(Welcome to the Entertainment Strategy Guy, a newsletter on the entertainment industry and business strategy. I write a weekly Streaming Ratings Report and a bi-weekly strategy column, along with occasional mini-dives into other topics, like today’s article. Please subscribe.)
Does it count as an “exclusive” if you’re just the first person to write about something?
Because my team (basically, my editor/researcher) uncovered a groundbreaking diversity report/academic paper that literally no other news outlet in the country has written anything about it. (I mean literally no one. Go check Google News.)
Here’s how Berkeley News describes this fantastic research:
“[David] Bamman [an associate professor at Berkeley] and his team used facial recognition technology to track the amount of time actors appear on screen in more than 2,300 Hollywood films released between 1980 and 2022 — a total of 4,412 hours of footage. They analyzed both “popular” films, defined as the top 50 box office earners each year, and “prestige” films, which are films nominated for “Best Picture” by at least one of six different organizations, including the Academy Awards and the Golden Globes.”
Now that’s a use of AI that I can get behind! It led to this amazing visual:
On the one level, this is great news! Especially for Black and Asian representation. The Ankler just re-ran my series on the Average American and if you go there, you can see that Hollywood’s racial representation now matches much of the country. (With the exception of Hispanic actors, which is what we warned about in that article…)
The bad news? Women are still underrepresented at 40% of screen time.
I’ve never discussed diversity reports in this newsletter before, usually because I have methodological and transparency issues with the most famous reports. (And airing those disagreements is probably just going to lead to fights that I don’t want to have.) But here’s what I love about this one:
- Sample size. This is an excellent sample size, but not in the way you think! In this case, smaller is better. Other diversity reports (or film analyses) look at 200 or more films each year. (Sometimes hundreds more.) 200 relevant films don’t come out each year. When half or more of your sample sizes includes films that don’t come from the major studios or never got a wide release, you can’t generalize about “Hollywood” or “the studios”—that gigantic sample set says more about indie/foreign film production—but many people still do. Now, this sample set could probably go up to 100 films or include every major studio release, but I’m quibbling. This sample size is great.
- This survey analyzed screen time. A lot of other reports focus on categorizing roles as lead actor, supporting or other. How you define that (and whether you track it over years) could allow the researchers to fiddle with the data. (For example, one prominent report excluded a certain type of film, which meant excluding two films starring an underrepresented group, then the headline said Hollywood didn’t make any films featuring that underrepresented group. But you excluded that group’s two films from your dataset! Come on!) By focusing on screentime, this analysis solves that problem and provides the most accurate look possible.
- All of this data is publicly available. So anyone can check their work!
But here’s the craziest part. Since this paper came out on 4-Nov, not a single news outlet has covered it. No, literally, not one. Google search “David Bamman” or “AI diversity”. You’ll get no results since this paper was published.
And I don’t really get it. Is it because the report offers too much good news? Bamman is at pains to say that this survey isn’t all good news (Black representation for leading roles isn’t high enough, women are still underrepresented on screen, and so on.) Or is it because AI was used?
I’m not sure, but everyone should check out this analysis.
(I’m trying to work in a few smaller visuals like this one in addition to my regular articles—the Streaming Ratings Report, Most Important Story, and other deep dives—so let me know what you think!)
