Personalized Content Feeds Suck. Curation is a Better, Scalable Alternative
Personalized content recommendation algorithms fail to expand user interests despite the deluge of information on the internet
The internet creates more information per day than anyone could ever consume in a lifetime. For the average individual, the supply of information is not a problem. People with access to the internet can get instantaneous access to information for free in most cases. Twitter informs us about what's going on in the world before many news sites can write about it. Facebook lets us know what's going on with our friends before they pick up the phone to tell us about it. With just a few keystrokes, Google points us in the direction of whatever information we want whenever we want. All these at no monetary cost to their users. If the information is out there, a regular person can find it just as easily as someone with more resources. News sites may seem like an exception, but they aren't. People don't subscribe to Wall Street Journal or New York Times because they can't find the information elsewhere. They pay for curated content and to support a journalist or the publication. The same thing goes for scientific research journals. You can choose to pay to download research papers, but they are often available for free on other sites. Blog posts, podcasts, educational videos, and even entertainment media is widely abundant. The internet has eliminated the scarcity of information and abstracted the access layer through various mediums (distribution). That abstraction and abundance created a new non-scalable element.
Hence a wealth of information creates a poverty of attention and a need to allocate that attention efficiently among the overabundance of information sources that might consume it. — Herbert A. Simon
Information, and access to it, became abundant. As a result, attention became the new scarce, non-scalable element. Ok, maybe not "new." The attention economy existed before the internet. But the proliferation of the internet made attention even more scarce. Internet content giants like Google and Facebook want users to spend more time consuming information on their platforms and are prepared to monetize that attention. In 2019, Netflix's shareholder letter for the year prior stated that Fortnite was a bigger competition compared to HBO or Hulu. It's simple. Time spent playing Fortnite is time that could've been spent watching shows on Netflix. Subscribers to both Fortnite and Netflix have to split their time between these platforms. When attention became (even more) scarce, companies got creative and figured out ways to make it abundant.
Creating More Attention
As a content company, there are a couple of obvious ways to get more attention without spending money to acquire new customers:
- Make more subscribers
- Give existing people more free time
- Make people spend more time on YOUR platform
For my sake and yours (taking a wild guess here), let's skip the first point. We'll assume content companies aren't creating more people for the sake of consumption. That leaves us with the second and third options. Naturally, technology makes things more efficient, thereby creating time. I'll skip the long history and focus on recent events.
- Uber creates time for would-be drivers
- Amazon creates time for would-be physical shoppers
- Doordash saves time from cooking or going out
And that's just the demand side. On the supply side, most of these services let people leverage their time or expertise. Content companies have also created more time by creatively modifying consumption. TikTok (and Vine) tackled short-form videos because people weren't willing to spend more than a few seconds of their attention watching videos. YouTube is also experimenting with Shorts. It's not just videos; music is also getting shorter thanks to streaming. With an unlimited library and effortless track-seeking, listeners don't mind skipping songs that don't grab their attention in the first few seconds. The same thing happens on Netflix. Movies and shows may be hours long, but they play clips and trailers every chance they get. Those clips serve the same purpose as the first few seconds of a video on TikTok or music on Spotify: to convince you to pay attention. If it works, you'll end up spending minutes or even hours watching the full video.
The content we consume has adapted to our shorter attention span. Consumers aren't wasting time on content they don't find interesting. Naturally, this should free up more attention for consumers to spend as they wish. But as Netflix noted, the time spent consuming other content could be spent consuming Netflix's content. Now that consumers aren't wasting attention on things they don't like, anything they spend their attention on becomes a data point. While writing this paragraph, I opened Netflix to confirm that they still auto-played clips and trailers. I was greeted with a massive banner for Yasuke, which is something I would watch. After adding it to my list (another data point), I checked the other profiles on the account to see what Netflix recommended. Sure enough, it wasn't Yasuke. The chance of me clicking on any of the shows on the other profiles was slim. Each profile is personalized, giving each person their own Netflix library. Companies like Netflix prioritize feeds (personalized content recommendations) because it helps them capture more consumer attention. The better their recommendation algorithms, the more time consumers spend on their platform. Content companies create more attention by making it extremely easy for consumers to find what they like. However, they end up capturing even more attention by only surfacing personalized content to consumers.
When companies prioritize feeds, they push out some users. Feeds are good at showing more of what you like. They work by including content similar to those you've enjoyed in the past and excluding things close to what you've disliked. However, when it comes to uncategorized or category-ambiguous content, feeds are terrible. Whenever content doesn't fit in the like or dislike buckets, feeds tend to exclude them. Consumer's needs aren't met when feeds fail to surface content that doesn't fit. YouTube is one of the few feeds that do a good job of this. Its algorithm sometimes throws a curveball by recommending videos that don't fit. Oftentimes the videos aren't your taste. But sometimes, the curveball recommendation opens a new world of content. This happened to me very recently. I watched a YouTube video from 2015 of an unmanned lawnmower attached to a rope. I'm not into mowers or lawns, but I can proudly say that I've watched more lawnmower videos in 2021 than all previous years combined. I wouldn't be this versed in lawnmowers had youtube excluded that video from my feed. Other than YouTube, content companies aren't great at throwing these interest-expanding curveballs. Spotify is horrible at recommending music that sounds different than what you usually listen to. Netflix is only marginally better. When algorithms fail to expand our knowledge and interests, we turn to content curators.
Curation: A Better Way To Discover Content
When I say "curator," I'm referring to individuals (or a group) acting independently of a platform. Though people curate Snapchat's live stories (if they're still called that), those people are still subject to Snapchat's platform rules and standards.
