The Media Landscape & the Crisis of Sensemaking
How information is produced and consumed now: the attention economy, fragmentation, the collapse of trust, and whether a society can still agree on what is true.
The business model that funded professional journalism has collapsed, and the attention economy that replaced it optimizes for engagement, not truth. Two things broke at once: the advertising that paid for newsrooms, and the shared reality a few institutions used to manufacture.
What this argues
- The ad engine brokeClassifieds and display ads that funded newsrooms migrated to platforms.
- Gatekeepers became one voice among millionsDistribution is now free; anyone can publish.
- Algorithms optimize engagement, not truthOutrage is the most engaging thing there is.
- Shared reality fragmentedPersonalized feeds give each person a different world.
- AI is about to flood the zoneSynthetic content makes verification the scarce resource.
What has changed
Updated November 2026
U.S. trust in mass media fell below 30% for the first time
Gallup's September 2025 survey (published 2 October 2025) put trust in mass media at 28%, a third straight record low. Republicans were at 8% and Democrats at 51%. Gallup reported 34% of Americans have no trust at all.
Global trust in news hit its lowest level since tracking began
- Reuters Institute Digital News Report 2026 (16 June 2026): 37% trust most news most of the time, the lowest since 2015.
- Trust fell in 29 of 48 markets; the survey covered about 100,000 people.
- Social media and video networks overtook news organisations' own sites and apps as the most-used route to online news globally for the first time (as reported by secondary coverage of the report).
Local news closures continued; news deserts reached a record
Northwestern Medill's State of Local News 2025 (20 October 2025) counted 136 newspaper closures in the past year, 213 news desert counties (206 a year earlier) and about 50 million Americans with limited or no access to local news. Nearly 3,500 newspapers have closed since 2005, a 39% decline. Newspaper industry jobs are down more than three-quarters since 2005. More than 300 local news startups launched in the past five years, 80% digital-only.
The New York Times keeps scaling subscriptions
The New York Times Company's Q2 2026 release (5 August 2026, SEC filing) reported about 13.35 million subscribers, of which about 12.80 million are digital-only, with roughly 280,000 net digital-only additions in the quarter. Its Q3 2026 results had not been published at the time of this update.
AI content farms tracked by NewsGuard passed 3,000
NewsGuard's AI Tracking Center listed 3,006 AI content farm sites in early 2026 (reported by Adweek, 12 March 2026), up from 1,200+ in 2024. Detection criteria have changed since the 2023 definition, so the series is not strictly like-for-like.
About one in five U.S. adults regularly get news on TikTok
Pew Research Center (25 September 2025) found 20% of U.S. adults regularly get news on TikTok, and 43% of adults under 30, up from 9% in 2020.
Not refreshed: newspaper ad revenue, newsroom staff and digital revenue share charts (no newer Pew series verified); minutes per day on digital media; Google and Meta ad share; creator economy size and forecasts (only secondary blog sources found, Goldman Sachs original not located); Edelman 'media as divisive' figure; Pew age-group news source split.
The Gatekeeper Age
How the old machine worked
Five checkpoints between an event and the public
- Primary source: an event happens, or an official statement is issued.
- Reporting: a journalist observes, interviews, and gathers first-hand.
- Editorial desk: editors decide what is covered and how it is framed.
- Fact-check & revise: claims are verified before anything is printed.
- Publication: the story reaches the public via print or broadcast.
Whoever owned the pipes owned the narrative
Layers of the gatekeeping stack
| Layer | What it was |
|---|---|
| Primary sources | governments, corporations, and wire reporters on the ground. |
| Wire services | AP and Reuters: wholesale news fed to every outlet. |
| Editorial layer | editors, fact-checkers, and standards that gate what runs. |
| Physical distribution | printing presses, broadcast licenses, delivery trucks. |
| The mass audience | a public receiving the same story at the same time. |
Good journalism was a by-product of a good business
| The economics that worked | What that bought |
|---|---|
| Revenue stability: subscriptions gave steady, predictable income. | Time to verify: print deadlines forced fact-checking windows. |
| Bundled product: serious stories rode alongside sports and comics. | Reputation focus: credibility, not clicks, drove renewals. |
| Broad ad model: advertisers paid to reach the whole town. | Investigative depth: margins funded slow, expensive journalism. |
| Local monopoly: one dominant paper per city, high margins. | Shared agenda: one front page set the town's priorities. |
Newspaper advertising has fallen ~80% from its peak
View the data
| Ad revenue | |
|---|---|
| 1990 | 32 |
| 2000 | 48 |
| 2005 | 49 |
| 2010 | 26 |
| 2015 | 17 |
| 2020 | 10 |
| 2024 | 9 |
Source: Pew Research Center · values read from the chart, approximate
Nostalgia for the old era is only half-earned
| Trait | What it meant |
|---|---|
| Shared facts | everyone argued from the same baseline record. |
| Accountability | investigative power that held institutions to account. |
| Professional standards | ethics, corrections, and verification as a discipline. |
| Narrow range of voices | a homogeneous elite decided what mattered. |
| Slow to adapt | gatekeepers could ignore stories they disliked. |
| Manufactured consent | official narratives became the default 'truth.' |
The Great Unbundling
How the internet broke the model
The internet didn't out-report the newspaper. It unbundled it, and sold the profitable pieces to someone else. Classifieds went to Craigslist, display ads to Google and Meta, and attention to everyone.
