Curators vs. the Algos
push to talk #49 // to find truly great games, music, movies, or books, you've gotta exit the algorithm and find a human with good taste
Today I’m trying out a more casual, observational riff. I hope you dig it. –RKR
I keep writing about how games marketing is being reshaped by discovery algorithms.
We’ve all seen this on a practical level: You’re playing a game and Steam goes “oh word, you liked that? I got another 10 games that are kinda like that.”
The same thing is happening across all of media. Spotify can spin up a themed radio station based on anything. Amazon wants you to know that Customers Also Bought [some similar garbage]. And God forbid if you ever look at a tweet for too long, because the algo will forcefeed you a million more just like it.
In opposition to this trend are human curators—smart people with good taste and maybe a little reach. Savvy players, listeners, and readers don’t want to be fed the same algo-approved slop as their peers. So they look to even savvier tastemakers to put them onto the really good stuff.
I’ve been thinking about this dynamic lately—about how the “curators vs. algos” relationship is playing out not just in the games industry, but across other forms of creative media.
A few observations about the role of curators in an algorithm-driven world:
1) Recommendation algos have burnout baked in
I’ve been a Spotify subscriber since 2011, the year it launched. And the more I used it, the worse my taste in music got. This was my fault more than the app’s.
I’d allowed my music listening habit to devolve like this:
1) Hop in the car 2) pull up the Spotify app 3) realize I don’t actually want to spend more than 10 seconds searching for music 4) pull up the first song or artist that pops into my head 5) play it and then let the “radio” station kick in when the music ends 6) listen to slop that sounds exactly like the same songs I’ve heard a million times.
Finally, like a year ago, I decided to try to get more serious about finding new music. Spotify’s pre-made playlists were fine, but the more interesting rabbit holes were on YouTube, where anonymous curators compile mixes based on specific genres.
To get their mixes seen, these uploaders have made a game of iterating on eye-catching titles and imagery for their vids, like this one called Breakcore mix to dissociate:
You can find a ton of great music this way—but there’s a real risk that suddenly every video in your feed becomes some even more esoteric breakcore-themed mix. That’s cool for like three weeks. But eventually you get burnt out on the stuff.
The algorithms are a little too good at feeding you “more like this” even when what you really want is “something cool that I didn’t even know existed.”
At some point, you start to get desperate for something novel again, and most discovery algorithms (at least for now) are weirdly bad at surfacing novelty.
2) A recommendation from someone w/ good taste is like finding hidden treasure
One of my favorite Substacks is Read Max by Max Read. His free posts are a smart opinion/commentary column, and each of his weekly paid posts are just roundups of stuff he recommends likes books (nominative determinism at work: the guy reads a lot), old movies, and usually a few songs.
Something I like a lot about Read’s music recs is that most of the time he gives you little or zero context—instead, at the bottom of his posts he has a section labeled something like “songs I like right now,” followed by embedded YouTube videos.
His taste is super broad, so clicking play feels like opening up a little musical lootbox—you have no clue what you’re getting into.
And occasionally, he introduces you to pure magic, like Strawberry by Roy Blair:
As of this writing this song has barely 5k views on YouTube. How does he find this stuff? Maybe not knowing is part of the fun.
You can’t get further from the Customers Also Bought approach to content recs than this. The lack of connection to your existing listening habit gives it a little mystique, especially when the thing being recommended is an under-the-radar banger. Because then you’ll get to be the person recommending that cool things to your friends. It’s a gift, in that way.
3) The best curators map out meaningful connections
Finding hidden gems is just one thing curators are better at than algos. Another is their ability to articulate and summarize knowledge about the history of a creative tradition—e.g. how does this cool thing I already know connect conceptually to other works?
This was part of what you used to get from old-school music magazines: A deep-in-the-weeds lore dump on the artist and their inspirations, brought to you by someone with enough time on their hands to become informed.
This tradition is still alive. Recently I’ve gotten obsessed with this Instagram account called Earfeeder. The account posts these incredible maps exploring extremely specific music subgenres, like this one focused on “spiritual jazz:”
The people running the account intentionally position it as an old school music mag that just happens to publish via Instagram.
For each music map they post, they include these hilariously baroque layouts with extra historical context on the genre they’re discussing. Like in their map of Southern Hip-Hop, they follow up with a breakdown of the differences between the genres descendants, like Memphis Rap and Trap:
Part of what I like about this is how open-ended it feels. It’s not like a “top ten” list, focused on ranking the works mentioned. There’s no implied hierarchy sorting the artists, so their recs are less corrupted by old debates about who the “best” artist is.
