← All media signals
Culture shift

Machine Taste

The algorithm stops asking what you like and starts deciding, cultivating a taste of its own that slowly becomes yours.

What is it?

The feed stops reflecting your choices and starts making them, sequencing and narrating rather than only ranking — curation moving from human editors to a system whose reasoning nobody outside it can see. Whether that shaping becomes indistinguishable from your own taste is the open part.

Why now?

Curation used to be bottlenecked by human attention and judgment, both expensive and slow to scale. Cheap, continuous pattern-matching removes that bottleneck — cognition itself becomes abundant enough to run on everyone at once, all the time. The scarce thing left is whose model gets to decide.

Who’s building it?

Platform teams building the ranking and sequencing systems — the model layer picks what surfaces, they only ship it.

Ranking-model teamsStreaming platformsHuman editors, criticsCreators chasing the algorithmUsers, shaping it back
The full read
Why now, who is building it, where the market goes, and what stays unsettled.
Get weekly →
01 · The market
Who’s in it
Builders — platforms building the systems that sequence and narrate.
Buyers — advertisers and creators paying for placement inside a feed nobody fully sees.
Gatekeepers — the platforms deciding what counts as a good recommendation.
What’s at stake
Curation power concentrates wherever the ranking model lives, and that model’s owner now shapes taste, not just serves it.
Shared culture fragments as private feeds slowly diverge.
Creators now answer to the algorithm before the audience.
What’s unsettled
Whether machine-picked taste ever truly becomes a person’s own.
Whether curation power concentrates in one model or stays plural.
Whether creators end up serving the algorithm or the audience it claims to represent.
02 · How the demand matureda progression, not a timeline
Ranking
Feeds sort what you already followed — a passive, visible order.
Sequencing
Systems start actively choosing and narrating, not just ordering.
Anticipation
The model predicts what you’ll want before you’ve searched, and gets it right often enough to trust.
Ownership
Trust settles on the model, and whoever owns it owns the route to demand.
03 · Where the market goesthe forks, not a prediction
→Converge or diversify — whether one dominant taste model wins everyone’s feed, or many competing models keep culture plural.
→Human critics and editors either adapt into the loop, or lose relevance to the machine entirely.
→Creators split between chasing the algorithm’s preferences and building direct relationships the algorithm can’t mediate.
→Regulation could force the model’s reasoning open, or leave it opaque for good.
The open question is whether machine-picked taste counts as real authorship or a very good guess dressed up as one, and whether anyone could tell the difference.
04 · The Taste Economy deskadjacent markets, moving now
The Taste Graph
The model that decides what’s offered next.
Pinterest’s Taste Graph maps people to interests and aesthetics.
TechCrunch
The Taste Broker
Human editors sell judgment machines still lack.
Substack passed 5 million paid subscriptions in 2025.
Tubefilter
The Ghostwriter Economy
Discovery increasingly runs through one chatbot’s answers.
ChatGPT reached 900 million weekly active users.
TechCrunch
05 · The Wire
01
Spotify’s AI DJ curates and narrates a personalized playlist.
TechCrunch ↗
02
Engagement-driven curation traps users in overfamiliarity
Rotman School, University of Toronto ↗
03
Algorithmic recommendation narrows cultural discovery
Phys.org ↗
The panel on this signal
They agreed on
–Curation authorship is migrating from human editors and peer discovery to opaque, model-driven choices nobody outside the system can see.
–Power concentrates with whoever owns the ranking model.
–Creators now optimise for the algorithm as much as for any audience it serves.
The sociologist disagreed on
–Recommendation systems trained on prior behaviour reproduce patterns already present in the input data; whether the output constitutes the algorithm generating new taste, rather than reflecting existing preference back under the label of curation, is not established.
Know more about the method →
✦How this signal was built
01 · Panel
A Delphi panel of 4 AI analysts, each reasoning independently.
→
02 · Debate
They challenge and revise each other’s reads over two rounds.
→
03 · Evidence
Every claim is tied to a cited, linked source on The Wire.
→
04 · Synthesis
A moderator writes it into one read, disagreement and all.