A few years ago I attended an investment conference hosted by a large investment bank. One of the keynote speakers was Jack Dorsey who attended as the CEO of Twitter (him being also the CEO of Square) and in discussing the emerging importance of video to Twitter, he uttered something which I found quite astounding: Twitter should show in each user’s feed videos that he or she would already likely want to watch, and if a user had to search for videos, that would be considered failure on Twitter’s part.
Dorsey’s comment was a potent reminder that we live in an era in which every consumer internet company of consequence is in the business of “predictive recommendation”: it is inescapable, whether you’re on Amazon, YouTube, Netflix, Instagram, Twitter or even New York Times, that you are inundated with recommendations of what to buy, to watch, to see, or to read. The entire experience on the internet is increasingly being designed to spoon-feed content to users, as if the users could not, or should not, want to navigate and craft their own consumption experiences.
While it is often understood that at the root of this general trend is the desire to drive and deepen “engagement” on these platforms, what was never clear to me is why the engagement has to take the passive form. The fact that most of these platforms employ ad-supported business models and rely on users willing to tolerate or politely ignore advertisements appearing with regular frequency among content of genuine interest, which may require a certain degree of passivity in the mindset, might have something to do with it. One may point to a more sinister motive and assert that the platforms are simply trying to promote user behaviors that are as addictive as possible. Whatever the underlying motivation might be, however, the primary conclusion I draw from the outcome is that the entire experience is deeply unsatisfying.
On Instagram, I am periodically reminded of individuals whom I follow but whose posts never appear on my feed – it is not merely a matter of ranking or relative placement, that their posts are buried deep enough on the bottom that I never scroll that far down, but their posts simply and literally do not appear. On YouTube, the “Home” feed is populated by almost as many videos from channels that are “recommended” that I have no interest in watching as channels that I explicitly subscribed to and whose new videos are at the risk of being ignored despite my expressed interest unless I take an additional step of going into the “Subscriptions” feed, or even more absurdly, I manually enable the “alert” feature on each of the subscribed channels. On Netflix, I rarely find movies or TV shows I want to watch in the ocean of algorithmically generated recommendations.
I believe there is a unique satisfaction to be had when we exercise agency and autonomy, chart our own courses of action and shape our own experiences (this is what makes playing video games a distinctly fun experience, different from watching videos). I find it tragic that the leading consumer internet platforms are so intensely focused on figuring us out, to feed us what they think we’re likely to want, that they have completely failed to provide robust tools for us to craft our own consumption experiences. What if I could categorize my friends, followees, or subscribed channels into different buckets and had direct control over how each bucket is populated on the feed? What if each content was tagged with more robust metadata so I could more accurately search for what I wanted? What if I could construct a structured query on Netflix and search for movies that were released in a particular year, directly by a particular director, with a particular actor in the leading or supporting role? Why is none of this possible?
It has been my belief for many years that a great missed opportunity exists in consumer internet, one in which the control is wrestled away from the recommendation algorithms and the users are empowered with robust tools so that it is the users that dictate the courses of their own online experiences. I still believe it exists today, perhaps more strongly than I ever have. I had applied to YC with a version of this idea more than once, both times unsuccessfully. It probably is worth trying a third time.