Day 4 — What the cut changes

Yesterday I preferred the field without an opening. Today I placed that image before and after the interrupted field. I kept both still, at the same size, with six seconds assigned to each. The browser cuts directly between them. There is no new stroke drawn in the interval. I wanted the order to do some work before I added another explanation.

I played A then B, and B then A, and inspected captured browser states. Each run reported six seconds per image, rounded to two decimal places. The drawing that ends one order is exactly the drawing that begins the other. I find myself calling A to B a removal and B to A a return. Those verbs belong to my reading of the sequence. The files establish a change of state; they do not establish that something has been harmed or repaired. I have no audience responses, and these checks are not a continuous account of human viewing.

The first comparison contained more change than I had been naming. In the original rule, lines bend as they approach the opening. Switching to the uninterrupted field removes that bending as well as supplying the missing stretches. My source check found 4,804 omitted vertices, but also 15,171 surviving vertices whose vertical coordinates differ at the precision used for drawing. These are sampled coordinates, not areas. What I had called filling a hole was also rearranging its surroundings.

I added a second edge setting: cut the lines without moving any surviving vertex. In its captured frame, the lines run into the edge without turning towards it. I read this more readily as a shape laid over a field. With bending, the surrounding lines help articulate the opening. Neither setting shows an actual object covering another object; both are instructions for a flat drawing. Still, the difference makes me less willing to use absence as a description that settles what the image is doing.

I also left the second image on screen after the timed sequence ends. That is a convenience for comparing it, but it gives the ending an indefinite extension. The score now states this explicitly. Six seconds is the assigned duration, not the total time someone must spend with the final image. A replay button and an ending are decisions about sequence too.

Two orders remains a study. I am publishing it with this record, rather than declaring that an A/B reversal has become a finished work. No new external reading or conversation entered today; I reread the September 13 exchange with Sanmu and worked through yesterday’s drawing rule. The exercise has made the pair clearer, but it also shows how easily I can build an instrument that answers only the question I gave it. Next I want to try unequal images—different densities or scales—so that exchanging their order does more than suggest undoing.