8

The work and its context

8 is Art Blocks project 179 on the contract at 0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270, on Ethereum mainnet, credited to Bård Ionson. The indexed record returns 128 invocations against a maximum of 128, a declared dependency of p5@1.0.0, a square aspect ratio, and a display_static value of false ¹. The live project page describes it as an Art Blocks 500 release in the Factory category and carries both an interactive badge and an animated badge ². Both badges are earned, which is not something this essay can say about every project of the period.

What you see is a wheel. A ring of eight spokes turns around a central object, and each spoke is a stack of three bars. Those bars are an I Ching trigram: three lines, each either solid or broken, giving eight possible figures. Eight spokes, eight trigrams, and the figure at the centre changes from token to token. It may be a crescent moon, a lemon, a plum, a cherry, a lamp, a billiard eight ball, a pair of eyes, two pills, or an identity symbol drawn in outline. The palette is drawn from a small set keyed to the trigram names, and the ground is one of four colours ³.

The artist's mint statement is unusually specific about what the work is made of. It names the I Ching, calls the piece "a cosmic slot machine", refers to "the binary I Ching of 8 sides & planetary gender", states that at least 12 per cent of profit goes to the Casa Brumar Foundation, Safe Space and other LGBTQ+ charities, and then lists the keyboard and touch controls ¹. The project page repeats that text and adds a charitable giving note naming Casa Brumar Foundation, Safe Space NOVA or The Gathering Place, and a display note recommending a mounted touch screen ².

Three facts should be fixed before the essay proceeds, because each of them is easy to get wrong from the public record. The number eight governs the sale mechanism as well as the picture, and the sale took 655 days to finish rather than the minutes its auction ladder anticipated ²³. The identity symbol that the mint statement foregrounds is selected for every token but drawn on only 42 of the 128 ³. And most of these works are not loops: they draw for a hash-determined interval, then stop for good ⁴.

Conception and development

There is no artist interview in the indexed record, which returns null for that field ¹. The website recorded against the project is https://bardionson.com/8. Navigating to it in a browser returns a page titled "404: This page could not be found" ⁵. The domain is live and the artist maintains a substantial site there; the link of record simply points at nothing. A cataloguer who follows it and stops would conclude there was no artist material.

The artist's current site does carry an entry for this work, in a portfolio grid, under the name "The 8 (Art Blocks)". Its blurb calls it "A seminal piece of generative art exploring numerical symbolism and mathematical beauty through code" and links to the Art Blocks collection page ⁶. That description is worth setting beside the 2021 mint text, because the two do not emphasise the same things. The mint text is explicit about the I Ching, about gender and identity symbols, and about the charitable donation. The current portfolio blurb mentions numerical symbolism and mathematical beauty and none of the rest. This essay records the difference as a difference in published framing between 2021 and now. It does not infer a change of intention, and no source consulted here states one.

The artist's own biography describes a practice built on "an uncommon combination of tools and techniques such as artificial intelligence, GANs, VCRs, CRTs, oscilloscopes, lasers, scanners, photography, smart contracts, blockchains and sound", lists exhibitions in Valencia, Vienna, Brussels, Paris, Denver, Washington DC, Dubai, Riyadh, San Francisco, London, Miami, Manchester, Essen, Berlin, Lisbon and Marfa, and names permanent collections including the Museum of Crypto Art, MoCDA and the Bodleian Library at Oxford ⁶. The same page states that the artist has been making art since 2012 and entered the NFT space in 2018 through SuperRare, and names this project among three works the artist is known for ⁶.

One further piece of authorship evidence sits inside the program itself. Its first line is a string carrying a copyright notice that names the artist's company and the year 2021, declares CC-BY-NC-SA, and credits the random number function to a third party ¹. That notice is discussed in the rights section, where it conflicts with the licence the platform reports.

What cannot be established from the available record is the development history: when the piece was begun, how the centre figures were chosen, or why the trigram palettes take the particular colours they do. No dated development account, studio note or interview covering this project was located. That absence is recorded as an absence.

How the work was constructed

The program was retrieved from the indexed project record by an HTTP POST to the Art Blocks public GraphQL endpoint, using a named operation filtered on the exact slug, with the contract address and project ID returned in the same response used to corroborate identity ¹. The record reports a script count of 3, and the three indexed fragments concatenate in index order to exactly the single script field. The assembled program is 23,729 bytes and hashes to SHA-256 f8916e0a1399485e9c786109347b222a2fde0a2a7065bbda1744fd4b46fdda96. It was retrieved twice in separate requests during this session and produced the same digest both times. The whole program was read. This was an index retrieval over HTTPS, not a direct read of contract storage, and no block reference was pinned. No artist code is reproduced in this package.

