The Man Who Counted the Cuts

In the 1970s, in the viewing rooms of Britain’s National Film Archive, a man sat at a Steenbeck, the flatbed editing table, winding old films through it. He paid no attention to the story, the stars or what the director meant. Every time the picture cut, he read the table’s footage counter and wrote down how long the shot had run. When the film ran out, he boiled the whole of it down to a single number.^2,4

His name was Barry Salt, and he was hard to classify.

Born in Melbourne in 1933, Barry had danced in ballet companies in Australia and England, then turned to theoretical physics, writing a doctorate on superfluid helium under David Bohm, one of the century’s great physicists.^1,3 “When it became clear to me that I was not another Einstein,” he wrote, “I returned to film-making.”¹ At film school, in a diploma thesis of 1968, he proposed a way of studying film style that film studies had not taken up: measure it.¹

The number was the average shot length, “the length of the film in seconds divided by the number of shots in it.”⁴ Nobody had collected it in any systematic way. Filmmakers didn’t need it, and the people who wrote about films weren’t interested in numbers. Salt was about to discover how strongly they felt about numbers.

01The market for being right

Every field, sooner or later, gets someone who does the job its insiders think beneath them: keeping score. Baseball’s was Bill James, who in 1977, working nights at a pork and beans cannery in Lawrence, Kansas, sold a booklet of statistics through a small ad in The Sporting News. About seventy-five people bought it.⁵ Most of the people who ran baseball teams ignored him for a quarter of a century. When they finally listened, it changed the sport.

Salt sent his first paper of measurements to Sight & Sound, the most prestigious film magazine in Britain. Its editor, Penelope Houston, turned it down. She had, Salt said, “an aversion to numbers and graphs, like many people.”² Screen, the Marxist journal at the center of academic film theory, would print nothing that strayed from what he called “the party line,” he concluded.² The American Film Quarterly took it in 1974, because its editor, Ernest Callenbach, “had a scientific education before he moved into writing about film.”²

Then he wrote a book. Film Style and Technology was a history of how films had actually been made: which lenses, lights and editing patterns were in use in which years, backed by his measurements. It opened with chapters attacking the film theory of the day. By Salt’s own account, the publishers’ readers in England and America said the rest was fine, provided those chapters came out. “A pile of reports from their readers an inch thick saying the same thing,” he wrote. “I was not going to be silenced by the academic Thought Police, so I set out to publish it myself.” He bought a Sirius 1 computer, a daisy-wheel printer and a copy of WordStar, and named his imprint Starword.² The book came out in 1983.

Reviewing it in 1986, the critic Jonathan Rosenbaum called it “conceivably the most detailed account of film technology that we have had to date,” noted that it “had to be brought out at the author’s own expense,” and suggested that “the refusal of the academic world to deal with his arguments has probably brought a certain stridency” to Salt’s writing.⁶ Salt sold all two thousand copies he had printed.¹

Stridency was one word for it. Salt was combative in print and stayed that way, even about compliments. When Liz-Anne Bawden, editor of The Oxford Companion to Film, congratulated him on an article for being “really scholarly,” he wrote, “I could feel the skin crawling up my backbone, but I forced out a polite ‘thank-you’ with great difficulty.” It was the word he couldn’t stand. He had “complete contempt” for the way people in the humanities wore “scholarly” as “a badge of value.” In the real sciences, he wrote, “the important thing is how good your work is.” As for himself, “I think of myself as a researcher or historian.”²

02The great speed-up

What Salt had been counting turned out to be one of the biggest changes in the history of the medium.

Over the following decades he worked out the average shot length of more than twelve thousand films, first in archives and later in front of a row of television sets and VHS recorders, where he “got down to collecting ASLs for everything that passed across their screens.”^2,7 He called what he found the Great Speed-Up. It ran “from 1951 to 2002. Over that period, the number of shots in the average feature film more than doubled.”⁸ The typical American film of the 1940s held each shot for about nine seconds; by the early 2010s, about three and a half.⁷ The fastest he found was Derailed (2002): 3,007 shots in eighty minutes, a new one every 1.53 seconds.⁸

One minute, by decade

1940s9.0 s

1950s9.0 s

1960s7.3 s

1970s6.4 s

1980s5.6 s

1990s4.9 s

2000s3.9 s

2010–163.4 s

And here is the strange part. The people making the films didn’t see it happening.

