The short version
Throw out an enormous number of claims, examples, statistics and accusations in one burst. Any single one might take ten minutes and two sources to properly knock down. Making it took four seconds.
That asymmetry is the entire weapon. Your opponent can answer some of it or none of it, and either way there’s a pile left over — which an audience will quietly score as the part he couldn’t answer.
Where the name comes from
Duane Gish was a creationist who spent about thirty years — from the early 1970s to around the turn of the century — debating evolutionary biologists, and he was extremely good at exactly this: rattling off far more claims than any opponent could address inside a debate format, then treating the leftovers as unrefuted. He did it, by his own side’s count and by his critics’, more than three hundred times. Eugenie Scott of the National Center for Science Education gave the technique his name in 1994, and it has outlived the debates by a wide margin.
What it looks like
- Your numbers are from a discredited study
- The trial had eleven participants
- The lead author later retracted a different paper
- Three other countries tried this and reversed it
- You’re ignoring the 2011 data entirely
- The funding came from an interested party
- Nobody in the field actually agrees with this
- The effect vanishes if you control for income
- That’s not what the original paper said
- You said the opposite of this two years ago
Answering any one of these honestly takes ten minutes. You have four.
Why it works, and why it isn’t an argument
Alberto Brandolini put the mechanism best: the energy needed to refute nonsense is an order of magnitude greater than the energy needed to produce it. That’s true accidentally most of the time. A gallop is what it looks like when somebody uses it on purpose.
Volume isn’t evidence. Fifteen weak claims don’t sum to one strong claim, and they don’t sum to anything at all — but they feel like they do, because a listener following an argument is counting exchanges rather than weighing them. Ten unanswered points reads as a rout even when all ten are junk.
The bind for whoever has to respond is genuinely nasty. Answer everything, and you spend your entire turn on rapid-fire corrections, sound defensive, bore the room, and still run out of clock. Answer some, and the rest becomes “he never addressed the funding.” There is no version of chasing it that comes out well.
This is not a Gish Gallop
- A complicated case with a lot of parts. Plenty of real arguments have twelve moving pieces, and plenty of good ones are piles rather than chains. A dozen independent lines converging on one conclusion is how a circumstantial case works, and how most of science works — good reasoning is a cable of many slender fibres, not a chain no stronger than its weakest link. So the difference isn’t the number of parts, or how they’re joined. It’s whether the parts are any good, and whether the person will stand behind them.
- Somebody who just talks a lot. Verbosity is a personality, not a tactic.
- A lot of evidence for one claim. That’s not a gallop, that’s a well-supported argument, and you should be pleased to get one.
The test: pick one item and make them defend it. A real case has a best piece, and its owner is glad to be asked for it. A gallop has no best piece — press on any single item and you get a new item instead of a defence.
How to call it out
The only response that doesn’t lose is refusing to chase. Pick one item — ideally their strongest, not their weakest — and stay on it.
- “There’s a lot in there. Let’s do the biggest one properly. If I can’t answer that one, I’ll take the rest seriously.”
- “I’m not going to answer fourteen things badly. Pick your best.”
- “That was ten claims in ninety seconds. Any one of them takes me about ten minutes to check. Which one do you want?”
Say the last one to the room, not to the person — because the room is where this move is actually aimed.
And accept the cost up front: you are going to look like you’re dodging, at least for a minute. Do it anyway. Answering everything badly looks worse than answering one thing well, and it’s the outcome the gallop is built to produce.