The central idea is that speech rules should be designed for cases where we cannot reliably trust authorities to draw boundaries without bias or expansion. Once a system gives institutions the power to restrict speech based on its “harmful” or “hateful” nature, the key problem becomes not just whether hate speech is bad, but whether that standard can be applied consistently and safely over time. The argument is that it cannot, at least not in a stable way. Categories like “hate” or “offensive” are inherently interpretive and socially shifting, which means enforcement depends heavily on who is interpreting them and under what political conditions. From this perspective, the concern is structural rather than moral. Even if one agrees that hateful ideas are undesirable, granting the power to suppress them creates a tool that does not reliably stay limited to that purpose. Historically, restrictions on speech tend to expand outward from the original target group to other controversial or dissenting viewpoints. The logic here is that once the principle is accepted that the state may restrict speech based on perceived social harm, the boundary between “dangerous speech” and “unwelcome opinion” becomes politically contestable rather than objective. Another part of the argument is that the best long-term response to bad ideas is not suppression but exposure. The assumption is that ideas compete in a kind of marketplace: if harmful beliefs are openly expressed, they can be challenged, criticized, and shown to be false or destructive. Suppression may reduce visibility, but it does not necessarily reduce belief; instead, it can increase distrust and reinforce the idea among adherents that their views are being unfairly silenced. In that sense, restriction can unintentionally strengthen the very ideas it aims to weaken. Finally, the argument often rests on a principle of individual autonomy: if a society values individuals as capable of reasoning for themselves, it must tolerate a wide range of expression, including offensive expression, because the alternative is to replace individual judgment with institutional filtering. Even when some speech is harmful in a social sense, the cost of empowering authorities to decide which ideas people are allowed to hear is considered higher than the cost of tolerating the speech itself. This is the strongest internal logic of the “allow hate speech” position: not that hate speech is harmless or good, but that the mechanisms required to suppress it reliably tend to create greater long-term risks for open discourse and political freedom than allowing it does.

I agree speech has consequences. But if your principle is “restrict whatever speech causes social harm,” then we also have to examine the harms caused by censorship, because those are real-world consequences too. Your neo-Nazi example actually proves a narrower point than you think. Threats, intimidation, and direct incitement are already restricted in many legal systems because they involve specific, targeted harm. But you’re expanding from “credible intimidation” to “speech judged harmful by authorities.” That jump matters. You cite Germany. But if hate-speech restrictions are the decisive solution, why do countries with strong speech laws still struggle with extremist violence? Germany has maintained strict bans on Nazi symbolism, Holocaust denial, and hate speech for decades, yet authorities have continued reporting thousands of politically motivated far-right offenses annually. The existence of strict laws alone does not prove censorship solves extremism. Using your own public-health logic: suppressing ideas does not necessarily eliminate them — it can drive them into closed networks where they become harder to monitor, challenge, or debunk. Studies of extremist movements and online radicalization have found that deplatforming can reduce visibility but may also fragment communities into less transparent spaces. The misinformation example cuts both ways too. If harmful falsehoods justify broad restrictions, what happens when institutions are wrong? During COVID, some topics dismissed or heavily restricted early on later became subjects of legitimate scientific and public debate. A system that grants authorities broad power to define “dangerous misinformation” carries its own measurable risk: suppressing accurate or partially accurate information. And statistically, societies with stronger protections for free expression and press freedom often score highly on transparency, corruption control, and government accountability. That makes sense: open criticism helps societies detect mistakes. If your standard is minimizing harm, then you have to weigh the harms prevented by censorship against the harms caused when institutions gain greater control over acceptable speech. By your own logic, the question isn’t “Does speech have costs?” Of course it does. The question is: Which system produces less total harm — open debate with narrow limits for threats, fraud, and incitement, or broader censorship that can be misused, politicized, or wrong? That burden of proof still remains.