The Science of Typos: Why Your Fingers Hit the Wrong Keys

You have typed “teh” instead of “the” a thousand times. You have typed “adn” instead of “and,” “thr” instead of “the,” and “wrold” instead of “world.” You are not careless. You are not a bad typist. Your fingers are obeying the laws of physics, and the errors they produce are as predictable as the orbit of a planet—if you know the math.

Typing errors are not random. They are governed by motor control noise, key proximity, finger biomechanics, and the physical properties of the input device. Decades of research in human-computer interaction have documented these patterns with rigorous precision, and the findings overturn most of what people assume about why typos happen.

The Motor Control Explanation

Every keystroke begins as an intention in the brain. The motor cortex plans the movement, sends signals through the spinal cord to the arm, hand, and fingers, and a specific finger travels to a specific key. The problem is that this signal chain is noisy. The human motor system does not execute movements with mechanical precision. It operates with variability—what researchers call motor control noise.

This noise means that your finger does not land exactly where your brain intended it to land. It lands near that position, within a probability distribution centered on the target key. Most of the time, the error is small enough that you still hit the right key. But sometimes the finger drifts far enough to land on an adjacent key instead. This is why the vast majority of typing errors are adjacent-key substitutions, not random character replacements.

Research from the CHI 2025 conference confirms this model. Shi et al., in their paper “Simulating Errors in Touchscreen Typing,” developed computational simulations of finger movement noise during typing. Their model treats each keystroke as a targeting action subject to Gaussian noise in finger position. The further a neighboring key is from the target, the less likely it is to be hit accidentally. The closer it is, the more likely. This simple principle—proximity governs error probability—explains the overwhelming majority of human typing errors.

Why Adjacent Keys, Not Random Characters

Think about the physical layout of a keyboard. On a standard QWERTY layout, the letter “e” is surrounded by “w,” “r,” “d,” and “s.” When your finger aims for “e” and drifts, it will almost certainly land on one of those four neighbors. It will never land on “m” or “p” or “z”—those keys are too far away for a small motor error to reach them.

This is not intuitive to most people. When asked to imagine a typo, many people picture a completely wrong character appearing—as if the keyboard were a roulette wheel and any letter could come up. In reality, the keyboard is a physical space, and errors are local within that space. Research by Pereira et al. (2013), published in the journal Human Factors, demonstrated that reducing key spacing by just a few millimeters significantly increased error rates, precisely because smaller gaps make adjacent-key interference more likely.

The implication is profound: typing errors carry spatial information. You can look at a typo and, in many cases, determine which key the typist was aiming for based on which wrong key they hit. “thr” means the typist aimed for “e” and hit “r” (right neighbor). “thw” means they hit “w” (left neighbor). Each error is a fingerprint of a physical movement that went slightly off course.

What 136 Million Keystrokes Revealed

The most comprehensive typing study ever conducted was published at CHI 2018 by Dhakal et al. at Aalto University. They recruited 168,000 volunteers through an online typing test and recorded over 136 million individual keystrokes. The dataset is unprecedented in scale, and its findings have reshaped our understanding of how people type.

Eight Distinct Clusters of Typists

The study identified eight distinct typing strategies, ranging from two-finger hunt-and-peck to fluid ten-finger touch typing. But the most surprising finding was that conventional typing classes do not determine typing speed. Many fast typists had never taken a typing course. They had developed their own idiosyncratic techniques—using six fingers, or eight fingers, or unique hand positions that no typing instructor would recognize.

What separates fast typists from slow typists is not which method they learned but two specific factors: the number of fingers they use (fast typists average 8.4 fingers versus 5.3 for slow typists) and the degree to which they alternate hands between keystrokes. Typists who frequently alternate hands type faster because the unused hand can pre-position for the next key while the active hand is still pressing.

Speed, Accuracy, and Error Patterns

Here is where it gets interesting for understanding typos. The study found that fast typists do not simply make fewer errors—they make different errors. Typists who use more fingers and type faster tend to produce more rollover errors (pressing the next key before fully releasing the previous one) and transposition errors (swapping two adjacent characters). Slower typists produce more substitution errors (hitting the wrong key entirely).

The study also documented that transposition errors are strongly bimanual. When two characters get swapped, it happens most often when the two characters are typed by different hands. The leading hand hits its key slightly too early, before the other hand finishes its keystroke. This is not a random phenomenon—it is a coordination failure between two limbs that are executing movements in parallel.

The Speed-Accuracy Tradeoff

One of the most fundamental principles in motor control is the speed-accuracy tradeoff, formalized by Fitts’ Law. The faster you move, the less accurate your movement becomes. This applies directly to typing: the faster you type, the more motor noise is introduced into each keystroke, and the more likely your finger is to drift to an adjacent key.

