# Research notes

Series: Stroming by Bart Simons. Assignment AB5D-SERIES-088 in the pack's ordering. Package status draft-for-human-review.

## Sources and access

Ten sources were inspected and all ten are cited. Source identifiers are not contiguous; one planned identifier was not reused after a source was dropped.

Identity confirmed before research: chain 1, contract `0xa7d8d9ef8d8ce8992df33d8b8cf4aebabd5bd270`, project id 86, slug stroming-by-bart-simons. The atlas position for this series is also 86, which is a coincidence and was not relied on. The supplied metadata snapshot was treated as a lead and every field re-read live; refreshed values agreed with the snapshot.

Access route: the cloud container has no egress to artblocks.io, the Art Blocks index, the media proxy or the generator, all failing at the outbound proxy with 403 on CONNECT. All live observation ran through the browser pane on the operator's computer.

The artist website recorded on the project is an Instagram profile rather than an artist-controlled site. It was not treated as a source, because it serves nothing substantive to an unauthenticated reader, and no attempt was made to work around that. A web search for artist writing about this project returned only platform listings, the artist's Art Blocks user page and secondary roundups; that search and its outcome are recorded as a source for the negative result rather than the positive one.

No interview exists in the index and none was located.

## Full script inspection

Retrieved from `projects_metadata` filtered on the exact slug, with contract address and project id checked in the same response. Length 6,820 bytes. SHA-256 `9cdc8edd366835b058de3c2f97da7497e53c980f4e049d23e276daf40e9bb148`, computed over the UTF-8 bytes via the Web Crypto API in the browser. Script count 1 and the single fragment reproduces the script field exactly.

Coverage: the entire script was read in one pass. Structures traced: the sixteen literal HSB palette arrays; the hash decomposition into thirty-two bytes and the two seed derivations; setup in full, including the loop disable, canvas sizing, seeding, the scene function array, the view-mode bands, the palette selection and inversion, and the construction of the 180 by 180 Perlin angle grid; the draw routine including scene dispatch and the two texture passes; all sixteen scene routines; the stream field and stream drawing routines; the two mapping helpers; and the split and quad view dispatchers.

Direct findings recorded: the animation loop is disabled in the first statement of setup; the palette's first two entries are always background and underlay; scene routines interleave geometry with stream drawing rather than layering them; one grid routine occupies three of sixteen selection slots; the split and quad view bands are six and three hash values wide respectively; the stream mark grows along the stream and its colour walks with each step; and two noise textures are composited in multiply mode at the end, leaving three offscreen buffers in the document.

One finding is recorded with particular care because it would trip anyone reimplementing the work. The random seed is derived by parsing hexadecimal characters 16 to 30 of the token hash in base fifteen. Hexadecimal digits run to `f`, which is not a base-fifteen digit, so JavaScript's parser stops at the first `f` and returns only the prefix before it. The seed is therefore truncated, and the truncation point varies by token. Outputs remain fully deterministic and nothing is broken. No claim is made about whether this was intended.

External dependencies: declared `p5@1.0.0`, external asset dependency count zero. No network requests, no audio, no event listeners.

A third-party note is recorded in the rights block rather than here: one scene routine reproduces the behaviour of the one-line Commodore 64 random-diagonal program and the traits name it 10PRINT. What is reused is an algorithm and a piece of folklore, not code or text, and no third-party source is present in the script.

No block reference was pinned; this was an index retrieval. Script content is not redistributed.

## Token identity and visual observation

Five token identities were confirmed on the exact contract before use, plus a full sweep of all 256 rows. All four candidate identifiers supplied in the assignment exist. No token in this edition carries an empty trait object.

Two deliberate substitutions were made at the example level and the rationale is recorded because it matters. The edition divides into 249 Single Scene tokens, 5 Split Scene and 2 Quad Scene. All four suggested candidates are Single Scene. Adopting them unchanged would have documented only the common configuration and left the two rarest in the project unexamined. Invocation 249, one of the two Quad Scene tokens, was substituted in, and invocation 6, a 10PRINT scene with circular stream marks, was added to cover both the Commodore reference in the scene library and the minority stream shape. The records for invocations 1, 10 and 20 are retained in the evidence file.

Token page checks: five token URLs and the collection URL were requested and all returned status 200 at 2026-09-18T00:31:19Z. No status has been fabricated.

Images: four were loaded at native 2400 by 2400 through the platform media proxy and viewed.

## Motion and audio

Motion was tested rather than assumed, and the test mattered here because the index claims the work animates.

The live renderer for invocation 80 was opened and allowed to settle, then the artwork canvas was read directly three times at four second intervals. All three signatures were identical, each sampling 160,000 non-transparent pixels, so the readback was of a genuinely drawn canvas rather than a blank one. Four canvas elements were present: the artwork at 1324 by 1324 and three offscreen buffers at 1320 by 1320 left from the texture passes.

That matches the script, whose setup routine calls the p5 loop disable before anything is drawn. The work is static. The `display_static` field in the index is false, which asserts the opposite, and is recorded as conflicting with an explicit note. The project page carries no animated badge and is correct.

