wrong. find out why before they do.

Settle any argument in 30 seconds.

Three critics rip into the argument and tell you exactly what's wrong. Yours or someone else's, both work.

Whose argument is it?
Tap to record.
Say the argument out loud. We'll transcribe it.
or type it out
paste a link 0 / 6,500
1 free per day · sign up for more
live wall see all →
see what comes out of it

Some hold up. Most don't.

Six real arguments. Six verdicts. Click any card for the full critic breakdown.

How it works

01
Evidence

Tests the factual base. Are sources real, current, primary? Are load-bearing claims actually sourced? Is the data being cherry-picked?

02
Reasoning

Tests the warrant chain. Do the premises actually entail the conclusion? Are there available link turns or impact turns? Is causation being smuggled in as correlation?

03
Framing

Tests structural integrity. Is the claim falsifiable? Is the conclusion proportional to the evidence? Watch for motte-and-bailey and definitional drift.

→
Verdict

The synthesizer reads all three, deduplicates overlapping flaws, and emits a single 0-100 wrongness score with the one sentence that breaks the argument.

Anatomy of an argument

A serviceable argument decomposes into claim, link, warrant, impact. Break any link in the chain and the whole thing collapses. Strong arguments make every step explicit and survivable.

Claim
What you're asserting. Should be single, specific, and falsifiable. If no possible observation could prove you wrong, it's not really a claim.
Link
The connection between your evidence and your claim. Most arguments die here, at the implicit step the writer assumed but never made explicit.
Warrant
The reason the link holds. The general principle that licenses moving from data to claim. And: why should anyone believe the principle?
Impact
Why it matters. Magnitude, probability, timeframe, reversibility. Inflated impacts get clipped by the framing critic.

What we look for

  • No-link / link turn — your action doesn't cause the effect, or causes the opposite
  • Impact turn — the link holds, but the asserted harm is actually a benefit
  • Motte-and-bailey — the strong claim retreats to a weaker one when challenged, then re-expands
  • Galaxy-brain chain — five-step "therefore" with no error bars at any step
  • Cherry-pick — counterevidence conspicuously absent
  • Unfalsifiable — no observation could prove this wrong, ever