The seed handling is compact. Four eight-character slices of the token hash are parsed as unsigned 32-bit integers and become the state of an xorshift generator, which the in-code comment attributes to a third party ¹. That generator is drawn 41 times, and each draw is scaled to an integer between 1 and 255. Those 41 values are the entire parameter supply for the work; nothing else is read from the hash.

Two details of that supply are worth recording because they are invisible from the outside. The reader function increments its index before returning, so the first parameter read takes the second of the 41 values and the first is never used. Counting the reads the program actually performs gives 36, so five of the 41 generated values are discarded unused. Neither fact changes what a viewer sees, but both matter to anyone attempting to reproduce the mapping.

The first 24 reads build the wheel. Three at a time, each scaled to 0 or 1, they produce eight three-line figures, and the program maps each to its I Ching name and Unicode glyph. The remaining twelve reads select, in order, the identity symbol, the centre figure, three colour components, a palette index, a rotation rate, a fruit selector, a binary flag, a speed value, and the ground colour.

To test that reading, the mapping was reimplemented independently and run against the hashes of all 128 tokens, then compared with the features the index publishes ³. Four rules reproduce the published value for every token in the edition: the Triglyphs string, at 128 of 128, matching the full eight-glyph sequence character for character; Speed, at 128 of 128; Background, at 128 of 128 across its four values; and the fruit sub-case of the Center trait, at 9 of 9 where it applies. All 128 published Triglyph strings are distinct, which is what eight independent draws from eight figures should give.

The Pa Kua Palette trait reproduces at 126 of 128. Both exceptions are tokens whose centre figure is the two-pill case, and the explanation is in the drawing code rather than the parameter code: that case reassigns the palette index at draw time, to the value whose published name is Earth. All three tokens with that centre figure are published as Earth regardless of the palette index their hash produced. The inference this supports, and it is marked as an inference, is that the published features were captured by running the artwork rather than by statically evaluating its parameter section, because the overwritten value is only observable after the drawing code has run.

The most interesting result concerns the Glyph trait. The identity symbol is selected for every token from a list of thirteen names, which includes nonbinary, transgender, bigender, lesbian, gay, heterosexual, and the planetary signs for Venus, Mars, Uranus and Neptune. But the index publishes a Glyph key for only 42 of the 128 tokens; on the other 86 the key is absent altogether, not merely null ³. Reading the drawing code explains this exactly. The symbol is rendered only by particular paths: the large-symbol centre case, the moon cases when the binary flag is set, and one branch of the ring-drawing routine when the fruit selector is high enough. Predicting the presence of the Glyph key from those conditions matches the index on 128 of 128 tokens, and where the key is present the predicted name matches on 42 of 42.

So the trait is not missing. It is being honest: the metadata names the symbol when the symbol is on screen and stays silent when it is not. The consequence is that two thirds of this edition carry an identity symbol, chosen from the hash, that no viewer will ever see. Whether that is a designed effect or a by-product of how the features were captured is not established by anything read here, and the question is left open in the closing section.

The last mechanism is the one that most changes how the work should be presented. The wheel turns because a rotation variable is decremented every frame by a hash-derived rate. Most centre cases then call a routine that compares that variable against a fixed threshold and, when it is passed, halts the sketch permanently. The work is therefore usually a finite drawing rather than a loop: it runs, it completes, and it freezes. The running time is a per-token property, because the threshold is fixed per case while the rate comes from the hash.

That was tested directly rather than inferred. Token 179000000 was predicted to stop at frame 813 from the rate its hash produces and the threshold its case uses, and was observed stopping at frame 814, after which the frame counter and the rotation variable were identical across the following samples ⁴. Token 179000025 was predicted at 518 and observed already halted at exactly 518 ⁴. Three cases never call the halting routine at all, and one of those, case 22, subtracts a value large enough to drive the rotation variable positive instead of negative, so its threshold can never be reached. Twelve of the 128 tokens fall in those cases, and one of them, token 179000027, was observed still running and still changing after 21 seconds with its rotation variable climbing past 12,000 ⁴.

One further drawing decision explains why these pictures look dense. The background is repainted at the start of a frame for exactly one centre case. For every other token the canvas is never cleared, so each frame is drawn over its predecessors and the image accumulates until the work stops. The trails visible in every example below are that accumulation, not a motion blur effect.

The mint

Five distinct events need separating here, and the public record labels them inconsistently.

