{"path":"research/legacy-object-model-harvest.md","content":"# What to harvest from the legacy object model\n\n**17 October 2026.** The founder asks that `object-model.md` clearly identify its legacy status and\nthat its useful ideas be recovered for current ontology work. This is an audit, not a schema ruling.\nIt reads the whole inventory against the recovered drawings, the Meteor prototype's `links.js`,\n`evidences.js` and `lib/enums.js`, and the current parts, reasons, relevance and concept notes.\nThe [image catalogue](../sketches.md#the-recovered-photographs) separates recovered originals from unlocated drawings.\n\n## 1. What was already there\n\nThe old drawings already contain the distinction the gate example risked rediscovering. In the\n[Socrates sheet](../sketches.md#additional-recovered-sheet-socrates-inference-and-navigation), *Socrates is human* and *All humans\nare mortal* join beneath a VALID badge and the sentence *Assuming premises are true, conclusion makes\nsense*. A case, a general rule, their application and the conclusion are visually distinguishable.\nThe same sheet offers *4 more reasons*: the collective case can grow without changing the dependence\ninside this argument. These are observations of the drawing; it does not settle today's calculus.\n\n![The Socrates sheet: case, rule and validity, with navigation (held privately)](../../.private/sketches/socrates-inference-and-navigation.jpg)\n\nThe current homes already say more precisely what the sketch suggests:\n\n- [A part and a reason](a-part-and-a-reason.md) § 5 separates the content of one claim, the premises\n  of one argument, and the collective case for a conclusion. § 11 applies this to an inline *because*.\n- [Retrofitting the reasons channel](retrofitting-the-reasons-channel.md) § 1 relocates the\n  `necessary` / `alternative` / `corroborative` vocabulary after reasons were mistaken for parts.\n  Joint premises, separate arguments, and corroborating reports are different jobs. Grouping alone\n  does not settle aggregation or independence; the proposed arithmetic has its own standing.\n- [The chain inside a link](the-chain-inside-a-link.md) § 10.3 records the ruled floor: one stated\n  rule applied once to one stated case, with every remaining doubt addressable. Vertical exposure\n  of inferences and horizontal separation of asserted content complement each other.\n\nAn ordinary composite claim can state a combined reason, as the founder suggests. Its parts expose\nwhat was asserted; the inference beneath its link exposes why it bears. A composite and a\ncase-plus-rule view can describe **the same application** and must not be counted as two independent\narguments. The remaining choice is their representation and interaction, not a new discovery that\nsome premises depend on each other. [Worked gate example and correction](one-challenge-one-home.md#2a-a-shared-case-and-a-reason-that-states-its-combination-17-october).\n\n## 2. The inventory, harvested rather than reinstated\n\n| Legacy item | Useful idea and current home | What does not follow |\n|---|---|---|\n| Contention / Claim; immutable prototype text | Keep assertions addressable and preserve their wording and attribution. [Store the uses, read the meaning](store-the-uses-read-the-meaning.md) and [reversible sameness](soft-canonical-clustering-and-reversible-merge-semantics.md) are the richer current account. | A prototype deny rule does not settle present revision, identity or supersession semantics. |\n| Argument bundle; premise | A reader can focus on one argument while seeing other arguments for the same conclusion. This is the line/application distinction above. | A bundle-shaped enclosure does not require an ArgumentBundle database type. It can be a view over claims and their inferential uses. |\n| Counterargument | Make what is challenged visible, including whether it is a premise, a rule, an application or a conclusion. [One challenge, one home](one-challenge-one-home.md) § 3. | Attack is a role relative to a target, not necessarily a separate species of claim. The sketch's label does not warrant a new content store. |\n| Evidence and Source; found/missing votes | The prototype deduplicates a source–proposition pair and separately asks whether the proposition was found there. Preserve the distinction between locating a statement, faithfully representing it, its truth, and its relevance. [What counts as evidence](what-counts-as-evidence.md) and [extraction](../extraction-pipeline.md). | A found citation is not proof of its content. Neither a found/missing vote margin nor the mere presence of a URL is a truth score. Repeated citations need not be independent evidence. |\n| Link; New/OldFor and New/OldAgainst; endpoint preference fields | One stored pair exposes different directional views from each endpoint. Harvest that navigation pattern for shared challenge homes: entering from either end need not duplicate the underlying contribution. | Support is not symmetric, and the old per-pair record does not prove that there can be only one inference or context for that pair. Distinct applications can share