Startups fail. The flipside of the explosive growth of a new Facebook or Google are the many other companies that do not pan out. Since big risks are the complement of big rewards, Silicon Valley is famous for its cultural acceptance of risk, with admonishments like “fail fast.” A common saying is that 9 out of 10 startups fail.
It’s a statistic that seems to have no basis in real data - a growing Quora thread has yet to find one - but it represents an ethos. The uncertainty derives from defining failure. Does failure mean closing down completely, failing to meet predictions of a billion dollar business, or somewhere in between? In a field aiming for the next Google, is being acquired for $20 million success or failure?
The Wall Street Journal, for example, writes of the following “rule of thumb” among venture capitalists who invest in startups:
Of 10 start-ups, only three or four fail completely. Another three or four return the original investment, and one or two produce substantial returns. The National Venture Capital Association estimates that 25% to 30% of venture-backed businesses fail.
Less generous research from Shikhar Ghosh of the Harvard Business School finds that 75% of venture backed businesses don’t return investors' capital and that 95% miss the high growth rates they target.
Critics of Hollywood complain that it churns out the same movies over and over again. Yet these numbers from Silicon Valley are applicable in Hollywood.
Out of every ten movie projects launched by a studio - by buying a script, acquiring the movie rights to a novel or comic book, or starting work on an internal idea - only one will actually be made and released. The other nine will be victims of “development hell,” the years spent working up from a script or concept through script changes, casting, and so on. The potential for the film to be pulled or frozen in development indefinitely is a constant threat.
Without even looking at performance in terms of profit or return on investment, the odds of a Hollywood movie making it into theaters are the same as Silicon Valley’s 9 out of 10 figure and much longer than startups’ actual failure rate.
Hollywood and venture capital have some similar dynamics. Both are in the business of finding a few extremely profitable hits. Paul Graham writes from his experience providing seed funding to early stage startups at Y Combinator “that effectively all the returns are concentrated in a few big winners.” The variation in outcomes between startups is “1000x” and just two companies are responsible for ¾ of the value of all Y Combinator companies. Hollywood producers are less transparent, but a 2002 analysis found that “of the 450-odd Hollywood films released in any year, the top ten bag about 30% of all box-office takings.” The concentration of profits has intensified since then largely thanks to the globalization of film profits.
The two industries also look to data but cannot necessarily trust it. Hollywood producers can look at the past success of similar films when deciding whether to fund a new project, but as Adam Davidson of Planet Money has written, it can be hard to disentangle whether, say, The Hunger Games was successful because of actress Jennifer Lawrence, the “built-in audience” from the book, or its spring break release date. Venture capitalists can calculate market size or look at the success of similar ventures, how do you determine that Google or Facebook will outperform its many competitors?
Studios can show advance screenings to focus groups. But if Regency Enterprises followed their negative audience research, Fight Club would never have been made. Similarly, Paul Graham describes metrics like the amount of funding raised by startups as actually misleading for predicting success.
Despite their similarities, however, Silicon Valley and Hollywood have similar failure rates for opposite reasons. Silicon Valley doesn’t know what a future success looks like - every investor has a story about passing on a company like Facebook, Google, or Pinterest because it seemed unlikely to succeed or even stupid - so it accepts risk and celebrates it.
Hollywood knows what success usually looks like (it’s a sequel and tends to wear a cape). But pure repetition is not a sure thing (see the recent flop The Lone Ranger - a reboot of a classic story that was also Pirates of the Caribbean retreated as a Western) and it will miss out on surprise hits like Forest Gump.
Accordingly, Hollywood producers and decision makers play a risk averse dance away from boldness and creativity. The result is development hell.
The stories of development hell are of the many film ideas that never made it through to production. The Beatles were once tapped to star in Lord of the Rings as Frodo, Gollum, Sam and Gandalf. Darren Aronfsky, the director known for gritty, dark films like Black Swan and Requiem for a Dream, once intended to direct a Batman reboot involving:
A teenage Batman living in a basement garage, who doesn’t know he’s part of the Wayne dynasty, driving a converted black Lincoln Continental with a school bus engine and Boss tires, having an affair with one of the girls in the cat-house across the street.
Part of the reason films get stuck in development hell is that the number of moving pieces means that the stars need to align to get a film made. Studios could never make Total Recall 2 because star Arnold Schwarzenegger turned down the role every time he was shown a new script. A studio ceased production on a sequel to Forrest Gump after the September 11 attacks because it was incongruous with scenes involving Forrest speaking at the scene of the Oklahoma city bombings.
But the main reason that so many movies get stuck, as related by David Hughes, author of Tales from Development Hell, is the need for decision makers in Hollywood to look good:
“Films carry with them a certain amount of fear because if you say ‘Yes’ to something and you’re wrong, you’re out on your ear, whereas if you say ‘No’ to something, you’re never going to get into trouble, if everything is always defensible. So you wind up in development with people trying to make things more like things they know, because that is a defensible position: you will probably not get fired for it. Unfortunately that’s why you wind up with films that look like other films.”
Hughes gives the example of the attempt to adapt a comic called The Sandman to the big screen. The comic had been a huge hit both in terms of sales and accolades from both the comic book and literary world. While adapting a comic book was a natural move for a studio, the comic was different from many Superman type comics. Rather than a supersized version of dueling criminals and cops, the protagonist of The Sandman is the lord of the dream world and the villain’s scheme takes advantage of the connection between the dream world and reality.
While a script bounced around and saw changes in reaction to producers' differing opinions and egos, Hughes makes clear the real cause of its failure. A producer saw The Sandman as his next Batman - his next big “meal ticket” - and wanted to turn The Sandman into a cape and tights wearing superhero beating up his antagonist - something completely at odds with the concept. The film was never made.
There are similar dynamics at play in Silicon Valley and Hollywood. Both pour money into a large number of projects in the search for a few huge successes that fund nearly the entire industry. The startup world has accepted that success is hard to predict and learned to tolerate and even celebrate failure. In contrast, movie producers and investors express interest in bold endeavors but favor projects that look like past successes, eschewing any risk that could make them look bad or get them fired.
This author once had a professor who said that great artists must constantly court failure. To do any less meant playing it too safe to produce anything of artistic value. But it's the investors in Palo Alto that follow that creed - not Hollywood's.
April 5, 2014 · 4,042 views
Here's a radical idea. Instead of companies doing pranks for April Fools Day, maybe they should do nice things instead?