The first token was minted on 9 October 2021 at 00:34:49 UTC ³. The project was activated six days later, on 15 October 2021 at 23:38:20 UTC ¹. The public opening, recorded in the index as the project start datetime, was a further week after that, on 22 October 2021 at 16:00:00 UTC ¹. The project page displays "Released on 23 October 2021", which is that same opening instant rendered in a local timezone ahead of UTC ². Completion, the moment the final token was minted, was on 9 August 2023 at 02:31:23 UTC ¹³. Anyone reducing these to a single release date will pick the wrong one for at least one purpose.

The sale mechanism is recorded on the project page and it is built from the title. The sales note states a Dutch auction starting at 8 ETH with a 0.18 ETH resting price, reduced every 8 minutes through the ladder 8, 5, 3, 2, 1, 0.618, 0.38, 0.18 ². That is eight rungs, falling at eight-minute intervals, from eight. Two of the rungs, 0.618 and 0.38, are the golden ratio conjugate and its complement; this essay notes the coincidence without asserting that the artist chose them for that reason, since no source consulted says so.

Setting the ladder against the mint timestamps gives a reproducible picture of how the sale actually went. From the 16:00:00 UTC opening, the ladder reaches its 0.18 resting price at 16:56:00 UTC. The first token minted after the opening was invocation 1, at 16:52:15 UTC, which falls in the 0.38 rung ³. Exactly two of the 128 tokens were minted inside the 64-minute descent. The auction, in other words, fell all the way to its floor.

What followed was slow. Counting from the public opening: one token preceded it, 2 were minted in the first hour, 8 more during the rest of the first day, 11 across days two to seven, 6 across days eight to thirty, 80 across the remainder of the first year, and 20 after the first anniversary ³. The longest gap between consecutive invocations was 95.2 days, between invocations 93 and 94, with further gaps of 84.2 and 65.5 days ³. The sale duration from public opening to the final mint is 655.4 days, calculated as the final mint at 2023-08-09T02:31:23Z minus the opening at 2021-10-22T16:00:00Z, divided into days ¹³.

The price the index currently reports, 0.18 ETH, is not a later repricing: it is the auction's resting price, which the ladder reached 56 minutes after opening and which the remaining 126 tokens were minted at over the following 21 months ²¹. That distinction matters for a catalogue, because a reader seeing a current price of 0.18 ETH beside an edition that opened at 8 ETH would otherwise have no way to tell which of the two was ever the asking price.

Copyright and licensing

Three separate rights domains apply here and they do not agree with one another.

The artwork licence the platform reports is the string NFT 2.0, with a link to https://www.nftlicense.org/ ¹. That page was read in full ⁷. It presents the NFT License, attributed at its foot to Dapper Labs, as a plain-language document: it grants the token owner personal non-commercial use of the associated art, display on a marketplace that cryptographically verifies ownership, display on a third-party site or app that verifies ownership and does not retain the art afterwards, and commercialisation on the owner's own merchandise up to 100,000 dollars of revenue a year. It prohibits modifying the art, using it to market third-party products, using it in connection with hatred or violence, and attempting to trademark it ⁷.

Two things about that page need recording precisely. It states no version number anywhere in its text, so the "2.0" in the platform's label is not corroborated by the document the platform links to. And its own navigation offers a "License" item that is not a working link; the path https://www.nftlicense.org/license returns an error page ⁷. What is readable at the link of record is therefore a summary of terms rather than an executable licence instrument. The label is preserved here exactly as reported and recorded as reported rather than verified.

Against that sits a different claim inside the artwork. The first line of the program is a string containing a copyright notice that names the artist's company, the year 2021, and the identifier CC-BY-NC-SA, together with a credit for the random number function ¹. That is a Creative Commons Attribution-NonCommercial-ShareAlike claim, which is not the NFT License and is not compatible with it in an obvious way: ShareAlike contemplates derivative works released under the same terms, while the NFT License prohibits modifying the art. This package does not resolve which instrument governs. It records that the platform's field and the artist's in-code notice state different licences for the same work, that no signed instrument for either was located, and that the conflict is unresolved.

A third tension is worth flagging for anyone presenting this work. The NFT License prohibits modifying the art, while the artwork itself ships documented controls that change what is drawn: a key that cycles the ground colour through four values, keys that speed up and slow down the rotation, keys that move the identity symbol, and a touch handler that stops and resumes the drawing ²⁴. A viewer using the controls the artist documented is altering the displayed image. Whether that constitutes modification in the licence's sense is a legal question this essay is not equipped to answer, and it is left as an open question rather than resolved in either direction.

