Typing Speed: From 1888 Contests to Modern WPM Calculations

The International Typewriting Contest of 1888 — won by Frank McGurrin at 95 WPM — established both competitive typing and the WPM standard, with 1 WPM defined as 5 keystrokes per minute based on the average English word length of the typewriter era. Gross WPM counts all keystrokes divided by time, while net WPM deducts errors, a distinction that produces dramatically different scores for the same typing performance. Text difficulty in typing tests is determined by word frequency, character diversity, and sentence complexity — standard test texts control these variables while adaptive tests intentionally vary them to measure different skills. Monkeytype (2020) calculates WPM by dividing keystrokes by elapsed time with per-key timing, TypeRacer (2007) measures time to complete a fixed text passage against opponents, and Keybr (2011) uses adaptive word generation that adjusts difficulty based on real-time error patterns — all three report different speeds for the same typist.

What Typing Speed Actually Measures

Typing speed appears straightforward at first glance: how many words can you type in a minute? But the moment you look under the hood, the simplicity unravels. The measurement depends on what you count as a word, how you handle errors, whether you include correction time, and what text you are typing in the first place.

At its core, typing speed quantifies the rate at which a person transcribes written text using a keyboard. The raw output is measured in keystrokes — each letter, space, punctuation mark, and modifier key press counts as a discrete input event. Those keystrokes are then converted into a speed metric using one of several competing formulas. The choice of formula fundamentally changes what the resulting number actually represents.

The most fundamental distinction is between gross speed and net speed. Gross speed, sometimes called raw speed, simply divides total keystrokes by elapsed time and then converts to words per minute using the standard of five keystrokes per word. Net speed starts with the same calculation but then applies a penalty for each error made during the test. The size of that penalty varies between testing systems, which means the same performance can yield different net speeds on different platforms.

Beyond the gross-versus-net question lies the unit of measurement itself. Words per minute remains the dominant unit in English-language typing tests, but characters per minute (CPM) is common in non-English contexts where average word length differs significantly. Keystrokes per minute (KPM) appears in data entry and transcription work where the content is not natural language. Each unit serves a different purpose and produces numbers that are not directly comparable without conversion.

Standardized typing tests attempt to control for these variables by using fixed text passages with known difficulty characteristics. When you take a typing test that claims to measure your "true" typing speed, you are actually measuring your performance on a specific text under a specific scoring algorithm on a specific platform. The number is meaningful for comparison only when the testing conditions are identical.

The 1888 International Typewriting Contest and WPM Origins

The history of competitive typing begins in the 1880s, a decade that saw the first typing contests emerge alongside the commercial spread of the typewriter. These early competitions were not the polished online affairs of today — they were public exhibitions held in auditoriums, with stenographers and clerks competing on identical machines under the watch of official judges. The atmosphere was closer to a sporting event than a skill assessment.

The pivotal moment came in 1888 with the International Typewriting Contest held in Toronto. The contest pitted typists using the newly popular QWERTY layout against those using competing keyboard arrangements, most notably the Caligraph. The winner was Frank McGurrin, a federal court clerk from Salt Lake City who had taught himself to touch-type on the QWERTY keyboard using what would become the modern home-row technique.

McGurrin's winning score was 95 words per minute, an astonishing speed for the era. At a time when most typists still used the hunt-and-peck method and rarely exceeded 40 to 50 WPM, McGurrin's touch-typing technique gave him a decisive advantage. He typed from dictation, not from a prepared text, which added an additional layer of difficulty — he had to listen, process, and transcribe simultaneously, a skill that even modern typists would find challenging under live competition conditions.

The 1888 contest did more than crown a champion. It established the standard for measuring typing speed that persists to this day. The International Association of Typing Examiners defined one word as five keystrokes, a convention based on the average length of English words in the typewriter era. This five-character standard was not arbitrary — it emerged from empirical studies of typical business correspondence and newspaper copy, which showed that the average English word, including the space that follows it, came to approximately five characters.

This definition has survived remarkably intact for over 130 years. Every modern typing test that reports words per minute uses the five-keystroke standard, whether the platform's developers know its history or not. The standard has been criticized for being imprecise — average English word length varies by genre, register, and time period — but no alternative has achieved broad adoption. The five-keystroke word is the meter of the typing world: an imperfect but universally accepted convention.

