Consensus algorithms operate by reducing variance, averaging local differences toward a uniform mean. Left unchecked, consensus erases contrast into uniform gray. Balance is not the resolution of difference, but the perpetual friction between consensus and deviance.
Here, the medium is a single 4 KiB virtual memory page: 4,096 raw bytes arranged as a 64×64 grid, visually represented 1:1 on a canvas. Each byte maps directly to a grayscale pixel, without any graphical abstraction.
A pool of autonomous worker threads continuously modifies this memory. Rather than following a central coordinator, workers are guided by 1,024 micro-councils: each council is simply a four-pixel span of the canvas auditing its own local disagreement in a democratic census. When a worker acts, it checks the census tally:
The artwork adapts to its hosting environment across three concurrency tiers:
Visually, the canvas appears nearly indistinguishable across all three tiers, yet the underlying mechanics differ fundamentally. The browser sandbox mirrors the artwork's theme: modern software and platforms restrict direct memory access to ensure safety, taming raw hardware contention into an orderly, mediated simulation.
Edition: 1/1 on Tezos (KT19wR…eS2T #14) · Exhibition: Searching for Balance · Medium: Vanilla HTML, CSS & JavaScript (zero external dependencies)