I am not against knowing your material. I am against the specific act of committing a written text to memory word for word, because of what it costs and what it does to your delivery on the way.
The arithmetic nobody does
Take a three-minute video. That is roughly 450 words. To hold 450 words reliably enough that you will not stall in the middle, most people need several passes and some spacing — and however long it takes you, it is time that produces nothing but the ability to say one specific text once, after which the value goes to zero. You cannot reuse it. Next week's video is a different 450 words.
Put that against the alternatives available in the same slot: rewriting the weak middle section, recording a second video, or simply publishing on time. Memorisation is the only one of those that leaves nothing behind.
And the cost is not only the hour. It is that the hour comes out of the same limited evening as the writing, so the writing gets cut short to make room for learning it. You end up rehearsing a worse script very thoroughly. That is the trade that actually happens, and nobody frames it that way at the time.
What memorising does to your face
Here is the part people do not expect, because it contradicts the whole premise.
Recall is visible. When you are retrieving a memorised line rather than reading it or thinking it, your eyes go up and to the side, your blink rate changes, and there is a small hesitation before clauses you are less sure of. Viewers do not identify it as "he is remembering". They read it as slightly absent, or slightly rehearsed.
So the thing you spent an hour on to look natural has its own tell. It is a different tell from reading, and often a subtler one, but it is not zero — and it gets much worse under the conditions solo creators actually record in, which is tired, at the end of a day, on take nine.
There is a second failure mode that is worse. A memorised script has a single correct path through it. Lose the path and there is no graceful recovery, because you have no text in front of you. So you stop and go back to the top. Memorisation removes your safety net at exactly the moment you need one, which is the fastest route into a retake spiral.
What to commit to memory instead
Not the words. The shape.
Before you record, be able to say, from memory, in one breath each:
- What this video is about, in a sentence.
- The three or four moves it makes, in order.
- The first line of each move.
- The last line of the whole thing.
That is maybe forty words of actual memorisation and it takes a few minutes. It buys you the two things that matter: you always know where you are and where you are going, and if you come off the text you can rejoin at the next move instead of restarting.
First lines matter more than the rest, because the start of a section is where you are re-orienting and most likely to stall. Last lines matter because an ending you have to find on the spot is the most common reason an otherwise good take is unusable.
Read the middle, and read it well
Everything between those anchors, read. That is what a prompter is for, and reading is not the compromise it is treated as — badly set up reading is.
What separates the two is unglamorous:
Write it out loud in the first place. Read the script aloud once before you record and cut every sentence you stumble on. A stumble is not a memory failure, it is a sentence built for the eye. Two ideas in one sentence, a stacked noun phrase, a clause that resolves too late. Written and spoken rhythm are genuinely different and a script that reads well silently can be nearly unspeakable.
Narrow the column and enlarge the text. Short lines, one idea each, generous space between blocks. This is what keeps your eyes still, and still eyes are most of what people mean by natural. There is more on the geometry of this in the piece on where to put the text so reading does not show.
Know the material, which is not the same as knowing the text. If you could explain the topic to a friend without notes, you can read a script about it and sound like you mean it, because you are re-saying something you believe rather than reciting a document. If you could not, no amount of memorisation will save the take, and that is the real signal to go back to the writing.
The reason people memorise anyway
Almost everyone who memorises tells me the same thing, and it is not about looking natural. It is that reading from a fixed-speed scroll is unpleasant. You are managing a timer while also trying to perform, and after enough evenings of that, spending an hour learning the words starts to look like a reasonable price to escape it.
That is a rational response to a badly designed tool, and it is the wrong problem to solve with your time. The thing making reading unpleasant is not reading. It is that the text is moving on a clock instead of moving with you — which I unpacked in why a fixed scroll makes you sound like you are reading.
Remove the clock and most of the reason to memorise goes with it. If the scroll follows the words you have actually said, you can pause where the meaning is, take the sentence at your own pace, and the text is simply there when you look at it. You can also come off script entirely, say something better than what you wrote, and rejoin — because the position is determined by your voice, not by elapsed time. That is what my own app does, and if the hour you spend learning lines is the hour you would rather have back, the 7-day trial is enough to try one video the other way and compare what it cost you.
Where memorising is genuinely right
Three cases, so this is not a blanket rule:
- Very short pieces. Under about thirty seconds, learning it is faster than setting up anything.
- Walking, moving, or on location. No screen is available and that is that.
- A talk you will give many times. Here the hour amortises, and it stops being a per-video cost.
Notice what those have in common: memorisation is worth it when there is no prompter available, or when the text will be used more than once. For the weekly video that will be said exactly one time, in your own room, with a screen right there, it is an hour bought at full price and consumed immediately.
Spend it on the script. That is the part the audience actually receives.