qi3 — the transactional substrate ("QI realized")

The next engine after link3. Design lineage: CDT (discrete causal lattice), 't Hooft (deterministic local rules → emergent QM), Kastner TI (emission and absorption as mutual transaction completion — offer + confirmation, not just photon impact).

What link3 proved (inherited mechanics)

• Local rules only: site reads own registers + channel neighbors, c = 1 hop/tick
• ν multi-role: energy (E=hν), mediator mass (range ∝1/ν), momentum (p=E/c), fusion yield (E=mc²)
• Parked flux stays put: wound mass is structure, lUP carries transport
• Session lifecycle + broadcast announcement — a node announces on all ports until a peer answers
• Conservation ledgers: LIPH + event counters auditable every tick

What qi3 adds (the absorber side — where QM lives)

link3's gap was absorber response: lumps had ONE excitation register (iE — a scalar bank). Real absorption is selective: a transition either completes at a matching energy or doesn't happen. qi3's sites carry an internal LEVEL STRUCTURE:

The edge-first rule (fundamental design decision)

Structures form BETWEEN nodes — in channel state — the way link3's exchange bond lived in the gap traffic, not in either lump's internals. A node's internal registers may be complex (they have to drive complex edge states), but the goal is to DERIVE the resulting state — never enumerate it. No hardcoded level table (n=1,2,3); the internal dynamics produce configurations, and the observable level is read off the structure by measurement.

The heartbeat model: edges react first. An offer arrives on a link; a confirmation arms on the reverse link; when the confirmation pulse returns, the node's edge PATTERN flips — the "level" changes because the circulation changed, not because a counter incremented.

Registers (spelled-out naming — long names document the ontology;

inner-loop helpers may stay shorthand)

• qWinding[site][c] — the site's internal circulation: winding/ momentum population per channel. Complex machinery that drives the edge states — the level is DERIVED from the pattern, never an enumerated value stored here
• qCharge[site] — bound charge (deposits from completed transactions)
• qConfirm[site][c] — pending confirmations per channel: offers received, responses armed (the TI's confirmation wave made a register)
• qLevel[site] — DERIVED, not stored: read off the edge configuration by measurement. The spectrum is generated by the dynamics, not hardcoded — we observe which stable patterns emerge, we don't enumerate them in advance

The transaction (replaces absorb-on-contact)

Absorbers are per-EDGE, not per-site, and statistical in extent: we don't know how large an absorber is at Planck granularity — it may span one edge or twenty. Everything is in a way an absorber; "where the absorber is" is a derived/statistical question per tick. One tick evaluates ALL edges on ALL nodes simultaneously — each channel's offers judged independently against that edge's own state, decisions buffered (no order dependence within the tick).

1. Photon (offer, ν) transits a site on edge c
2. The EDGE evaluates: does ν match a transition this edge can express? (derived from that channel's winding/mass density — an extended absorber is a contiguous region of absorber-active edges)
   • match → arm a CONFIRMATION on the reverse channel
   • no match → photon passes (the offer declined — no partial fills)
3. Confirmation returns / completes the transaction: photon absorbed, the edge's state transitions (level change = circulation change)
4. Emission: an excited edge can fall to a lower stable configuration, emitting ν = the transition (spectrum = structure)

Why this could produce 1/n²-like series (candidates to measure)

• If site internal state = closed loop of length L (integer cells), and energy ