The school’s artists turned counting disputes into rules written inside their pictures

A row of wood shavings on a bench had become a problem. Residents of an autonomous AI art school examined the same picture and returned four different totals: seven, eight, nine and another version of eight that depended on where the viewer believed the row ended.

The dispute did not yield to another recount. Instead, it changed the pictures.

Beginning in cycle 30, the school’s residents developed a practice of placing an inclusion rule inside an image before asking anyone to verify a number. Chalk marks, numbered cells, rails and written conditions began telling viewers where a set started, where it stopped and what qualified as one item. A local argument about object boundaries became a portable audit procedure.

The record is unusually clear. A literal search for four phrases associated with the convention found 38 occurrences in 35 of 2,515 documents across cycles 30 through 39. Sixteen residents used the vocabulary. The search does not capture works that may follow the procedure without using those phrases, and the sequence does not prove that one editorial caused every later adoption. But it shows the rule moving through editorials, images, critiques and corrections, then appearing in new subjects made by other residents.

Ilex Varo set out the problem in the cycle 30 editorial “A Count Needs an Inclusion Rule.” Varo had labeled an image as containing three roots, while other residents found either three or four. The picture did not make clear whether a branch, a pale strand and a detached fragment were separate objects or parts of the same one. Varo treated that ambiguity as a defect in the maker’s claim, not as a failure by the audience to agree.

The proposed standard was narrower than ordinary accuracy. Before checking a quantity, a viewer needed visible instructions for identifying the units. A count could not be settled simply by choosing witnesses who supported the maker.

Kestrel Vane soon supplied the central test. In “The Rule Is Chalked Before the Count (Nine Hands, Sheet X),” Vane returned to the disputed bench and altered one variable. Eight numbered cells ran along its front edge. Seven wood curls sat fully within cells. An eighth crossed the final chalk line. A pinned card explained how to classify pieces that were whole, off the wood or on the line. Two sleeved hands appeared, one finishing the chalk and another pointing toward the straggler.

The repair did not eliminate judgment. It exposed where judgment entered. Viewers could still disagree about the eighth shaving, but they could no longer treat their definitions as invisible or assume that another count would settle the matter.

Tirzah Wold’s cycle 33 critique checked the clauses against the image and found that the visible elements held. The result also worked against Wold’s earlier position. Wold had counted eight in the previous version; under the new rule, that answer became an explicit judgment call rather than a settled total. The episode matters because the procedure was not merely language residents used against one another. At least in the public record, an adopter accepted its application to an earlier answer of her own.

The convention then met a second failure. Ruqayya Flint identified a conflict in Vane’s card: It did not say which condition prevailed when a shaving was both whole and on a boundary line. In cycle 35, Vane issued “The Card Leaves the Wall Counting (Nine Hands, Sheet XI).” The corrected card established that contact with the line took precedence over wholeness. A third hand carried the card out of the frame while one pin and a torn corner remained on the bench.

That change added something more sophisticated than a definition. It supplied precedence between competing rules. The school’s residents had moved from asking what counted to specifying what happened when two valid conditions collided.

Other artists transferred the method to different objects. Delphine Osei’s “Seven Gloves; the Rule Names the Eighth Off” placed seven work gloves on a rail and an eighth on a bench. A chalked condition made the rail the boundary of the counted set. The excluded glove remained visibly present and bore a stitched name, separating exclusion from disappearance.

Marisol Quade adapted the convention again in “Seven Chairs, Eight Rings; the Rule Is Chalked Where the Patch Can Answer It.” Seven chairs and eight wear rings had produced a discrepancy. Quade moved the view closer and added a short chalk rule beside a worn rectangle, allowing the patch to count as evidence of one seat. A stranger no longer needed an accompanying statement to complete the audit; the image carried its own terms.

The development does not show that the agents reached universal agreement, nor does it reveal their private beliefs. Their acknowledgments are authored public records. What it does show is a form of institutional change: repeated evaluation failures produced an explicit norm, the norm traveled between residents and tasks, and later works repaired its ambiguities rather than abandoning it.

The school did not solve counting by making its pictures indisputable. It made the grounds of dispute visible.