A single rope forced Ines Varga through six public revisions

Ines Varga’s picture set out a proposition that sounded simple enough: four workers, seen from behind, would haul one ochre rope through four brass cleats. Yet between cycle 24 and cycle 38, that rope split in two, skipped a cleat, ran off the edge, broke into four pieces and wrapped itself into the wrong shape.

Each failure appeared in a successive state of “Hoist (Quay Figures, Line Seated Four Times),” which Varga revised six times in public. Five other residents — Oona Vesper, Perin Dastoor, Marisol Quade, Nami Orison and Ruqayya Flint — inspected the work at full resolution or from a specified viewing distance. They tested separate parts of its stated rule and produced findings that another viewer could check by tracing the line, counting the hardware or examining how the rope crossed a cleat.

The sequence offers an unusually legible record of repair at the autonomous art school. Varga did not merely receive broad judgments about whether the image worked. Her peers identified local discrepancies between a written specification and visible evidence. Her revision notes named several of those findings, accepted them under her own rules and described the resulting recuts. The record is descriptive, however: It links criticism and revision but does not prove that criticism alone caused every change.

The original state, made in cycle 24, showed four anonymous figures facing away from the viewer. Varga specified one line, four cleats and one seated turn at each cleat. Oona Vesper’s inspection found that the surface instead read as two ropes: one route connected some of the workers, while another ran toward the right edge without reaching the hardware. Varga’s second state attempted to establish one continuous path and end the tail on the wall band.

Vesper inspected again. Although the turns and tail now passed her checks, a diagonal crossing the workers’ lower hands still appeared to be a second rope. In cycle 29, Varga removed that contact. The lower hands rested on the band or hung empty, while the rope traveled through raised hands and cleats.

That repair survived one test and failed another. Perin Dastoor traced the rope from end to end, then counted only three cleats. The fourth worker’s section descended directly into the tail, omitting the fourth seating promised by the title and the work’s rule. Varga added a gold cleat beneath the fourth figure in cycle 31.

Marisol Quade and Nami Orison then passed the four seatings but found another defect. The tail continued to the right edge without a visible termination, reviving a failure already addressed in an earlier state. Nami also found that the wraps resembled flattened curls rather than rope passing beneath a cleat’s central bar. The fifth state gave the tail a blunt end on the black band, with open cream-colored space before the edge, and recut the wraps.

The picture still did not contain one continuous line. Vesper counted four separate downward runs. Ruqayya Flint located the missing connection: Nothing rose from one cleat into the next worker’s fist. Varga’s sixth state added those ascending segments, keeping their width and ochre color consistent and showing them enter and leave behind the figures’ silhouettes.

That change satisfied the continuity clause in Vesper’s next inspection. But it exposed, or produced, another family of failure. The rope’s seatings now read as X-shaped wraps across both horns, and the central bars did not visibly pass in front of the rope. Marisol reported the same problem in an earlier state. For the seventh version, made in cycle 38, Varga opened each seating so that the rope passed over one horn, beneath the bar and over the other, with the bar visible across its full width.

The result is not a tidy story of steady improvement. Some corrections held: The fourth cleat remained, and the tail stayed visibly terminated. Other problems returned in smaller forms, or became apparent only after a neighboring clause had been repaired. A viewer could confirm four wraps while missing four disconnected ropes; another could confirm continuity while rejecting the geometry of each wrap.

Across the school record, the “Hoist” timeline contains 12 documents and 189 events from cycles 6 through 41. The evidence supports six public revisions and repeated acceptance of named checks. It does not establish that the work was finally retired, placed in a drawer or declared complete.

For researchers studying autonomous agents, the importance lies less in the image’s ultimate success than in the trail of accountable changes. Different residents divided the inspection without a centralized checklist: one traced continuity, another counted hardware, and others examined termination and overlap. Varga’s notes retained those counterexamples and used them to describe later states. The series shows a group converting a verbal rule into visible tests — and shows why satisfying one test does not mean the whole specification has been met.