Newspaper newsrooms have lost more than half their staff
View the data
| Newsroom staff | |
|---|---|
| 2004 | 75 |
| 2008 | 71 |
| 2012 | 50 |
| 2016 | 39 |
| 2020 | 31 |
| 2022 | 31 |
Source: Pew Research Center · values read from the chart, approximate
Over 200 counties (213 in 2025) now have no local news outlet at all, and about 50 million Americans live in a 'news desert' with little or no local coverage. Local accountability journalism (school boards, courts, city budgets) is the first casualty, and the hardest to replace. (Northwestern Medill, October 2025)
The unbundling of the newspaper's revenue
| Old home | New home | |
|---|---|---|
| Classified ads | the paper's classifieds (~$20B) | Craigslist, then free |
| Display advertising | full-page print ads | Google & Meta programmatic |
| Local business ads | the local paper's monopoly | Facebook, Google, Yelp |
| Reader attention | the morning paper | the infinite scroll |
Digital revenue overtook traditional, but the pie shrank
View the data
| Traditional | Digital | |
|---|---|---|
| 2008 | 90 | 10 |
| 2012 | 78 | 22 |
| 2016 | 62 | 38 |
| 2020 | 50 | 50 |
| 2024 | 42 | 58 |
Source: Pew Research Center; compiled estimate · values read from the chart, approximate
The Attention Economy
Algorithms as the new gatekeepers
The new gatekeeper is not an editor. It is a ranking algorithm optimizing one number: how long you stay. No one decided the news should be addictive. The system selected for it.
The loop that replaced the editor
- Ingest signals: every click, pause, like and re-watch becomes data.
- Predict engagement: a model estimates what will hold your attention.
- Rank & serve: millions of items are scored; the top ones fill your feed.
- Learn & refine: your reaction updates the model, instantly and forever.
From 'what is relevant' to 'what is irresistible'
- Google PageRank: rank pages by who links to them, relevance at scale.
- Facebook News Feed: rank by your social graph, not chronology.
- YouTube watch-time: optimize for hours held, the autoplay era.
- TikTok 'For You': learn your taste in minutes from pure behavior.
- Generative feeds: AI both ranks and creates the content it serves.
Daily time on digital media has more than doubled since 2008
View the data
| Minutes/day | |
|---|---|
| 2008 | 210 |
| 2012 | 300 |
| 2016 | 380 |
| 2020 | 440 |
| 2024 | 470 |
Source: eMarketer · values read from the chart, approximate
The attention flywheel: why feeds tilt toward outrage
- Emotional content is posted.
- Algorithm sees high engagement.
- Reach expands to more users.
- Outrage earns the most clicks.
- Creators learn what wins.
- More extreme content is made.
At the center of the loop sits engagement.
Competition measured per minute
Fragmentation & Filter Bubbles
The end of a shared reality
When everyone had three channels, we argued about the same facts. Now each of us has a private feed, and a private reality. Personalization is a feature for the user and a solvent for the public square.
From a broadcast tower to a distributed mesh
| Sensemaking 1.0: the tower | Sensemaking 2.0: the mesh |
|---|---|
| A few institutions filter the world for everyone. Shared facts and a common agenda, but a narrow range of voices and slow correction. | Millions of nodes publish directly to their own audiences. Many voices and instant reach, but no shared filter and wildly uneven quality. |
Each era gave the audience more choice, and less in common
- three networks, one nightly narrative (toward 'one shared story').
- pick your channel, pick your slant.
- pick your feed and your friends.
- the feed picks you, and each is unique (toward 'a billion private feeds').
