The Unbundling of YouTube
The 2010s were the unbundling of Craigslist. Will the 2020s be the unbundling of YouTube?
Balaji Srinivasan, tragic hero of last week’s episode of Three Quarks, had the kind of thought experiment last week that we should all be having: would this be better if it were decentralized? His subject was Twitter:
The thread is likely a reference to Jack Dorsey’s mea culpa that he’s more excited about Bitcoin than Twitter. But, of course, it’s also a throwback to Andrew Parker’s preview of the past decade from 2010, the infamous “Unbundling of Craigslist”:
Either way you look, the point is the same. Back in 2019, Jeff Jordan and D’Arcy Coolican took stock of the 2010s for a16z by describing the key insight of these charts:
“In all but a few circumstances, the broad horizontal verticals eventually break. They become a victim of their own success. As the platforms grow, their submarkets grow too; their product gets pulled in a million different directions. Users get annoyed with an experience and business that caters to the lowest common denominator. And suddenly, what was previously too small a market to care about is a very interesting place for a standalone newco. Like clockwork, a new wave of innovation begins to swell, picking off the compelling verticals the new horizontal players cannot satisfy.”
So what are the unsatisfying horizontals of 2019? They don’t pick Twitter—which enables verticals to free-associate rather than compartmentalize (arguably its moat)—but they do pick YouTube, LinkedIn, and, confusingly to me, Zillow.
I want to focus on YouTube in particular, partly because it features in that other classic Bundle-Text, Ben Thompson’s 2017 The Great Unbundling, as a disrupter and a winner. In Thompson’s telling, YouTube isn’t a doomed horizontal that can be unbundled into verticals, but a successful aggregator that has collected modularized video content and provided it to the masses. If we extrapolate that a bit through the lens of Thompson’s Aggregation Theory, we can see that YouTube’s strength against its incumbent (television) is that:
1) YouTube doesn’t have to create and distribute the content, but only has to distribute content that users supply (allowing it immense scale and low cost)
2) YouTube can distribute content under consumers’ terms (they can watch whenever they like, wherever they like, at whatever point in a video they like)
3) YouTube represents an Unbundling of Television. For Thompson, TV has broken up into five verticals, each represented by a different media platform—YouTube is just one.
So how do we reconcile these two visions of unbundling? Does YouTube exemplify successful unbundling or face its risks in the years ahead?
It’s tempting just to say that Thompson was right about the past (yes, YouTube disrupted TV), but Jordan and Coolican are right about the future (yes, YouTube has created the conditions for its patricidal heirs). If we wanted to make the case against Thompson, we would duly point out that the major media aggregators are trying not to remain aggregators, distributing free or licensed content at-scale for low-cost, but struggling to build empires on exclusive intellectual property as their moat. Indeed, Thompson’s description of Netflix from only four years ago that “Netflix bears no marginal costs in terms of COGS” and “does not create shows” has become unrecognizable as Netflix invests in original media. Meanwhile, Spotify has signed Joe Rogan and the Obamas, and even Substack pays to acquire writers.
Really, it’s tempting to say that when a company with a reputation for aggregation tackles an industry with a reputation for IP, it is the reputation of the industry that will remain intact.
Nevertheless, I’d rather argue that the a16z team is right for the wrong reasons—and that the real reason YouTube stands to be unbundled lies both in the heart of Thompson’s Aggregation Theory, as well as a development predating all of the cases above: the rise of the creator economy and Web 3.0.
Horizontals -→ Verticals
Let’s start by dissecting the easy case: Jordan and Coolican’s argument for Andreessen Horowitz that YouTube represents a crowded horizontal that would be better broken into verticals for its user segments. They write:
“A large number of newcos are trying to get a piece of that action, by targeting specific categories and audiences within the larger general platform. Twitch has seen a lot of success in gaming videos (Amazon acquired the company for just under $1 billion in 2014); TikTok has exploded in short-form mobile videos; MasterClass by targeting educational videos; Overtime by zeroing in on sports, with highlights and original content aimed at Gen Z; and many startups are looking now to claim the space of video content for kids.”