The 1880s contests also introduced the concept of accuracy scoring, though the methods were crude by modern standards. Judges counted uncorrected errors and applied penalties to the raw word count, a practice that directly prefigures the net WPM calculations used today. Contest rules typically required that errors be marked on the paper and that uncorrected errors result in a deduction of a fixed number of words from the final score. The parallel to modern net WPM calculations — which subtract a penalty per error from the gross word count — is unmistakable.

The legacy of these early contests extends beyond measurement standards. They established typing as a measurable, improvable skill worthy of competition and practice. Before the 1880s, typing was seen as a manual dexterity task with limited variation in ability. The contests proved that expert typists could outperform average ones by a factor of two or three, creating the incentive structure that eventually led to formal typing education and the modern typing test industry.

Gross WPM vs Net WPM: Two Scoring Systems

The distinction between gross WPM and net WPM is the single most important concept for understanding what a typing speed score actually means. These two systems answer different questions. Gross WPM asks "how fast can your fingers move?" Net WPM asks "how fast can you produce correct text?"

Gross WPM

Gross WPM, also called raw WPM, is calculated using a simple formula:

Gross WPM = (Total Keystrokes / 5) / Time in Minutes

Total keystrokes include every character typed — correct letters, incorrect letters, spaces, punctuation, and often even backspaces if the platform counts correction keystrokes. The division by five converts keystrokes into standard words. The division by time yields words per minute.

The advantage of gross WPM is that it captures peak finger speed with no penalty for mistakes. A typist who pounds out 600 keystrokes in one minute with 50 errors will have the same gross WPM as a typist who types 600 keystrokes with zero errors: both will show 120 gross WPM. This makes gross WPM a pure measure of raw physical speed.

The disadvantage is equally clear. Gross WPM tells you nothing about output quality. A gross WPM of 100 with 40 percent accuracy produces far less usable output than a gross WPM of 60 with 99 percent accuracy. In real-world typing tasks — writing emails, coding, transcribing — errors must be corrected, which consumes time and reduces effective throughput. Gross WPM ignores this reality.

Net WPM

Net WPM applies a penalty for errors, aiming to reflect the effective typing speed after accounting for mistakes. The most common formula is:

Net WPM = ((Total Keystrokes / 5) - Number of Errors) / Time in Minutes

In this formula, each error reduces the gross word count by one word before dividing by time. A typist who types 600 keystrokes in one minute with 50 errors would compute as follows: (120 words gross - 50 errors) / 1 minute = 70 net WPM. The same performance with zero errors would yield 120 net WPM.

Some platforms use a more aggressive formula that deducts the number of error words rather than individual errors. In this approach, each word containing a mistake is counted as an error word, and the total error words are subtracted from the gross word count. This produces a steeper penalty because a single error word might have been partially correct but is still discarded entirely.

Other platforms use a percentage-based approach, applying a penalty multiplier based on the accuracy rate. For example, a test with 95 percent accuracy might multiply the gross WPM by 0.95 to produce the net WPM. This method smooths out the penalty and avoids the discontinuities that can occur when a single error causes a discrete word deduction.

Which One Should You Use?

The choice between gross and net WPM depends entirely on the purpose of the measurement. For self-assessment and skill tracking within a single platform, either metric works as long as you apply it consistently. The trend over time matters more than the absolute number.

For comparing scores between platforms, gross WPM is slightly more reliable because the error penalty varies so widely between net WPM formulas. A score of 80 net WPM on one platform might correspond to 60 or 90 net WPM on another, depending on how errors are penalized. Gross WPM calculations are more standardized because they all use the same formula — though even here, differences in how keystrokes are counted (do backspaces count? does the space bar count?) can introduce variation.

For employer screening or certification, net WPM is standard because it reflects usable output. Most professional typing tests, including those used for court reporting, transcription, and data entry positions, report net WPM with specific penalty rules that are documented in advance so candidates know what to expect.

Text Difficulty: Why Word Lists Change Your Score

Two typing tests can report very different speeds for the same person, even when both use the same WPM formula. The reason is text difficulty. The content you type dramatically affects your achievable speed, and different tests use different content, making cross-test comparison unreliable without accounting for this factor.

Text difficulty in typing tests breaks down into three primary dimensions.

Word Frequency

Common English words — "the," "and," "that," "have," "with" — are typed faster than rare ones. This is not simply a familiarity effect at the cognitive level. Frequent words have more practiced motor sequences in the fingers. A skilled typist types "the" as a single ballistic movement pattern, whereas "thematic" requires individual letter-by-letter processing with pauses between syllables.

