How to Legislate in a Half-Known World

Why lawmakers should act on a “rational basis review” of the available evidence, rather than waiting for definitive proof.

Intro by Jon Haidt and Zach Rausch:

As legislatures around the world begin taking action — increasing the minimum age for social media access, setting bell-to-bell phone-free school policies, and adding warning labels on social media platforms — some critics have pushed back, arguing that the science is not settled and that policymakers should wait for more definitive evidence before passing laws.

We disagree. We've written about this elsewhere, most recently in our essay on the new consensus statement about social media’s effects on adolescents, and in a longer explanation of why definitively establishing causal effects is the wrong bar for policymaking in the real world.

Today’s essay goes much further. That’s in part because it’s written by someone who understands not just the evidence, but the actual process by which laws get made.

Casey Mock is an attorney who recently joined The Anxious Generation movement team as Senior Policy Director. He has spent his career designing and advocating for common-sense policies that respond to how emerging technologies shape society, most recently as Chief Policy and Public Affairs Officer at the Center for Humane Technology. Before that, he spent four years on the public policy team at Amazon, where he led national tax policy, and earlier in his career, he advised two governors — one from each party. Casey is also the creator of the Substack Tomorrow’s Mess, where he writes more broadly on tech, politics, and culture.

In this new piece, he reframes the evidentiary debate, showing why the heavy burden of proof pushed by industry and echoed by some academics is unworkable and ultimately unfair to the very people legislators are supposed to protect.

What should the standard of evidence be when considering health risks to children? And how should legislators navigate a world where the stakes are high, the evidence is debated, and the process is messy? Casey’s post is a guide.

– Jon and Zach

---

How to Legislate in a Half-Known World

By Casey Mock

On December 29, 1876, the Pacific Express train carrying 160 passengers approached the Ashtabula River bridge in northeastern Ohio during a fierce winter storm. The bridge, spanning 150 feet, had been designed by a businessman with no formal engineering training who had chosen an experimental design never before tested at such a scale. As the locomotive crossed the center span, the structure buckled and collapsed due to a flaw in the iron, sending the train plummeting seventy feet into the frozen river below. The wreckage caught fire, and more than ninety people died in what remains one of America's deadliest rail disasters.

Source: Wikimedia Commons (Ashtabula Bridge, Ashtabula Bridge Disaster)

With the benefit of hindsight, it seems obvious that the law should not have permitted the use of a bridge built on a novel design, without any prior testing, by someone unqualified to design bridges.¹

But is it really that obvious? Today, we routinely permit the same logic when it comes to children and technology. Chromebooks were distributed to millions of schoolchildren across the nation without any evidence that they would improve learning outcomes. Indeed, mounting evidence suggests they may harm educational achievement. Social media platforms, designed by computer programmers with no background in child development or psychology, have been thrust into the lives of American pre-teens and teenagers. And there is a growing recognition in the scientific community that heavy social media use can cause mental health and social harm to young people. “Ask for forgiveness rather than permission” is a corollary to the Silicon Valley motto of “move fast and break things.”

Yet we continue to hear arguments that the evidence is still inconclusive, insufficient, and that we "cannot definitively establish causal effects" between smartphones, social media, and declining youth mental health. These claims cut against what parents, teachers, school administrators, pediatricians, and legislators themselves observe unfolding around their kitchen tables, in their classrooms, and in their communities.

It’s as if we are all watching as the train is crashing into the ravine while an onlooker — sometimes with a megaphone provided by industry, by the way — insists we need more evidence that the bridge is unsound before any legislative efforts to address the harms can be justified.

Of course, we should want our lawmakers to be standing on firm ground before legislating on an issue. But how should legislators, and the voters who elect them, reconcile the fundamental tension between the insatiable demand for more evidence and the imperative to act?

