Context
Maria Ressa, Nobel Peace Prize-winning journalist and co-founder of Rappler, argues that the fight over democracy is now inseparable from platform design, information integrity, and AI governance. Speaking from years of direct confrontation with disinformation campaigns and state pressure in the Philippines, she describes how social products and business incentives can hollow out truth, then institutions, then accountability. For PMs, the episode matters because it treats trust and safety not as a policy side issue, but as a core product and market-design problem.
The Big Idea
When platforms optimize for attention without accountability, they do not just spread bad content. They weaken the public's ability to tell fact from fiction, and that failure cascades into broken institutions and weaker democracy.
Ressa's point is that the root problem is structural, not just moral. Product incentives, discovery systems, business models, and governance choices all shape whether technology strengthens society or extracts from it.
Key Insights
Truth failure breaks institutions
Ressa frames the problem as a chain reaction. First comes a crisis of truth, where people can no longer reliably distinguish fact from fiction. Then comes a crisis of order, because institutions and rule of law lose legitimacy when shared reality disappears. The last step is a crisis of accountability, because citizens cannot hold either governments or platforms responsible if the information environment itself is corrupted. For PMs, the lesson is that trust failures are upstream failures. By the time they show up as political instability or reputational harm, the product system has already done years of damage.
Platform incentives reward manipulation
- Ressa argues that lies spread faster on social platforms by design, because fear, anger, and hate drive engagement.
- In the Philippines, Facebook had system-level power because it effectively was the internet. In 2017, 97% of Filipinos on the internet were on Facebook, and Ressa says that figure is effectively 100% today.
- The harms are not evenly distributed. Ressa said women journalists in the Philippines were attacked at least 10 times more than men in 2017.
- For PMs, the point is that harmful behavior is often a consequence of the metric system, ranking logic, and distribution model, not only a moderation failure at the edges.
AI escalates from attention upward
Ressa describes AI as a three-layer escalation of the same underlying power problem:
- Social media and the attention economy hacked biology by exploiting human attention.
- Chatbots and large language models hack intimacy by entering the trust layer of human relationships.
- Agentic and multi-agent systems threaten agency itself, because they can act on behalf of users or institutions at scale.
She ties that product argument to geopolitics. The UN panel she co-chairs reported that about 75% of global compute power sits in the United States, 15% in China, and 5% in the EU. Her warning is not only that AI is powerful, but that power is concentrated and moving faster than governance.
News faces market failure
- Ressa says news organizations are in a period of market failure.
- Zero-click search has sharply reduced discovery. In the Philippines, referral traffic to news sites dropped about 85%.
- Facebook, once the largest distributor of news globally, had already cut referral traffic by 60% to 70% by 2024.
- AI makes the economics worse. Publishers pay higher server costs when models scrape their sites, which means the original content producer is subsidizing the extractor.
For PMs, this is a reminder that platform policy changes can destroy adjacent ecosystems long before the platform feels the full cost.
Survival needs business reinvention
Rappler survived not by waiting for platforms or governments to fix the environment, but by changing its business model. After the Philippine government tried to shut the company down, Rappler lost 49% of its advertising revenue within four months. It responded by building a sister company, Nerve, that used tech and data capabilities such as NLP to map attack clusters and distribution networks. That capability now contributes roughly 40% of Rappler's revenue. The PM lesson is that when a core channel or revenue source is structurally compromised, resilience comes from building new capabilities close to the mission, not from hoping the old model rebounds.
Agency starts with small actions
Ressa's answer to large-scale institutional failure is not abstract despair. It is local, practical action. She says the first fight is against learned helplessness: check how much time you spend on screens, identify which app makes you angriest, do a 15-minute security audit, support journalists, and build a civic circle of 8 to 15 people who can act together. For PMs, the deeper point is that agency is a product variable. Systems that train passivity, outrage, or dependence are not neutral. Systems that help people coordinate, verify, and act create a different kind of user.
Public interest tech needs builders
Ressa is skeptical that values talk alone will fix the problem. She asks where public interest tech is. Her examples are concrete: Rappler built a federated chat app on the Matrix protocol in 2023, now used across 65 Filipino news groups and expanding in Southeast Asia. She also pointed to ROOST, an effort to pool trust and safety tools from major tech companies and make them broadly available for free. The product lesson is that safer ecosystems often require shared infrastructure and open standards, not just better intentions inside a single company.
Mental Models & Frameworks
The three-crisis ladder
- Step 1: crisis of truth. Users cannot tell fact from fiction.
- Step 2: crisis of order. Institutions and rule of law lose the shared reality they need to function.
- Step 3: crisis of accountability. Bad actors, including governments and platforms, become harder to constrain.
Use this model when discussing trust and safety investments. It helps a team connect an apparently local product choice, such as ranking or recommendation logic, to much larger downstream effects.
