was rare in my projectionist days that they were longer than two hours. But, as I said, such a belief is not worth much. I want to crunch the numbers.
If you google the question you will get a number of hits on the subject but most of the ones I have checked have been questionable. There was for example an article in the Independent with the headline "Movies aren’t really getting longer, so why does it feel that way?" even though the graph by Randal Olson (they got it from here) that accompanied the article clearly suggested that they were getting longer, after a dip in the 1980s:
The most comprehensive deep-dive into this that I have seen so far is from a site called Towards Data Science. They tried to get to the bottom of it by using IMDb as the source, and published the result here. The result is summed up like this: "There is no trend in the movies runtime. The differences are too small to be noticed. We can say that for the last 60 years movies on average have the same length. No matter what criteria we take into account, the result is the same."
If this is true, it means that I am wrong in my belief that films now are longer, on average, than they used to be fifteen years ago. And this is plausible. However, there are many problems to consider here. Take for example the vagueness of the statement "Movies are longer now than they used to be." What is "now" in this sentence? The last year? The last decade? This century? And what kind of movies are we talking about? All kinds? Probably not. It is usually only successful mainstream films that people think of when they make these claims. (That Nuri Bilge Ceylan's films have gotten longer over the last decade is not a matter of dispute.) These things have to be considered when making graphs. It is also the case that facts and figures on IMDb have to be treated with caution. If you include all films you will not get a result that is relevant for most people's experiences, as it would for example include plenty of local films that few have seen and that might skew the result. If thousands of films that hardly anyone has seen have been getting shorter over the last decade, while popular films that have been global hits have been getting longer, the statistics would say that films on average are getting shorter even if most films people have actually seen are longer now than they were ten years ago. I am not saying this is the case, I am just saying that we would not know unless we examined the source material with care.
Let us get back to Toward Data Science. I should add that their grasp of film history is not that good, as
this quote illustrates: "Cinematography in the beginning of XX century
was still in its infancy. There were not many movies created back then
and most of them were just short presentations of new technology and
experiments." That is not an accurate description of the 1910s and 1920s, but either way they start with the year 1931 (source) and provide some graphs, divided by decades. This is based on 27,743 titles, because they wanted only fiction films
longer than 40 minutes and which have been rated by at least 1000 users. They emphasise that there "is a big jump between 1930’s and 1940’s, then a smaller one after 1950’s and since then the differences are marginal."
I disagree with their contention that the differences are marginal. The 1980s are considerably shorter than surrounding decades, and there has been an increase in the average film-length for each decade since then. But it is also the case that over half of the films included are from 2003 and after, so their sample pack from each year is not spread evenly. It might be that this has made the result inaccurate.
***
Now the result of my own analysis of some data. What I am interested in is the kind of films that people talk about when they complain that films have become longer, the mainstream successes, which are also the films I remember from my days as a projectionist. Therefore I will not use IMDb as my source but Box Office Mojo (which admittedly is owned by IMDb), and focus on the twenty most successful films in terms of box office figures for each year since 1995, and see how long they are. I use the domestic figures for the US, which is similar to each year's global figures, but by only looking at the data from the US I make it consistent and avoid the risk of a runaway local success somewhere which might affect the result.
I will look at five things. The average film-length of each year; the number of films longer than two hours for each year; longer than 135 minutes for each year; longer than 150 minutes for each year; and the number of films shorter than 100 minutes from each year. This should give an idea of whether the most successful films of each year are getting longer or not. The average length of each year is a figure that is not ideal as a very long film (such as Titanic (1997)) can disproportionally push up the average for that year. So with that caveat, here are the figures I collected:
I think we can say that films on the top 20 list have become longer, especially compared to before 2014. The average length is longer, and there are more films now above two of the three thresholds, of 120 minutes and 135 minutes, but about the same amount of films over 150 minutes. The only films now that are shorter than 100 minutes are animated films for children, and as there are more of these now than there used to be, they drag down the average length a little bit. But in previous years, any kind of film could be shorter than 100 minutes, not just children's films.
There are ways of fine-tuning this, for example by excluding animated films for children and only include the top 20 most successful live-action feature films of each year. Something that would increase the average film-length for each of the last five years, but less so for earlier years.
So I was right in thinking that it was more common for films to be shorter than two hours when I was a projectionist, and that now it is more common that they are longer than two hours. There are many other things to say about these top 20 lists, such as how the kinds of films that make it have changed over the years, and which studios appear and disappear, and the extent to which one year's hits have been more or less forgotten the next year, and the number of sequels, and many other things. I will return to these issues. Now that I have all this data, I should make some more use of it.