“As far as I am aware,” Salt wrote, “no-one commented on this increase in cutting rate at the time.” The average was falling by about a tenth of a second a year. “The film-makers had their noses so close to the trees they were hacking down that they could not see the whole Hollywood.” The first comments from filmmakers appeared only in the 1980s, three decades in.⁸

No editor ever decided that films should cut twice as fast. Each put every cut where it felt right, film after film, for fifty years. Sandra Adair, who cut Boyhood, once told a room of academics who had come to measure exactly this: “What guides me is my gut, my instinct, and my taste. I am rarely thinking about why I’m making the decisions I’m making.”⁹ That is the right answer for an editor. It is also why the pattern needed someone outside. From inside a cutting room the drift was too slow to feel. Seen across fifty years, it was impossible to miss.

03Then the field came to him

In Bill James’s story, the insiders eventually came around. Something like that happened to Salt, except the insiders who came were not in Hollywood. They were in the universities that had turned him away.

In 2002 David Bordwell, the most influential American scholar of film style, published “Intensified Continuity” in Film Quarterly, an essay that leaned on average shot lengths.¹⁰ In 2005 Yuri Tsivian, a film historian at the University of Chicago, started Cinemetrics, a website where anyone could measure a film by clicking at every cut while it played.^9,11 By 2022 it held more than twenty thousand titles, counting television programs and ads.¹² Salt’s lonely number had become a crowd sport.

In 2010 a Cornell psychologist, James Cutting, used shot lengths from 150 films to argue in Psychological Science that modern editing had drifted toward a 1/f rhythm, the pattern found in the ebb and flow of human attention.¹³ New Scientist headlined it “Solved: The mathematics of the Hollywood blockbuster,” a headline Cutting himself objected to.¹⁴ Salt complained that the paper gave “the impression that the editor of a film has complete freedom” over how long each shot runs.¹⁵

Then the field started arguing over the number itself, which is how you know it had arrived. The statistician Nick Redfern called the average “not a robust statistic of film style” and wanted the median instead, since a few very long shots can drag an average a long way.¹⁶ Salt replied that Redfern “surely can’t be serious.” Banning the average, he wrote, was “reminiscent of the Catholic church” keeping its ban on discussing the earth revolving around the sun. The median alone, he added, “can also be very misleading.”¹⁷ Forty years earlier he could hardly get the number published. Now people were fighting over how to calculate it.

04The floor

A tenth of a second a year is nothing. Over half a century it took away more than half of every shot.^7,8 Every cut was right for its moment, and a craft that runs on feel can still drift a very long way without anyone noticing. The only instrument that catches a drift is a record kept over time.

Then, by Salt’s count, the speed-up stopped. The average leveled off around 2002, he found, and films have bunched more tightly around it since.⁸ The floor, he argued, was dialogue. Even the most mindless action film has to stop and explain its plot, and a conversation cut too often stops making sense. Filmmakers looking for a style of their own turned to other things: movement inside the frame, stronger color, more words a minute.⁸

Whether it stays that way is another matter. Salt’s table has no American film after 2016, the year ByteDance launched Douyin in China. Its version for the rest of the world, TikTok, followed a year later.^7,19

Everyone feels their attention span shrinking, and in one sense it is. The psychologist Gloria Mark timed people at work. In 2004 they stayed on one screen for about two and a half minutes before switching to another; by the late 2010s, about 47 seconds.²⁰ That measures how often we switch, though, not how long we are able to look. Nobody has shown that we can no longer pay attention, and the famous claim that ours is now shorter than a goldfish’s has never been traced to any research.²¹ This July, The Odyssey, a nearly three-hour retelling of Homer, took $124.5 million in North America on its opening weekend, the biggest global opening of Christopher Nolan’s career.²²

The feeds themselves do cut fast. When engineers at Kuaishou, one of TikTok’s Chinese rivals, marked every cut in 853 short videos by hand, the average shot ran 2.59 seconds, against about three and a half in the last American films Salt counted.^23,7 Whether the films follow them is the next thing worth counting.

