Behind the official reports and policy briefs about disinformation, there is a quieter story: the people actually tasked with fighting it are often unseen, understudied, and caught between competing pressures. In France, a team of researchers set out to change that by surveying 106 practitioners working across the military, government, regulatory bodies, civil society, and the media. These are not academic researchers writing papers from a distance; they are the civil servants, intelligence officers, regulators, and specialist journalists who decide what to monitor, when to intervene, and how to protect democratic institutions from harmful falsehoods. What emerges from this survey is a portrait of a field that shares a common diagnostic instinct—focusing on the powerful platforms and algorithms that amplify falsehoods—but that is blocked by weak coordination, scarce resources, and a pervasive fear of being accused of censorship. They trust science in principle, but rarely read it firsthand. And they almost unanimously expect the problem to get worse, while only a tiny minority believe the institutions they serve will be able to keep up. In short, the people on the front lines of the information war are not naive about the scale of the threat, but they are also not optimistic about their own ability to meet it.
The most striking thing about this group is how much they agree on what the problem actually is. When asked to identify the major threats to the information environment, nine out of ten practitioners point to the owners of social media platforms. Ordinary citizens come last by a wide margin. This is what the researchers call a “supply-side” diagnosis: the problem is not that people are hungry for falsehoods, but that platforms and organized actors push them relentlessly into every feed. Large majorities believe that platform algorithms strongly shape political opinions, and that producers and platforms distribute misleading content on a massive scale. The idea that the public actively seeks out such content divides the group almost exactly in half. This matters because it shapes everything else: if you think the problem is the pipeline rather than the consumer, then your response will focus on filtering, detection, removal, and regulation—not on building public resilience or rebuilding trust in institutions. The survey also shows that practitioners see disinformation as a first-order societal threat. Ninety-six percent call it an important or very important problem, ranking it on par with democratic decline and climate change. They attribute enormous effects to it: 90 percent say foreign campaigns are a major threat to democracy, 87 percent believe disinformation sustains Russian support for the war in Ukraine, and 83 percent agree that a single disinformation campaign can decisively change an election outcome. Interestingly, when the question is narrowed to a concrete case—Trump’s 2016 election—agreement drops to 71 percent, suggesting that the threat feels larger in the abstract than in any specific instance. And when compared to academic scientists surveyed earlier, the practitioners diverge most sharply on exactly these contested issues: they judge electoral impact far higher, while granting much less weight to institutional distrust, which scientists consider central to vulnerability. The practitioners’ diagnosis is coherent, but it is also selective—it overweights the pumps and undervalues the soil.
If the diagnosis is shared, why does action lag so far behind? The survey’s most powerful finding may be that the biggest obstacle is not external resistance, but an internal gap between rhetoric and reality. Three out of four practitioners say that counter-disinformation is constantly declared a political priority without the concrete action to back it up—and this complaint comes from inside the state itself. Officials, officers, and regulators report that governments call for action but fail to provide coordination, resources, clear mandates, or shared indicators of success. The next most cited obstacles are reactive policies that always seem to chase after new tactics, and the influence strategies of large digital companies. A majority also cite insufficient budgets and staff, administrative slowness, lack of operational tools, unclear roles, and difficulty attracting specialists. Then there is the fear of censorship. Sixty-three percent rank the fear of being accused of censorship as a major obstacle. But here is the twist: only one in five practitioners says they actually encounter ethical dilemmas often in their daily work, while more than six in ten say they rarely or never do. What inhibits them is not an active moral conflict; it is the anticipation of an accusation. They are restrained not by doubts about whether what they are doing is right, but by the fear of how it might look. That anticipatory anxiety, the researchers note, may be even more powerful than genuine ethical wrestling, because it makes practitioners self-censor before they have even acted. The result is a field that understands the threat clearly, but is frozen by fragmented authority, ambiguous legitimacy, and the absence of enforceable safeguards that would make legitimate intervention defensible.
