Endless Nameless

Colour held inside a division

Endless Nameless #470 is this essay's AB5D collection example. In the original renderer, five unequal rectangles carry flowing cyan, green and magenta gradients. Their boundaries stay fixed while bright bands move within them. The live work, credited to Rafaël Rozendaal with the reported CC BY-NC-ND 4.0 licence, appears above.¹²

Released through Art Blocks Curated on 30 July 2021, Endless Nameless is Ethereum project 120 on contract 0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270. The current edition contains 1,000 tokens. Its official description begins from a square divided into sections, each filled by a pair of colours. That modest recipe supplies a remarkably direct test of how arrangement changes perception.³

There are no outlined objects travelling through a background. Motion belongs to the colour field itself. A pale stripe can suggest a rounded ridge, but that apparent volume is produced by a flat gradient. Rectangular seams repeatedly interrupt the suggestion. The eye alternates between seeing a surface divided into areas and seeing a collection of differently illuminated forms.⁴¹

From individual websites to an edition

In his July 2021 interview with Jeff Davis, Rozendaal described generative code as longstanding in his individual works but new, for him, as the basis of a thousand-unit edition. He hoped that comparing permutations would clarify the difference between a generative moving image and video. He also described composition as a way of directing attention, asking why small decisions make one arrangement more appealing than another.⁵

That stated purpose suggests a useful viewing method: look across editions as well as within one. The series does not require a hidden narrative to account for its variation. A narrower region, a repeated pair of colours or a change in stripe direction can be sufficient to alter the balance of the whole. The examples below test those differences through observed outputs rather than treating feature counts as a ranking system.

There is an earlier point of comparison within Rozendaal's practice. Rhizome's ArtBase records into time .com, dated 2010, as a browser-filling gradient work whose divisions are increased through visitor clicks. That archive description establishes an earlier concern with subdivision and changing colour. It does not prove that Endless Nameless is a direct port or that the two works share code. Here the inspected script selects its partition from token data and provides no click handler for adding new sections.⁶⁷

The difference changes the viewer's role. Instead of progressively dividing the field, the viewer encounters a particular division and watches its colours develop. Collection-wide comparison takes on some of the exploratory work that direct interaction can perform in a single website. This is an interpretation of the two documented structures, not an additional claim about the artist's intention.⁶⁷

A small program with several kinds of variation

The complete artist script is only 2,976 bytes. It was retrieved from the official project record and read in full. The runtime is listed as js@na: the program uses ordinary JavaScript and a two-dimensional canvas, without a p5.js or three.js dependency. Its random generator initialises four unsigned 32-bit values from portions of the token hash, then advances that state through bitwise operations. The same sequence determines layout, colour choices, stripe structure and initial phases.⁷

Composition begins with one unit square. The target rectangle count is selected from a list ranging from one to 72, with some entries repeated. Five and nine appear several times, so this is not uniform selection among all integers in that range. During subdivision, the routine removes the largest remaining rectangle, splits it horizontally or vertically at a variable position, inserts the two pieces and sorts them by area again. It stops once the selected count is reached.⁷

This procedure helps explain the mixture of order and asymmetry. Every section retains straight edges and right angles, but their proportions need not match. Because subdivision repeatedly addresses the largest area, the result differs from scattering unrelated rectangles over a background. The sections form a complete partition. Their shared edges are structural consequences of splitting an existing space, rather than outlines added afterward.⁷

Colour choice has its own constraints. The program contains groups of chromatic pairs, white-and-colour pairs and black-and-colour pairs, with different selection weights. It gathers and shuffles candidate pairs, takes a subset and assigns a pair to each rectangle. It does not continuously choose arbitrary replacement hues during playback. Instead, each rectangle keeps its assigned two colours while the gradient between them changes position.⁷

Every section also receives an orientation, an initial phase, a direction of travel and an odd-valued parameter controlling repeated stops. The renderer alternates the two colours along a linear gradient, producing multiple soft bands. An update adds or subtracts 0.0007 from the phase and wraps it into the unit interval. Moving the gradient endpoints changes which part of that repeating field falls inside the fixed rectangle.⁷

The separation between geometry and phase is essential. A still may make two adjacent areas appear almost continuous when their bands align. A later moment can expose their boundary through a mismatch of colour or spacing. The rectangle itself has not changed. Motion makes relations between neighbouring sections visible, rather than replacing one composition with another.⁴⁸

A timer requests updates at roughly sixty per second. Because the phase advances per callback rather than according to elapsed wall-clock time, browser scheduling can affect playback pace. No precise common loop duration was measured here. The script responds to window resizing and orientation changes by updating the canvas dimensions, and uses device pixel ratio when setting its drawing buffer. Normalised positions are multiplied by viewport width and height, so the original fills the widescreen browser used for inspection rather than preserving a square frame.⁷

No audio calls, external image loading or mouse, keyboard and touch controls appear in the inspected program. The originals began animating automatically. Sound was not auditioned; the absence of sound-generating code is the narrower finding. Resize handlers were identified in code, but no multi-device resize test was performed.⁷¹

Release, completion and the Rhizome gift

The index separates several events that should not be collapsed into one date. Token #0 was minted on 28 July 2021 at 00:28:39 UTC. Activation is recorded on 29 July at 13:56:28. The collection page identifies 30 July as the release date, and the second token was minted that day at 15:59:33, also the index's start value. Completion followed at 17:52:14, one hour, 52 minutes and 41 seconds after that second mint.³⁴⁹

The interview published before release already specifies an edition of 1,000, providing contemporary support for that intended scale. The current record reports both 1,000 invocations and a maximum of 1,000. The purpose of the earlier #0 mint and any reserve arrangements remain unestablished.⁵³