This is the reverse of the defect found in other projects in this batch, where genuinely animated works carry no animated badge on their pages. Taken together the two patterns suggest the flag and the badge are maintained independently, which is worth the coordinator knowing before relying on either.

Audio: no audio or video element in the rendered document and no audio anywhere in the script. No sound was heard because there is none. Autoplay policy never engaged.

Interaction: no event listeners in the script, no documented controls, nothing responds to input.

## Header ownership and media

The complete current AB5D wallet set was read from the platform's own profile record: `0xce6dc14c5494d8a4325a81d3b257d130fa2039d1`, `0x2247247ee0eee59557e6cd14f7e047f6a9309b5c` and `0xdd19308a9731cdecb67e0d80ca6823394208c045`. The ownership sweep returned token 86000080 held by the third address. The assignment's cached candidate named the same token and owner; the live check confirms rather than repeats it.

Because the work is static, the presentation file recommends a still rather than the live renderer, and records explicitly that the index flag must not be used to describe the work as animated.

## Mint chronology and calculations

Events distinguished: first indexed mint 2021-06-11T13:36:00Z; activation 2021-06-17T15:54:17Z; second indexed mint 2021-06-20T19:06:00Z, duplicated in `start_datetime`; completion 2021-06-20T19:49:08Z.

The owner of invocation 0 is the artist address on the project. This is one of only two projects in the batch where that is true.

One calculation is performed: second indexed mint to completion, 43 minutes and 8 seconds, covering invocations 1 through 255. Labelled inferred. Not presented as a public sellout time, and the three day gap between activation and the second mint is recorded because it shows activation did not open the sale.

Sampled mint times show invocations 1, 10 and 20 at 19:06:00Z, 19:06:08Z and 19:06:31Z, so at least twenty-one invocations went in the first thirty-one seconds, while the AB5D holding at invocation 80 was minted eight minutes in. The head of the sale was much faster than the tail. This is stated as an observation about sampled tokens.

Timezone handling: the project page displays 21 June 2021 when read from a UTC+10 browser, which is the second-mint timestamp rendered locally; in UTC that timestamp is 20 June.

Price: 85000000000000000 wei, that is 0.085 ETH, recorded as the currently configured value and as reported. No primary announcement of an original price was located. The page's present secondary floor of 0.1498 ETH is recorded as an observed figure only, with no inference drawn and no valuation commentary anywhere in this package.

## Rights and conflicts

The only rights statement located is the index field `license` reading "CC BY-NC 4.0" with `link_to_license` null. The project page publishes none, and the artist website of record is a social profile, so there is no artist-controlled page on which terms could be published; the absence there is not evidence of anything and is not treated as such.

The licence census across all 359 projects on this contract returns 34 distinct strings with only 87 carrying any link, which establishes the field as unvalidated free text.

Artwork rights recorded as category creative-commons, label preserved exactly, terms url null, status reported. Code rights unknown and kept separate. Two third-party components are recorded: p5, declared and used substantively, upstream licensing not inspected; and the 10PRINT algorithm, recorded for completeness with an explicit note that an algorithm and a piece of folklore are not a copyrightable text and that no third-party code is present.

## Unresolved questions and exceptions

Seven exceptions are raised, all pending, all assigned to the publication coordinator: the wrong index static flag; missing licence terms; unknown code rights; the artist record gap; the unattributable trait vocabulary; the base-fifteen seed; and the unverified original price.

Unresolved leads not converted into sources: no interview or artist writing was located; the artist website of record is a social profile and was not pursued; the trait extractor producing the Dutch palette names and the nineteen scene names is not in the retrieved script and was not located; upstream p5 licensing was not inspected.

## Actual model, tools, elapsed time and measured usage

Model: claude-opus-5, served through the Claude desktop application in Cowork mode on the Claude Agent SDK. The batch routing label was "claude", a routing label and not a model identifier. The agent is not Astra.

Tools actually used: HTTP POST to the Art Blocks public GraphQL index; a browser pane on the operator's computer for page rendering, text extraction, image inspection, live renderer execution and canvas sampling; the Web Crypto API for the SHA-256 digest; web search for artist material; local shell and file tools for package assembly and validation. No human reviewed or edited the prose.

Word count: the coordinator waived the 1,500 to 2,500 word limit for this batch during the session. This essay runs longer than the original cap by that instruction.

Elapsed work: produced within a single continuous session on 18 September 2026, Melbourne time.

Telemetry: no token usage telemetry is exposed to the agent. All counts are null in usage.json with the reason given. Nothing estimated and labelled measured; batch totals not divided into invented per-essay figures.


## Coordinator review

OpenAI Codex independently read the artist program, inspected four original renderers, refreshed all 256 token records and the AB5D wallet set, and located the artist project page. Revised technical claims include fourteen distinct scene routines in sixteen slots, possible NaN from base-fifteen parsing, non-monotonic mark size, and a discarded first texture. Corrected artwork observations, chronology and rights. Submitted signature sampling, broader licence census and earlier page tests remain attributed to Claude. No separately measured token usage. Standard artwork terms linked; historical terms and artist-code grant remain unresolved.