The book the publishers would not take now sells on Amazon for more than a hundred dollars.¹⁸ Salt gives the whole of it away as a PDF on his website.¹ It all began with a man at an editing table in an archive, who decided the thing worth knowing was the thing nobody was counting, and started counting.

Sources

1. Barry Salt, Film Style and Technology: History and Analysis, 3rd ed. (Starword, 2009), Introduction and Preface to the 2nd edition. Free PDF from the author: http://www.starword.com/fstyleTZ.pdf ("when it became clear to me that I was not another Einstein, I returned to film-making, doing a course at the London School of Film Technique for which, in a Diploma thesis written in 1968, I proposed the methods of stylistic analysis"; "I sold all the 2000 copies of it that I had printed"). Birth date and place: Wikipedia, https://en.wikipedia.org/wiki/Barry_Salt.
2. Barry Salt, Moving Into Pictures (Starword, 2006). Free PDF: http://www.starword.com/MovPicFin.pdf. Salt's account of Sight & Sound, Screen, Film Quarterly, the publishers' readers, the Steenbecks at the archive (data "from film prints viewed on Steenbecks or on the screen at the National Film Theatre"), the footage counters read "to the nearest half foot of film" (ch. "The Numbers Speak," p. 393), the VHS recorders, the Sirius 1 / WordStar / "Starword" name, and Liz-Anne Bawden's "really scholarly" and the skin crawling (p. 36).
3. Same, on the doctorate: "my work was judged good enough to warrant a doctorate in 1965."
4. Barry Salt, "Data and Method," starword.com ("I have always worked with prints of the films I was analysing ... on Steenbecks and other flat-bed editing machines," later DVDs and VHS through Adobe Premiere): http://www.starword.com/Data_Method/data_method.html
5. Michael Lewis, Moneyball (2003), pp. 65–66 (the 1977 booklet, the ad in The Sporting News, about 75 buyers), as cited in "Bill James," Wikipedia: https://en.wikipedia.org/wiki/Bill_James ; the Stokely-Van Camp pork-and-beans plant in Lawrence: The Lawrence Times, https://lawrencekstimes.com/?p=60800
6. Jonathan Rosenbaum, review of Film Style and Technology, Wide Angle 8:3-4 (1986): https://jonathanrosenbaum.net/?p=29936
7. Barry Salt's table of average shot lengths, SaltASL5.XLS (last modified March 2022): http://www.starword.com/Data_Method/Data_Tables/SaltASL5.XLS. About 12,260 distinct films by our count. Decade medians of US films, our calculation from that table: 1940s 9.0 s, 1950s 9.0, 1960s 7.3, 1970s 6.4, 1980s 5.6, 1990s 4.9, 2000s 3.9, 2010-16 3.4. The table has no American film after 2016.
8. Barry Salt, "The End of the Great Speed-Up — and After," Digital Scholarship in the Humanities 37:1 (2022, online May 2021), 218–228, doi:10.1093/llc/fqab049; author's copy: http://www.starword.com/Articles/Speedup13.pdf
9. University of Chicago News, "Cinemetrics quantifies creativity in film" (May 26, 2015): https://news.uchicago.edu/story/cinemetrics-quantifies-creativity-film
10. David Bordwell, "Intensified Continuity: Visual Style in Contemporary American Film," Film Quarterly 55:3 (2002), 16-28; the quoted sentence as given in Nick Redfern, Wide Screen 9.1 (2022): https://widescreenjournal.org/wp-content/uploads/2022/08/formatted-cutting-rates.pdf
11. "Cinemetrics," The Bioscope (July 22, 2012): https://thebioscope.net/2012/07/22/ ; Yuri Tsivian, "How to use": https://cinemetrics.uchicago.edu/how-to-use
12. Nick Redfern, Exchanges (2022), "over twenty thousand motion pictures": https://exchanges.warwick.ac.uk/index.php/exchanges/article/download/853/953/5870
