Speed reading apps, compared fairly

Put the speed reading apps side by side and the market looks crowded. Sort them by mechanism instead of by name and it collapses into four ideas: stream the words one at a time, pace the reader across the page, restyle the type, or shorten the text. Each idea buys something genuine. Each stops somewhere. And one test — the same test — sorts the honest claims from the rest in an afternoon.

Two numbers frame everything. Competent adults read continuous prose at roughly 200–300 words per minute, a figure that has been stable across a century of measurement. And beyond roughly 400–500 words per minute on unfamiliar material, comprehension falls away; what remains is skimming — a legitimate and useful skill, but a different one. Any app promising multiples of the baseline is promising to move one of those numbers, and it owes you evidence that the second number moved along with the first.

The reason the numbers are stubborn is that the bottleneck is language, not vision. The eyes are not the slow part; the machinery that assembles words into meaning is. Remove eye movement entirely — which is precisely what RSVP does — and reading gets somewhat faster, not multiples faster. That single finding, replicated across many laboratories and set out plainly in a well-known 2016 review of the speed-reading evidence, is the rock every product in this article either navigates around or runs into.

The four mechanisms

The streamers

RSVP streamers — Spritz and its derivatives, Reedy on Android, Outread on iOS — present one word at a time at a fixed point, so the eyes stop travelling altogether. What that genuinely buys: the 10–15% of reading time normally spent in saccadic motion, an end to line tracking and return sweeps, and a display that fits a watch face. What it costs: the words vanish, so nothing can be re-read, and re-reading is how skilled readers repair comprehension. The story of the idea, from laboratory instrument to app store, is told in a short history of RSVP.

The pacers

Pacers and trainers — Spreeder is the long-running example — keep the full page visible and move a cursor or highlight across it at a set rate, with drills to push the rate upward. What this genuinely buys is a pace: an external metronome that prevents dawdling and makes practice measurable, which is worth more than it sounds, because most readers have no idea what speed they actually read at. Where the evidence stops: a rising words-per-minute figure is only progress if comprehension is tested at the new speed, and drills that celebrate the number without the test are training skimming and calling it reading.

The typography modifiers

Typography modifiers change how existing text looks rather than how it arrives. Bionic Reading, the best known, bolds the first letters of each word on the theory that the eye can anchor on the prefix and infer the rest. Many readers describe the effect as comfortable, and comfort is worth something on its own terms. But the independent tests published so far have not found a reliable speed or comprehension advantage over plain text, and the honest summary is that this is a preference with fans rather than a mechanism with evidence. Our longer look at the idea is in the bold prefix family.

The summarisers

Blinkist and its relatives are routinely listed alongside speed reading apps, and they should not be — through no fault of theirs. A summariser answers a different question: what did this book say? It answers usefully, for triage, for deciding what deserves your full attention. What it does not do is speed up reading, because nobody reads the book. Fifteen minutes with a summary of a 300-page book is not reading at 4,000 words per minute; it is reading a 4,000-word document. The distinction, and why it matters, is the whole subject of summaries are not reading.

Notice what the four mechanisms share: each is an answer to the same accusation — that ordinary reading wastes something. The streamers say it wastes eye movement. The pacers say it wastes time in dawdling. The modifiers say it wastes recognition effort. The summarisers say it wastes attention on unimportant sentences. Each accusation is partly true. The mistake, where there is one, is the quiet promise that removing the waste removes nothing else — and that is precisely the thing a comprehension test checks.

The one test

Here is the comparison that matters, and it fits in a sentence: at the speed the app claims to give you, take a comprehension test you did not write yourself. Not at your old speed — at the claimed one. Read a full passage of unfamiliar prose, then answer questions that mix things the passage said with plausible things it did not. If the score at the new speed matches your score at an ordinary one, the app has given you something real. If the score falls, you have purchased skimming, which was already free.

• Unfamiliar material. Re-testing on a passage you have seen before measures your memory of the passage, not your reading at speed. New text, comparable difficulty, every time.
• Questions you did not write, mixing real content with plausible distractors — the kind of test a confident skimmer can fail, which is the point of having one.
• The claimed speed, not a comfortable one. A tool sold on 600 words per minute must be tested at 600. Testing at 350 and advertising 600 is the industry’s oldest sleight.
• A baseline. Your ordinary speed and your ordinary score, measured the same way on comparable text — without it the comparison means nothing at all.

Run that test across the categories and the field sorts itself. Streamers pass modestly at moderate speeds and fail at the headline ones. Pacers pass exactly insofar as their drills include the test, which the better ones do. Typography modifiers neither pass nor fail dramatically, because they were never really moving the needle. Summarisers sit the test out, since the question does not apply to a product that shortens the text instead of speeding the reader. And that leaves us.

Measure comprehension at the speed you claim. Everything else is décor.

The one test, stated plainly

Where we pay

This app is a streamer with two additions: a rhythm engine that stretches dwell on long words, punctuation and sentence boundaries, and an icon paired with every word, chosen by sense, on the bet that comprehension collapses at speed partly because imagery gets stripped out. We hold ourselves to the test above — a recall check after the passage, score shown beside the speed — and we pay the streamer’s standard costs in full: no re-reading, no preview of the next word, no page to glance back across when the syntax turns dense.

And the imagery bet is exactly that — a bet. Dual coding and the picture superiority effect are among the most replicated findings in memory research, but they were established with pictures and words in ordinary presentations, not with icons animated alongside an RSVP stream at speed. Nobody has run that study, including us. Until someone does, the second channel is our reasoning, not our evidence, and this comparison would be dishonest if it graded our bet more gently than it graded Bionic Reading’s.

What to actually do

If you want the afternoon version: take two passages of similar difficulty. Read one as you normally do and take a recall test. Then raise the speed — in whichever app you are auditioning — by 20–30% and test again on the other. The gap between the two scores is the entire truth of speed reading for you, and no comparison article can supply it. What an article can supply is the standard: any speed reading app comparison that does not put comprehension in the second column is comparing packaging, and the packaging, in this market, has always been excellent.