endpoints. |\n| New/OldBetter; Substitute; Opposite; Unrelated | The old enum already distinguished better wording, opposition and no relation. Compare it with the keeper's present task and [soft sameness](soft-canonical-clustering-and-reversible-merge-semantics.md). | Better phrasing is not logical equivalence. Opposition is not automatically exact negation. The legacy vote winner must not replace these distinctions. |\n| Definition; Value / Moral Weight | The *sweet* sense branches and *hot* example anticipate sense-per-use and the distinction between facts, classification and values. [Core, periphery and the concept layer](core-periphery-and-the-concept-layer.md), [the retrofit](retrofitting-the-reasons-channel.md) § 3. | Separate headings do not require separate node types. Changed evidence or a threshold disagreement need not mean a new sense. An old \"axiom 100%\" annotation is not a present immunity from challenge. |\n| Truthiness, relevance, controversy; follows/does-not-follow; affect | Distinguish a claim's standing, an inference's bearing, disagreement among readers and emotional tone. The recovered 30°C notebook draws several of these separately. | A majority is not a probability, polarity is not sentiment, and a VALID badge does not establish the truth of its premises. |\n| Full distributions; strongest pro/con; center of belief | Show what the summary conceals: disagreement and reasons can remain visible together. Treat this as a view to compare in Lins. | The >30% visibility threshold and highest-share winner are prototype mechanisms, not approved current thresholds or epistemic laws. |\n| Graph and Meteor collection list | Preserve as historical implementation evidence and as a source of questions. Current storage is in [architecture reference](../architecture-reference.md#graph-database). | The union of every historical collection, visual role and proposed concept is not the optimal ontology. Unknown historical `elements` remain unknown rather than receiving an invented function. |\n\nTwo corrections to the old summary come directly from code: `Opposite` and `Unrelated` were omitted\nfrom its link-vote list; `foundIndex` is simply `foundVotes - missingVotes`. Both are now explicit.\nThe audit checks these particular mechanics, not the correctness of the old prototype as a whole.\n\n## 3. The most useful visual comparisons\n\nThese are additions to the existing walkthrough's questions, **not yet added as Lins cards**:\n\n1. **One application, two views.** Show the gate's case and rule beneath its link, then the same\n   argument as a composite reason with exposed parts. Give the application one identity in the\n   example. Dispute the case; then dispute the rule; then point out a gap in its application. Show\n   the surviving camera argument in every view. If the objection has to be re-entered, the design\n   has duplicated a home. If the case fails and the rule still appears as a complete argument for\n   this occasion, the view has lost the dependency.\n2. **Enter from either endpoint.** Reach one objection from the claim, its reason and the relevance\n   panel. These are routes, not three contributions. Then draw two genuinely different inferences\n   between the same claims, so deduplication does not erase distinct reasoning.\n3. **Found, faithful, true, relevant.** Use a source that really says a false thing and a true\n   quotation that does not bear on the target. A failed citation lookup should have a different\n   visible target from a counterargument to its content. This is the old evidence object's useful\n   separation without committing to that old type.\n4. **One concept, changed reading or changed world.** Compare the notebook's 30°C → hot with the\n   comfort case. Change the sense, the occasion, the threshold and the evidence one at a time.\n   Record which statement or application changes. Avoid manufacturing a new sense to explain\n   away every substantive disagreement.\n5. **Direct checking alongside argument.** Give the date claim a clock/calendar result, then\n   contrast it with a human-supplied schedule. Change timezone and occasion. The requested automatic\n   check and proposed contract are in [the walkthrough](one-challenge-one-home.md#2b-directly-checkable-facts-time-is-an-instrument-input).\n\nThe point of these comparisons is to inspect the same reasoning through alternative representations.\nThey must not reward drawing more copies. No new ontology or scoring formula was adopted by this audit.\n\n## 4. What the recovered images change\n\nThe six recovered photographs cover four of the seven catalogue entries and add the Socrates sheet.\nTwo photos show successive states of the same concept-tracking sheet; six photos are not six\nindependent designs. They corroborate the long lineage of the current questions without proving any\nparticular implementation. The bundle-tabs, feed-comparison and felicific-calculus images remain\nunlocated, so their descriptions are still secondary records. The old \"~13 photos\" estimate is not\na verified inventory. [Recovery register and full-resolution files](../sketches.md#the-recovered-photographs).\n"}