After the Feed: A New Era for Music Gatekeeping Has Arrived
- George Ergatoudis
- May 19
- 8 min read
A third handover of gatekeeping power has just begun, and most of the music industry isn’t paying attention.
A few weeks ago I wrote ‘The Economics of Sameness’ examining why algorithms, AI and commercial pressure increasingly reward familiarity and predictability, making originality both rarer and more valuable. The next day my friend Will Page, the music industry’s most revered economist, told me I needed to talk to Eli Pariser, whose book The Filter Bubble first warned that algorithmic personalisation would fracture our shared reality way back in 2011.
Pariser, now co-director of US non-profit, New_ Public, has just published an important report: “After the Feed: Trust, connection, and the next era of social technology.” It barely mentions music. But the implications for our business are enormous, and worth examining in detail. So here goes…
The gatekeeper is changing again
I’ve been an industry gatekeeper for much of my working life, so I maintain a very close eye on trends in music discovery and recommendation. New_ Public frames the period since the mid-19th century as three eras of gatekeeping. Institutions - newspaper editors and radio programmers held the power until around 2010. Then platforms and algorithms dominated through to 2025. Now a third set of gatekeepers are emerging - agentic interfaces, the LLM-powered assistants people will increasingly ask, “What do I need to know this morning?”. I’ve been testing an AI-powered audio app called Huxe that has access to my calendar, email and news preferences. It doesn’t recommend music, but it does offer an early glimpse of how useful highly personalised agentic interfaces are going to be.
I changed careers because of the last shift. At BBC Radio 1 I watched the station’s power diminish as streaming grew, so I jumped to Spotify and then Apple Music. Gatekeeping moved from pure human editorial to a combination of algorithmic and human curation: discovery at scale, personalised, fast, and - frustratingly for artists and labels - far less transparent.
The incoming agentic shift may be slow at first, but over the next few years many listener journeys will begin outside DSP interfaces. Instead, they’ll ask an AI assistant to “play me something” or “tell me what’s worth hearing.” The route to discovery will be hyper-personalised, mediated by a model whose incentives the fan can’t see.
Playlists were at least known entities you could pitch to, study, or argue with. An agent’s recommendation is a black box with a conversational voice.
The strategic question is going to shift from “how do we get on the playlist?” to “how do we earn standing with the system fans now trust to filter the world for them?”
DSPs risk becoming infrastructure companies sitting underneath somebody else’s relationship with the listener.
Of course, this also has massive implications for all content companies, including record labels and how they sign, develop and market music. I’ll be returning to this theme over the coming weeks.
Fandom is migrating to where trust lives
For music, the most important word in New_ Public’s report is trust. For a decade the currency has been likes, views, followers and monthly listeners and every one of those metrics has been open to manipulation. When engagement can be synthesised at scale, engagement stops being a signal. What becomes scarce, and valuable, is trust: in the content, and in the creator. Spotify’s recently launched human artist verification system is an early acknowledgement of this shift. Apple Music must now follow suit.
A million monthly listeners, or a track with 10 million streams, tells you almost nothing about whether anyone would cross the street for that artist. What matters now is whether anyone will buy the vinyl, show up to the show, or defend the artist in a group chat.
One of the report’s key concepts is “thick reputation” i.e. a reputation earned slowly, through sustained contribution to a specific community and it’s a great description of how a real fanbase behaves. Not “10K followers”, but “showed up for two years.” For artists, thick reputation is analogous to building up the body of the iceberg I wrote about in my article ‘The Iceberg Principle of Curation’: the part below the waterline carries the weight.
Fans are already voting with their feet. The report describes the migration from the open, hostile internet, the “dark forest”, to the “cozy web” of trusted group chats. Anyone close to music has watched this happen. The centre of gravity for a serious fandom is no longer the artist’s public feed; it’s the Discord, the subreddit, the fan-run server, the close-friends story. That’s where the real conversation, the real intimacy and the real loyalty now sit. And it’s true whether you’re into One Piece or One Direction.
The opportunity is obvious: most artists still treat private community as an afterthought when it may already be becoming the primary venue.
And, ironically, the collapse of mass-feed culture may end up being healthier for artists. The last decade forced musicians into an exhausting cycle of perpetual visibility. Smaller, higher-trust communities will prove more sustainable, emotionally and creatively, than chasing infinite reach.
What slop and bots do to discovery
New_ Public’s second trend is blunt: AI is destabilising the big platforms, making them less safe, less social, full of slop, full of bots. “Rage bait” was the Oxford English Dictionary’s Word of the Year for 2025; Merriam-Webster’s was “slop.”
Automated traffic now rivals human traffic online.
Music is already living this. Streaming services are facing AI-generated tracks at industrial volume, and the economics are corrosive: in a pro-rata royalty pool, every fake stream and every synthetic track dilutes the pot for human artists. More subtly, slop degrades the discovery environment itself. When a meaningful share of what surfaces is machine-made filler, the cost of finding something real goes up, and fans respond exactly as the report predicts. They retreat to spaces they trust.