As Jordan and Coolican suggest, Craigslist got unbundled because it offered way too big a pool of offerings to serve its central function: matching a very small set of offerings with a very small set of fervent consumers, neither of whom could find each other without wading through a morass of irrelevant content. And as they also suggest, we could certainly make the same case against YouTube. Indeed, it’s hard to think of a user-generated vertical that wouldn’t be better served having its own platform to get content to viewers. Let’s take some still-untapped micro-verticals as examples:
High School Sports: Students and college coaches would benefit from having a dedicated online platform to share and find videos for all potential recruits, without students having to email coaches individually with a link.
College Sports: Teams would benefit from finding user content for games in one place to edit together promotional reels. (Enter Weplayed.)
Real estate: Potential buyers and renters would be better off posting their videos of apartments in conjunction with the online listing for others to see, rate, and comment (brokers might not love it, and competitive buyers wouldn’t want to share, but Zillow would likely benefit from UGC).
Weddings, Birthdays, Bar Mitzvahs: Parties often use dropbox to share videos that they can edit together—it’s unwieldy, and the key players would benefit from having an online video page to preserve their memories for years.
Journalism and Social Justice: Journalists would benefit from having a decentralized platform for witnesses to post scenes of violence and police brutality they’ve witnessed in order to get faster and more comprehensive attention.
And so on.
The point here is the same as it was for Craigslist: it’s massively inefficient if not impossible for small segments of passionate viewers to find user-generated content that’s relevant only to them. There are ways to hack the matching process—posters could use hashtags or keywords that users would know to search—but it’s telling that even hashtags have gone out of style as they lead to too many posts to be useful.
Still, there is one major difference between Craigslist and YouTube: Craigslist failed because its search capabilities were nearly non-existent while YouTube has been a way to search and find content from its beginning.
Why does this matter? Go back to Jordan and Coolican’s argument that horizontals become unbundled when “their submarkets grow too; their product gets pulled in a million different directions.” In the case of Craigslist, this was undoubtedly right: the more people that posted, the harder it became to find relevant information. But with YouTube, the opposite is true: the more people that post, the more robust YouTube becomes as a universal search engine and algorithmic recommender of content that can benefit user experience by showcasing the content that’s most relevant to them. Nobody ever scrolled through pages and pages of YouTube content to find what they were looking for because its goal from the start has been to direct users through search and recommendations. For YouTube, the more content, the better.
That also means that over time, YouTube could become far more effective at cross-posting than Craigslist was: if its function weren’t to host original posts, but to search and surface posts, it could integrate best-of reels from Twitch and Tik-Tok and Overtime (as it now, awkwardly) while plugging its own video-hosting services into smaller sites. The mini-verticals mentioned above, like high school sports and weddings, could integrate with YouTube’s technology to include their videos on their site, while larger media conglomerates, like NBC and Paramount, could offer subscriptions through YouTube (as Hulu and some sports leagues currently do). Monopolistic enterprises like Netflix, Amazon, and Disney+ obviously don’t want to integrate with YouTube, but that’s also why the classic Hollywood studios and TV stations might not have a choice: it’s unlikely that consumers will get subscriptions to all services, and these incumbents need a way to sell content. YouTube remains the best bet for videos content creators to host and sell their content.
So in other words: yes, YouTube will be unbundled as a video player, but it could still offer the technological infrastructure for its successors, continuing to aggregate and support its own unbundling.
And yet, it’s striking that with the compromised exception of YouTube TV, YouTube is doing little of the above. So why is this great aggregator failing to aggregate?
One answer is that it possibly can’t afford it. At a moment where individual and corporate creators of all stripes are insisting on direct distribution and monetization of their work—let’s include everyone from Warner and Disney to Cardi B (OnlyFans) and Gilbert Gottfried (Cameo)—YouTube is sticking to an opaque playbook of paying small commissions from ad revenue and allowing pay-per-view services. In other words, it is facing the classic Innovators Dilemma. Increasingly, media aggregators, whose sole function is distributing pre-existing material, are expected to give ways for consumers to directly monetize the supply . Yet it would cost YouTube a tremendous sum to revert to a traditional media model that represents its suppliers with high-paid contracts, as Netflix and Substack have done, since it’s not in a position to charge for these. And so it prefers to remain an aggregator with incredible margins.