Policymaking and Known Unknowns

As we know,
There are known knowns.
There are things we know we know.
We also know
There are known unknowns.
That is to say
We know there are some things
We do not know.
But there are also unknown unknowns,
The ones we don’t know
We don’t know.

— Donald Rumsfeld, Feb. 12, 2002, Department of Defense news briefing²

Former Secretary of Defense Donald Rumsfeld’s musings about knowns and unknowns capture the challenges that policymakers face when dealing with evidence. While we may want policy to be grounded in the world of known knowns, the reality is that policymakers often must act without full knowledge — and sometimes without knowing the full extent of what they don’t know.

Thinking about policymaking as a matter of navigating known unknowns feels instinctively unsatisfying. That’s partly because legislative proceedings bear a strong resemblance to court proceedings. Someone presides over a large room with an American flag and, if you’re lucky, some heavy antique wood furniture. There's testimony from various parties, and usually some grandstanding. Pomp and circumstance can confuse and frustrate the uninitiated. And sometimes — but not always — there's what we might call evidence. There’s a sense that there’s fact-finding happening, that open questions are being resolved, and that unknowns are being transformed into knowns through the esoteric process of hearings, protocols, and specialized language.

But here the similarities end. Why? Because each institution serves fundamentally different purposes.

Let’s start with the courts. The central questions a court seeks to answer are “What happened?” and “How does the law apply?” Trials are structured to establish facts upon which a judgment will be based, and different standards of proof are applied depending on whether it’s a civil or criminal case.³

Now compare that to the legislative branch. Legislators are fundamentally trying to answer different questions: “What might happen?” and “What should the law be?” They hear testimony not just about past events, but about what stakeholders want, and what experts think might happen under policy X versus policy Y. Legislative legitimacy is inherently forward-looking, which is part of why the Constitution prohibits ex post facto laws.

When it comes to evidence, lawmakers face an inherently different task from judges or juries. Yet too often, we treat legislative deliberations about keeping kids safe online as if they were trials. We have come to expect the same level of evidentiary proof that the Constitution demands for a criminal trial. That confusion is sometimes deliberately fostered by industry interests who benefit from legislative paralysis, but it’s also reinforced by well-meaning academics. These scholars, who attempt to import concepts from fields like epidemiology, are trained to demand strong causal evidence before making claims — for good reason. The cost of getting it wrong in those disciplines can be measured in lives.

But policy lives in the world of possible futures; inaction can cost lives, or it can save lives. Legislators are not judges or scientists. Their job is not to transform all uncertainty into certainty, but to act in the presence of knowns, unknowns, and unknowables alike. The world of policymaking will always be a half-known one, at best.⁴

Rational Basis: The North Star for the Half-Known World

If you’re lost (and your GPS isn’t working) how do you find your way? A little humility goes far – being willing to stop, check the map, or ask for directions. Instinct matters too; at a certain point, you have to act on the information you have, which might include a gut feeling. But rationality is how you find your way out of the wilderness. Streams lead to rivers, which tend to lead to cities. Humility, rationality, and instinct are the qualities we should hope for in our legislators as they confront the half-known evidence of policymaking.

American constitutional jurisprudence would agree. For the vast majority of legislation – anything that doesn't violate a fundamental constitutional right — courts apply what's called "rational basis review." Under this standard, legislators need only pass laws that are "rationally related" to a "legitimate" government interest.

When courts review statutes that do impact fundamental constitutional rights, they apply "strict scrutiny," asking whether the law is narrowly tailored to serve a compelling state interest. This more demanding standard reflects the judiciary's role as a check on government overreach. While it’s not identical to the evidentiary bar for a criminal prosecution, it reflects a similar philosophical principle we apply whenever the government threatens to take away something: the higher the stakes, the higher the bar. That bar is highest in protecting life and liberty.

