The Evidence

Using 2.7 million posts on COVID-19 and climate change, this study measures how Twitter's algorithm treats low-credibility content. Tweets sharing information from low-credibility domains perform better in the algorithmic feed, with the effect concentrated among high-engagement, high-follower accounts.

A theoretical and empirical analysis showing a fundamental trade-off: weighting engagement signals such as likes and shares increases platform engagement but also increases misinformation and polarisation. Provides evidence that Facebook's 2018 "Meaningful Social Interactions" update contributed to increased ideological extremism, and proposes an "engagement tax" as a policy mechanism.

Analyses news exposure across TV, web, and YouTube for roughly 55,000 panellists over 39 months. Television still dominates news consumption at a ratio of more than five to one over online sources, complicating narratives about algorithmic feeds replacing traditional media.

Annual survey tracking 8.9 million independent content creators in the US. The income distribution data is stark: 71% earn under $30,000 annually, only 9% exceed $100,000, and 34% make less than $5,000. Useful for grounding conversations about the "creator economy" in the reality that most participants are earning supplementary income, not a living.

Peer-reviewed study examining whether newsletter platforms like Substack represent a genuine alternative media space or reproduce existing journalistic hierarchies. The most successful newsletters are overwhelmingly authored by already-established journalists, and algorithmic recommendations increasingly shape subscription behaviour.

Pew's regularly updated fact sheet compiling industry data on newspaper circulation, advertising revenue, and newsroom employment over time. Print advertising revenue has fallen from roughly $49 billion in 2006 to a fraction of that, and digital advertising has never come close to replacing it. Valuable as a longitudinal data source.

Companion to the 2025 edition. Includes detailed data on willingness to pay for news across dozens of markets: only 17% globally say they are willing to pay, 57% will not consider paying anything, and among non-payers just 2% would pay the price of a full subscription. Suggests the news industry has already captured most of those willing to pay.

A global survey of over 240 publishers across 85 countries documenting the structural shift in news revenue. For the first time, print circulation and advertising account for less than half of publisher revenue, while digital revenue has crossed the 30% threshold. Essential for understanding the global picture beyond US-centric data.

Analyses news exposure across TV, web, and YouTube for roughly 55,000 panellists over 39 months. Television still dominates news consumption at a ratio of more than five to one over online sources, complicating narratives about algorithmic feeds replacing traditional media.

The bestselling book arguing that smartphones and social media have driven a rise in youth anxiety and depression. Strongest where it catalogues mechanisms of harm (sleep deprivation, attention fragmentation, social comparison) and weakest where it claims a single causal story. Should be read alongside Vuorre and Przybylski's more cautious findings.

A New York Times bestseller that identifies twelve structural causes of the attention crisis, from tech design choices to pollution and diet. Hari's central argument is that attention collapse is systemic, not a personal failing, and that solutions must be collective.

Distinguishes "dopamine-scrolling" from doom-scrolling as a distinct behavioural pattern driven by variable reinforcement schedules rather than anxiety. Identifies the pursuit of novel, entertaining content and rapid platform switching as a reward-mechanism-driven habit with public health implications.

The most high-profile institutional statement on social media and youth wellbeing, finding that teens spending more than three hours daily on social media face double the risk of depression and anxiety. Important not as the final word on the science -- the evidence is more nuanced than the advisory suggests -- but as the document that shifted the policy conversation.

A systematic review examining whether interventions that change social media use actually improve mental wellbeing. Finds that interventions are effective, especially for depression and when therapy-based approaches are used. Practically useful because it moves beyond correlation to test whether changing behaviour changes outcomes.

A rapid review examining cognitive decline from overconsumption of low-quality digital content. Synthesises research from 2023-2024 linking doomscrolling, zombie scrolling, and social media addiction to emotional desensitisation, cognitive overload, and psychological distress.

A peer-reviewed study in the International Journal of Research in Marketing mapping the roles platforms play in the creator economy: connecting actors, supporting content creation, and facilitating monetisation. Provides a useful academic framework for understanding creator-platform dependencies and the structural power platforms hold over funding flows.

Qualitative study based on interviews with 21 Patreon creators examining the relational labour required to sustain a patronage income. Creators construct relationships with patrons ranging from transactional to familial, and emotional support from patrons is as important as financial support for sustainability. Reveals the hidden labour costs of direct-funding models that aggregate statistics often obscure.

