Author: Indigo Curnick
Date: 2026-05-09
I haven't seen "magic words" as a common cognitive bias, but I thought it would be useful to expose this one. I also call this one "bargaining with reality". I can't decide which name is better! I like them both very much.
Magic words is the cognitive bias of thinking words warp reality, rather than understanding that words can be a description of reality when used correctly. Of course, used incorrectly, words won't describe reality. In that case, they will be misleading.
This is most close to the idea of the "map is not the territory"; the conceptual model has diverged from reality. However, what makes magic words distinct is the idea that the words are literally warping and modifying reality.
What are some examples of the magic words bias?
The most classic example of the magic words fallacy is prioritisation in the office. So common it's now a trope: presenting a manager with several options and asking what's high priority. The answer: "they're all high priority". This is a fundamental failure to understand the nature of ordinal numbers; the manager is engaging in magic words. "If everything is priority, nothing is" is mathematically unavoidable. Prioritisation is about ordering under scarcity, and saying words won't generate abundance. Suppose that the capacity of a team is N, if work sums to more than N, then at least one of the following will happen:
- Scope shrinks (actually, this is a legitimate solution. You could get everything by trimming the fat from tasks. Likely a lot of stuff you don't actually need!)
- Deadlines move (most common)
- Quality drops (do NOT do this! It's tempting but in the long run it's a bad, bad idea)
- People burn out (if you're a manager do NOT let this happen! This will kill every project ever)
A weirdly common magic word fallacy is villains in films. You see it all the time in James Bond style films, something like
Evil Boss: "How long will this take?"
Scientist: "Four hours"
Evil Boss: "You've got thirty minutes"
The implication being that the scientist underlying only has 30 minutes before the villain kills them, thus encouraging them to speed up. The villain is engaging in magic thinking (maybe why villains always lose?). Threatening someone doesn't actually make it happen any faster. If anything, it does mean they'll be so nervous as to make more mistakes. I usually don't watch many films because whenever I see this trope I roll my eyes.
One villain who doesn't do this, which makes him much more compelling and interesting is Saruman from Lord of the Rings. In the film, he converses with an Orc (the fanbase has dubbed him "Union Rep Orc"):
Saruman: "I want them armed and ready to march in two weeks."
Union Rep Orc: "But my lord, there are too many. They can not all be armed in time. We don't have the moons."
Saruman: "Build a dam, block the stream, work the furnaces: night and day."
Union Rep Orc: "We don't have enough fuel to feed the fires."
Saruman: "The forest of Fangorn lies on our doorstep. Burn it."
Union Rep Orc (menacingly): "Yes!"
I'm glad the workers and managers could work together to find a solution which solves everyone's problems. Competent villains are much more threatening than goofy magic thinkers.
Politicians are also prone to this. Without using specific examples to keep the blog neutral, I'm sure we can all think of a time a politician said something, but did not actually do something about it. In order for a wish to become real, something has to happen. In the case of a politician, that would be passing effective legislation, actually enforcing it and so on.
Security by policy is a classic. I remember once asking a system admin what was preventing users from doing something "well that would break policy". Okay, so nothing. Just having policies without at least enforcement or consequences is doesn't do anything. Even if you do have consequences, never underestimate the human ability to disregard the possibility of consequences (or simply not be able to keep up with policy). Rather than say "use secure passwords" a better design is to prevent a system from accepting passwords under 12 characters, or something like that.
AI is another place where this crops up. The idea here is "just throw an AI at it". If the solution to a problem is "just put it through AI", without specifying training data gathering, training data processing, compute, and evaluation, and also whether it's even theoretically possible for an AI to arrive at the desired output from the input this is just a magic word. This kind of thought pattern is only ever exhibited in people who don't know how AI works: people who have never trained an AI, never manually spent hours labelling data, never done a principal component analysis. To them, AI is magic. In reality, it's compute!
Another example is "jinxing" things. Obviously, saying "this is going well" does not impact the outcome of events. However, the belief is that the words are somehow magic and will. I suppose this is a kind of inverse magic words fallacy: the words make the opposite of reality. (This one is a tiny bit subtle, since if people believe in the fallacy they might deliberately or accidentally mess up upon hearing a jinx.)
Another way to look at this is that the magic words bias treats words as causal. So much of my intellectual career has been spent hammering this home: reality comes first; words come second. With Saruman, just saying "work the furnaces" doesn't magically make them burn. Saruman was able to realise this, and found a suitable fuel source.
Why do humans sometimes think that words shape reality? I think it's because humans are very social animals, and to some extent our words do have impacts, if indirectly. Words have the power to make other people do things, and so leading to a result. I see magic thinking the most in people like managers, politicians and influencers. If you're someone who only tells people to do things, you don't see the action. From your perspective, you say words and then sometime later "magically" a result has happened. The problem comes when you ask for something that can't be done. Since you've never physically done anything, the idea that action has to actually be taken is alien. I guess it's confusion on that part as to why the words no longer work. They never did.
How can you avoid the magic word fallacy? Honestly just listening is the best. I mean truly genuinely listen.
Failing that, I recommend doing things. Take on long term projects. Good projects to try are writing and publishing a book yourself, crochet or knit some clothing, do a large programming project like making a video game. Anything very difficult where you have to physically do things for a long period of time in order to get to the end result. Cold, hard reality will destroy magic words.