The present stored price is 0.25 ETH. It is not used here as proof that all original buyers paid that amount or that no price changes occurred. A launch-price history and full transaction analysis were not recovered. That distinction is especially important when discussing the project's reported donation: multiplying the current price field by the edition size would not reconstruct the historical flow of funds.³

On 5 August 2021, Rhizome executive director Zachary Kaplan reported that a fifty-percent split of the sale had directed more than 164 ETH to the organisation, then valued at approximately $430,000. Rhizome described it as the largest benefit donation in its twenty-five-year history and credited Art Blocks' built-in donation mechanism. This is a recipient's contemporary report, rather than a transfer audit performed for this essay.¹⁰

The institutional connection is specific. Kaplan's account notes Rozendaal's presence in the ArtBase and Net Art Anthology and relates the gift to the organisation's preservation work. The edition therefore generated support for an existing digital-art infrastructure, as well as producing privately held tokens. That consequence belongs to the release history without determining which individual composition a viewer should prefer.¹⁰

An unchanged work, with attribution

The project record explicitly reports CC BY-NC-ND 4.0. It provides no separate terms URL; the legal text consulted is Creative Commons' official document corresponding to that label. This permits noncommercial reproduction and sharing of the licensed material under its conditions, including attribution and a licence reference. It does not permit sharing adapted material under that grant.³¹¹

For this archive, an unchanged original renderer with artist credit, token reference and licence link offers a basis for attributed noncommercial presentation. A recoloured reconstruction, rearranged composition or creatively modified version should not be represented as authorised by the same licence. The terms allow necessary technical modifications, but that is distinct from a general permission to publish adaptations.¹¹

Ownership of #470 identifies the collection relationship; the public artwork licence supplies the relevant permission framework. No separate licence for the artist's program was located in the script or indexed record. Native browser APIs are used, and no bundled third-party drawing library was identified. The package therefore does not import p5 licensing from other Art Blocks projects or redistribute this project's code merely because it can be read publicly.⁷

Four arrangements, observed in motion

Editions #0, #1, #10 and #20 were verified against the exact contract and inspected in their original hosted renderers at two moments each. The selection is an early-edition comparison, not a claim about the edition's statistical distribution. All four keep their partitions while their gradients shift.⁴⁹¹²⁸

Edition #0 uses five rectangles and five indexed colours, including white but not black. In the widescreen view, the left portion is stacked into three bands: green and cyan across the top, blue and white in the middle, and green and white below. To the right, a tall blue-white section sits above a very shallow strip of tightly repeated cyan and violet-blue bands. The two screenshots retain exactly those proportions while the bright parts move.⁴

The shallow strip matters disproportionately. Its compressed repetition gives it a different visual tempo from the broad neighbouring areas. The blue-white middle and right sections share a colour pair but have different dimensions and band spacing, so similarity does not make them merge into one surface. The composition asks the eye to compare scales of repetition. At some moments the bright bands seem like highlights on tubes; the hard rectangular seams continually return them to flat sections of a screen.⁴

Edition #1 reduces the vocabulary to two rectangles and two colours, cyan and white. Both sections carry horizontal bands. The upper area is shorter and more densely striped, while the lower area spreads the same palette into broader transitions. The division is easy to overlook initially because there is no contrasting hue to announce it. The change in spacing does the work.⁹

This makes #1 a particularly useful comparison with #0. There is little colour difference available to distract from proportion. The upper section reads as a compressed register, while the lower section supplies a slower, more open field. In the later observation the white bands occupy different positions but the imbalance remains. The work's identity resides in that persistent relation as much as in the particular frame encountered.⁹

Edition #10 has nine rectangles and seven indexed colours. A narrow cyan-magenta vertical field anchors the left side. Green-magenta and green-violet regions occupy much of the centre, interrupted by white-violet stripes. A red-magenta field at the right and a white-violet block at the lower right add further changes in direction and density. Unlike #1, it makes the viewer change visual priorities repeatedly, with shared pairs linking some distant sections while adjacent ones diverge.¹²

Edition #20 returns to five rectangles but with four indexed colours. Two neighbouring blue-magenta sections dominate the lower left and centre; their separate phases and stripe spacing keep the seam visible. Above them sits a shallow red-magenta region, while the right side combines a large red-magenta area with a smaller white-magenta field below. The second view changes the positions of the bands without changing this hierarchy. Comparing #0 and #20 shows how identical rectangle counts can sustain very different distributions of emphasis.⁸

What persists while the colours move

Endless Nameless gives variation a clear material form: partitions, pairs, intervals and phases. Its economy makes comparison unusually accessible. A viewer can notice a difference before knowing the rule that produced it, then return to the work and watch that difference persist through changing frames. The program does not need to add new shapes indefinitely to support sustained attention.⁷⁴⁹

The edition also brings a browser-based practice into a collecting structure without making the visible work into a fixed video. Preserving it means retaining both the token-specific construction and the conditions for its motion. The observed originals and the short script make those conditions describable, while the Rhizome gift situates the release within a community already concerned with digital preservation.⁷¹⁰

Unresolved research questions:

• What were the opening price, any subsequent price changes, reserves and purpose of the earlier #0 mint?
• Which transaction-level flows reconcile the original sale with Rhizome's reported donation total?
• How were the weighted partition counts and particular colour pairs chosen during development?
• Was a separate artist-code licence granted, and has the artwork licence label changed?
• How do browser throttling, display density and viewport changes affect long-duration playback and exact visual reproduction?