In May 2026, New Public published research called “After the Feed.” The lead author is Eli Pariser, the researcher who coined “filter bubble” in 2011, years before most people understood what recommendation algorithms were doing to how we read the news. That track record matters. Pariser is not an alarmist. He is someone who has named structural platform problems correctly, and early.
“After the Feed” argues that the algorithmic social feed era (roughly 2010 to 2025) is ending. What replaces it is what the report calls the Agentic Interface Era: LLM-powered tools that curate, synthesize, and deliver information personalized to a single person. The shift moves power from platform algorithms to AI agents sitting in front of every user. The report calls these agents “the new new gatekeepers.”
For communications directors and community managers, the practical implication lands fast: the distribution channels many organizations spent years building are losing structural value. Facebook Pages, LinkedIn company presences, Twitter followings. Not because of a bad policy decision at Meta or X. Because the underlying mechanism that made those channels work is being replaced.
The Open Social Platforms Are Losing the War Against Bots
In 2024, 51% of internet traffic was generated by bots. That is not a footnote in the New Public report. It is the primary driver of the cultural shift they are mapping.
Yancey Strickler’s “Dark Forest Theory” describes what happens next. As open platforms fill with automated content, surveillance, AI-generated slop, and cheap harassment at scale, real users retreat to smaller, hard-to-find, trust-filtered spaces. Private Discord servers. Closed Slack channels. Branded apps where every member was invited and every editor has a name.
This is a rational response. When most activity on a platform is automated, the social promises it makes are fraudulent. Users who can tell the difference between a real community and a space that only looks like one are opting out of the latter. The organizations that built closed, owned platforms, where content editors are identifiable people, channels are structured, and the app runs on the organization’s own infrastructure, made the right call before the research confirmed it. The research is now confirming it.
The trend will not stabilize. Every AI tool that makes it cheaper to generate convincing-looking content makes the bot problem worse on open platforms. The Dark Forest grows.
Trust Is the Scarce Resource Now, Not Eyeballs
The attention economy framing of the last 15 years treated reach as the primary metric: impressions, follower counts, view totals. New Public’s report identifies the replacement: trust.
Content from known, credible sources holds value that AI-generated volume cannot manufacture. The employee who sees a message in a branded internal app, from an editor they recognize, on a channel their team uses, trusts it differently than the same message arriving in a general WhatsApp group alongside 200 other notifications. The community member who has read a specific columnist’s work for two years trusts that voice differently than an algorithmically surfaced content recommendation.
This has a direct implication for how communications teams answer the “why not just use WhatsApp or a LinkedIn Group” question. The old answer was features: better notification controls, richer content formats, analytics. The better answer now is trust architecture.
Both WhatsApp and LinkedIn Groups are broken as high-trust communication channels at organization scale. AI-generated content floods LinkedIn feeds. WhatsApp groups become noise. The question is not whether those platforms have the features. It is whether their structure supports trust, or undermines it. The answer has become clear.
Thick Reputation Is the Moat That AI Cannot Copy
New Public draws a distinction between thin reputation and thick reputation. Thin reputation is platform-native: 10,000 followers, a verification checkmark, an algorithmic amplification boost. Thick reputation is community-native: contributed here for two years, vouched for by people this community trusts, has a visible track record of work anyone can check.
The AI era will collapse the value of thin reputation. Anyone can generate a verified-looking presence at scale. Generating a two-year contribution history inside a specific community, under a real name, with a track record that members remember, is harder to fake.
For communications and community teams, thick reputation is the asset that compounds. The internal editor who has published credible communications for three years holds organizational authority that no AI content agent can replicate by producing more volume. The community contributor whose comments shaped how a topic gets discussed in a closed forum is irreplaceable in ways that a high-follower public account is not.
This is a product argument, not just a content argument. Community platforms that track contributor history, give editors visible identity, and let consistent voices build standing inside their specific community are building infrastructure for thick reputation. Platforms that flatten everyone into the same feed are not.
What This Means for Internal Communications Teams
For communications directors at enterprise organizations, the Dark Forest dynamic is familiar terrain. The move toward closed, branded internal apps was underway before this analysis. The reasoning was practical: reach stays inside the organization, no third-party algorithm interferes with what employees see, data does not flow to Meta servers.
What “After the Feed” adds is the external context that explains why that choice compounds now. As open platforms fill with bot traffic and AI-generated noise, the gap between “we have a WhatsApp group for internal news” and “we have a branded employee app with structured channels and editorial ownership” will widen. The WhatsApp group gets harder to use. The employee app, maintained with editorial discipline, does not.
The daily decision for internal comms directors is straightforward: continue investing in a channel built on someone else’s algorithm and someone else’s data policy, or invest in the channel the organization owns and controls. New Public’s research makes the structural direction of that tradeoff explicit. It does not guarantee outcomes. A badly run owned platform is worse than a well-used public channel. But the ceiling and the floor are different.
What This Means for Publishers, Brands, and Membership Organizations
New Public’s report contains one line that captures the community argument more precisely than a paragraph could: “AI can help you find info, but you probably don’t want to hang out with your best friend’s agent.”
Platforms built on human-to-human relationships have a structural anchor that AI curation cannot remove. A local news publisher that has built an engaged community around named journalists and known local voices holds something that AI content flooding the same topic space cannot replicate. The community knows who reports for the publication. They have read those journalists’ work over years. That trust is thick, not thin. AI cannot manufacture the two-year reading relationship that makes a reader trust the source.
For brand communities, the same logic holds. The community that recognizes its contributors, rewards consistent engagement, and gives members a reason to stay beyond passive content consumption is the one that survives the shift to agentic content delivery. AI agents will route people to information. They cannot route people to belonging.
The organizations that treated their community as an audience to be reached built thin reputation. The organizations that treated their community as a group of people who know each other built thick reputation. The AI era does not reward them equally.
The Infrastructure Question
The structural shift New Public describes, from open platforms toward high-trust private spaces, does not happen without infrastructure. It requires a platform the organization controls, channels the editorial team configures, reputation mechanics that recognize contributor history, and a technical architecture that keeps data inside the organization’s perimeter.
tchop.io gives communications and community teams the platform to build those owned spaces without custom development or ceding control to a third-party algorithm. Editorial ownership and structured channels are where trust compounds.
See how tchop helps you build an owned community
The Teams That Act on This Early Have a Structural Advantage
Pariser named the filter bubble problem in 2011. Most organizations took another decade to internalize it. “After the Feed” names the next structural shift. Organizations that treat it as abstract platform theory risk making the same mistake twice.
The communities that last through the agentic era are the ones where members know each other, contributors have visible identities, and the trust that accumulates over years cannot be replicated elsewhere. That has always been true. The AI era makes it the only viable long-term strategy.
