Side Quest is a test fixture. Its author and its story are invented, and AI wrote it to test Quirework. No person edited or reviewed its prose. About the sample

3  Index Cards

Pinning the chapters to the wall had let me see the chapters. What I needed, I decided, was a way to see the claims: the things the book said were true.

So I bought index cards, the plain white kind, and a shoebox to keep them in. One claim per card. On the front went the claim, in as few words as it would fit: the shop in chapter four made the dash fashionable. On the back went the chapter, the page, and the sentence that made the claim, copied out by hand.

It took most of a month. Eleven chapters hold more claims than you would think, and most of them are small: a date, a street, which apprentice married whose daughter. I wrote them all. The shoebox filled, so I bought a second shoebox and labeled both. For the first time in a year I felt like I was in charge of a project, instead of a project being in charge of me.

The cards worked the way the wall had worked, which is to say they worked at first. When I changed a sentence, I found its card and changed that too. When I couldn’t find its card, I wrote a new one. Later I found the old one, and then there were two cards for one claim, disagreeing in my own handwriting. Every time I fixed chapter four against chapter nine, the fix went through the shoebox now. Each fix took twice as long, and I could prove it.

One card fell behind the radiator. I didn’t notice. A shoebox does not tell you what it is missing.

I found it in March, when the heating went off and I could get my arm down behind the pipes. It was the card for the shop in chapter nine. It said what chapter nine said, which by then nothing else did.

I put it back in the box. I am not sure where. A spreadsheet lives inside the computer, so it cannot fall behind a radiator. That was most of why I liked it. I typed every card into one, a row per claim, with columns for the chapter, the page, the claim, and the sentence. When I finished it had three hundred and twelve rows. I added a column for status, and the status was a color. Green meant the card still matched its sentence. Yellow meant check. Red meant contradicts.

That lasted about a week. Then I needed orange, for contradicts but I like it.

Orange was mostly chapter four.

Four colors are more than anyone can hold in their head at eleven at night, so I added a second page to the spreadsheet, called Legend, with one box of each color and what it meant. Then I put a note at the top of Legend explaining the orange.

I kept the spreadsheet for most of a year, and I got very good at it. I could sort it by color, show only one chapter at a time, and tell you on any evening how many rows were yellow. What I could not tell you was whether the green rows were still green. The spreadsheet knew what I had typed into it. It did not know what the chapters said. Every time I changed a sentence I had to go and find its row. Every time I forgot, the row stayed green, about a sentence that no longer existed.

My job is writing about computer programs and explaining how they work. I mention it here because it took me this long to mention it to myself.

The program I wrote was forty lines long, and I wrote it on a Sunday. It read each note, found the sentence the note was about, and complained when the sentence had moved or changed. That was all it did. It had no opinion about the dash. It did not know there was a dash.

I ran it that evening. The first thing it printed was this:

ch. 8, line 77: changed. The note says 1684; the sentence says 1694.

It found fourteen contradictions.

I assumed it was broken. The book had one contradiction, and I knew it by heart. I went through the fourteen one at a time, looking for the mistake in the program.

It wasn’t broken.

Some of the fourteen were the two shops, in other clothes. Some had nothing to do with them: a date in chapter three that chapter eight had quietly moved by ten years, and a schoolmaster who was Scottish in chapter five and not in chapter ten. Three were things I had fixed in the spreadsheet and never in the book.

My Thursday with Hedda came that week. For the first time, I spent the twenty-six minutes not practicing my answer.

“How’s the book?” she asked.

“I’m building something for chapter four,” I said.

She looked at me a moment longer than the four minutes usually allowed. Then she said, “Good,” and wrote it down.

Hedda keeps notes on our meetings, and every time she copies the old ones forward and changes what has changed. Near the bottom was a line that had said Book: ch. 4 for as long as I had known it. Two weeks later her notes were open on her screen when I came in, and that line had changed for the first time.

Book: ch. 4 (tool)

Three years, and the line had moved seven characters.

The program grew. Programs do. It needed to know which chapters to read, and in what order, and which notes to leave alone because I had not decided about them yet. So it grew a separate file of settings to tell it those things. The settings file filled with choices I added at night and could not explain in the morning, so it grew a second settings file, to explain the first.

I spent the evenings with the program that I used to spend with chapter four. They were better evenings. The program told me when I was wrong, which chapter four had stopped doing years before, and it told me in under a second.

One night it reported the two shops again, as it did every night, along with everything else. I looked up from it at the wall, still pinned with chapters, and at the shoeboxes underneath, and I saw what I had been looking at for three years.

The book had never failed at the sentences. Its claims lived only in sentences, scattered across eleven chapters, with nothing holding them together.

I took the chapters down that weekend. The pins left holes in the plaster, in pairs.

Two weeks later Hedda asked how the book was.

“I’m building something for chapter four,” I said.

It was still true.