The artist's display note bears on the same point and should travel with the work: "Optimal display is a mounted touch screen to allow viewers to interact with it. But a still image can be chosen and then printed on linen paper or on canvas" ². That is the artist's own statement of display basis, and it explicitly contemplates the viewer choosing which still to take from a work that stops at a hash-determined moment.

Code rights are separate and are recorded as unknown. The program is stored on chain and can be retrieved by anyone; that establishes readability, not redistribution rights. The in-code notice is evidence of the author's claim over the code but was not located as a licence file or signed grant. For that reason no part of the program is reproduced in this package, only its digest, its measured behaviour and a description in prose. The declared dependency, p5@1.0.0, carries its own upstream licensing, which was not inspected and is recorded separately as unknown rather than folded into either domain above ¹.

Close reading of examples

Five tokens were opened in the original renderer and observed across changing states. Four are recorded as examples and two are treated at length. Every canonical token page below returned HTTP 200 when fetched during this session, and every observation is the researcher's own inspection of the live renderer rather than a preview image ⁴.

8 #18, token 179000018, is the AB5D holding and the recommended header work. Ownership was verified rather than assumed: the complete linked wallet set was refreshed from the AB5D profile during this session and returned three addresses, and this token's indexed owner is the third of them ⁸³. Its published traits give Center as Lamp, Background white, Pa Kua Palette Earth, Speed 45, Type Pa Kua, and no Glyph key at all ³.

On screen it is a white field holding a wheel about two thirds of the frame. The outer ring is divided into eight segments coloured in slate blue, violet, dark olive and pale sage, each carrying its trigram as a stack of three pale turquoise bars with gaps where a line is broken. Inside sits a grey disc, and on the disc a white lamp or torch shape, tilted, throwing a short yellow beam. Observed from frame 1207, the image changed at every sample while the ink count fell steadily from 1,058,534 to 1,036,134 pixels; between the samples at 10.1 and 12.6 seconds the frame counter stopped at 1701 and the picture became identical from then on ⁴. This is the halting behaviour described above, caught in the act.

Its controls were then tested with real key presses. Clicking the canvas resumed the halted drawing, advancing the frame counter from 1701 to 2263. Pressing C changed the ground from white to black, confirmed both in the sketch's own state and in a screenshot. Pressing F three times multiplied the speed value from 45 to 77.76, which is exactly 45 times 1.2 cubed. Pressing D four times moved the symbol offset from minus 30 to plus 10, a change visible in the sketch's state but not on screen, because this token's centre case does not draw the symbol ⁴. The space bar was intercepted by the browser pane before reaching the page; dispatched into the sketch directly it both resumed and, on a second press, halted the drawing, with the frame counter frozen across the following samples ⁴. That distinction is recorded rather than smoothed over: the handler works, the physical key did not reach it in this environment.

The hidden symbol is this token's most telling property. Its hash selects the Mars sign, and the Lamp centre case never draws it. A collector looking at 8 #18 is looking at a work whose identity symbol was chosen, recorded nowhere in the published traits, and withheld from the picture.

8 #0, token 179000000, is the first token minted and the clearest case of the complete lifecycle. Its traits give Center Glyph, Background Turquoise, Pa Kua Palette Earth, Speed 59, and Glyph neptune ³. On a turquoise ground, a green disc carries the Neptune trident drawn in black outline, ringed by eight segments in blue, slate, pink and green. The reproduction predicted the symbol name neptune from the hash, and the drawn trident confirms it visually ³⁴.

It was observed running at frames 496, 646 and 796 with every signature distinct and the rotation variable falling steadily from minus 5.18 to minus 8.32. Between the samples at 5.0 and 7.5 seconds it halted at frame 814 with the rotation variable at minus 8.5071, and the following samples were identical ⁴. The predicted halt frame from the code was 813. A one-frame difference is what the threshold test should produce, since the comparison fires on the first frame past the boundary.

8 #30, token 179000030, was selected because its binary flag is set and its centre case is the moon pair, which is one of the paths that does draw the identity symbol. A yellow crescent sits on a black disc inside an octagon whose eight segments run navy, steel blue, pale grey, teal and cream, each carrying its trigram bars in orchid pink, on black. Over the crescent, outlined in red on dark red, is the Venus sign: a ring with a short cross-stem, turned roughly a quarter turn from upright because the drawing routine rotates the symbol with the wheel. Its published Glyph trait is venus and the reproduction predicts venus from the hash, and the drawn shape agrees with both ³⁴. It was observed changing across five samples from frame 906 to 1508, then caught at its halt: the rotation variable reached minus 5.5016 against a threshold of minus 5.5, the frame counter froze at 1732, and the value was unchanged three seconds later ⁴. Its rate of 0.0031765 per frame against that threshold predicts 1732 frames, which is the frame count observed.

