The imperative of the video business is simple: grow the audience. The health of the individual content provider and the greater video economy depend on an active, satisfied and increasingly large viewing audience. Technology is really helping here.
Under the old studio/network model, scheduled programming sprang from the forehead of some executive god. But Big Data and the algorithm now determine what you watch next (past practice will be an indicator of future performance.) Science! Progress!
But this tech-driven future — relying on data to grow the audience — has a dystopian aspect to it that poses a threat to the video ecosystem if we are not mindful.
I argued in a previous post that the algorithm is a necessary tool to support an effective consumer video search and discovery S&D experience, with the caveat that algorithms should always be used in conjunction with human editorial/promotional efforts. Until the point of singularity, human intelligence trumps machine intelligence (with possible exceptions like chess, and medical diagnosis as will be noted below.) But I would now like to add an additional caveat.
Differentiation via mutation is behind the glorious diversity we see in life, which creates endless wonders waiting to be discovered. As far as we know, evolution does not work according to a fixed mathematical formula, with entirely predictable results. The natural genetic mutation occurs not because of something you did. Action or behavior does not a) cause the natural mutation, or b) determine what that mutation is. It seems to be random. Evolution has been successful only in that it incorporates the random mutation that makes accommodation for exogenous environmental factors. That's how the herd stays healthy.
But this is not the way we currently think about video discovery. Here, the algorithm that determines what happens next is prefigured by what has happened already:
1. What you did last.
2. What is already trending.
3. What your social “friends” recommend.
What is in danger of being lost is the element of random differentiation, the surprise, the accommodation, the adoption of the new that ensures a vibrant and diverse viewing experience and a healthy and thriving viewing audience.
This present reality in video S&D is hardened by:
This rigid algorithm-based reality, results in a kind of Frankenstein imitation of real life: horribly predictable. Made of old parts. So the extended caveat is that the algorithm also needs to preserve the random in our video lives. The algorithm should throw a change-up pitch every now and then.
This can be done by drawing a title from the catalogue completely at random. Behind-the-scenes editors can create a pool of quality, but perhaps overlooked, titles that can be randomly fed into the results pages. Or just as Google developed paid search results — not always the best result but hey, they are paying us for it — qualified content providers looking to find and build an audience can pay/rev share for a limited-in-number search result placements.
Increasingly, the algorithm recommendation engines return results that I have already seen. “Those who watched that, also watched this.” Really? What a surprise.
Not every random video will be a winner, but there will be unexpected winners — like the small Indie film that blows a hole in the conventional wisdom every now and then and opens up a whole new sub-genre/species to keep the movie industry from stultifying into a giant special effects trade show.So steal a page from Darwin. Keep it random.