Typing tests that draw from a small vocabulary of common words will produce higher speeds than tests that use a broad dictionary. This is why a test based on the 200 most common English words can yield scores 10 to 20 WPM higher than a test using the full English lexicon at the same nominal difficulty level. The finger memory for common sequences reduces cognitive load and allows faster execution.

Character Diversity

The specific letters and letter combinations in the text matter enormously. Typing "asdf asdf asdf jkl; jkl; jkl;" is fast because all characters are on the home row and alternate between the two hands. Typing "qax zyp wuv mib crs" is slow because most characters require reaching to the top or bottom rows and the hand alternation pattern is broken.

Character-level difficulty factors include:

• Hand alternation: Text that alternates between left and right hand is typed faster than text that favors one hand. The word "lollipop" is slow because it stays on the right hand. "street" requires both hands and is faster.
• Row reach: Home row keys (ASDF JKL;) are fastest. Top row (QWERTYUIOP) is next. Bottom row (ZXCVBNM) is slowest for most typists.
• Bigram frequency: Common two-letter sequences like "th," "he," "an," "re," "in" have practiced motor programs. Rare bigrams like "xg," "zq," "vw" require individual keystroke planning.
• Same-finger bigrams: Sequences typed with the same finger (like "de" or "lo" on QWERTY) create timing bottlenecks because the same finger must strike two keys in sequence.

Sentence Complexity

At the sentence level, complexity affects reading speed, planning load, and error rates. Key factors include:

• Punctuation density: Sentences with multiple commas, semicolons, parentheses, and quotation marks slow typing because punctuation keys are on the periphery of the keyboard and require hand repositioning.
• Capitalization: Proper nouns, acronyms, and sentence-initial capitals require shift key coordination, adding cognitive overhead and physical timing complexity.
• Numeric content: Numbers require reaching to the top row and are typically typed slower than letters, especially when mixed with alphabetic text.

Standard typing tests attempt to control these factors by using curated text passages with known difficulty profiles. But the very act of controlling for difficulty introduces another problem: it limits the representativeness of the test. A typist who scores 100 WPM on standard test texts may score significantly lower on real-world text that contains unusual vocabulary, mixed case, punctuation, and numbers.

Standard Test Texts and Measurement Comparability

To make typing speed scores comparable across time, platforms, and individuals, the typing community has developed several standard test texts. These passages are carefully designed to have predictable word frequency, character diversity, and sentence complexity characteristics. They serve the same function as standardized test passages in reading assessments: they control for content difficulty so that differences in scores reflect differences in skill rather than differences in text.

The Most Common Standard Texts

The current standard corpus for most English typing tests draws from several sources:

The "common words" lists used by Monkeytype and similar platforms select words from the most frequent entries in large English corpora such as the Google Books Ngram dataset or the Corpus of Contemporary American English. These lists typically contain 200 to 1000 words, selected for frequency, length distribution, and character diversity. A 200-word list produces markedly easier tests than a 1000-word list because the restricted vocabulary creates repeated practice effects.

Standard paragraphs — fixed, published passages — are used by TypeRacer, Typing.com, and many certification tests. These passages are selected or written to have specific characteristics:

• Average word length of 4.5 to 5.5 characters
• Grade-level reading difficulty appropriate to the target population
• Balanced character distribution that exercises all keyboard rows
• Inclusion of common punctuation without overwhelming density

The most famous standard passage in typing history is "The quick brown fox jumps over the lazy dog," a pangram containing every letter of the alphabet. While used primarily for keyboard testing and display purposes rather than speed measurement, it established the concept of a representative test text.

Comparability Problems

Even with standard texts, comparability remains imperfect. Two platforms using the same word list may still produce different scores because of differences in:

• Keystroke counting: Does pressing backspace count as a keystroke? Does holding shift for capitalization count as one keystroke or two? Platforms differ.
• Word boundaries: Is a word considered correct only if every character matches, or are partial credit approaches used?
• Test duration: A 15-second test produces higher WPM than a 2-minute test because fatigue and concentration lapses accumulate with time. The same typist will typically score 10 to 20 percent higher on a 30-second test than on a 5-minute test.

The practical implication is that typing speed scores should be treated as platform-specific measurements. A score from one platform should not be directly compared to a score from another without understanding the methodological differences. The number is more useful as a trend indicator within a single platform than as an absolute measure of typing ability.