The most rigorous large-scale academic study of Patreon, analysing the complete set of pledges exceeding $2 billion from 2013 to 2020. Examines the membership platform from three perspectives -- patrons, creators, and the platform as a whole -- and finds that patrons who pledge to a narrow set of creators are more loyal but churn off the platform more often.

Goldman Sachs estimates the total addressable market for the creator economy at $250 billion in 2025, growing to $480 billion by 2027. Only about 4% of global creators earn more than $100,000 annually, with brand deals constituting roughly 70% of revenue. Useful for scale and growth data, though its framing is oriented toward investor opportunity rather than creator welfare.

An OECD working paper providing the most comprehensive institutional overview of platform cooperatives as employment alternatives to conventional gig platforms. Documents working conditions, challenges in scaling, and policy recommendations. Moves beyond individual patronage models to examine structurally different ownership forms for digital platforms.

Peer-reviewed study analysing 3,229 crowd patronage projects (Patreon-style recurring funding, as opposed to one-off Kickstarter-style campaigns). Finds that incentivising mechanisms (tiered rewards, exclusive content) have a long-term impact on patronage sustainability, while project characteristics have only short-term effects. One of the few rigorous empirical studies of the recurring funding model that underpins much independent media.

Detailed data on one of the largest digital public-good fundraising operations in the world. Over 8 million donors across 200 countries gave an average of $10.58 in FY 2023-24, sustaining $168 million in revenue. The declining average donation size and the operation's reliance on banner campaigns provide a real-world case study in both the potential and the fragility of grassroots digital giving at scale.

Deloitte's annual consumer survey covering streaming, social media, and digital entertainment. The 2025 edition finds that 47% of consumers feel overwhelmed by subscription management, 75% are frustrated by recurring price increases, and 61% would cancel their favourite streaming service over a $5 monthly increase. Documents the broader subscription fatigue environment in which news competes for wallet share against entertainment.

Peer-reviewed study examining the psychological mechanisms behind subscription cancellation. Identifies two distinct factors: "subscription fatigue" (the burden of managing multiple platforms) and "mental coupling" (the constant cost-benefit evaluation users perform). Coupling is the stronger driver of cancellation, suggesting that perceived value relative to cost matters more than sheer subscription volume.

Companion to the 2025 edition. Includes detailed data on willingness to pay for news across dozens of markets: only 17% globally say they are willing to pay, 57% will not consider paying anything, and among non-payers just 2% would pay the price of a full subscription. Suggests the news industry has already captured most of those willing to pay.

Peer-reviewed study analysing 3,229 crowd patronage projects (Patreon-style recurring funding, as opposed to one-off Kickstarter-style campaigns). Finds that incentivising mechanisms (tiered rewards, exclusive content) have a long-term impact on patronage sustainability, while project characteristics have only short-term effects. One of the few rigorous empirical studies of the recurring funding model that underpins much independent media.

The foundational report that framed open source software as public infrastructure. Commissioned by the Ford Foundation, it argued that digital infrastructure should be treated as a public good and that money alone would not solve sustainability problems without understanding open source culture. A decade on, the problems it identified have only intensified.

The most comprehensive survey of how organisations fund open source, finding that responding organisations collectively invest $1.7 billion annually, with 86% of that value being employee labour rather than direct financial contributions. Only 4% of financial contributions go directly to maintainers. Makes visible the enormous gap between the economic value open source creates and the resources directed to sustaining it.

The most comprehensive measurement of which open source libraries are actually used in production, based on over 12 million observations across more than 10,000 companies. Confirms that much of the most widely used open source software is maintained by only a handful of contributors. Essential for understanding where open source infrastructure is most vulnerable.

A detailed forensic analysis of the XZ Utils backdoor (CVE-2024-3094), in which an attacker spent 2.6 years building trust in an under-resourced open source project before injecting a critical backdoor into Linux compression infrastructure. Makes the abstract problem of open source underfunding viscerally concrete: this is what happens when critical infrastructure depends on overstretched volunteer maintainers.

Annual survey of open source maintainers covering compensation, burnout, and sustainability. The data on how many maintainers are unpaid — and how that affects the software infrastructure everyone depends on — is sobering and concrete.

