It usually starts small. Someone posts a remark about a local council decision, a tax, a protest, a border, a school policy. You disagree. You type a reply. Twenty minutes later you are forty comments deep, the original point is nowhere in sight, and you have written three paragraphs to a stranger who has already decided you are dishonest. The argument was not lost because one of you was wrong. It was lost because both of you used moves that make agreement impossible — and most of us use them without noticing.
None of this is about being nicer, or about pretending every position deserves equal weight. Some claims are false and should be pushed back on firmly. But pushing back well is a craft, and the craft has known failure modes. Here are the five that do the most damage, and what to do instead.
Someone writes: “I think the city should spend more on protected bike lanes.” The reply arrives instantly: “So you want to ban cars and trap elderly people in their homes.” No one said that. The original claim was a budget preference; the reply attacks a fantasy.
Straw-manning is comfortable because the invented position is easy to defeat, and it feels like a win. It is also the fastest way to make a reasonable person stop reading. People do not defend the view you invented — they defend the view they actually hold, so the conversation immediately becomes about whether you are arguing in good faith.
The fix is mechanical, and it works even when you are annoyed:
This takes an extra sentence. It saves entire threads.
There is a moment in many arguments when one person stops making arguments and starts making piles. Four links, a screenshot, a substack post, a video timestamped at forty-one minutes. The implication is: read all this and then we can talk. Nobody reads all of it. The conversation dies of weight.
Links are not evidence in themselves; they are containers. What matters is the specific claim inside, and whether it survives contact with the objection on the table. A link volley signals that you have outsourced your reasoning to a reading list.
Better habits:
Tone policing is the move where the content gets skipped because the delivery was too sharp, too loud, too impatient. “I’d take you more seriously if you weren’t so angry.” Sometimes that is a fair observation. Often it is a way of avoiding a point that landed.
The mirror-image mistake is just as common: using your own anger as proof of sincerity and treating any request for calm as an attempt to silence you. Neither version moves the argument forward.
A practical rule: deal with the claim first, the tone second, and only if the tone is genuinely blocking progress. “I want to answer your point about the budget, and I’ll do it properly — but I’m going to stop if the next reply is another insult” is a boundary, not policing. It names the behaviour, keeps the substance on the table, and gives the other person a clear way back in.
It also helps to be honest about your own register. Frustration reads differently on a screen than it sounds in your head. Sarcasm that would be obvious face to face looks like contempt in text, and contempt is what people remember afterwards.
Public threads have an audience, and the audience changes how people write. Replies get longer, wittier, more performative. You start writing for the people who already agree with you, because their approval arrives instantly and the person you are supposedly talking to never gives any.
This is how a two-person disagreement turns into a campaign speech. The giveaway is a reply that contains no reference to anything the other person actually said — just a well-turned line aimed at onlookers.
Two small corrections help. First, write as if only the other person will read it, because in the long run that is the only reader who might change their mind. Second, remember that screenshots travel. A clever line at the expense of someone’s spelling will outlive the argument about the policy.
If you genuinely want an audience — sometimes you do, and that is legitimate — then make the case rather than the kill. Explain the reasoning so a bystander can follow it. That is persuasion; the other thing is just noise with a scoreboard.
The last mistake is the quietest: needing the exchange to end with a declared winner. That pressure produces some familiar bad moves — refusing to concede anything, even the small true parts; redefining the point after it has been answered; piling on new objections each time one is resolved so the conversation never lands.
You can break the pattern by conceding out loud, early, and specifically. “You’re right that the figures I used were from an older report.” It costs almost nothing and it changes the temperature of everything that follows. Then state plainly where you still disagree, and stop there if you have both been clear. An argument that ends in two precisely defined disagreements is a good argument, not a failed one.
Being right is cheap online. Being understood takes work, and it is the only version of this that anyone remembers a week later.
Most of the damage happens in the ten seconds before a reply goes out. A short checklist catches a surprising amount of it:
If you cannot get past the first question, ask it instead of arguing. If you can, you are already ahead of most of the thread. And if a conversation has gone three rounds without either of you repeating the other’s point back accurately, it is not a disagreement anymore. It is two monologues sharing a comment box. Close the tab, make a cup of tea, and come back to it tomorrow — or don’t. Some arguments are better left unanswered, and recognising which ones is its own kind of skill.
Photo: geralt / Pixabay