A smart character feels smart when the reader can follow their reasoning, not when they reach a conclusion no one else could. The trick is not raising the character's IQ; it is lowering the gap between what they notice and what the reader is allowed to notice with them. Watching a mind work is the payoff, not watching a mind win.
This piece is for writers who have decided that one of their POV characters is going to be the sharpest mind in the room and are now staring at a blank page wondering how to actually pull that off without sounding like they cheated. There are no universal rules here, and just about everything I say below has been successfully contradicted by some published author somewhere, but I think the patterns underneath good "smart character" writing are consistent enough across genres to be worth naming. The examples are drawn from fantasy and crime, because that is where the trope shows up most often, but the techniques apply equally to a contemporary literary novel where a single character needs to feel two beats ahead of everyone else.
Why "intelligent character" writing usually fails on the page
Most attempts at a genius protagonist collapse because the writer mistakes intelligence for prescience. The character announces a conclusion that no one in the scene, and crucially no one in the audience, could have reasonably reached, and then the narrative rewards them for being right. This is not intelligence; it is omniscience laundered through a character's mouth. Sherlock Holmes is the textbook offender people invoke, but the books actually mostly avoid this; Conan Doyle shows Holmes pointing at a callus and a tan line and reasoning out loud, even if he does it faster than any human ever could. The bad imitations of Holmes are the ones that turned the deductions into a magic trick, and that magic-trick template is what gets copied badly across modern fantasy, mystery, and progression fantasy alike. I think the diagnostic question to ask yourself, before any technique, is this: if you stripped the conclusion out and gave the reader only the inputs the character had, could a careful reader plausibly get within striking distance of the same answer? If no, you are not writing intelligence. You are writing a deus ex machina with a smug delivery.
The other failure mode is the calculator. This is the character whose intelligence is demonstrated by reciting prime numbers, doing chess in their head, or running probabilities mid-sentence. It is a static trait, not an active one. The reader registers "smart" and then waits, because nothing about that calculator brain is going to actually do anything in the story until the next puzzle arrives. Tony Stark works as a smart character not because he is canonically a genius but because his intelligence shows up as restless modification; he is always rebuilding the suit mid-fight, always taking apart what he just built. The trait does work inside scenes. A static genius is a label; a working genius is a verb.
Show the reasoning process, not the conclusion
The single most useful technique is to externalize the steps. A reader who watches a character notice, then connect, then test, then commit, will accept a leap at the end of that chain that would have looked absurd in isolation. This is the Holmes pattern done honestly, and it is also why Tyrion Lannister in the early A Song of Ice and Fire books reads as believably brilliant. George R. R. Martin lets you see Tyrion notice that someone said the wrong name, then turn it over, then test it with a probing question, then commit to the implication. By the time he plays his card you are nodding along, because you got to walk most of the road with him. My feeling about this is that you do not actually need to slow the prose down much to do it; you can compress the chain into half a paragraph of close third, as long as each link is visible. He had heard the steward say 'my lord' to the wrong man at supper, and the slip had not left him; if Lyman knew the boy was the heir, then Lyman was no neutral party, and the letter he had pressed into Tyrion's hand that morning was not a courtesy. That sentence is doing the entire trick: three observations, one inference, one commitment, no magic.
The corollary is that you should occasionally let the smart character be wrong about a step on the way to being right about the conclusion. Locke Lamora in The Lies of Locke Lamora does this constantly. He plans, the plan goes sideways, he improvises, his improvisation has the seams showing, and yet the overall caper lands. Scott Lynch is showing you that Locke's intelligence is adaptive rather than predictive, which is much closer to how real cleverness functions and much harder for the reader to dismiss as authorial cheating. A smart character who is never wrong is a savant; a smart character who is wrong about details and right about the shape is a person.
Build intelligence out of attention, not output
A trick I think is underused is to spend the character's POV time on what they are noticing rather than on what they are concluding. Hilary Mantel's Thomas Cromwell in Wolf Hall is the cleanest case study I know. Page after page, Mantel describes the room Cromwell is in: who is angled toward whom, whose hands are still, what was said before the meeting that the speaker is now contradicting, who has not eaten. None of these observations are flagged as clever. They simply accumulate, until you realize, about fifty pages in, that you have been watching the most ruthless mind in the kingdom assemble its model of the room in real time, and the actual maneuvers later in the chapter are almost an afterthought. Cromwell never tells you he is intelligent. The prose is the intelligence.
This is the same move Terry Pratchett used for Lord Vetinari across the Discworld books, although in a tighter package. Vetinari rarely explains himself; he just answers a question two beats before it would have been asked, or directs a meeting toward a conclusion the other characters did not realize was on offer. The reader extrapolates the model behind the moves, and the gap between what is on the page and what must be in his head is the entire effect. My advice, then, is that if you want to write a character who is smarter than you, do not try to write their conclusions. Write their attention. The conclusion can be one short sentence; the attention is where the page count goes.
