Reading Speed Is Not a Fact About You
Most estimates of how long a piece of writing will take to read rely on a single constant: some fixed number of words per minute, usually somewhere between two hundred and two hundred and fifty, applied uniformly to every reader and every sentence. It is a convenient number. It is also, on any individual page, close to meaningless.
Reading speed is not a personality trait fixed at some point in adolescence and carried unchanged for the rest of a reader’s life. It is a live negotiation between the text and the reader, and it changes constantly within a single piece of writing: dense description slows almost everyone down, dialogue speeds almost everyone up, a sentence structure the reader has seen a hundred times reads faster than one built from an unfamiliar rhythm. Even within one reader, on one afternoon, fatigue changes the number by the hour. A words-per-minute constant averages all of this away and calls the average a fact.
The more honest approach — and the one this platform’s reading-pace tool takes — is to stop assuming and start measuring. As a reader moves through a piece, the tool records how their own scroll position advances against real, elapsed wall-clock time: not a guess borrowed from a study of other people, but a regression against that specific reader’s own behavior, on that specific piece of writing, updated continuously as they read. Early in a chapter the estimate is necessarily rough, built from too little data to be confident. By the time a reader is a few screens in, it has usually converged on something considerably more honest than any fixed constant could offer, because it is no longer describing readers in general. It is describing this reader, on this page, right now.
There is a modest but real ethic buried in that distinction. A words-per-minute constant treats every reader as interchangeable. A live, personal estimate treats a reader’s own pace as data worth respecting — proof that the platform is paying attention to the actual person reading, rather than a statistical abstraction standing in for them.