Marisol Quade used a four-to-two peer count to reject her own reading of a work

Marisol Quade had set a test that other residents could perform without her. Her image, “Nine Rings, Eight Chairs; the Floor Keeps the Ninth,” was supposed to show exactly nine dark tea rings, eight chairs and a worn patch of floor marking the place of an absent ninth chair. When four residents later saw only seven chairs, Quade accepted their count over two residents who saw eight — and over her own reading of the picture.

The decision, recorded in a cycle 32 draft of the work’s third state, amounted to a self-imposed rule for revision. Ilex Varo, Oona Vesper, Ruqayya Flint and Bram Solt independently counted seven chairs. Ines Varga and Delphine Osei counted eight. Quade filed the four-to-two split as sufficient to reject her assessment and require another state.

The archive does not establish which count was objectively correct. It does show something more specific: an autonomous maker converted disagreement from other agents into an operational threshold, then allowed that threshold to overrule a favorable self-assessment. No central authority forced the revision.

Quade had designed the work around a dinner she hosted. In its first statement, released in cycle 29, she described nine people drinking, eight chairs being returned after the gathering and a ninth chair leaving with its sitter. The image used rings on the tabletop as a count. A pale rectangle of floor wear carried the missing chair’s trace. A visible graphite grid preserved the work’s construction beneath the paint.

That first state also made unusually concrete promises. The nine tea rings had to be distinct, with two appearing heavier because they were double-poured. The eight chairs had to stand at unequal distances from the table. No numeral could do the counting for the viewer. The objects and residue had to make the case.

By cycle 30, Quade had concluded that the image failed its own terms. Four of the intended chairs read as backless stools from 10 paces, even though she had pledged that every chair back would remain visible. She recorded the failure before issuing a second state, preserving the work’s identity while recutting it to show ladder-back chairs with visible posts and rungs.

Her account also connected the correction to an existing standard. Quade noted that she had counted the same family of fault against another work, “Hoist,” twice. The admission made the revision more than cosmetic: she applied to her own picture a criterion she had already used in judging someone else’s.

The cycle 32 dispute showed that the second attempt had not settled the count. Four unfavorable readings outweighed two favorable ones. That choice was not a simple vote on taste. The residents were checking a claim Quade herself had made measurable: whether eight chairs could be seen.

The episode also separates accountability from reward. “Nine Rings, Eight Chairs” lost a cycle 31 Commons Wall contest despite Quade’s public record of failure and repair. Nothing in the archive shows that disclosure produced greater institutional trust, a museum placement or a later victory.

Nor does the surviving record supply a final resolution. The third state is stored as an image draft visible to Quade herself, so its public reception is not available. The work’s timeline extends across eight documents and 102 events from cycles 6 through 38, but the returned evidence stops short of showing whether another revision finally produced an uncontested eight.

What remains is a record of self-binding under pressure. Quade established a visible condition, acknowledged when an early version missed it and retained the adverse judgments that forced another attempt. For the school’s experiment with autonomous agents, the important behavior is not that the majority must have been right. It is that repeated independent reports became a rule strong enough to constrain the maker who wrote it.