{
  "_meta": {
    "source": "ui_leviathan/web/content/conversations.md",
    "version": "0.3",
    "status": "DRAFT",
    "synced_at": "2026-10-04T02:34:45.771Z",
    "content_hash": "3a1ec65745443a23e62b13562e71bb51c124f22f0915413a002da3c31681cc26",
    "generator": "web/scripts/sync-docs.mjs"
  },
  "frontmatter": {
    "title": "What we learn from conversations",
    "version": "0.3",
    "status": "DRAFT",
    "updated": "2026-10-04"
  },
  "body_md": "\n# What we learn from conversations\n\nSomeone who lives with a problem may notice what a researcher misses. A builder may know a method that opens a new possibility for a care group. We want conversations to make those connections useful: a clearer question, an experiment worth trying, a better design, or a reason to change course.\n\n“What is Leviathan?” belongs in these conversations too. It is a system we are developing, and another person's experience could change its purpose, language, or way of learning. We want to hear from caregivers, researchers, builders, institutions, and people whose experience rarely reaches such discussions.\n\nA conversation can begin with your own assistant, within an independent group, or directly with us. With permission, we want to share useful accounts here, including criticism and reasons someone chooses not to take part. Different conclusions can remain visible alongside work people choose to do together.\n\n## What we want to learn\n\nHow could this work be useful to someone who does not share our starting views? Where could another person's observation or method change what we are able to do? What do existing tools already make possible, and which need remains unmet? [What is Leviathan?](/what-is) explains the proposal; the [Whitenote](/whitenote) presents its personal starting point.\n\nOur [animal-welfare work](/animal-welfare) makes these questions concrete. Which interests are missing from an account of care? What evidence could change an intervention? What would make an experiment worth doing, and who can help examine its effects? A useful conversation could connect a question in one field with a method from another.\n\nFor example, a caregiver could describe a situation our proposed sensor records would fail to explain. A researcher could offer a comparison that would help distinguish possible explanations. The next useful output might be a revised observation plan, including what it would and would not tell us. Neither contribution needs to begin with an objection to an existing claim.\n\nWe also want to understand what makes participation possible: relevant expertise, time, resources, or a different way of working. A conversation may uncover a better approach elsewhere or a reason to change our own plans. Agreement with the architecture is not a condition for contributing to that learning.\n\n## How a conversation becomes a public record\n\nA conversation can remain entirely private. Before publishing an account of it, we will agree with the participant on the name or attribution to use, the exact excerpts to quote, and the summary to share. Agreeing to meet, offering support, and agreeing to publication are separate decisions.\n\nEach published record should make the question and date clear, present the approved view, and show our response: what we learned, what remains unresolved, and whether a decision changed. Where a claim refers to an available public source, we will link to it. Corrections will stay connected to the record they revise.\n\nFor a discussion of the architecture, the account should name the version being examined. A criticism of one proposed rule should not quietly become a verdict on a later design. If participants use the same word differently, we should preserve that difference and the limits of the summary. An agreed account establishes what may be published; it does not make every claim in it true or everyone in agreement.\n\nAn unresolved question or objection should remain available for later inquiry. In the language we are developing, such a record can be a *shadow*. A new method or observation could reopen it, produce a test, and change the design. We can already document questions and responses in ordinary records and the local inquiry; automated learning from these exchanges and transfer between independent Leviathans remain development goals.\n\n## Disagreement and support\n\nSomeone can help us think without endorsing Leviathan. Support can also be limited to a particular question or project. We want the record to make that scope clear and give criticism room alongside encouragement. A group can retain its own commitments or build an independent approach; a conversation does not join it to a common authority.\n\nIf someone declines an invitation or support request and wants to share their reasons, we can publish the wording they approve. A refusal alone does not establish a scientific objection, and silence tells us no one's opinion. We will not turn unanswered invitations into a public list.\n\nQuestions about an institution's public claims or decisions should be examined through evidence and their effects. Its willingness to support Leviathan is a separate matter.\n\nMaterial support should identify its purpose and recipient, and must not buy a favorable conclusion or silence an objection. Expertise, code, care, equipment, or time may be useful contributions. No partnership, funding channel, or commitment follows merely from being mentioned in a conversation.\n\n## The record so far\n\nNo agreed accounts of conversations with external participants have been published on this page yet. The [local development inquiry](/research) is documented separately: its review exchanges are attributed to Codex development roles, not independent outside communities.\n\nYou can begin with a short account of a question, experience, or method you would like to discuss and explain what the conversation could help produce. The [participation guide](/participate) gives the current draft and forum paths. The research form only downloads a file; it does not request a meeting or deliver a message, and a response is not guaranteed.\n\nYou can also explore the [Markdown introduction](/what-is.md) with your own assistant. It provides the same overview as the website. An assistant's draft remains a draft until you choose to share it; providing a link does not make your conversation public.\n\n*Revision 0.2 · 2 October 2026: frames conversations as a way to examine and change the proposed language, learning methods, and independent structures. Adds version-aware accounts and a path from unresolved objections to further inquiry. No new conversation, endorsement, or partnership is reported.*\n\n*Revision 0.3 · 4 October 2026: begins with the useful work a conversation could make possible and gives a concrete care-and-method example. Clarifies the difference between future external conversation accounts, the documented local inquiry, and preparing a draft for possible sharing.*\n"
}