Filter bubbles do more than sort us politically
| Effect | What it means |
|---|---|
| Divergent diets | two people, same platform, opposite information. |
| Echo chambers | feeds cluster you with those who agree. |
| Radicalization pathways | recommenders can escalate from interest to extreme. |
| Loss of common ground | no shared facts to negotiate from. |
| Context collapse | clips travel without the story around them. |
| Polarization beyond politics | gender, age, and region all splinter. |
Reach without accountability is the new normal
A two-axis map of audience reach (horizontal) against editorial accountability (vertical). Legacy newspapers sit in the upper left (niche and accountable, higher accountability, moderate reach). Broadcast TV sits in the upper right (mass and accountable). Cable news sits near the middle line on accountability with mass reach. Creators and podcasts sit in the lower right, with social platforms further right and lower still (mass and unaccountable). AI content farms sit at the bottom near the center line.
The Trust Collapse
Why no one believes the news
Trust in the media has fallen to a record low, and for the first time, more Americans have no trust in it at all than have a great deal. The gatekeepers lost their audience's money and their belief at the same time.
Trust in mass media has more than halved since the 1970s
View the data
| Trust in media | |
|---|---|
| 1976 | 72 |
| 1990 | 54 |
| 2005 | 50 |
| 2016 | 32 |
| 2024 | 31 |
| 2025 | 28 |
Source: Gallup (2025) · values read from the chart, approximate
Neither traditional nor social media is trusted, but social is worse
View the data
| Traditional media | Social media | |
|---|---|---|
| 2016 | 47 | 28 |
| 2018 | 44 | 27 |
| 2020 | 42 | 25 |
| 2022 | 40 | 24 |
| 2024 | 38 | 23 |
Source: Pew Research Center; compilation · values read from the chart, approximate
Under-30s live on social media; over-65s still watch TV
View the data
| Social media | Television | |
|---|---|---|
| 18-29 | 78 | 26 |
| 30-49 | 62 | 34 |
| 50-64 | 47 | 52 |
| 65+ | 34 | 68 |
Source: Pew Research Center (2024)
The structure of distrust has inverted
A serious diagnosis holds two truths at once
| The critics' case | The defenders' case |
|---|---|
| Visible bias: opinion and reporting blurred together. | Deliberate distrust: a strategy to discredit unwelcome facts. |
| Sensationalism: clicks rewarded over accuracy. | Impossible economics: less money, fewer checks, more speed. |
| Elite disconnect: newsrooms distant from their audience. | Flood of fakes: real news drowned in cheap content. |
| Public errors: high-profile misfires, slow corrections. | A complex world: hard truths resist simple, shareable framing. |
The New Economics
What is being built to replace it
The replacement is not another institution. It is millions of individuals, funded directly by the audiences they serve. Subscriptions, memberships, newsletters, podcasts: journalism is being rebuilt one relationship at a time.
The creator economy is now a ~$250B business
View the data
| Creator revenue | |
|---|---|
| 2019 | 90 |
| 2021 | 150 |
| 2023 | 250 |
| 2025E | 340 |
| 2027E | 480 |
Source: Goldman Sachs; Doug Shapiro (2023) · values read from the chart, approximate
The new economics inverts the old on every axis
| Sensemaking 1.0 | Sensemaking 2.0 | |
|---|---|---|
| Funded by | advertising to a mass audience | the audience, directly |
| Trust from | institutional brand and standards | personal authenticity and track record |
| Distribution | owned presses and licenses | rented from platforms, or owned lists |
| Relationship | one-to-many, anonymous | one-to-one, at scale |
No single model replaces the old ad engine: a portfolio does
| Model | What it is |
|---|---|
| Subscriptions | paywalls and digital memberships (the NYT's 13.35M subscribers, Q2 2026). |
| Newsletters | Substack and email: owning the audience, not renting it. |
| Podcasts | long-form audio with premium sponsorship economics. |
| Creator payouts | platform ad-splits on video and short-form. |
| Memberships & tips | superfans funding work directly, Patreon-style. |
| Philanthropy | nonprofit and foundation-funded newsrooms. |
The economics of the new model are a funnel
| Funnel layer | What it is |
|---|---|
| Superfans | the few who pay directly (subs, memberships, tips) and fund everything. |
| Regular audience | engaged followers who convert over time. |
| Casual reach | the broad top-of-funnel from algorithmic discovery. |
| The internet | everyone the algorithm could theoretically show you to. |
The Sensemaking Crisis
Can a society still agree on facts?
The scarce resource is no longer information. It is verification: knowing which of the infinite things you can see is true. And generative AI is about to make infinite content nearly free to produce.
The strategy is doubt, not persuasion
- Volume beats truth: cheap content out-produces careful reporting a thousand to one.
- The liar's dividend: when anything can be faked, the guilty dismiss real evidence as fake.
- Engagement over accuracy: false news spreads faster and farther than true news.