Content curators are human-led recommendation engines that fill the gaps created by feeds. Curators are made up of a single individual or a community of individuals with a sole purpose: to bring together content that interests them. They are often solving problems for themselves; people just happen to like their taste. Unlike feeds, curators are often specialized. Instead of covering a broad range of interests, curators focus on small niches. Usually, this would be unviable, but thanks to the internet, members of small niches and communities can connect seamlessly. The decentralized nature of curation allows them to work together on similar interests, work separately on unique takes on similar interests, or work on entirely separate interests. Their specialized nature often makes them experts in their niche. When it comes to tasteful recommendations, general-purpose algorithms can't compete with curators. Consumers get to choose their curators but not their algorithms. When tastes change, customers can easily switch to a different curator. This proves more difficult with algorithms. TikTok has one of the better feeds with their For You page, but it too struggles when a consumer gets bored of the persona the engine has built for them.
Let's say you're looking into learning about a new topic, say blockchain. You listened to an expert on your favorite podcast and want more. Spotify sucks at this. The next thing Spotify plays is another episode from the same podcast. Now you've gone from listening to and getting excited about the potential future of blockchain to listening to an episode with a real estate expert. Spotify doesn't recommend anything remotely related to the topic you just listened to. The platform treats podcasts like music albums. This happened to me recently. I listened to The Potential of Blockchain Technology on Invest Like The Best. Patrick O'Shaughnessy interviewed Chris Dixon in what soon became my favorite episode of the year. After listening, I had to leave the platform to dive deeper into the topic. I could have searched for "Chris Dixon" to find more podcast episodes, but then I'd risk listening to an interviewer I didn't like. As it turns out, that wasn't Chris Dixon's first time chatting with Patrick O'Shaughnessy. Spotify knew how often I listened to the podcast. It knew I didn't skip around and was hooked on the entire episode. It had more to offer but didn't bother to surface it.
I left the platform and ended up on Colossus — Patrick O'Shaughnessy's curated site. From that podcast episode, Colossus recommended twelve more content for me to dive into. Ten articles, a book, and another podcast episode. A feed would have suggested more content of the same type on the same platform. Curators don't have that limitation. By being platform & medium agnostic, curators can recommend more targeted content. The best part is that consumers will continue to go back for more content as long as they continue to enjoy the recommendations.
As Usual, Gaming is a decade ahead.
Gaming is a perfect example of the problem with information on the internet. There are a lot of games being published every year. More so now than a decade ago. As far as attention goes, gaming is one of the costlier mediums of consumption. They often take hours, if not days, to finish, and you need all but two of your senses to get through many games. It makes decisions difficult and often results in gamers deciding what to play instead of simply playing. Steam, the largest digital PC game distribution platform, introduced curation in 2014. Curators review and recommend what to play or what to avoid. Consumers can follow curators and get their recommendations alongside Steam's recommendation engine.
- Gamers win when they're able to find what to play without spending too much time debating. They are more likely to trust reviews from groups they are already a part of.
→ Gamers read reviews → purchase and play → go back to curator to find more games of similar taste →
- Curators win by growing an audience on an additional platform. Steam also lets them attach youtube videos which drive additional views and loyalty to the curator. This particular group started on Reddit, but they're also on Twitch, Youtube, and Twitter.
→ Curator posts reviews → grows audience on- and off-platform → posts more reviews →
- Most importantly, Steam wins. They win when the gamers win (revenue) and win when the curators win (engagement, loyalty, and more revenue). And what about game developers? Unlike many triple-A titles, most game developers can't afford to spend millions of dollars on advertising. Many great indie games never make it due to a lack of awareness. Curation also solves this. On Steam, there are many curators with thousands of followers dedicated to reviewing these lesser-known gems. In gaming, it isn't uncommon for communities to form around games created by a small team or a single person. Steam helps connect those developers with relevant Curators. They can send games directly to curators for a review.
Curation on Steam is good, but it is still limited to the platform and medium even though gaming is increasingly becoming platform-agnostic (I have written about this. I think Microsoft is at the forefront).
More creators are starting to go direct to consumers. We've seen a surge in the popularity of platforms like Substack and OnlyFans. Older players like YouTube also continue to be dominant. It is becoming increasingly difficult to expand your interests on the internet despite the deluge of content. Attention will continue to be a scarce resource, even more so with short-form content and personalized feeds. Consumers looking for new interests or more profound knowledge are severely underserved by feeds. They are also underserved when looking for new mediums of consumption. At the same time, they are overserved by the overwhelming amount of information on the internet. Content discovery is a problem that requires solving. I believe curation will play an essential role in figuring it out.
- The internet made attention more scarce by making information more abundant and abstracting distribution
- Internet content companies (& creators) created business models around this scarce resource
- shorter content duration
- content designed to grab consumer attention in the first few seconds
- personalized feeds that filter what consumers are exposed to
- Feeds aren't good for discovering content that expands consumer interest
- they're centralized and limited to the platform
- they're rigid and don't adjust well to changes in human taste
- Curation solves most of these problems
- they are decentralized, specialized, and scalable
- distribution is abstract
Sources, Links, and Recommended Reading
Emergent Layers - https://medium.com/swlh/emergent-layers-chapter-1-scarcity-abstraction-abundance-5705666e4f15
Lawnmower Video - https://www.youtube.com/watch?v=dmCQkosIa2k
Colossus - https://www.joincolossus.com/episodes/22848496/dixon-the-potential-of-blockchain-technology?tab=blocks
More of my writing (Subscribe For More) - Essays | Decision Making | Investing
Also on Twitter (@tolusnotes)