The Third Mode
From Simile to Symmetry
Aa
The prior anchors defined objects, obligations, and constructions. A32 names the discipline they collectively constitute.
That discipline is the Third Mode. It is not a new theorem placed on top of the others, nor a slogan attached after the fact. It is the form a system takes when it can move from proposed resemblance to certified sameness without hiding the conditions of the move. The earlier anchors established what a witness is, what a scope is, what gluing requires, which preservation obligations a new predicate must meet, and what work a declared checking regime requires. A32 gathers those results into a claim that the Third Mode requires: some systems do not merely store assertions or enforce schemas. They can carry a claim across borders of vocabulary, context, and time while making the terms of that passage inspectable.
The historical argument of Volume I approached this problem through merchants, notaries, ledgers, and signatures. The formal argument approaches it through witnessed equivalence, context sites, sheaf gluing, conservative extension, and coherence accounting. Both ask the same question: what grounds may a recipient inherit, and which additional claims does the receiving use require?
The Composite Discipline
Each of the pivotal anchors answered one part of that question.
A10 established that sameness is not a bare declaration but a witnessed, scoped relation. A12 and A13 specify the structures under which exact compatible local data glue uniquely. A17 and A17b distinguish proposed admission requirements, logical conservativity and preservation of operational queries. A21 identifies work and budgets for specified checking regimes. Its rejected refinement-monotonicity claim supplies no ordering of all those costs. A30 closed the loop by insisting that every substitution remains answerable to the scope within which it was licensed. A31 belongs with the demonstrations rather than the prerequisites: n-ary event structure is the capstone applied to event collections whose grouping is lost by the particular unkeyed projection examined in Chapter 29, not a condition on the Third Mode itself.
Taken one by one, these anchors define objects, prove scoped results or state implementation obligations. Appendix K.8 identifies which results the machine checks. Taken together, they define a style of system behavior. A Third-Mode system does not confuse proposal with certification. It permits invention, but invention must pass through explicit tests. It permits substitution, but substitution must cite a witness and a scope. It permits local variation and requires the applicable comparison before claiming that views compose. It permits growth through time, but version change must remain legible to the claims already in circulation. A changed use remains possible, but the record must identify the additional work or authority that supports it.
This is why the Third Mode is a mode rather than a module. The conditions belong together in the proposed design because a reliable local check can be undone by an unexamined transition. Other systems can meet particular evidentiary needs without implementing this whole design. Nor must every justified revision preserve the old theory: a discovered error may require changing it, with its actual consequences traced. The discipline concerns carrying those consequences through, not protecting an inherited vocabulary from correction.
Anchor A32: The Third Mode
For this specification, Third Mode names a system satisfying the following four requirements. This is a definition of the proposed design, not a theorem that these requirements are necessary or sufficient for every adequate institution:
- Predicate invention under disciplined extension. The system can introduce new predicates, test them, and establish the required preservation or record a proved failure or an unfinished obligation. A breaking revision requires the applicable authorization and migration account.
- Witnessed equivalence under declared scope. The system can represent equivalence claims with witnesses and scopes, and it requires transport discipline before substitution.
- Composable view structure. The system represents local views and the maps used to compare them. Exact gluing requires its sheaf and matching hypotheses; a justified approximate evaluation retains its own stated guarantee. Composition must preserve those differences.
- Explicit coherence accounting. The system accounts for the checks it undertakes and exposes additional evidence, disclosure or repair obligations. When grounds change, that account follows the premises actually consumed: withdrawn support, adequate surviving grounds, unfinished reassessment and actions requiring a further authorized process remain distinct.
A32 states the design requirements. Appendix I specifies the ten-operation implementation contract used to assess compliance. Listing these four requirements is not that compliance test. A refusal needs its grounds; a version change needs a supported compatibility result, a demonstrated break or an incomplete assessment; an accepted predicate carries the evidence and scope of its admission.
The four requirements describe this fuller design. A system with a fixed vocabulary may still perform adequate scoped reuse, retain alternative support and propagate a correction through the premises actually consumed. It has not thereby implemented predicate invention or every operation in Appendix I. Its success should be judged by the obligations of the use it supplies, rather than denied because it does not instantiate the whole Third Mode.
A completed local check can save another inquiry, yet lose its warrant at the next transition. The proposed design places that transition inside the obligation to be examined. A different use may need further evidence; a correction may withdraw one ground while leaving another adequate. The receipt must make those differences available to the next recipient. Its existence cannot establish that the transition was justified.
Why This Is A Mode
A simile says that one thing is like another. It explores a resemblance without yet accepting the liabilities of identity. A symmetry says that two configurations remain the same under an admissible transformation. It bears the stronger burden. A useful resemblance can suggest an operation without supplying the grounds for substituting one thing for another.
The Third Mode exists in that interval. It gives proposal a path to earn the right to travel. Embeddings, heuristics, and statistical patterns may still generate the first suggestion that this is like that. Its permitted use depends on what the applicable contract establishes. A justified statistical evaluation can remain approximate; an exact substitution needs the grounds for that stronger claim. The point is to discipline the relation between the flexible and the exact so they can compose without smuggling one another's failures across the boundary.
The notarial analogy concerns a claim’s use among strangers who did not witness the original transaction. The analogy identifies work an institution must arrange: preserving the claim, establishing its grounds, determining its permitted use and dealing with a failed comparison. A32 proposes one organization of that work. Its mathematical definitions do not confer jurisdiction, and pricing the operation does not authorize the action it supports.
A refusal or fork can leave useful work to do. The receiving operation must expose that work when it cannot complete the proposed reliance. It can say what failed to glue, what witness was missing, what predicate could not be extended, what checking cost was incurred and which charge remains to be explained. A21 does not identify a provider’s rent by subtracting a supposedly universal verification fee. The recipient can then decide whether further checking, another source or a separately justified course is available.
What A32 does not provide is a total formalization of judgment. Appendix K separates machine-checked statements from written proofs and proposed contracts; Appendix L identifies the unfinished constructions. A21 can make a provider’s claimed checking work inspectable, while the grounds for its price and authority require further examination. The Third Mode does not mechanize mercy, settle every question of institutional legitimacy, or eliminate the need for human judgment at the boundary cases where the formal apparatus runs out. Its consequential demand is narrower and firmer: an institution cannot lend a decision assurance from a result that did not establish its premises. When a proof fails, its dependent uses must stop advertising it as support and reassess their remaining grounds. Some conclusions may survive that examination. Keeping the failed proof’s receipt while continuing to use its guarantee would preserve the record and abandon the discipline.
The refusal interlude that follows leaves one proposal partly repaired and still awaiting admission. That unfinished result belongs to the design as much as its successes: the missing work remains identifiable without becoming somebody else’s supposed evidence.