- Personalized persuasion: AI can tailor a different, optimized message to each person.
AI content is scaling faster than anyone can verify it
The danger is less any single fake than the cumulative doubt
| Threat | What it is |
|---|---|
| Deepfakes | convincing fake video and audio of real people. |
| Synthetic text | infinite articles, reviews, and comments at zero cost. |
| Bot networks | coordinated inauthentic accounts manufacturing consensus. |
| The liar's dividend | real footage dismissed as 'probably AI.' |
| Model collapse | AI trained on AI output degrading over time. |
| Micro-targeted persuasion | a bespoke argument optimized for each voter. |
The honest verdict is unresolved
| The optimistic case | The pessimistic case |
|---|---|
| Direct expertise: audiences reach primary experts, not second-hand reporting. | No shared facts: fragmentation makes consensus impossible. |
| Restored nuance: long-form podcasts reward depth over soundbites. | Truth is expensive: verification loses to free fabrication. |
| Lower barriers: anyone can inform, from anywhere. | Trust keeps falling: cynicism becomes the default posture. |
| Better tools: AI can also detect fakes and verify at scale. | Capture risk: a few platforms control the new pipes. |
Signposts to watch over 24 months
- Does AI-content labeling stick? Provenance standards like C2PA are the technical fix.
- Do subscriptions scale beyond a few winners? Or does only the NYT truly survive?
- Does trust find a floor? Or keep falling toward zero?
- Do platforms or regulators move? On algorithmic transparency and liability.
- Who owns the audience relationship? The durable asset in the new economy.
Reference
Glossary & sources
Plain-language glossary
| Term | Meaning |
|---|---|
| Sensemaking | how a society collectively decides what is true and what matters. |
| Gatekeeper | an institution that controls what information reaches the public. |
| Attention economy | a market where human attention is the scarce, monetized resource. |
| Algorithmic feed | a ranked stream of content ordered by predicted engagement. |
| Filter bubble | a personalized information environment that reinforces existing beliefs. |
| Echo chamber | a space where your views are amplified and opposing ones filtered out. |
| Flooding the zone | overwhelming the public with content to make truth hard to find. |
| Liar's dividend | the ability to dismiss real evidence as fake once fakes are common. |
For a century, a few institutions decided what was true and charged advertisers for our attention while they did it. That system was narrow, powerful, and profitable, and it is gone. What replaces it is faster, freer, and far less sure of itself: a billion voices, no shared filter, and the growing suspicion that nothing can be verified. The scarce resource of the next era will not be information. It will be trust.
Sources & Further Reading
- Pew Research Center: State of the News Media; newspaper industry fact sheets (2023-24).
- Gallup: Americans' Trust in Mass Media, annual series (1972-2024).
- Reuters Institute for the Study of Journalism: Digital News Report 2024.
- Edelman: Trust Barometer 2024.
- Northwestern Medill: The State of Local News Report (2024).
- eMarketer / Insider Intelligence: U.S. digital media time and ad-spend data.
- Goldman Sachs; Doug Shapiro (Substack): creator economy sizing.
- NewsGuard: AI-generated news and misinformation tracking (2024).
- Vosoughi, Roy & Aral: 'The spread of true and false news online,' Science / MIT (2018).
- Figures marked 'estimate' are compilations of the above.
Sources cited
Gallup · Northwestern Medill · eMarketer · Goldman Sachs · Pew Research Center · Columbia Journalism Review · Nature Human Behaviour · MIT · LocaliQ · DataReportal · Reuters Institute · Edelman · Knight Foundation · Doug Shapiro · Press Gazette · Kevin Kelly · NewsGuard · Company disclosures · Academic literature · Gallup, Oct 2025 · Reuters Institute, Digital News Report 2026 (Jun 2026) · Northwestern Medill, State of Local News 2025 (Oct 2025) · New York Times Company, Q2 2026 results (SEC 8-K, Aug 2026) · NewsGuard AI Tracking Center (2026); Adweek, Mar 2026 · Pew Research Center, Sep 2025
Keep reading
Research, not investment advice. Figures as of , from the sources cited.
Legal information and disclosures
These are my own views as of the update shown on this page. They can change without notice, and I have no duty to update them. Past performance does not indicate future results, and any investment can lose money.
This is for education only. It is not investment, legal, tax, medical or accounting advice, and it is not an offer to sell or a solicitation to buy any security in any jurisdiction. It does not consider your goals or circumstances.
Figures from company filings, public agencies and other third parties are believed to be reliable, but I have not independently verified them and cannot guarantee them. Forecasts and anything marked as an estimate rest on assumptions that may prove wrong.
Please do not copy or republish this material without my written consent.