Formats›Reading for silence, in two passes

Reading for silence, in two passes

A reusable format for reading one text in your own field. Take it, fill it, use it again — once per text, and the last line is the reason it exists.

What this is for

This is for a text you already agree with. A text you disagree with gets read carefully by itself; the argument does the work and you will find its gaps without help. The dangerous text is the one that sounds like the field talking — the introduction everybody was given, the principles everyone cites, the piece so settled that it reads as description rather than as argument. Those are the texts whose assumptions get inherited whole, because nothing in them announces itself as a position.

A silence is not an error. It is what the text does not treat as a question. Errors are arguable and get argued; silences are invisible, because the reader supplies the missing part out of their own assumptions and never notices the transaction. Finding one is not a criticism of the author. Every text has them — this one has them — and the useful move is not the finding but what you do with it afterwards.

The last pass is the whole point. It is easy to name what somebody else took for granted and comfortable to stop there, and a session that stops there has taught the room to be clever about other people. The turn is asking whether the same silence is in your own argument. It usually is, because you were trained by texts like this one.

The format

THE TEXT        one published thing in your own field, short enough
                to read twice in an hour

FIRST PASS      what it says — the argument, in its own terms
SECOND PASS     what it is silent about — not wrong, unasked
THE COST        who is not in the text, and what it would have to
                concede to let them in

YOUR TURN       whether the same silence is in your own argument

A text above, a reckoning below, and three passes between. The last line is not a summary of the others.

What each line is asking

THE TEXT. One published thing, short — a thousand words is plenty and more is worse. Choose one you agree with. If you picked it because it annoys you, you will produce a rebuttal, which is a different and easier document. The best candidates are the pieces so established that quoting them feels like citing the weather.

FIRST PASS. What it says, in its own terms, as its author would recognise it. Do this properly and do it first. A second pass run on a text you have not yet stated fairly is not a reading, it is a search for ammunition, and it will find some.

SECOND PASS. What the text does not say. Not the counter-argument — the question it does not treat as a question. Who its reader is assumed to be. Which value is left unmarked, so that it functions as a fact. Which category is used without being opened. What kind of person, body, place or arrangement does not appear anywhere, and would break something if it did. Write the silences as a list, plainly, without indignation. Indignation is a tell that you have moved back to arguing.

THE COST. Who is not in the text, and what the argument would have to give up to let them in. This is the line that separates a real silence from a trivial one. Some silences cost nothing — a text about typesetting is silent about agriculture and no harm follows. A silence that matters is one where admitting the missing party would force a claim to be weakened, scoped or withdrawn. If nothing would have to change, you have found an absence rather than a silence.

YOUR TURN. Whether the same silence is in your own argument. Write it about a specific thing you have made or are making, not about your practice in general. The honest answer is usually yes, and the interesting part is where: which claim of yours assumes the same reader, leaves the same value unmarked, or would have to be scoped the same way. If the answer is genuinely no, say what in your work made it no — that is worth more than the finding above it.

A worked one

THE TEXT        A widely-taught one-page introduction to usability.
                About a thousand words. Cited everywhere, argued
                with almost nowhere.

FIRST PASS      Usability is a quality attribute with components you
                can name and measure — learnability, efficiency,
                memorability, errors, satisfaction. It is testable,
                cheaply, on a handful of people, and testing it early
                is much cheaper than fixing it late. The argument is
                clear, practical, and correct on its own terms.

SECOND PASS     Its user is literate, alone, sighted, at a desk, and
                using the thing to accomplish a task of her own
                choosing. Efficiency is unmarked — faster is better
                is never stated because it is never doubted. The
                category "error" is used throughout and never opened:
                an error is a departure from what the designer
                intended, so a person using a thing in a way nobody
                planned is producing errors by definition. Nothing
                about oral cultures, about a device used by several
                people at once, about a task somebody else set you, or
                about bodies that vary.

THE COST        Admitting collective use costs the measurement: the
                metrics are per-person and there is no per-room
                version of "time on task". Admitting a task somebody
                else set you costs "satisfaction", which assumes the
                user's goal and the task are the same goal. Both are
                real concessions rather than footnotes, which is what
                makes these silences and not absences.

YOUR TURN       My own tools measure a person alone, at a screen, with
                her own goal. I have the same reader. The place it
                shows is that when two people use one of them, I have
                no vocabulary for what happened and describe it as
                noise in the data — which is the error category doing
                exactly what it does above.

Note that the second pass did not find a mistake, and the fourth line is not a confession. It names one place where a claim would have to be scoped, which is a piece of work rather than a feeling about oneself.

How to use it

1. State the argument fairly before you look for anything. If you cannot write the first pass in a way its author would sign, you are not ready to do the second.
2. Leave a gap between the passes. Even ten minutes. Running them together turns the first pass into a preparation for the second, and the fairness goes out of it.
3. Write the silences flat. A list of plain sentences. The moment the prose gets warm you have started arguing, and an argument is not what this produces.
4. Test each silence against THE COST. Most of what a first attempt finds is absence rather than silence, and this is the line that sorts them.
5. Do YOUR TURN in writing, about one named thing of yours. Held in your head it becomes a general resolution to be careful, which changes nothing.

Three ways it goes wrong

It becomes a rebuttal. The second pass turns into a list of what the author got wrong, which is a normal and useful activity and is not this. A silence is something the text never treated as a question; a mistake is something it answered badly. Only the first kind gets inherited without being noticed.

The last pass is skipped, or generalised. I should be more aware of my assumptions is the shape this failure takes, and it is indistinguishable from having done nothing. The turn only works when it names a claim of yours that would have to change.

The text is chosen for being bad. A weak text has obvious silences and finding them teaches nothing, because nobody was going to inherit them. The format is worth running on the text you would have cited without thinking.

Prayas Abhinav. Licensed CC BY 4.0 — copy it, change it, build on it, including commercially. Say where it came from, and if you changed it.

Take a copy: reading-for-silence-format.md. Yours to alter.