I kind of want an Earfeeder for everything. Somebody give me a map for horror movies, or for weird game subgenres. I wanna know more about the connections between the old stuff, not just who is currently considered the “best of all time,” as if that means anything.
4) The most powerful recommendation is an interesting person saying “this is peak.”
That being said, the appeal of those “best of all time” lists is undeniable, which is probably why so much curation takes the form of ranked lists. If you’re writing for a general audience that’s interested in an art form, it makes sense to try to establish a canon, even if it’s inherently nonsensical to compare works from different subdisciplines or artistic schools.
What does it really mean when IGN says that Super Mario World is the second best game, while Portal 2 is the third best. The only connection between those two works is that they’re both video games. It’s a little silly, right?
But I still read that list when it was published. It’s fun to see someone try to justify their rankings.
There’s something else really powerful about these lists. Because they’re considered to be authoritative, they have an influence on the minds of everyone who reads them—almost like they carve out a groove in the meme-space that we all live in. If you’ve encountered a bunch of these lists, your thoughts are guided along that groove from the phrase “best games of all time” to a bunch of specific titles that are often included in these lists like Super Mario World and Ocarina of Time and Resident Evil 4. Even if you don’t personally think of those games as being the best, the idea of them being widely considered to be the best is etched into your brain, in a way that influences your thinking (even if only to give you a starting point from which you articulate your own disagreement with the canon).
We see the same thing with albums, films, and novels. Every “best novels of the 20th century” list is obligated to include To Kill a Mockingbird and 1984 and The Great Gatsby. There’s a deeply carved groove in our shared meme-space that leads from “best novels” to those books. Is it possible to dislodge them? Maybe, but it’d take an enormous effort.
It’s not just “best of” lists that have this power. Individual influential writers sometimes have the ability to carve these grooves in the meme-space by giving a really powerful endorsement an artistic work. An example: If you ever start reading the works of Friedrich Nietzsche you almost inevitably get led to the Russian novelist Fyodor Dostoevsky, because Nietzsche (who rarely spoke highly of any other writer) famously said that Dostoevsky was “the only psychologist from whom I had anything to learn.” In his book Twilight of the Idols, Nietzsche also calls Dostoevsky a “profound human being,” and says that discovering Dostoevsky was “one of the happiest accidents of my life.”
This quote works as a sort of permanently grooved tunnel in the history of philosophy and literature that leads generations of readers toward Dostoevsky. Basically anyone who gets deeply interested in Nietzsche ends up reading Crime and Punishment to see what the fuss is about, all because a German guy from 1888 said “this boy Fyodor is peak.”
5) The curators are under pressure
My buddy Will Starck has been an influencer marketing lead for a bunch of crazy companies (Spotify and Riot Games included) and recently he was making the case to me that in the algorithm vs. curators war, “humans are back” because more and more people don’t want to be a midwit algo consumer.
“People are bombarded by recommendations from the algorithms and it’s so hard to sort through it, so you need a curator,” Starck says. “Why do people look to content creators/influencers/etc… now for basically everything?”
The natural result of this is that marketers are trying to influence the influencers (a term that’s dying out in any case). “Record labels for instance now all try and get their songs to be used by either TikTokers or on big influencer playlists. There’s BookTok, obviously. And in games that’s a lot of YouTubers.”
We all know this—that the relationship between creators and marketing teams has evolved in strange and sometimes tense ways. And audiences are getting savvy to the ways sponsored creators and curators work. “It’s just too much noise for people and they crave a connection to a person,” Starck told me.
I’ll end with a prediction: I think individual curators are only going to become more important, even as they become intertwined with the companies that run the recommendation algos in ever more confusing ways.
For the people doing the recommending and curation, then, their personal credibility and integrity will be what matters most over the longterm.
That’s it for this week. I’m gonna look up what mean things Vladimir Nabokov said about my favorite writers.
I’ll see you next Friday.
Very thought provoking, thanks.
This is something Ted Gioia is also saying, and I quite like the idea too.
There's one more thing to this: subcultures. 30 years ago people often identified with this group or another. I dare to say those were the collective curators, where the final vector of the subculture is an effect of thousands of tiny curators: leaders of the local branch of punks or Depeche Mode people or the ravers.
It's almost gone now unfortunately, but I think subcultures may be coming back to fight the algos.
And there are still subcultures that thrive almost completely under the radar. Since you like breakcore, I would recommend researching Free Party, or Free Tekno movement. This is unique to Europe I believe and most of that stuff is not to be found on Spotify.