13. James E. Cutting, Jordan E. DeLong and Christine E. Nothelfer, "Attention and the Evolution of Hollywood Film," Psychological Science 21(3):432-439 (2010). PubMed 20424081.
14. Ewen Callaway, "Solved: The mathematics of the Hollywood blockbuster," New Scientist (February 18, 2010); Cutting's comment on the headline, on Cinemetrics: https://cinemetrics.uchicago.edu/article/590235ff-85aa-4714-9553-71d4c44fd61c
15. Barry Salt, "Comments on Attention and Hollywood Films," Cinemetrics: https://cinemetrics.uchicago.edu/article/7026830d-73a7-4b27-b8fe-41ebe823589b
16. Nick Redfern, "The average shot length as a statistic of film style," Cinemetrics: https://cinemetrics.uchicago.edu/article/2732e556-9e15-4d0b-b864-e08625fb6acf
17. Barry Salt, "Graphs and Numbers" (2012), Cinemetrics ("he surely can't be serious"; "reminiscent of the Catholic church continuing its ban on the discussion of the idea of the earth going round the sun, even after the concept was in wide use"): https://cinemetrics.uchicago.edu/article/4bae8cac-cceb-452a-b6e6-1aa2df4fa76a
18. Amazon.com listing for Barry Salt, Film Style and Technology: History and Analysis, 3rd ed. (ISBN 9780950906652), paperback from $103.34, checked 2026-10-30: https://www.amazon.com/dp/0950906654. Prices move (it was $260.48 on 2026-10-22).
19. Douyin's launch by ByteDance in September 2016 (as A.me, renamed Douyin that December) and TikTok's international launch in 2017: "TikTok," Wikipedia, https://en.wikipedia.org/wiki/TikTok ; History of Information, https://www.historyofinformation.com/detail.php?id=5298 ByteDance launched Douyin on 20 September 2016 and TikTok outside China in September 2017.
20. Gloria Mark, Attention Span (Hanover Square Press, 2023); Mark on the 2004 and recent figures, Steelcase, "Our 47-Second Attention Span With Gloria Mark": https://www.steelcase.com/research/articles/our-47-second-attention-span-with-gloria-mark-s5-ep3-transcript/. The 2004 CHI paper says "somewhat more than two minutes"; "two and a half" is Mark's later wording. The studies followed different people. UC Newsroom on the 2004 study (two and a half minutes) and the 2016-2020 studies (47 seconds): https://www.universityofcalifornia.edu/news/how-sharpen-your-attention-and-meet-your-goals-2024
21. The 8-second "goldfish" figure, from a 2015 Microsoft Canada report that credited "Statistic Brain": Simon Maybin, "Busting the attention span myth," BBC News (More or Less, March 2017), which found no research behind it: https://www.bbc.com/news/health-38896790 ; WARC, "The goldfish myth": https://www.warc.com/newsandopinion/opinion/the_goldfish_myth/2806
22. The Odyssey (Universal, 2026), 172 minutes: AP, “A look inside Christopher Nolan’s ‘The Odyssey’ by the numbers,” https://wtop.com/entertainment/2026/07/a-look-inside-christopher-nolans-the-odyssey-by-the-numbers/ ; $124.5 million North American opening, $264.1 million worldwide, Nolan’s biggest global opening: The Hollywood Reporter, https://www.hollywoodreporter.com/movies/movie-news/the-odyssey-box-office-opening-christopher-nolan-1236651507/ ; TheWrap, https://www.thewrap.com/industry-news/business/the-odyssey-264-million-global-box-office/
23. Wentao Zhu et al., "AutoShot: A Short Video Dataset and State-of-the-Art Shot Boundary Detection," arXiv 2304.06116 (April 2023; Kuaishou Technology, UT Austin, Cornell), Table 1: 853 short videos, average length 39.5 s, average shot length 2.59 s (ClipShots, a set of longer videos, 15.34 s). The videos come from "one of the most widely used short video platforms" and include ads and game videos. https://arxiv.org/abs/2304.06116