Rational basis is the standard legislators should start from when thinking about evidence on kids, screens, and online safety.⁵ It leaves a lot of room to legislate in areas filled with reasonable doubt — full of known unknowns and unknown unknowns — and the Supreme Court made clear that this flexibility is a feature, not a flaw. In the Supreme Court's 1993 decision in FCC v. Beach Communications, Inc., Justice Clarence Thomas wrote that "a legislative choice is not subject to courtroom fact finding and may be based on rational speculation unsupported by evidence or empirical data." This principle exists to "preserve to the legislative branch . . . its ability to function" because otherwise, legislative fact-finding could continue indefinitely, paralyzing democratic governance.

Courts are meant to settle disputes. But what is a legislature for? What is the “legitimate” government interest that legislation must be rationally related to?

John Adams, in his 1776 Thoughts on Government, wrote that "[T]he happiness of society is the end of government,” and our legislative assemblies were designed by the Framers to bring about that end — particularly at the state level. Justice Louis Brandeis famously called states "laboratories of democracy," a phrase rooted in a fundamental constitutional principle: the power reserved to state governments by the Tenth Amendment. States have broad authority to regulate for the health, safety, and welfare of their citizens, a power that predates the Constitution itself and forms the bedrock of local self-governance.

As James Madison explained in Federalist No. 46, "The federal and state governments are in fact but different agents and trustees of the people, constituted with different powers, and designed for different purposes." In Federalist No. 45, Madison noted that state governments, being closer to the people, are better equipped to address local concerns and reflect the values and preferences of their constituents.

The Declaration of Independence speaks of the pursuit of happiness as an unalienable right, a principle that has never been, and could never be, evidence-based in any scientific sense. We pursue happiness because we believe it's worth pursuing, not because randomized controlled trials have demonstrated its efficacy. The pursuit of happiness is inherently forward-looking, qualitative work. It resists traditional forms of evidence. You can’t quantify it like an epidemiologist might quantify the spread of a virus. If comprehensive evidence for a policy already existed, we'd likely already have already adopted and implemented that policy.⁶

When parents decide they want their children to experience childhood without constant digital surveillance and manipulation, and they communicate that preference to their representatives, that preference deserves respect and accommodation regardless of whether it satisfies the methodological standards of academic researchers.⁷ That input is a known known, and an important one.

So what do we do with the known unknowns and unknown unknowns?

Understanding the American constitutional framework is one thing. Grappling with the practical realities of lawmaking is another entirely. The average state legislator is what we call a "citizen legislator" — often someone with a day job who serves part-time, equipped with minimal staff resources and operating under severe time constraints. Most legislative sessions last just a few months each year. These lawmakers must pass a balanced budget while reviewing hundreds, sometimes thousands, of bills on everything from highway funding to professional licensing to educational policy.⁸

That leaves precious little time for the kind of exhaustive fact-finding that academics might prefer. A hearing on a phone-free schools bill might allow just two minutes of testimony per stakeholder, barely enough time to state a position, let alone debate methodology or statistical validity. There's rarely time for legislators to read academic papers or evaluate research design, even if they possessed the technical training to do so effectively. And because legislative testimony is rarely given under oath, it can be difficult to separate fact from spin.

This environment generates confusion and hesitation. Legislators often find themselves lost in a thicket of competing claims: from families, from researchers, from advocacy groups, from industry. How should they decide whom to trust? How do they weigh action against First Amendment concerns? And how can we know what’s working?

Here are five practical tips for legislators navigating the challenges of policymaking in the half-known world:

1. Consider carefully the incentives of anyone offering compelling testimony.

It takes a special kind of person to show up at the state legislature during their free time to testify. If you hear a surprising or counterintuitive claim – say, that smartphones actually improve children's mental health – ask who is offering this perspective and why. Are they affiliated with industry? Do they receive funding from companies with business interests at stake? Are they trying to raise their public profile by staking out a contrarian view? What other professional or financial incentives might shape their public positions? None of this means the claim is wrong, but it’s crucial context. Too often, legislators adopt talking points without understanding the motivation behind them.