The practitioners’ relationship with science is another source of tension—not because they reject evidence, but because they rarely consume it directly. Seven in ten say they trust peer-reviewed research to guide their decisions, but only about a third say they keep up with recent academic work. When they do read, they rely mostly on press coverage and gray literature—reports from think tanks, government agencies, and advocacy groups—rather than the studies themselves. Language plays a quiet but powerful role here: 88 percent primarily read about disinformation in French, while the scientific literature on the topic is overwhelmingly in English. The barriers they report are not skepticism or intellectual resistance, but organizational ones. Lack of time, limited access to publications, and weak institutional incentives all rank above perceived irrelevance or methodological difficulty. The researchers caution that these self-reports may flatter the respondents: admitting that you struggle with statistics or scientific English is more face-threatening than saying you are too busy. But the deeper point is structural. The knowledge that could broaden the practitioners’ selective diagnosis rarely reaches them intact. It is filtered through intermediaries, translated into simplified takeaways, and stripped of its caveats. This creates a quiet gap between what practitioners believe the evidence shows and what it actually supports. A practitioner who trusts peer-reviewed science but never reads it may end up with a confident but distorted picture of the field—one that fits the institutional worldview of supply-side threats, but misses the subtle social and institutional conditions that make certain messages persuasive in the first place. The researchers are careful not to blame the practitioners. They point out that a literature dominated by Anglophone, platform-centric, population-level studies may not speak directly to the targeted foreign operations that many French practitioners confront daily. And they note a deeper structural bias: institutions tend to measure what they can see. Harm that never happens leaves no trace, so prevention is invisible, while detected campaigns are countable. This pushes the whole field—practitioners and researchers alike—toward a supply-side focus that makes the diagnosis feel self-evident, even if it is incomplete.
Looking forward, the pessimism is hard to miss. Nearly all practitioners expect foreign interference to increase over the next three to five years, and the vast majority expect false or manipulated content to flood ordinary users’ feeds. More than eight in ten expect trust in media, science, and institutions to decline. And while six in ten think state capacity will increase, only 13 percent believe that capacity will actually be sufficient to meet the threat. In other words, even the people working hardest to counter disinformation see no credible path to closing the gap under current arrangements. This is not a counsel of despair, but it is a warning. The survey also reveals that the challenges are not experienced equally across the field. Educational background is the most consistent source of difference: those trained in computer and data sciences tend to see obstacles as less severe, while those in social sciences and communication see more of them. Fear of censorship varies by role too—it is highest among regulators and public-protection officials, and lowest among cybersecurity specialists. These differences matter because they imply that a one-size-fits-all response will fail. An educator may define success as accurate beliefs, a regulator as compliance, an intelligence officer as attribution. The evidence base itself is just as fragmented: psychology and communication study corrections and literacy; security and computer science study attribution and deterrence; and some crucial questions, like how to evaluate institutional responses, belong to no field at all. The remedy, the article suggests, is not another universal best-practice guide, but a translation effort that speaks to each profession in its own language—a legal memo for regulators, a threat report for intelligence analysts, a classroom resource for teachers—alongside a small set of shared indicators that measure what actually matters rather than what is easiest to count.
Of course, this study is not the final word. The sample is non-random, centered in France, heavily weighted toward state institutions, and recruited partly through defense and security networks, which may make supply-side and foreign-threat framings more prominent. All measures are self-reports, and the term “disinformation” was deliberately left undefined, which respects the diversity of institutional usage but limits comparability. The comparison with academic scientists is descriptive, not confirmatory, because the two surveys used different methods, scales, languages, and timings. Still, France is not merely a niche case. As a European Union and NATO member, it is covered by the Digital Services Act, has built one of Europe’s most institutionalized counter-disinformation apparatuses, and increasingly serves as a reference point for partners. Its problems—fragmented responsibility, high security salience, and uneven access to an Anglophone literature—are the problems of many democracies. The deeper lesson of this survey is human as much as institutional. The practitioners fighting disinformation are not faceless censors or naive true-believers; they are people doing a hard, politically sensitive job with imperfect tools and limited support, under the shadow of accusations they fear more than failure. They share a clear-eyed, if somewhat one-sided, understanding of the threat. They are not hostile to science—they are simply too busy and too organizationally constrained to engage with it deeply. And they are almost alone in believing they will succeed. If democracies want to do better, the study argues, they will need to build bridges between research and practice, measure outcomes that matter, and create safeguards that make legitimate intervention defensible—before the pessimism of the front line becomes a self-fulfilling prophecy.