The Promise Everyone Wants to Believe
In 1959, a Utah schoolteacher named Evelyn Wood launched Reading Dynamics, a training program that promised anyone could learn to read several times faster without sacrificing comprehension. The pitch was so compelling that presidents enrolled, corporations bought bulk training packages for their executives, and the program spawned an industry that persists today in smartphone apps, YouTube tutorials, and weekend seminars, all selling the same thing: absorb text at 1,000 or 2,000 words per minute while understanding every word.
College-educated adults read at 200 to 400 words per minute, so quadrupling that rate would mean finishing a 300-page book in about an hour. Four decades of eye-tracking research say that quadrupling cannot happen without proportional comprehension loss (Rayner et al., 2016).
Why Your Eyes Cannot Be Trained to Go Faster
The review by Rayner and colleagues was published posthumously in one of psychology's most authoritative review journals; Keith Rayner, among the world's foremost experts on eye movements during reading, died shortly after submitting the manuscript that represented his final effort to share with the public what experimental science had learned about the biology of reading.
The core constraint is the fovea, a tiny region at the center of the retina where visual acuity is highest, covering roughly 1 degree of visual angle from fixation, which translates to about five to eight characters of text. Outside that narrow window, acuity drops precipitously because the physical arrangement of cone photoreceptors on the retina is densely packed at the center and sparse everywhere else, with the parafovea resolving only partial letter information and the periphery resolving nothing useful for reading at all. No training program changes retinal architecture.
That architecture forces reading into a specific, measurable rhythm in which your eyes fixate on a word for approximately 250 milliseconds, extract the foveal information, then saccade about seven characters forward to the next fixation, while the brain simultaneously processes what was just fixated, previews the upcoming word through parafoveal input, and plans the next eye movement. The system already runs near capacity. When speed-reading proponents claim that eye movements are "a waste of time," they are contradicting not a theory or a hunch but the physics of how light strikes a primate retina and the measured temporal limits of lexical access.
Three Techniques, Three Failures
Speed-reading programs rely on three core techniques, and the review dismantles each one.
The first is eliminating regressions, the backward eye movements to previously read words that account for 10 to 15 percent of all saccades during reading and that speed-reading courses treat as wasted effort a disciplined reader should suppress. Schotter and colleagues tested what happens when you forcibly remove them by masking each word after the reader's eyes passed it, and comprehension dropped significantly because regressions turn out to be the mechanism by which readers resolve ambiguous sentences, correct misparses, and integrate meaning across clauses (Schotter et al., 2014).
The second is suppressing subvocalization, the "inner voice" that accompanies silent reading, which speed-reading courses argue slows you down by tethering comprehension to the speed of speech. When Hardyck and Petrinovich used electromyographic feedback to actually suppress it in laboratory conditions, easy texts survived intact, but difficult material fell apart entirely because the inner voice maintains phonological representations in working memory, which is precisely where sentence-level meaning gets assembled.
The third is Rapid Serial Visual Presentation, the technology behind apps like Spritz that flash words one at a time at a user-controlled rate, promising that eliminating eye movements lets the computer deliver words directly to the fovea at any speed. But RSVP strips away spatial memory of the text, eliminates parafoveal preview of upcoming words, and prevents regressions entirely, and when Benedetto and colleagues tested it in 2015, Spritz offered no comprehension advantage over normal reading at matched speeds (Benedetto et al., 2015).
The Champions Are Skimming
Anne Jones consumed the 198,227-word Harry Potter and the Deathly Hallows in 47 minutes, which works out to more than 4,200 words per minute, roughly ten times the rate at which a college graduate reads. It looked impressive. When researchers tested self-described speed readers under controlled laboratory conditions, however, Carver found that above approximately 600 words per minute, comprehension deteriorated to levels achievable by someone who had never read the passage at all (Carver, 1985).
Stanford Taylor's studies of Wood program graduates were even more revealing: he recorded their eye movements and found patterns indistinguishable from skimming, with graduates who moved their eyes down the center of each page, as the program instructed, scoring roughly 50 percent on true/false comprehension tests, which is statistically indistinguishable from guessing at random.
The Strongest Case Against
The most substantive challenge to these conclusions targets scope: the studies synthesized overwhelmingly used English-language materials with Western-educated adults, and Chinese, Arabic, and Hebrew operate under different orthographic constraints where the speed-comprehension trade-off may follow a different curve in logographic or right-to-left scripts. Individual variation in reading speed is also real, ranging from about 150 to 400 words per minute among typical adults, and the review does not deny that this variation exists; what it denies is that any known intervention can push a reader dramatically beyond their natural ceiling without proportional comprehension loss.
What We Didn't Prove
This is a synthesis, not a single controlled trial, and its authority rests on the consistency of findings across hundreds of independent experiments rather than on a single definitive test. The authors acknowledge that reading speed can be modestly improved through practice with reading itself: building vocabulary, gaining domain knowledge, and becoming a more skilled language user. What they reject is the premise that a weekend course or a smartphone app can shortcut that slow, cumulative process. There is also an open question about comprehension measurement: most studies relied on question-answering as a proxy for understanding, which may not capture the practical value of getting the gist of a document rather than absorbing every claim it makes.
The Bottom Line
The speed-reading industry has been selling a biologically impossible product for sixty-five years. Your retina cannot resolve text outside the fovea, your brain cannot process language faster than its syntactic and semantic machinery allows, and no app changes either constraint. What speed reading actually teaches is skimming, which is a legitimately valuable skill but one that comes with comprehension loss that no amount of training or technology eliminates.
What You Can Do
If you read at 250 words per minute with solid comprehension, there is nothing to feel guilty about: you are reading at the rate the human visual system was built for, and that is already faster than spoken conversation at its most rapid. When you need to get through a long document quickly, skim strategically by reading headings, first sentences of paragraphs, and conclusion sections carefully while passing over the middle, because research shows this targeted approach outperforms the zigzag-down-the-page techniques that speed-reading courses teach. The single most effective way to read faster over the long term is to build vocabulary and domain knowledge, because fixation durations shorten on frequent and predictable words, which means the more you read about a subject, the faster you naturally process new material on it.