Showing your work: documenting your writing process

Somewhere in the last few years, a question was added to the ones a paper has to answer. Beside what do you claim and what is your evidence there is now, sometimes spoken and sometimes not, how did you write this?

Students who wrote their paper themselves find this question insulting, and it is. It is also, for the moment, a question that gets asked, and the honest student is worse placed to answer it than they should be — because the honest process leaves no trace. The paper is finished; the drafts were overwritten; the reading notes are in a notebook somewhere; the thinking happened in your head. All that remains is the document, and a document cannot say how it was made.

Cards can. Not because they prove anything about you — nothing can — but because the process of working from cards produces a record as a side effect, and the record is the kind of thing a fair reader finds convincing: not a claim about how you worked, but the work.

This guide is about keeping that record without doing any extra work, and about what it can and cannot say.

What a record of process is

It is not a statement. I wrote this myself over three weeks is a statement; anyone can make it, and it is exactly what a person who did not write it would also say.

A record of process is the material the process left behind, with dates: the reading notes, the cards as they were made, the claims as they changed, the orders that were tried and abandoned, the cut pile. It is convincing for the reason a lab notebook is convincing — because it shows the wrong turns. A finished paper has no wrong turns in it. A process has dozens, and only a process leaves them.

Step 1: Date the cards

Every card gets the date it was written, in the corner. Source cards, quotation cards, claims, counterpoints — all of them. It takes two seconds and it is the whole method.

A dated stack of cards is a timeline. Sources first, quotations as the reading went, claims appearing partway through the reading and thickening at the end, counterpoints arriving late — usually the night you realized the argument had a hole. The thesis card dated after the claims, not before. That timeline is what a research process looks like from the outside, and it is almost impossible to fake convincingly, because faking it means inventing a plausible three weeks of wrong turns.

Step 2: Keep what you cut

The pile of cards you did not use — the evidence that supported no claim, the claim you conceded, the two sources that contributed nothing — is part of the record. Keep it, with a rubber band and a card on top saying not used, and if you can, a line on each about why: overlaps with S5, couldn't support it, interesting but off the point.

This is the pile students throw away and the one that says most. A paper that used everything the writer found is either a paper that was written to use its findings or a paper whose findings were selected to fit it. The cut pile shows the reading was wider than the paper, which is what honest reading looks like.

The reasons are worth the effort. Overlaps with S5; keeping for the next paper is a sentence that shows judgment, and judgment is precisely what a reader of the record is looking for.

Step 3: Keep the old orders

Each time you rearrange the row (guide 6), photograph the old arrangement before you break it. Date the photograph. Three or four of these across the project are a record that the structure was worked on — that the order in the finished paper was chosen from alternatives, not arrived at once.

Step 4: Keep the draft's history, lightly

On paper, this means: do not overwrite. Write the draft in a fresh document each time you make a structural change, and keep the old ones with dates in the filename. Not every sentence-level edit — that is noise — but each time the shape changed.

Step 5: Write a sentence about how you work

Every process has a pattern, and patterns differ. Some people read for two weeks and write in two days. Some write a little every morning. Some take notes on paper in the library and type them up on Sundays, so the cards all carry the same date and the reading looks like it happened in one sitting. Some share a computer and work in short bursts when it is free.

None of these patterns is better than another, and none of them is what a paper looks like when it was not written by its author. But a record shows the pattern, and a reader who does not know your circumstances may read a pattern as something it is not — a week of no cards as a week of no work, when it was a week of reading.

So write one sentence, on a card, dated, saying how you work: I read on paper during the week and make cards on Sunday. Put it on top of the stack. It is not an excuse and it is not required. It is the context that turns a pattern into a person.

What the record can and cannot say

Be clear-eyed about this, because the people asking will be.

It can show that the paper was assembled from pieces over time; that the reading was wider than the paper; that claims were formed after evidence and revised; that structure was tried more than one way; that some things were cut, and why.

It cannot show who held the pen. A record of process documents a process. It does not prove authorship, and you should not claim it does, because the claim is false and a careful reader will know it is. What the record does is answer how was this written with something better than trust me — and for the honest writer, that is the only question that needed answering.

It must not be graded. If an instructor takes your record, the right thing for them to do with it is read it. The wrong thing is to score it — to count your cards, compare your timeline with another student's, or decide that a "good" process has a certain shape. Working patterns vary for reasons a record does not show, and a record read as a score punishes the student who reads on paper, or shares a laptop, or writes in bursts. If a policy asks for process, it should ask to see it, not to measure it. You are allowed to say so.

When to hand it over

Only when asked, and then the parts that answer the question. A summary — the timeline, the cut pile's card, the sentence about how you work — answers how was this written. The full stack answers a specific challenge, if one comes, and it should not have to.

Keep it either way. A stack of dated cards is the most useful thing you take out of a paper, because the next paper in the field starts from it. The record was never mainly for the marker. It was for you, in March, wondering what you found in October.

Doing this in software

Software that works from cards can keep this record without the two seconds per card, because every card, link and rearrangement is dated when it happens, and the dates are kept apart by how much the software could vouch for them — recorded on a connected device, recorded offline and verified later, or brought in from a file the software did not make — so that nothing is rounded up to more certainty than it has. From that history it can produce a document, when the writer asks for one, that describes the process the way this guide describes it: what was made when, what was cut and why, which orders were tried. Such a document should carry the writer's own sentence about how they work, should say plainly that it documents process and not authorship, and should say nothing that evaluates the pattern — no score, no comparison, no adjective. If it does any of those things, it is not a record; it is a verdict, and you should not accept one from a machine.