People throw too much magic in their explanations. How does an individual detect magic in their explanations?
What is Magic?
I think of certain technical jargon as “magic” because its intended function doesn’t educate the audience and often acts as a curiosity stopper. I call these words Magic Words because you can substitute any magical term with the term “magic” and still walk away with the same level of understanding.
To use jargon and acronyms that people don’t understand can cause poor side effects. Using Magic Words can make people feel dumb for asking and mental models built on Magic Words as premises can be faulty and lead to errors.
We’ll talk about how technical words can sometimes unintentionally become Magic Words and how we can detect it and combat it.
Magic Words as Knowledge Asymmetry
Technical words abstract complexity so that two people with the same level of understanding of something can use the word as a shorthand in conversation. Technical words become magical words when there is there is no previous agreed-on understanding of what a word means.
When someone learns a new word without given a formal definition, they tend to contextually guess what the word means, which can lead to poor abstractions of the word and a faulty mental model. For example if I don’t understand what the acronym ACID means and heard this sentence:
“Two-phase commit maintains ACID properties so we don’t have to worry about data integrity.”
Neophytes to the definition would think “ACID” is all about data integrity (the “C” in consistency) without understanding the other three (A – Atomicity, I – Isolation, D – Durability).
It takes more effort to assume knowledge asymmetry than to assume everyone is on the same page as you.
Magic Words as Curiosity Stoppers
Caching and distributed systems are explanations that often come up when a bug surfaces on engineering teams. It is true that sometimes these do cause the issues we experience, but I often think we are too quick to forego investigation because we have a scapegoat explanation. The problem with using “caching” and “distributed systems” as explanations is that we tend to be satisfied with the answers we come up with without confirming our hypotheses. “Caching” and “failed requests” have acted as curiosity stoppers.
For example, describing your bug as “transient” makes you look smart and gives you a lot of space for your bugs to exist, but doesn’t really help better our understanding of a situation. Anything that’s confusing to us we blame distributed systems, caches, and bad networks without continuing our investigation to further our understanding.
Magic Words as Intuition, not Understanding
The feeling of understanding often precedes real understanding, and Magic Words often mask misunderstanding behind words. If we don’t know something we put a label on so we can connect a word to an external object. The problem is that we like to shove everything we don’t know into a magical word, and use it as if we knew what it meant. “Microservices” may sound like a sensible term, but unless you know what how to differentiate it from a regular service, the terminology means nothing and adds no value in a conversation.
Magic Words as Pulls
In social interactions, “pulling” is a maneuver that prompts other parties to ask a question that a person wants to answer. An obvious example:
Me: Hey! Do you have a twin sibling?
Friend: No. Do you?
Me: Yes! Of course, Robert is the best twin brother I could ask for. We went to the beach yesterday …
In this example, I wanted to talk about my twin brother, but have no avenue to talk about it without prompting. So I use a pull to guide the conversation in that direction so I could talk about it.
Sometimes explainers will also have information they want to present, but have no prompting question to dive further into a particular area. So they use Magic Words as pulls to prompt a discussion or explanation. Assuming you don’t know anything about Cassandra, could you detect the magic in this sentence?
“Oh yeah, even though data is denormalized across multiple tables, you can use batch queries to maintain consistency across multiple tables.”
As an explainer myself in this situation, I would love to talk about batch queries and how they work, but if people treat “batch queries” as an explanation and not a Magic Word, then they won’t ask the question I intended on answering. I’m not even given the opportunity to explain something that can clear confusion, even though I really, really want to.
Magic Words as Attire
Like how we dress up ourselves to impress others, it’s very natural for people to dress up their explanations with Magic Words. It makes them look smart and cool. Not only is this a disservice to your explainee, it can set off a condescending tone, which people will definitely not appreciate.
How I’m combating Magic Words
Once I detect magic, the way I’m combating Magic Words is by focusing on the magical word. Some phrases I like to use are:
“What do you mean by
“Could you substitute the definition for the word here?”
<magic word>, do you mean
<my understanding of magic word>.”
“When you say , I’m thinking . Is that correct?”
To combat looking or feeling dumb, I remind myself that I value truth and understanding more than I care about looking dumb or feeling insecure. I adopt an omnivorous attitude towards valuable information. I don’t really care why someone is using Magic Words. I’ll extract knowledge in whatever form I receive it in.
As much of a problem magic words are, I want to re-emphasize that using Magic Words is natural and easy, and I’m not faulting individuals who use them. Let’s all be better at detecting magic so we can help each other become better thinkers!
“Mysterious Answers” — Good sequence of articles that codify lots of different types of magical explanations.
“Semantic Diffusion” — When words are used in more and more unsuitable contexts, it suffers from semantic diffusion. The worst transgression of this (in my opinion) is the term “love.””Love” is used for anything and everything that the word by itself has very little precision in meaning.