The problem
Two hours into the argument, you realize you've been fighting about five different things, none of them the real one. You're trading examples, scoring points, getting louder — and somehow the actual disagreement keeps sliding out from under the conversation. You both leave more entrenched, and neither of you could say precisely what you disagreed about.
That's the default failure of disagreement: people swing at surface positions while the load-bearing belief underneath never gets named. It's like arguing about which exit to take without agreeing on where you're going. You can fight forever and never touch the thing that's actually splitting you.
The cartographer's move is different. Before you argue, you map. You find the crux — and a named crux turns an endless fight into a finite, navigable disagreement.
The mechanism
A crux is the specific belief that, if it flipped, would change your conclusion. It's load-bearing: pull it out and your position falls. Most arguments never locate it because people argue proxies — surface claims that stand in for the real disagreement — and you can win every proxy battle while the crux sits untouched.
Crux-mapping fixes this by first taking the disagreement apart into three different kinds of claim, which get fought very differently:
Facts — what is actually true (resolvable, in principle, by evidence). Values — what matters, what's good (often a genuine difference, not an error — see moral foundations). Predictions — what will happen if we do X (uncertain, testable over time). Most heated arguments tangle all three together, so a factual dispute gets fought like a values war and never resolves.
Once separated, you look for the crux on each side — and check the key question: is the real disagreement empirical (we differ on a fact or prediction, which evidence could settle) or a values difference (we want genuinely different things, which is to be understood and respected, not won)? Both answers are progress. Naming the crux out loud — even with no agreement — produces a shared map: "we agree on the facts; we differ on how much to weight liberty versus safety." That sentence is worth more than two hours of proxy combat, because now you both actually know what you disagree about.
The operating system
Five steps to map a disagreement instead of just having one.
Notice you're fighting a proxy
The tell is a disagreement that keeps shape-shifting — you resolve one point and the objection just moves. That's the sign you're swinging at proxies, not the crux. Stop and say it: "I don't think we're actually arguing about this. What are we really disagreeing about?"
Separate facts, values, and predictions
Pull the tangle apart. Which parts of this are claims about what's true, which are about what matters, and which are about what will happen? Most arguments fuse all three. Sorting them is half the work, because each kind needs a different move and fighting them as one guarantees a stalemate.
Find your own crux first
Ask yourself the honest question: what would have to be true for me to change my mind? If you can't name anything, that's worth knowing — your position may be identity, not analysis. If you can, you've found your crux, and stating it models the vulnerability that invites them to find theirs.
Find theirs, and check the type
Help the other person locate their crux too, with the same question. Then check: do your two cruxes meet on an empirical point (a fact or prediction evidence could settle) or a values difference? If empirical, you now know what evidence would matter. If values, you've found a real, principled difference to respect rather than a stupidity to correct.
Name the crux out loud — the shared map
State the map plainly, together: "We agree on X and Y; the real disagreement is Z, and it comes down to [a fact we could check / a value we weight differently]." That shared sentence is the win, with or without agreement. You've replaced a fog of proxy battles with an accurate picture of exactly where two reasonable people part ways.
The printable: the crux map
Print it. Before the next big argument, map it instead of winging it.