A philosophy paper in Ethics and Information Technology arguing that platform decay constitutes genuine cognitive and moral harm. Using the extended mind framework, the authors show that when platforms degrade, they damage users' cognitive capacities and undermine conditions for developing virtuous character. A rigorous philosophical complement to empirical accounts of platform decline.

A peer-reviewed study in the International Journal of Research in Marketing mapping the roles platforms play in the creator economy: connecting actors, supporting content creation, and facilitating monetisation. Provides a useful academic framework for understanding creator-platform dependencies and the structural power platforms hold over funding flows.

The most rigorous large-scale academic study of Patreon, analysing the complete set of pledges exceeding $2 billion from 2013 to 2020. Examines the membership platform from three perspectives -- patrons, creators, and the platform as a whole -- and finds that patrons who pledge to a narrow set of creators are more loyal but churn off the platform more often.

A theoretical and empirical analysis showing a fundamental trade-off: weighting engagement signals such as likes and shares increases platform engagement but also increases misinformation and polarisation. Provides evidence that Facebook's 2018 "Meaningful Social Interactions" update contributed to increased ideological extremism, and proposes an "engagement tax" as a policy mechanism.

Goldman Sachs estimates the total addressable market for the creator economy at $250 billion in 2025, growing to $480 billion by 2027. Only about 4% of global creators earn more than $100,000 annually, with brand deals constituting roughly 70% of revenue. Useful for scale and growth data, though its framing is oriented toward investor opportunity rather than creator welfare.

Proposes a model for understanding attention as a currency that can be accumulated and exchanged, distinguishing between "calcified" attention (stored metrics such as follower counts) and "flow" attention (real-time engagement). Updates the theoretical framework of attention economics for the social media era.

Published in the British Journal of Industrial Relations, this peer-reviewed study extends the concept of enshittification from social media platforms to labour conditions. Based on 30 interviews with long-term gig workers, it identifies three mechanisms of value extraction over time: burden shifting, feature alteration, and market manipulation. Moves the enshittification framework from polemic to empirical labour research.

Annual survey tracking 8.9 million independent content creators in the US. The income distribution data is stark: 71% earn under $30,000 annually, only 9% exceed $100,000, and 34% make less than $5,000. Useful for grounding conversations about the "creator economy" in the reality that most participants are earning supplementary income, not a living.

Peer-reviewed study examining whether newsletter platforms like Substack represent a genuine alternative media space or reproduce existing journalistic hierarchies. The most successful newsletters are overwhelmingly authored by already-established journalists, and algorithmic recommendations increasingly shape subscription behaviour.

An OECD working paper providing the most comprehensive institutional overview of platform cooperatives as employment alternatives to conventional gig platforms. Documents working conditions, challenges in scaling, and policy recommendations. Moves beyond individual patronage models to examine structurally different ownership forms for digital platforms.

A philosophy paper in Ethics and Information Technology arguing that platform decay constitutes genuine cognitive and moral harm. Using the extended mind framework, the authors show that when platforms degrade, they damage users' cognitive capacities and undermine conditions for developing virtuous character. A rigorous philosophical complement to empirical accounts of platform decline.

The bestselling book arguing that smartphones and social media have driven a rise in youth anxiety and depression. Strongest where it catalogues mechanisms of harm (sleep deprivation, attention fragmentation, social comparison) and weakest where it claims a single causal story. Should be read alongside Vuorre and Przybylski's more cautious findings.

A New York Times bestseller that identifies twelve structural causes of the attention crisis, from tech design choices to pollution and diet. Hari's central argument is that attention collapse is systemic, not a personal failing, and that solutions must be collective.

Proposes a model for understanding attention as a currency that can be accumulated and exchanged, distinguishing between "calcified" attention (stored metrics such as follower counts) and "flow" attention (real-time engagement). Updates the theoretical framework of attention economics for the social media era.

Distinguishes "dopamine-scrolling" from doom-scrolling as a distinct behavioural pattern driven by variable reinforcement schedules rather than anxiety. Identifies the pursuit of novel, entertaining content and rapid platform switching as a reward-mechanism-driven habit with public health implications.

A rapid review examining cognitive decline from overconsumption of low-quality digital content. Synthesises research from 2023-2024 linking doomscrolling, zombie scrolling, and social media addiction to emotional desensitisation, cognitive overload, and psychological distress.