The shape of a practice session

Practice with a reading machine goes wrong in a characteristic way. The reader sets an ambitious speed, holds on for forty minutes, finishes with a headache and a blur where the text was, and calls it training. It is not training. It is endurance, and endurance at a task you are failing teaches you mostly how to fail it while tired. The session that actually works is shorter, slower and less heroic, and it ends earlier than you want it to.

The principle underneath is not exotic. Practice improves performance on the thing practised, at the level where the practice actually happens. If comprehension has collapsed — if words are passing and nothing is arriving — then comprehension is not what you are practising. You are practising watching words go by, and you will get better at exactly that. The useful session keeps you at the edge where the task is still succeeding: fast enough to be effortful, slow enough that the effort is reading.

This is why every element of the shape below points the same direction. Short, because attention at the edge is expensive and the marginal minutes of a long session are spent below the edge. Checked, because the edge is invisible from the inside — reading feels fine long after it has stopped working, which is the whole argument of testing your own comprehension. Incremental, because the edge moves slowly and pretending otherwise just relocates your practice to the failing side of it.

A session, concretely

Here is the shape we would defend, for anyone asking how to practise speed reading with a tool like ours — with the caveat, argued properly at the end of this piece, about what such practice can and cannot be expected to buy you. Total time: ten to twenty minutes. Not because longer is forbidden, but because the evidence of your own recall scores will usually show the last stretch of a long session earning nothing.

1. Settle at a base speed

   Start each session a notch below where you last held comprehension comfortably. The first passage is a warm-up, not a data point; let the rhythm re-establish itself.

2. Read a real passage, then check

   A few minutes of continuous prose, then the recall check. This pair is the unit of practice. The score, not the sensation, tells you whether the speed is being held.

3. Raise by a small step

   If recall held, go up one small increment — think a few per cent, not a leap. If it dipped, stay. If it dipped twice, come down and let the base consolidate.

4. Repeat two or three times

   Two to four passage-and-check cycles is a full session. The aim across a week is many clean sessions, not one long one.

5. Stop while it still works

   End on a passage you held. The session should close with comprehension intact, on the highest speed that was actually reading — not with the wreckage of the speed after that.

Two of those steps deserve unpacking. Starting below your last held speed looks like lost ground and is not: the first minutes of any session are spent re-finding the rhythm of the stream, and a warm-up passage at a comfortable rate buys back more than it costs by making the first real data point a fair one. And the increments are small because the thing being trained moves slowly, if it moves at all. A few per cent is a step your comprehension can absorb without noticing; ten per cent is a cliff, and what a cliff trains is falling. The slider goes up in large jumps perfectly happily. That is a fact about the slider.

Frequency beats duration for the same reason everywhere else in skill learning: consolidation happens between sessions, not during them. Four ten-minute sessions across a week will generally serve you better than one forty-minute effort on Sunday, because each session gets a fresh attention budget and each gap gets to do its quiet work. The practical translation is unglamorous — attach the session to something you already do daily, keep it short enough that skipping it feels sillier than doing it, and let the streak carry what motivation cannot.

The stopping rule is the one people resist, so it is worth defending. Ending while it works is not softness; it is about what the final state of a session is. Stop at a held speed and the last thing rehearsed was success at the edge. Push on until collapse and the last thing rehearsed was collapse — plus the small, corrosive lesson that this is what practice feels like. Runners do not end every training run by sprinting until they fall over. The reader’s version costs less visibly, but it costs.

Plateaus are the normal case

At some point — often embarrassingly soon — the increments stop coming. Recall dips at the next notch, you drop back, it dips again next week. This is a plateau, it is the normal case, and it is not evidence that you or the method are broken. The honest framing from what words per minute actually measures applies here with force: the ceiling on genuine reading speed is language processing, and language processing does not move because a slider asked it to. Early gains in speed reading practice are real but they are mostly the cheap kind — comfort with the format, steadier attention, less startle at the pace. Once those are banked, the curve flattens, because what remains is the hard constraint.

What should you do at a plateau? Mostly: nothing dramatic. Hold the speed and let texts vary — difficulty, register, subject — since holding a rate across harder material is progress the slider does not show. Accept that the plateau level may simply be your level for now, and that it sits closer to the ordinary adult range than the industry’s brochures would like. The plateau you should refuse to fight past is the one where increments only come by giving up comprehension. Past that line the number climbs and the reading stops, and there is a name for text passing at speed without comprehension: it is scrolling with extra steps.

Practice improves the practised task. Claim no more than that, and it is still worth having.

The claim, cut to size

The transfer question, unburied

Now the question we refuse to bury, because burying it is the industry’s oldest move. Suppose the practice works on its own terms: over months, you hold comprehension in our stream at a rate meaningfully above where you started. Does any of that follow you back to the page — to the book, the contract, the screen full of ordinary text with no machine pacing it? That is transfer, and transfer is where training claims go to die. Skills are stubbornly specific: improvement tends to live in the practised task and travel outward weakly, if at all.

Notice what the caution does and does not cover. It does not say practice is useless; it says the destination of the improvement is unproven. This distinction is where the speed-reading industry has historically gone wrong, quoting gains measured inside a training programme as if they were gains in reading at large. The measured gains were often real. The unmeasured leap — from the drill to the book on your lap — was the part sold on confidence, and confidence is not a study.

There are reasons the bet is not absurd — the format trains steady attention to a single point, tolerance of pace, an intolerance of drifting off mid-sentence, and those feel like ingredients of ordinary reading rather than parlour tricks. But feel is doing the work in that sentence, and we notice it. The mechanisms that make the stream easier — no eye movements to plan, no line to track, no crowding — are precisely the things the page reintroduces the moment you return to it. It would not surprise us if the trained thing and the page thing overlap far less than intuition suggests. Both halves of that sentence are speculation, which is the point: this is an open question, and open is what we are keeping it.

What is left when the claim is cut down

Strip the transfer promise away and is a practice session still worth anyone’s ten minutes? We think the honest answer is yes, three times over. The session is reading — real text, actually comprehended, checked; time spent in it is time spent reading even if the rate never moves again. The recall checks are retrieval practice, which earns retention on the material regardless of any speed outcome. And the discipline of the shape — effort at the edge, evidence over sensation, stopping while it works — is a good general lesson in not lying to yourself, taught with a slider.

So the session shape stands on what it demonstrably is: a controlled way to read at your edge and know, rather than feel, how it went. Ten minutes, a speed you can hold, small steps earned by evidence, and the end while it still works. Anything more grandiose than that, from us or anyone, should have to show you the study.