Gish gallop

The three-minute limit

At a neighborhood association meeting, members debate replacing the community pool’s chlorine system with a saltwater one. Each side gets three minutes. One speaker uses his three minutes to list fifteen objections: saltwater corrodes every metal fitting, it costs three times as much to run, it gives children rashes, the county has banned it, the pool would have to close for a whole summer, the insurer won’t cover it, and nine more. The other speaker spends her three minutes answering the first three. In his closing, he says: “Notice that she couldn’t answer twelve of my points.”

Some of the fifteen may be true; several are the kind of claim that takes a phone call or a document to check. The format guarantees that most can’t be answered, and the closing line turns that into apparent evidence. “She didn’t answer it” is a fact about three minutes, not about saltwater pools. This is the clear-cut case.

The document before the vote

The day before a team votes on switching project-management tools, a colleague circulates a document titled “Forty problems with the new tool.” Two of the items are serious: the tool can’t export data, and it lacks a permission setting the team needs. The other thirty-eight are minor, duplicated or unchecked (“I heard it’s slow on some laptops”). At the meeting he says: “Nobody has addressed most of these, so we’re clearly not ready to switch.”

No one is talking fast, but the same asymmetry is at work: forty items take an evening to write down and days to check, and the vote is tomorrow. This version is easy to miss because a long list looks thorough. It also shows a cost that falls on everyone, the colleague included: the two serious problems, which deserved a real answer, are buried among the thirty-eight that didn’t.

• Spoken: rapid delivery in a timed exchange, where the clock does the work.
• Written: a long list of objections, links or questions, sent where the other side has too little time or attention to go through them.
• Sealioning: Mark Satta describes this related tactic as repeatedly and unreasonably demanding evidence for even ordinary claims. The gallop overloads the other side with claims to answer; sealioning overloads them with requests to fill.

Satta groups both with other “burden-shifting” tactics that push an unfair share of the work onto the other person. The asymmetry they exploit is often summed up as Brandolini’s law, attributed to Alberto Brandolini and also called the “bullshit asymmetry principle”: the energy needed to refute nonsense is an order of magnitude bigger than the energy needed to produce it. Satta quotes it from a 2016 column in Nature by Phil Williamson.

Named after a person. The term was coined by the anthropologist Eugenie Scott, then executive director of the National Center for Science Education, after Duane Gish (1921–2013), a biochemist and vice president of the Institute for Creation Research who took part in many public debates with scientists about evolution. Scott used it for unstructured debate formats that allow the “presentation of misconception after misconception”.

A tactic rather than a classical fallacy. Most entries on this site describe a pattern of reasoning that fails however it’s delivered. The Gish gallop is a way of conducting a debate, and it needn’t contain any single bad inference. It is listed with the informal fallacies because it works by inviting two that are: treating an unanswered point as conceded, and treating the number of arguments as a measure of their strength.

Making many points, or making them quickly, isn’t a Gish gallop in itself.

• Many reasons that are each good. Some cases really do rest on a long list of independent reasons, and each good one adds weight. What matters is whether the points hold up one at a time.
• A quick summary backed up elsewhere. An opening statement that runs through points in brief, each supported in a handout, a report or earlier discussion, lets the other side check them later.
• A format with time to respond. A long written critique sent weeks before a decision gives the other side a fair chance to answer it.
• Unanswered points aren’t counted as conceded. A speaker who says “I don’t expect you to answer all of these now” isn’t relying on the clock.

The test: if the other side had all the time it needed, would each point survive checking, and is the speaker relying on them not having that time?

Twelve reasons with a handout

At the same pool meeting, a member who supports the saltwater system lists twelve reasons in her three minutes. She hands out a one-page sheet with the maintenance log, two contractors’ quotes, and the name of a nearby pool that switched last year, and says: “Any of these can be checked, and I’m happy to take questions on any of them afterward.”

Twelve points in three minutes looks like the first example. But each point is backed by something the audience can check, she doesn’t treat silence as agreement, and she invites follow-up. This is the backed up elsewhere and not counted as conceded conditions.

The closing recap

At the end of a two-hour planning session, the facilitator runs through ten decisions in one minute: “We agreed on the budget cap, the vendor, the launch date, ...”

That’s a lot of points delivered fast, but each was discussed at length already, and the recap exists so people can object if one is wrong. Speed here serves checking instead of preventing it.

The tactic works because of an asymmetry in effort. Satta puts it plainly: rattling off multiple claims is easy, and responding point by point is difficult, especially when a useful response requires research. A speaker who ignores accuracy pays nothing for a false claim, while the person answering pays for every one.

The audience’s position makes it worse. Watching a live exchange, most people can’t check claims either, so what they see is one speaker who made many points and another who answered few, and the unanswered points look like concessions.

It doesn’t always need bad faith. Someone who has collected every objection they’ve ever heard, and has never sorted the strong ones from the weak, can gallop sincerely.

None of these has been tested experimentally as a remedy. Where a suggestion comes from an argumentation theorist or philosopher, it says so.

• Say openly that you won’t answer everything, and why: “There are more points here than anyone could check in three minutes. I’ll take the one that matters most.”
• Answer the strongest point, not the easiest. Picking off the weakest items and implying you’ve dealt with the rest is its own mistake (see Straw man). If the strongest point holds, say so.
• Ask for one claim at a time. “Which of these would you most want to stand on? Let’s look at that one.” A case that rests on its best point can be discussed; a pile of twenty can’t.
• Group what can be grouped. Krabbe and van Laar describe clustering minor objections and responding to what unites them as one strategy against the Gish gallop. Several “it’s too expensive” items may be one question about cost.
• Check a few. If two or three points picked at random turn out to be wrong, that’s a fair reason to lower your confidence in the rest. It isn’t proof that the rest are wrong too.
• Name the pattern without insulting the person. Satta suggests that one way to handle exhausting conversational tactics is to point out the dynamic itself, for example that one side is making claims faster than the other can check them.
• Choose where to spend your effort. Satta also recommends being selective: energy is limited, and not every exchange is worth it.
• Move to a format with time. “Can you send me the two points you think are strongest? I’ll look into them properly.” This takes the clock out of it, which matters most when the other person has more authority in the room than you do.