Platform Divergence: Monkeytype, TypeRacer, Keybr

The modern typing test landscape is dominated by three platforms, each with a fundamentally different approach to measuring speed. Understanding their differences is essential for interpreting what their scores actually mean.

Monkeytype: Timer-Based with Per-Key Precision

Monkeytype launched in 2020 and rapidly became the default typing test platform for enthusiasts and speed-focused typists. Its approach to WPM calculation is the purest expression of gross speed measurement among the major platforms.

Monkeytype calculates WPM by dividing total correct keystrokes by elapsed time in minutes, using the standard five-keystroke word. The timer starts on the first keystroke and runs continuously until the test duration expires — typically 15, 30, or 60 seconds. The key innovation is per-key timing: each keystroke's timestamp is recorded, allowing Monkeytype to display per-character speed variations and identify which keys or bigrams are slowest for the typist.

The platform uses local storage in the browser to save all test history, including raw key logs, accuracy data, and speed trends over time. This data never leaves the user's machine unless they explicitly export it. The local-storage approach means that test history is device-specific — switching browsers or clearing local storage resets the history, which can be frustrating for users who expect cloud persistence.

Monkeytype's text generation uses word lists of configurable length and source. The default list draws from 200 common English words, though users can select larger lists, custom texts, or programming languages with punctuation. The short-word-list default produces the high scores that Monkeytype is known for — many users report scores 10 to 20 WPM higher on Monkeytype than on other platforms.

One notable feature is Monkeytype's handling of errors. The platform offers both "stop on error" mode (where the test halts until the correct key is pressed) and "off the wagon" mode (where the test continues and all characters are counted). In the default mode, errors must be corrected to proceed, which means backspace keystrokes are included in the total keystroke count, reducing gross WPM for inaccurate typists.

TypeRacer: Competitive Multiplayer with Fixed Passages

TypeRacer launched in 2007 and pioneered competitive multiplayer typing. Unlike Monkeytype's solitary timer-based approach, TypeRacer pits multiple typists against each other on identical text passages.

TypeRacer's WPM calculation is fundamentally different from Monkeytype's. Instead of measuring keystrokes per minute, TypeRacer measures the time required to complete a fixed text passage. The WPM is calculated as:

WPM = (Total Characters in Passage / 5) / Completion Time in Minutes

This is a word-completion-based calculation — the numerator is fixed at the passage length, not the number of keystrokes the typist actually made. An error that is corrected still counts in the total completion time, but the passage character count remains the same. This means TypeRacer inherently penalizes errors through the time spent correcting them, but the WPM formula itself does not include a separate error penalty term.

The competitive format introduces psychological factors that affect performance. Racing against other typists creates pressure that can either improve performance through adrenaline or degrade it through anxiety. TypeRacer's user interface — showing opponents' cars advancing in real time — adds a visual urgency that is absent from solitary typing tests. Many typists report that their TypeRacer scores are lower than their Monkeytype scores even on similar text difficulty, partly because of this competitive pressure and partly because of the different calculation method.

TypeRacer's text passages are drawn from a curated library of published works — song lyrics, movie quotes, book excerpts, and famous speeches. These passages vary significantly in difficulty. A passage from Dr. Seuss is notably easier than a passage from a technical manual, which means a typist's score on TypeRacer depends partly on the luck of the passage draw. The platform attempts to normalize for this by showing a passage's average completion statistics, but individual race outcomes remain passage-dependent.

Keybr: AI-Adaptive with Skill-Focused Measurement

Keybr, launched in 2011, takes yet another approach. Rather than measuring peak speed on fixed text, Keybr adapts the text to the typist's weaknesses, generating words that target the letters and letter combinations where the typist makes the most errors.

Keybr's adaptive algorithm works by maintaining a statistical model of the typist's performance on each letter and bigram. When a typist consistently makes errors on a particular letter — say, pressing "x" instead of "z" — the algorithm increases the frequency of words containing that letter in the generated text. This creates a training loop: the typist is forced to practice exactly the skills they are weakest at, which produces rapid improvement in overall speed and accuracy over time.

The WPM calculation on Keybr uses a timer-based approach similar to Monkeytype, but with an important difference: Keybr measures the time between individual key presses and uses this to compute instantaneous speed, not just average speed over the test duration. This allows Keybr to identify which specific keystrokes are slow and which are fast, feeding this data back into the adaptive algorithm.