2. When faced with good-faith critics who raise legitimate concerns, press them for alternatives.

Anyone can identify problems. Good critics also propose solutions. If someone argues that a bill is flawed but offers no workable alternative, it may be obstruction, not improvement. Conversely, critics who come prepared with specific amendments or alternative approaches are engaging seriously with the issue. In 2024, Maryland legislators managed this well in hearings about social media design legislation. They challenged NetChoice lobbyists to come forward with solutions rather than just saying no (NetChoice, for the record, offered little beyond vague calls for “a Genius Bar for parents” and parents taking kids’ phones away).

3. Know your constitutional burden of proof – and get help drafting from knowledgeable lawyers.

Most laws are safe as long as they are rational and a legitimate use of state power. But that doesn’t stop industry lobbyists from trying to scare legislators with constitutional threats. If a bill does touch a fundamental right, the bar is higher — but even then it’s not an impossible one. Even under strict scrutiny, the government can regulate if it demonstrates a compelling interest and narrowly tailored approach. When protecting children from digital manipulation and exploitation, that compelling interest is obvious. Thoughtfully-drafted regulations that focus on design features (rather than speech) can satisfy constitutional requirements.⁹

4. Build in budget and process for evaluation and revision after a bill becomes law.

Some countries, like Australia, have built in processes to evaluate and revise laws a few years after they pass. In the U.S. we often let outdated laws stay on the books long past their usefulness. If your bill addresses an issue that would benefit from more study, write in the resources and process for revisiting it. Allocate funding¹⁰ for monitoring and evaluation. Include language requiring the law to be re-evaluated after a fixed period (of course, such provisions should be drafted to prevent industry attempts to capture and manipulate this process).

5. Most importantly, remember: companies asking to enter your state's market are asking for a privilege, not exercising a right.

State legislatures not only have the power, but the responsibility, to place conditions on that market access. If a product could harm the citizens you represent, it’s reasonable to demand that it demonstrate safety before deploying at scale. This is a legitimate and constitutional use of state legislative power. We must stop assuming that we have to prove harm before acting. When it comes to children’s minds, companies should have to prove safety before operating.

Breaking the Cycle

The United States has a particularly brittle policymaking process, shaped by a system of checks and balances that Francis Fukuyama has called a ‘vetocracy.’ Companies deliberately exploit this. They fund academic research meant to muddy the waters and weaponize legitimate concerns about government overreach to prevent any meaningful accountability. Meanwhile, they continue running uncontrolled experiments on our children while demanding that parents and legislators prove harm according to standards that would make a pharmaceutical clinical trial look easy by comparison.

Legislators aren't peer reviewers; they're trustees of the public will. Their job isn't to conduct original research but to translate community values into workable policy while operating within constitutional constraints. When parents raise concerns about their children's digital well-being, when teachers report classroom disruptions from devices, when pediatricians document rising rates of anxiety and depression among young people — these observations deserve a legislative response, even if they don't meet the methodological preferences of some researchers.

Consider how the burden of proof works in other contexts. Pharmaceutical companies must demonstrate safety and efficacy before the FDA approves their drugs for market. Food manufacturers must follow safety protocols and labeling requirements. Even something as mundane as a new car model must meet safety standards before dealers can sell it to consumers. Yet somehow, when it comes to digital products designed specifically to capture and manipulate children's attention, we've allowed companies to deploy their products first and – maybe – deal with consequences later.

The rational approach isn't to wait for more evidence of harm. It's to demand evidence of safety before permitting the experiment to continue.

The American constitutional system is built to function in a half-known world. The most profound unknown unknown may be what we stand to lose when we allow profit-driven algorithms to colonize the interior lives of developing minds. On the one hand, the evidence seems clear to anyone spending time around young people today. On the other hand, we may never fully grasp how much our world will be altered by it.