Attention, intimacy, agency stack
Ressa's AI framing is useful because it shows escalation by layer, not just by model capability. Start by asking what part of human behavior a product system reaches: attention, trust, or action. A recommendation feed acts mostly on attention. A chatbot acts inside trust and relationship. An agent system can take actions and therefore shape user agency itself. The closer a product moves to agency, the higher the bar for control, auditability, and governance.
Spider-web coalition model
Ressa describes the emerging global order as a spider web. Any one strand may look weak, but the web can still hold if many countries and institutions connect. Use this model when a problem is too cross-border for one company or one regulator to solve. It shifts the question from "who can fix everything?" to "which interoperable alliances make the whole system stronger?"
Tactics before grand strategy
Ressa deliberately structured her book around tactical steps before strategic goals. Her reason is practical: asking people to solve the entire system at once produces paralysis, while a sequence of small actions produces movement. For PMs, this is a good rollout principle for any trust or governance initiative. Start with actions users and teams can actually take, then connect them to the larger strategic vision.
Decision Principles
Principle: Fix incentive loops
- When: a product grows by maximizing attention, sharing, or low-friction distribution without measuring what kinds of behavior those loops reward.
- Why: Ressa's argument is that manipulation is often a feature of the system's incentives, not a bug added later by bad actors. If fear and anger are the cheapest growth levers, the product will keep producing civic harm even with sincere values statements.
Principle: Build owned channels
- When: a business or community depends heavily on search or social platforms for discovery.
- Why: Rappler's experience shows how fast distribution can disappear when platforms change behavior. A product team that lacks direct relationships with users has far less resilience when platform rules or economics shift.
Principle: Share safety foundations
- When: a safety problem is common across companies, such as moderation tooling, abuse detection, or provenance checks.
- Why: Ressa's ROOST example shows that some safety work should become common infrastructure. Shared tools create a floor faster than asking every company to reinvent the same protections in isolation.
Trade-offs & Nuance
Engagement versus information integrity
High engagement is not automatically bad, but Ressa's warning is that some of the most reliable growth inputs are also the most corrosive. Fear, anger, and hate travel well. If a team optimizes for short-term attention without distinguishing healthy engagement from manipulative engagement, growth can come directly at the expense of truth and trust.
Convenience versus trustworthy networks
Ressa is blunt that federated, non-manipulative systems may feel boring compared with engagement-maximizing feeds. That is a real trade-off. Products that preserve information integrity may initially look less sticky or entertaining. The question for PMs is whether convenience and stimulation are being bought by quietly degrading user judgment and shared reality.
Open access versus creator survival
AI systems benefit from large-scale access to public information, and Ressa does not deny that technology can be used for good. Her warning is about extractive economics. When models scrape publishers, raise their costs, and return little value to the creators who produced the facts, the knowledge ecosystem that AI depends on becomes weaker over time.
Common Mistakes
Mistake: Waiting for self-reform
Ressa openly questions whether Silicon Valley can reform itself through values talk alone. The failure mode is assuming the existing incentive system will correct itself without new governance, new infrastructure, or outside pressure. Her preferred alternative is to build public-interest tech and collective safeguards now.
Mistake: Confusing values with design
A company can say it cares about democracy, truth, or community and still ship mechanics that reward abuse. The better test is whether ranking systems, distribution loops, and monetization choices actually support the stated values. PMs should judge integrity by product behavior, not branding.
Mistake: Outsourcing trust to platforms
News organizations that relied on Facebook for distribution paid a large price when the platform pulled back. The broader mistake is treating trust, discovery, or safety as something another platform will reliably provide. Ressa's story argues for owning more of those foundations yourself, or building them through shared public-interest infrastructure.
Practical Application
Audit anger-driven product loops
Before the next roadmap review, inspect which product surfaces correlate with outrage, pile-ons, or low-quality sharing. Look at ranking inputs, notification design, and growth experiments. If a feature wins by increasing anger or impulsive sharing, write down that trade-off explicitly instead of treating the metric lift as a pure good.
Measure platform dependency now
List every channel that drives user discovery, activation, or retention. Then estimate what happens if a major platform cuts that channel by 50% or more. Rappler's traffic and revenue shock shows that dependency can become an existential risk faster than teams expect.
Fund trust as infrastructure
Treat trust and safety capabilities as product infrastructure, not just policy overhead. That can include provenance, abuse detection, account security, or community moderation tooling. Ressa's core argument is that information integrity is upstream of almost every other social outcome, so underfunding it is a strategic choice, not a neutral one.
Contribute shared safety tooling
Identify one safety component your company could share or help standardize, such as moderation rules, evaluation methods, or open safety tooling. Ressa's ROOST example suggests that some of the fastest progress comes from making baseline protections widely available, not proprietary.
Model AI scraping economics
If your product depends on original content, calculate the cost of model scraping on servers, discoverability, and monetization. Ressa's warning is that extraction can look invisible until the economics are already broken. A PM should know whether the business is subsidizing someone else's model.
Questions to Consider
- Where does your product currently reward attention without distinguishing healthy engagement from fear, anger, or hate?