It’s not quite clear yet what will be the OnlyFans to YouTube’s Pornhub—it may in fact be OnlyFans itself!—but YouTube’s unbundling won’t just happen because it’s a horizontal better broken into verticals. Rather, YouTube likely can’t afford to enable the kind of interactive engagement and monetization between creators and consumers that has led to the dominance of Twitch, Tik-Tok, OnlyFans, and the like.
Aggregation Theory in a Time of Disaggregation
Combine the points above, and we start to conclude that Web 3.0 will bring a return to the p2p marketplaces of Web 1.0 in the 90s—allowing users to contract content directly from suppliers while bypassing an intermediary that keeps most of the money. Much of the early success of Ebay and Napster, for example, was that they were quite decentralized, allowing consumers to get content directly from those who had it. Today, the same could be said for Substack, Mirror.xyz, or, for that matter, Taylor Swift, whose battle against record companies seems like a winning battle in the war Napster started 22 years ago. The difference for Web 3.0 isn’t only that it will allow more ways for consumers to sponsor the generation of content directly, however, but that it will also let them invest in and even earn rewards on the content.
On the surface, this puts aggregators in a difficult spot. Since aggregators don’t own the rights to their supply, they’re primarily powerful as marketplaces to connect content to consumers—which leaves them vulnerable to decentralization. The rise of Coinbase, Uber, and Airbnb over the 2010s attests to the power in the Web 2.0 era of centralized disintermediaries, but this is a contradiction in terms: while these marketplaces have powerful network effects and branding, they take huge cuts from parties they refuse to represent. In the world of blockchain, what’s to prevent them from being disintermediated?
The case seems especially dire for YouTube if we compare it to similar platforms like Shopify, Doordash, and Stripe, which have also enabled the rise of individual micro-services to manage business through their platform. Like YouTube, all of these serve not only as distributors of sorts for suppliers to represent and transact their business online, but as incubators as well, actively leveraging data, loans, and representation to help the businesses grow. Unlike shops, restaurants, and online businesses, however, influencers and media companies don’t need YouTube to handle their product fairly exclusively; it’s only one of many distribution channels for them, and unlike Shopify, Doordash, and Stripe, it doesn’t allow them many options for setting their terms of sale.
You could say that, well, influencers and media companies only represent a small subset of users on YouTube: yes, they might go elsewhere, but 99% of content is generated by users looking to post video references (here’s the problem with my toilet guys). For that, we can go back to the argument above that these “reference” videos are best served by micro-verticals. Note the key value of Shopify, Doordash, and Stripe for businesses: they offer wide access (for broad audiences around the world) or they offer immediate access (for local audiences with urgent requests). YouTube offers wide access that micro-verticals don’t need while their own platforms would offer immediate access that YouTube can’t (here’s a grouping of user-generated videos from the big event to edit for social media).
And here we get to the larger point: increasingly, both influencers and everyday users are aggregators themselves, drawing on a pre-existing supply of material that they repackage and release with their own brand. The examples that a16z gives of YouTube disruptors—Tik-Tok, Overtime, Twitch—all help enable this kind of user-remixed-content, just as Spotify’s user-generated-playlists are much better designed than YouTube’s (which offers no easy way to go back to the lists after moving onto new videos). We can see this in writing and NFTs, too: as Packy McCormick put it earlier this week, his immensely popular essays are “remixes powered by other peoples’ ideas,” with whom he’s looking to share financial proceeds. And we can even see it in documentary filmmaker, where found footage aggregation has become the norm for artists from Garrett Bradley to Adam Curtis to Ja’Tovia Gary.
So here’s a prediction for the next decade of p2p: individual aggregators will disrupt corporate aggregators. Or to be more precise: platforms that let users aggregate rather than create content will be the ones to disrupt the traditional aggregators.
The 2010s was the decade that unbundled Craigslist because people found that digital platforms could be far more convenient and trustworthy for facilitating real-world, person-to-person interactions. Finding a place to live, finding a romantic partner, finding a person to fix that toilet, and finding a ride—these were Craigslist’s domain, and platforms that improved the search improved individuals’ ability to find each of these.
In the post-pandemic 2020s, though, we primarily interact outside the real-world through digital text, audio, and with the rise of VR and AR, video. YouTube stands to be unbundled over the next decade not only because vertical platforms will improve video search, but because there are far more dynamic ways to interact and transact with video online than just posting content. We’re all aggregators, remixers, and our own video avatars—now we just need a way to monetize.