But we don't need to quantify it to recognize our obligations to younger generations. The pursuit of happiness depends on preserving the capacity for genuine human connection, critical thought, and authentic self-discovery. The evidentiary standard for that pursuit isn't perfect knowledge. It's the wisdom to act when the stakes are high – not after disaster has already struck.

1. In the year after the Ashtabula crash, the Ohio legislature convened a committee to investigate these issues – the sort of reaction we in the policy world now call government by train wreck, waiting until after a disaster has happened to act on it. See also Dodd-Frank after the 2008 crash, or the reorganization of federal agencies into the Department of Homeland Security after 9/11.

2. The formatting in verse is courtesy of Slate, “The Poetry of Donald Rumsfeld,” by Hart Seely.

3. In criminal court, the standard for a conviction is "beyond a reasonable doubt" – a deliberately high bar designed to protect individual liberty against state power. In civil court – when one party sues another – the standard is “preponderance of the evidence,” meaning that the parties attempt to demonstrate their claim is more likely true than not. The standard for a civil suit is a lower bar than a criminal prosecution because the cost of answering the question of “what happened” is much lower compared to when someone’s physical liberty is at stake, and because if the standard were too high, it would discourage use of the civil courts altogether. Evidentiary standards in court aren’t arbitrarily handed down from on high, but have developed to support courts’ function in the American constitutional system: as a bulwark against tyranny and prosecutorial overreach on the criminal side, and to balance efficiency against trust and fairness on the civil side.

4. The Half-Known World by Robert Boswell is a book taught in creative writing classes, and the title refers to the idea that in both writing and life, there are aspects that are familiar and understood, as well as elements that remain mysterious or unknown. This duality is particularly relevant in creative writing, where authors often draw from their own experiences and knowledge while also venturing into uncharted territory to create compelling narratives. Making policy is not so different.

5. Despite the best efforts of industry lawyers to argue otherwise in court, the Bill of Rights does not include a fundamental constitutional entitlement for adults to have a digital platform to harass teenagers with, for companies to show ads for fad diets to teenagers, or to extend First Amendment rights to chatbots.

6. States do learn from each other and adopt proven approaches, but innovation requires someone to go first, to act on rational speculation about what might work.

7. For some issues, clear data exists and should guide policy. The Constitution provides for a national census every four years because knowing how many citizens live in a jurisdiction proves crucial to basic governmental functions. But as any state legislator learns during their first budget cycle, hard data and evidence prove far more elusive than outsiders might assume. When states fund education or corrections systems – budgets that depend on the number of people moving through these institutions – legislators often lack precise figures and must instead rely on complex funding formulas based on actuarial projections and educated guesswork.

8. This essay was written days before the horror and tragedy that unfolded this past weekend, when an assassin – apparently motivated by politics – murdered Minnesota state Representative Melissa Hortman and her husband in the middle of the night, critically wounded another state senator and his wife, and attempted to target other state legislators. Rep. Hortman exemplified what too few Americans know about our state legislators: they are our representative democracy’s stewards, shouldering responsibility for the parts of government that, often unseen, touch our daily lives. They work long hours with minimal resources and little recognition. I had the privilege of serving as a director of budget policy for the state while Rep. Hortman served as Speaker of the Minnesota House and ushered through a transformational state budget. Her commitment to serving her neighbors – whether they agreed with her positions or not – reflected a profound understanding of her assignment. Minnesota has lost not merely a skilled legislator but a dedicated public servant who grasped that democratic governance requires principle and pragmatism, conviction and humility.

9. Although it’s possible to draft legislation that meets the strict scrutiny standard, it’s challenging. For legislators reading this post who are interested in that topic, please reach out to the Anxious Generation team: we can make sure you have the expertise you need to get this right.

10. As most savvy legislators know, state agencies loathe unfunded reports – particularly where they just collect dust, so ensuring the monitoring has funding and there’s a process for reviewing the results and incorporating them into changes in the law is essential.