Keybr's scoring is more conservative than Monkeytype's. Because the generated text is deliberately skewed toward the typist's weak areas, the text difficulty is higher than a random word list of the same length. A typist who scores 80 WPM on Keybr might score 100 WPM or higher on Monkeytype, not because of a calculation difference but because Keybr's adaptive text is genuinely harder to type.

Keybr does not use error penalties in its displayed WPM — it shows gross speed based on correct keystrokes divided by time. However, the adaptive algorithm internally tracks accuracy for each letter and uses this data to adjust text generation. The platform also displays a separate accuracy percentage, and the relationship between speed and accuracy is explicitly visualized to show the typist how their speed changes as they focus on accuracy versus raw speed.

Mobile vs Desktop Speed Differences

A factor that all three platforms handle poorly is the difference between mobile and desktop typing performance. Touchscreen typing on a mobile device produces speeds that are typically 30 to 50 percent lower than physical keyboard typing, even for experienced mobile typists. The lack of tactile feedback, the smaller key targets, and the autocorrect behavior of mobile keyboards all fundamentally change the typing task.

Some platforms attempt to account for this by offering mobile-specific modes with adjusted scoring. Most platforms do not. A user taking a typing test on a phone will receive a score that appears to indicate poor typing ability, when in reality they are performing a different task altogether.

The platforms diverge in how they handle this. Monkeytype detects the device type and shows a notification but still reports WPM using the same formula. TypeRacer's mobile interface is limited and most competitive races happen on desktop. Keybr recommends desktop use and does not optimize its interface for mobile. For reliable self-assessment, taking a typing test on a physical keyboard remains the standard.

Platform Comparison Summary

Frequently Asked Questions

What is the difference between gross WPM and net WPM?

Gross WPM divides total keystrokes (divided by five) by time, with no penalty for errors. It measures raw finger speed. Net WPM subtracts a penalty for errors before dividing by time, reflecting the usable output speed. The typical net WPM penalty is one word per error, but this varies between platforms. A typist with 100 gross WPM and 90 percent accuracy might have a net WPM ranging from 70 to 95 depending on the penalty formula used.

How did the 1888 typing contest establish the WPM standard?

The International Typewriting Contest of 1888, won by Frank McGurrin at 95 WPM, created the need for a standardized measurement unit. The International Association of Typing Examiners defined one word as five keystrokes based on empirical studies of average English word length in business correspondence. This definition spread through typing schools, certification programs, and competitive typing, becoming the global standard that persists today.

Why does my typing speed vary between Monkeytype and TypeRacer?

Three factors cause the variation. First, the WPM calculation differs: Monkeytype uses timer-based keystroke counting while TypeRacer uses word-completion-based timing. Second, text difficulty differs: Monkeytype's default 200-word list is easier than TypeRacer's varied passage library. Third, the testing context differs: Monkeytype is solitary and self-paced while TypeRacer is competitive with real-time opponent pressure. The same typist can expect 10 to 30 WPM variation between the two platforms.

What is a good typing speed by profession?

General guidelines for net WPM on standard test texts: 40+ WPM is the minimum for most administrative roles, 60+ WPM is expected for transcription and general office work, 80+ WPM is competitive for data entry and customer service positions, and 100+ WPM is professional-level for court reporting, closed captioning, and real-time transcription. Touch typists average 50 to 70 WPM, while hunt-and-peck typists average 20 to 40 WPM.

How do adaptive typing tests work?

Adaptive typing tests like Keybr maintain a performance model for each key and bigram. The algorithm tracks speed and accuracy per character, then generates text that disproportionately includes the letters and combinations where the typist is weakest. Over time, as the typist improves on those weak points, the algorithm shifts focus to new weaknesses. This creates an efficient practice loop that addresses the typist's specific skill gaps rather than providing random practice.

What text difficulty factors affect typing test results?

The three major factors are word frequency (common words like "the" and "and" are typed faster than rare vocabulary), character diversity (home-row letters are fastest, top-row is moderate, bottom-row is slowest, and rare bigrams create bottlenecks), and sentence complexity (punctuation, capitalization, and numeric content add cognitive and physical overhead). A test using only common short words on home-row letters can produce scores 20 to 30 percent higher than a test using varied vocabulary and full keyboard coverage.

Related Topics

• Typing Practice
• WPM Calculator
• Words Per Minute Test
• Keyboard Tester
• Text Generator