The report makes another useful point: trying to police precisely where AI ends and the human begins is ultimately unwinnable, because AI is now woven through every step of creation and distribution. The answer isn’t a purity test. It’s provenance and relationship: who is the human behind this, what’s their track record, which community vouches for them? That’s a trust question, not a detection question.
Artists: build the room, don’t just rent the stage
The report’s fourth trend is “software abundance.” I see this every week - I’m regularly meeting start-ups building products with Claude Code that wouldn’t have been viable a year ago. New_ Public’s consequence: the “scale trap” is broken. A platform for 500 people, or 5,000, can now be built and run sustainably, where before only billion-user platforms made economic sense.
For musicians this is genuinely new and exciting. For twenty years the deal has been that an artist rents space on someone else’s platform, on someone else’s terms, measured by someone else’s metrics, with the relationship to the fan permanently intermediated. Software abundance means an artist, or more realistically a manager, or a label’s innovation group, can now build the room itself: a bespoke space for a specific fanbase, with its own norms, its own membership, its own economics. Not a landing page full of links. An actual product.
I hear WMG have shelved their superfan app, but the question they were asking is the right one: what happens if a record label starts thinking more like a gaming or community company than a traditional music company? Last week I saw an early demo from musician Frank Hamilton of an artist and fan-community platform that felt less like social media and more like a living digital world. I was seriously impressed and engagement in beta has been remarkable; he now plans to expand the test to about 50 artists. This kind of experimentation is going to explode.
Two enablers matter here. First, decentralised social protocols - AT Protocol, ActivityPub - let communities retain more ownership of identity, relationships and data, rather than rebuilding every time platforms change. Second, AI-supported stewardship means communities can scale without burning out the humans who run them. AI can absorb routine work while people focus on judgement, care and culture.
In New_ Public’s “jobs to be done” analysis, the human advantage sits exactly where music’s deepest value has always lived: connecting and belonging, expressing and creating, presence. The functions most exposed to AI are informational: quick answers, summaries, research. The functions that are protected are built around experience and being there. This maps almost perfectly onto the difference between just consuming a track - and being a fan. Recorded music as pure information becomes more replaceable; the artist as a human presence at the centre of a community becomes more valuable.
The industry must stop optimising for the old currency
The existing core metrics are depreciating assets. The industries that adapt fastest will be the ones that learn to read the new currency of thick reputation, i.e. depth, durability, community standing, and not just reach. That’s harder to measure, which is exactly why it’s about to become a competitive advantage for whoever builds the capability first.
The gatekeeping fight has also moved. The streaming era was fought over editorial playlists and algorithmic placement. The next fight will be over reputation inside agentic interfaces - because that will determine what an AI assistant surfaces and recommends when curating someone’s cultural world. New_ Public’s most urgent point is that the standards and protocols for this hybrid human-and-agent world are being written right now, in technical working groups most of our industry has never heard of. Civil society and the creative industries need a seat at that table before the defaults are set for us. We were largely absent when the recommendation systems of the last era were designed. We are paying for that absence still. We should not repeat it.
A word in defence of the job I used to do and the skills I built my career on. New_ Public borrows Ethan Mollick’s framing: “people will use AI when it’s better than the best available human”. AI will handle a lot of discovery: the obscure, the functional, the 'find me something for this moment' - and we should be honest about that. But curation at its best was never just retrieval. It was a person with taste, accountability and a relationship to an audience, putting their name on a choice. That is a trust act, and trust is the scarce resource now. The human curator, the tastemaker, the steward of a scene: those roles don’t disappear in the agentic era. They become more valuable than they’ve been in years.
So, what to do next?
• If you’re an artist or manager: stop treating private community as an afterthought. The Discord or fan-run server is no longer marketing - it is a key venue.
• If you’re a label: start measuring depth, not just reach. Begin the internal conversation about whether you’re a distributor or a community company. Build the infrastructure that many artists can’t do alone.
• If you’re a DSP: look hard at the human-curator question before you automate it away. Reputation inside agentic interfaces is the next battleground, and it will not be won by recommendation algorithms alone. DSPs should do more to amplify their human editorial credentials.
• If you’re anyone in the industry: find out who is in the room writing the standards for agentic interfaces, and get yourself, or someone you trust, into that room.
After the feed, the fans are still there
The report’s closing line is “after the feed, connection survives.” For music I’d put it differently: the feed was never where fandom really lived - it was just where we could most easily measure it. As that surface fragments, the work gets harder and more honest. As a curator, creator or musician you have to actually build trust, actually steward a community, actually be a human presence worth being a fan of.
That was true at BBC Radio 1. It remained true at Spotify and Apple. It will remain true after the feed: the tools change, but connection, belonging and creative expression endure. The people and companies that organise themselves around those will come through this era stronger, not weaker.
If you’re building in this space - whether you’re an artist experimenting with community, a label rethinking its model, or a team working on agentic music interfaces - I’d like to hear from you. Reach out in the comments or directly.
Source: Pariser, E., Quicksey, A., & Morelix, A. (2026). “After the Feed: Trust, connection, and the next era of social technology.” New_ Public. newpublic.org
Links:
The New_ Public Report is downloadable here:
Read more about Frank Hamilton’s project here:
Comments