Give them a domain, and give them friction inside it
Intelligence reads as flat when it is universal. The character who is the best swordsman, the best politician, the best alchemist, and the best reader of people in the room comes across as a writer's mascot, not a person. Glokta in Joe Abercrombie's First Law trilogy is brilliant at reading interrogation rooms and political alliances, and visibly mediocre at managing his own body, his finances, and his social standing. The asymmetry is what sells the character; we believe he is a frighteningly intelligent interrogator partly because we watch him fail at things any competent adult should be able to manage. Kaz Brekker in Leigh Bardugo's Six of Crows gets the same treatment in a different key. He outplans everyone in Ketterdam and cannot bear to be touched. The intelligence is not weakened by the limitation; it is anchored by it.
This is also where the "smartest person alive" framing tends to wreck a book before it starts. If the character is the apex of every relevant skill, there is no axis on which they can struggle, and without struggle the intelligence has nothing to push against. Better, I think, to pick a domain. Cromwell is brilliant about people and procedure, and indifferent to theology. Tyrion is brilliant about politics and bad at reading the people who actually love him. The domain gives the intelligence a shape, and the shape gives the reader something to anticipate.
Let other characters be smart too, then beat them honestly
A genius surrounded by incompetents is not a genius. They are an author surrounded by puppets. The fastest way to make a clever protagonist look unearned is to give them a foil who is described as cunning, and then have that foil consistently miss things any moderately careful reader would catch. The reader sees the gap, the reader stops believing the foil was ever a threat, and once the foil is downgraded the protagonist's wins downgrade with it. The way around this is to write the supporting cast as competently as the main character, and then let the main character win by margins. Orson Scott Card does this in Ender's Game; the other kids at Battle School are not stupid, they are very good, and Ender's victories read as victories specifically because the people he is beating are also drawn as smart. The same goes for the Bene Gesserit and the Mentats around Paul in Dune; Frank Herbert is careful to give every faction a credible plan before he lets Paul break it.
If you write the foils honestly and your protagonist still wins, the wins land. If you write the foils as cardboard, no amount of clever dialogue is going to save the effect. This is one of the cases where the discipline is mostly about restraint, in that you do not let your protagonist score off characters who never had a chance, even if it would be funny.
A short closing rule of thumb
The reason all of the above works, when it does, is that intelligence on the page is a relationship between what the character notices, what the supporting cast notices, and what the reader is permitted to notice along with them. Move any of those three and the effect shifts. Hide the reasoning and you get magic. Hide the supporting cast's competence and you get a coronation. Hide nothing from the reader and you get the rare, addictive sensation of watching a real mind at work, which is the thing you actually wanted in the first place. I think most writers can in fact write a character smarter than they are, but only by spending less time on the conclusions and more time on the small accumulating observations that make a conclusion feel inevitable in retrospect. Everything else, the wit, the chess metaphors, the dry one-liners, is decoration on that core.
Common questions about writing intelligent characters
How do you write a smart character if you are not that smart yourself?
You write the reasoning instead of the answer. A character who is observed working through a problem in visible steps will read as more intelligent than a character who blurts the answer, even if the answer is the same one you arrived at while plotting. You have the advantage of unlimited time at the page, so the character appears to do in seconds what you can construct over days, and the reader experiences the construction as live thinking.
What is the difference between a smart character and a Mary Sue?
The smart character has a domain and visible costs; the Mary Sue is universally competent and frictionless. If your character is the best at everything that matters in the story and nothing in their life is harder than it should be, you are writing competence porn, not intelligence. Glokta cannot climb stairs, Cromwell cannot stop the king from being the king, Tyrion cannot make his father love him. The asymmetry is what saves the portrait.
How do you avoid the "Sherlock effect" where the genius just announces conclusions?
Slow down the leap and show two or three of the observations that fed it. Even one inline sentence of the cuff was frayed where a left-hander would fray it, and the man at the table held his fork in his right converts a magic-trick conclusion into a deduction the reader can follow. The point is not to be exhaustive; it is to make the inference auditable.
Should an intelligent character ever be wrong?
Yes, and ideally about the details rather than the shape. Locke Lamora is wrong about almost every plan he makes and right about most of the cons; that combination of tactical error and strategic correctness is what makes him feel like a real clever person rather than a plot device. A character who is never wrong is not intelligent. They are the author wearing a costume.
Do I need to outline a character's reasoning in advance, or can I write it in revision?
I write the conclusion first and the reasoning in revision, almost every time. The conclusion is a writerly choice about plot; the reasoning is a craft layer added on top once you know what the character was supposed to figure out. The trick is to give yourself permission to do both passes, and not to let the first draft's missing reasoning convince you the scene does not work. It almost always works once the chain is in.
by Jacob Tam · May 20, 2026
I run IlorisNovel, a platform for writers of web fiction. If this kind of craft writing is your thing, the rest of the blog lives here.