What shared work leaves behind
The history of a creative process is not contained only in its final performance. How an invitation is made, participation is described, suggestions become tasks, and work reaches an audience also deserve to be preserved. Organizing documents around events reconnects photographs, recordings, texts and works through these questions. This essay sets out the method to be applied in stages across the library’s subsequent project research. Adopting a method does not mean that research on every project is complete.
The public recruitment notice for Becoming Wind, dated 1 February 2023, presents writing, directing, rehearsing, designing, performing and documenting as different ways to join the work. It also makes room for participation behind the camera. It proposes presenting an outcome at the end of the month while leaving the eventual direction to collective discussion. These are invitations and plans for future work, not evidence that the proposed outcomes were achieved. [S1]
Start with a question, then define the event and scope
Begin with a question the materials can answer, then choose the scope: how a proposal receives a response, how limits on participation affect arrangements, or whether a task leaves evidence of completion. The event is the main unit of narration; files provide access to it. One meeting may span several files, and one work may involve several meetings. Processing windows must not automatically become separate events. Identify the beginning, local changes and the outcomes currently visible, leaving gaps where evidence is missing.
Bounded reading starts with metadata and existing text, then adds selected stills, audio checking or continuous viewing as the question requires. Record which text was actually read, which audio was checked and which images were viewed, with separate ranges and quality notes. Repetitive, low-quality, out-of-range or unintelligible machine text should be marked unusable, rather than repeatedly processed to inflate coverage. Stop at the boundary of personal narratives unrelated to the work question instead of extracting them merely to cover the whole recording.
Make sources, timelines and coverage checkable
Keep a stable identity and version for each source, together with its relationship to original recordings, parts, exports, subtitles and republications. A different path does not necessarily identify a new source, and a repost or excerpt is not independent corroboration. A sample checksum can help check a file location; claiming identical copies requires a full checksum or a reliable production record. Keep work, recording, publication, planned and retrospective dates distinct, retaining conflicts when they exist.
To assess adjacency, compare available recording logs, container times, the preceding file’s duration, boundary scenes and unfinished topics. Consecutive numbers or modification times do not establish continuity. Calculating gaps from device times also requires stated assumptions about clock synchronization and what those timestamps mean. Until audiovisual alignment is checked, retain independent timelines rather than constructing a falsely precise combined one. Label subtitle, source-recording and public-edit times separately, and create cross-edition links only after verifying their mapping.
Track coverage by source, stream, interval, observation method and method version, distinguishing processing completed, text read, audio checked and material rejected. A transcript’s existence does not mean it has been closely read; reading is not audio verification, and sampled frames are not continuous viewing. Each article should state what was examined and what remains unread, rather than using one percentage to represent every activity.
Write proposals, responses, actions and outcomes as a process
Describe how different positions meet while distinguishing the kinds of claims being made. If someone hypothetically offers to edit, that is initially an intention. A claim that the task was undertaken or completed needs checkable production or completion evidence. Permission is not proof that an action occurred, and a recollection of the past is not direct observation of that past. An outcome assessment also needs an identified evaluator and time. If no response appears within the examined range, say so; do not turn that into a claim that nobody responded.
Keep local anonymous labels for statements whose speakers are unverified. Without reliable speaker separation, even one passage must not be assumed to contain only one speaker. Personal attribution requires non-biometric evidence such as explicit credits, an audio-checked self-introduction, production records or explicit confirmation, not faces, voices, clothing or filename prefixes. Preserve separate sources for proposers, respondents and absent people represented through others’ accounts. Distinguish conflict between fictional characters, personal recollection and practical work negotiation.
Make concepts testable and revisable
A concept should explain a specific change rather than give an event a grand name. State its evidence anchors, the researcher’s interpretation, alternative explanations and the new evidence that would invalidate the claim. For example, treating shared creation alongside documentation as a proposed working method requires separate evidence for the creative activity and the recording arrangement. If later evidence shows only fixed interviews without the proposed activity, revise the description to an unimplemented or changed proposal. This is an illustration of testing a concept, not a factual claim about a private recording.
A process unit may contain a statement, response, operation and outcome, but the surviving material may not preserve all four. Show readers how evidence supports the account and where interpretation extends beyond direct observation. Retaining counterexamples, differences and unresolved questions matters more than fitting fragments into a smooth story of progress.
Compare working mechanisms without presuming influence
The BODY interview study offers one comparison: how abstract character relationships become workable supports for performers. It draws on several published retrospective accounts, not a continuous conversation in one session. The forum-theatre study considers the announced rule allowing spectators to stop a second run and try another response; the announcement does not establish that a particular intervention occurred. These studies help formulate questions without supplying missing actions for another project. [R1][R2]
The Middle production study separates planning proposals, public schedules and pre-performance credits, reminding us that neither a proposal nor a credit establishes completed work. The Four Eyes transition study retains a candidate process of proposing, questioning, choosing and retrying. It remains an anonymous method analysis without audio verification; it cannot establish actions in another event or turn anonymous positions into named people. [R3][R4]
The existing GIF-edition document for Shadow Counterpoint suggests another comparison: how live collaboration becomes reviewable moving images, animated images and text. Different media provide different observational scope. This is a comparison of mechanisms only. Shared themes, institutional affiliation or chronological order do not establish influence. A claim of continuity or influence requires evidence such as explicit references, production documents or actual reuse of material. [S2]
Complete the account, then review what can be published
Each research batch should produce a continuous event account, a paragraph-level evidence table, changes in coverage and the next unresolved questions. Structured indexes support retrieval and deduplication; they do not replace the article. Edit the public text separately instead of publishing a private account with names removed. Review publication status, quotation boundaries, identity, privacy and editing context. Unverified harmful allegations, personal histories, unreviewed images and unchecked machine transcripts do not become publishable merely because processing is complete.
Ordinary work facts suitable for publication, explicit public plans and evidence-labelled interpretations still have different evidentiary strengths. Images need an accurate work title, edition, time and factual caption. Do not manufacture a viewing link when public access or time mapping is unverified. Translations must preserve the same factual scope and uncertainties rather than becoming more definite than the source text. A designated editor checks the actual payload and authorized scope before each publication; processing pipelines do not publish automatically.
Keep revisions retrievable and the next question explicit
Changing a source’s location does not change its identity, and citing an earlier claim again is not a new discovery. New support, counterevidence or revision should identify its relationship to the earlier claim and explain the change while retaining the previous text and citations. Each public edition preserves its text, evidence locators, translation and revision note at a retrievable fixed address. Later material may complete an account or overturn an earlier interpretation.
Prioritize responses or actions that could change a conclusion, important counterevidence and missing perspectives, followed by source-version questions and contextual additions. Record why a task is needed, its entry point, previously examined ranges, its processing limit and its stopping condition. When successive reading only expands personal background without answering the work question, stop and report what was not found instead of substituting a longer transcript for clearer research.
The scope of this publication
This is a public editorial edition of version 1.1 of the continuing research guidelines. It uses ordinary research methods, planning content from a verified public recruitment notice, and comparisons with existing public studies. It does not claim a complete reconstruction of a particular Becoming Wind meeting and publishes no new recording excerpts, private statements, machine transcripts or photographs. Specific event research will require separate verification of sources, identities, times and publication scope before publication. The questions remain open for further work rather than presenting unpublished research as established conclusions.