Most online arguments aren't really arguments. They're performances aimed at a third party — the bystanders reading the thread. That isn't cynical. It's the structural reality of public discussion. And once you accept it, the goal of an argument shifts.
The goal isn't to make the other person concede. People almost never concede in real time, especially in front of their own friends. Pride gets in the way. Identity gets in the way. The feeling that giving any ground will be treated as total surrender gets in the way.
The goal is to make the conversation more honest — so anyone reading walks away with a sharper view of the question. That's a tractable thing to aim for. And it changes how you argue.
The audience is the match
Every public argument is performed for someone other than the person you're arguing with. Losing sight of this produces the most common mistake online: debating to defeat your opponent rather than to clarify the question. Mockery defeats. Clarity persuades. They look similar from the inside. They read very differently from the outside.
If bystanders watch an exchange and understand the question better than they did before, the conversation went somewhere. The opponent doesn't have to concede. They don't even have to respond.
The three patterns to watch for — in yourself first
Most arguments come apart in the same three ways. The easiest mistake is to look for these patterns in the other person while doing them yourself.
Straw man. Watch whether the version of your opponent's argument you're responding to is the version they actually made. If you find yourself replying to a worse version than the one on the table, the fix is to restate their claim in language they'd accept and respond to that. You'll lose some of the rhetorical force; you'll gain everything else.
Ad hominem. Watch whether your reply is about the argument or about the person making it. Their character, their qualifications, their team affiliation — none of those settle whether their claim is true. Drop the person, answer the claim.
Whataboutism. Watch whether your response engages with the critique or just swaps in a different critique. Their hypocrisy might be real and worth raising, but it doesn't resolve the point they raised. Answer the point first; bring up the other thing separately, if at all.
None of these are hard individually. The challenge is that they happen fast and in combinations. Spotting them in real time, in yourself, before you hit send, is the actual skill.
Your own bias is the bigger problem
The hardest move is the inward one: noticing when you're doing it.
Confirmation bias is the version of this that matters most online. The other side's evidence feels weak. Your side's evidence feels decisive. The asymmetry is almost never real — it's almost always bias dressed as judgment.
The test: before you read a source you agree with, ask yourself how hard you would scrutinize the same study if it showed the opposite result. If the answer is "much harder," you're not evaluating evidence. You're shopping.
When to stop
The last move is the most useful one. Most online arguments shouldn't be happening at all. If the person you're arguing with has no stake, no stopping condition, and no real interest in your view, you're feeding a thread that feeds on attention. Walking away isn't losing. It's identifying which games aren't worth playing.
Save your argument for the ones where someone is actually watching and might actually be moved. Those are rarer than they look. They're the only ones where any of this matters.