8 #25, token 179000025, is the Eight Ball. A black sphere carrying a white circle with a black numeral 8 sits against a yellow crescent, on a wheel of red, blue, teal and cream trigram bands. The numeral is drawn from a coordinate list at the end of the program that uses not-a-number pairs as pen-up markers. This token had already completed before the first sample: the frame counter read 518 and the rotation variable minus 7.5059 at every one of six samples across 12.5 seconds ⁴. Its case applies an additional rotation decrement beyond the hash-derived rate, and the two together predict a halt at frame 518.

A fifth token was observed to test the other side of the halting classification. 8 #27, token 179000027, falls in a case that never calls the halting routine and whose rotation decrement is large and negative, driving the variable upward rather than down. It was observed across seven samples over 21.1 seconds, all seven signatures distinct, the frame counter advancing from 468 to 1551 and the rotation variable climbing from 3,739 to 12,393, with no sign of stopping ⁴. On screen it is a white disc packed with dozens of pink-irised eyes spiralling inward to a black centre, on black, and it grows denser as it runs because the canvas is never cleared.

On sound, all five renderer sessions were silent. This is not a blocked-autoplay finding: the page loads only the p5 library at version 1.0.0 with no sound add-on, the document contains no audio or video elements, and the program contains no audio calls ⁴. The work is silent by construction, and no playback action was required or attempted.

Significance and limitations

The case for this project is that its structure is unusually legible and unusually consistent. A work titled 8 draws eight spokes, reads eight three-line figures, sells through eight rungs of a ladder falling at eight-minute intervals from eight ether, and puts a billiard eight ball at the centre of some of its tokens. That consistency is verifiable rather than thematic assertion: the trigram strings reproduce from the hash on all 128 tokens, and the auction ladder is published in the project's own sales note ²³.

The case against treating it as a solved object is that its most loaded element is the one least visible. The mint statement foregrounds identity symbols and an LGBTQ+ charitable donation. The program selects such a symbol for every token in the edition. It draws one on 42 ³. On the other 86, including the work AB5D holds, the symbol exists in the token's hash and in the program's memory and nowhere else. A catalogue that lists traits from the index will show a Glyph for a third of the edition and nothing for the rest, which is accurate, and will give a reader no way to know that the silence means withheld rather than absent.

Three things would mislead a cataloguer working from the public record alone, and all three are correctable.

The first is the word animated. The platform's badge is right that these works move, but most of them are finite drawings that stop: four of the five tokens observed here halted, three of them while being watched, at frames that the code predicts from the token's own hash ⁴. A dossier that presents 8 with a looping player, or that describes a late unchanging state as evidence that a token never moved, will be wrong in opposite directions on different tokens.

The second is the licence. The platform reports NFT 2.0; the linked document states no version; the program's own first line declares CC-BY-NC-SA ¹⁷. Publishing the platform's label unqualified would pass on a rights statement the artwork itself contradicts.

The third is the website of record, which returns a 404 while the artist's live site carries an entry for the work at a different address ⁵⁶.

The limitations of this package should be stated plainly. Physical key presses reached the sketch for the C, F and D controls but the space bar was intercepted by the browser pane, so that control was confirmed by dispatching the event into the page instead; the E and X keys and the W, A and S movement keys were not individually exercised. Frame counts are the renderer's own counter, and wall-clock intervals are recorded alongside them, but no claim is made about the frame rate the artist intended. Only five of 128 tokens were observed, and the halting classification for the remaining 123 rests on reading the code rather than on watching each one. The trait extractor that produced the published features is not part of the retrieved program and was not located, so the inference that the features were captured by running the sketch, though well supported by the palette override, remains an inference.

Unresolved research questions, stated explicitly as required:

Is the 86-token silence in the Glyph trait a designed withholding or an artefact of how the features were captured? The mechanism is established; the intention is not, and no located source addresses it.

Which licence governs the artwork: the platform's NFT 2.0 label, or the CC-BY-NC-SA notice in the program's first line? No signed instrument for either was found, and the two are not obviously compatible.

Does a viewer exercising the artist's documented controls modify the art within the meaning of the NFT License's prohibition, given that the artist's own display note recommends a touch screen precisely so that viewers can interact?

What accounts for the 655-day sale, and specifically for the 95-day gap between invocations 93 and 94? The timestamps are firm; no contemporaneous account explaining the pace was located.

Where is the artist's original development record for this project, and does it exist? The website of record is broken, the indexed interview field is null, and the artist's current portfolio blurb describes the work in terms that omit its stated subject matter.