This tradeoff is not linear. At low typing speeds, accuracy is high and errors are rare. As speed increases, errors increase gradually at first, then more steeply. There is a sweet spot where typists achieve a comfortable balance between speed and accuracy, and most people unconsciously settle into this zone. Push past it—because you are in a hurry, because you are emotionally agitated, because you are fatigued—and error rates climb sharply.

Research by Pimenta et al. (2020), published in PLOS ONE, tracked office workers over six weeks and found that typing performance degrades measurably over the course of a workday. Afternoon typing is slower and more error-prone than morning typing. Mental fatigue manifests directly in finger control. The errors are not random—they follow the same adjacent-key pattern but occur more frequently as the day wears on.

The Four Categories of Typing Errors

Research across multiple studies has identified four major categories of typing errors, ordered roughly by how frequently they occur:

Substitution Errors

The most common type. Your finger hits a neighboring key instead of the intended one. “the” becomes “thr” or “tue.” The substituted character is almost always physically adjacent to the correct one on the keyboard layout. This is the direct consequence of motor control noise acting on finger position.

Insertion Errors

An extra character appears in the output, usually adjacent to one of the surrounding keys. This often happens during fast typing when a finger brushes a neighboring key during transit. “the” becomes “thre” or “tthe.” Doubled characters (pressing the same key twice) are a common subtype.

Omission Errors

A character is missing from the output. The finger aimed for the key but did not press it with enough force, or the keystroke was not registered. This is more common on touchscreens, where the absence of physical key travel makes it harder to confirm that a press was registered. “the” becomes “te” or “th.”

Transposition Errors

Two adjacent characters are swapped. “the” becomes “teh,” “and” becomes “adn.” This is the error type most people think of as a “typo.” As noted from the 136-million-keystroke study, the majority of transpositions occur across hands—the fingers on different hands lose synchronization.

Beyond these four primary categories, there are secondary error types that emerge from the interaction between device, speed, and context: spacing errors (missing or extra spaces), punctuation errors (wrong or missing punctuation), word-level errors (repeated or omitted words), and capitalization errors (accidental shift key activation). Each follows its own physical logic, and each varies in frequency based on the device, the typist’s speed, and their emotional and cognitive state.

Device Physics: Why Your Phone Betrays You

Everything discussed so far applies to physical keyboards, but the physics shift dramatically on touchscreens. A study of over 37,000 volunteers found that smartphone users make roughly five times more typing errors than desktop keyboard users. The average mobile typing speed is around 36 words per minute with a 2.3% uncorrected error rate—compared to desktop typing at 52 WPM with roughly 0.5% uncorrected errors.

The reasons are physical. On a touchscreen, your thumb or finger pad covers a much larger area than a single key. The lack of physical key boundaries means there is no tactile feedback to confirm which key you are pressing. Portrait-mode phone keyboards compress the key layout into a space too small for the human finger to target precisely. The motor noise that causes adjacent-key errors on physical keyboards is amplified on touchscreens because the target area is smaller relative to finger size.

Tablets occupy a middle ground. The larger screen provides bigger key targets, but the hand position is different from both phones and keyboards. Tablet typing tends to produce more omission errors (the flat surface provides ambiguous feedback about whether a key was pressed) and more spacing errors (the spacebar is harder to isolate on a flat surface).

What This Means

The science is unambiguous: typing errors are not random events. They are the predictable output of a physical system—human fingers interacting with physical or virtual keyboards under the influence of motor control noise, speed-accuracy tradeoffs, fatigue, emotion, and device form factor. Every error carries information about the physical act that produced it.

This is why random character mutation produces unconvincing errors. If you replace characters arbitrarily, you are ignoring the physics that generate real errors. A “p” replacing an “a” is physically implausible on any standard keyboard layout—those keys are nowhere near each other. But a “s” replacing an “a” is perfectly natural—they are adjacent on QWERTY. The difference between realistic and unrealistic errors is the difference between physics and randomness.

LikelyTypo models these physics directly. It calculates key adjacency on the actual keyboard layout, applies device-specific touch radius models, and generates errors whose distribution matches what the research describes. The result is errors that look like they came from a real person on a real device—because they follow the same rules that govern real human typing.

The next time you type “teh” instead of “the,” know that it was not your fault. Your fingers were executing a motor plan with sub-millimeter precision requirements, on a device whose keys are barely larger than your fingertips, at a speed that pushes the limits of neuromuscular coordination. The real surprise is not that you make typos. The real surprise is that you do not make more of them.