- If search engines or social platforms cut referral traffic to your product by 60%, what owned channels or direct user relationships would still remain?
- Which part of your product touches attention, which part touches trust, and which part can act on a user's behalf? The closer a feature gets to agency, what extra controls should it need?
- What safety or provenance tool could your team contribute to a shared industry floor instead of building only for itself?
- Where has your team slipped into learned helplessness by talking about platform harm as inevitable instead of naming one concrete action to take this quarter?
Bottom Line
Maria Ressa's message is that democracy does not only fail at the ballot box. It can fail much earlier, inside ranking systems, discovery models, business incentives, and AI products that erode the public's grip on reality. For PMs, the job is not just to grow usage. It is to design systems whose incentives, economics, and governance do not quietly make truth and accountability harder to sustain.
Resources Mentioned
| Resource | Type | Why it was mentioned |
|---|---|---|
| The Fight Back: A Citizen's Guide to Standing Up to Dictators | Book | Maria Ressa uses the book to structure her answer around tactical steps, strategic goals, and ways citizens can defend themselves, their communities, and institutions. |
| Universal Declaration of Human Rights | Foundational document | Maria Ressa invokes it as the world's response to earlier global catastrophe and as a benchmark for thinking about today's crisis of truth and accountability. |
| UN Independent International Scientific Panel on Artificial Intelligence report | Report | Maria Ressa cites the panel's July report to explain how compute power is concentrated and why AI governance cannot be left to individual companies alone. |
Case Studies Mentioned
Rappler's forced business pivot
When government pressure caused a 49% advertising revenue drop in four months, Rappler should have been in existential trouble. Instead, the company built Nerve, a sister company that used tech and data work, including NLP on attack clusters and distribution networks. That pivot now contributes roughly 40% of revenue. The lesson is that mission-driven organizations can survive structural shocks if they turn adjacent capabilities into a real business.
Matrix-based news federation
Rappler built a chat app on the Matrix protocol in 2023 and rolled it out across 65 Filipino news groups, with expansion into Southeast Asia. Ressa's point is not that federated systems are automatically more exciting. It is that they can preserve information integrity because they are not built to manipulate attention in the same way as engagement-maximizing platforms.
ROOST safety commons
Ressa describes ROOST, robust open online safety tools, as an alliance that asked tech companies to contribute trust and safety tools, improved them, and then rolled them out for free. Her example is a concrete model for how industry-wide safety baselines can emerge even when no single company has enough incentive to build the whole stack alone.
Tools & Products
| Tool / Product | What it does | Why it was mentioned |
|---|---|---|
| Social platform and major news distribution channel | Maria Ressa describes Facebook as effectively the internet in the Philippines and as a system whose design helped spread manipulation and later cut news referrals sharply. | |
| Professional social network | Maria Ressa jokingly exempts LinkedIn when criticizing social media's tendency to reward harmful engagement, which helps clarify that her critique is about incentive design, not every network equally. | |
| Matrix protocol | Federated communication protocol | Rappler used it to build a news-focused chat system designed around information integrity rather than manipulation. |
| ROOST | Shared online safety tooling effort | Maria Ressa uses it as an example of pooled trust and safety infrastructure that can be distributed broadly for free. |
| Nerve | Rappler's tech and data sister company | Nerve became part of Rappler's survival strategy by turning internal data and NLP capabilities into a new revenue stream. |
People to Follow
Maria Ressa
Nobel Peace Prize-winning journalist, co-founder of Rappler, and one of the clearest thinkers on how platform design, disinformation, and democracy interact. PMs who work on social products, AI, media, trust, or policy will find her work especially relevant.
Camille François
Researcher and trust and safety expert mentioned in connection with ROOST. Her work is relevant for PMs who want a deeper view of online influence operations, safety tooling, and cross-company responses to platform abuse.
Experiments to Try
Experiment: Outrage audit
- Hypothesis: one or more high-growth product loops are succeeding partly because they amplify anger or impulsive sharing.
- Change: pick one feed, notification, or sharing surface and add a quality guardrail, such as reduced distribution for likely rage-bait or stronger friction before reposting.
- Test: run the change on a limited segment for two to four weeks.
- Measure: compare engagement metrics with abuse reports, misinformation flags, negative sentiment, or user trust survey results.
Experiment: Referral shock drill
- Hypothesis: the product is more dependent on third-party distribution than the team currently believes.
- Change: model a scenario where one major platform reduces referrals by 60% and identify the fallback journeys users would use instead.
- Test: for a small market or cohort, shift effort from platform acquisition to owned channels such as email, direct subscriptions, or community spaces.
- Measure: track activation, retention, and acquisition efficiency from owned channels versus platform channels.
Notable Quotes
"When you lose your rights, be careful. It takes a while to claw them back." (Maria Ressa)
"Get off your seats and stop with learned helplessness." (Maria Ressa)
"Hope comes from action." (Maria Ressa)
"When you go on something that's federated that isn't manipulating you, I warn you, it will be boring. But you will have information integrity." (Maria Ressa)
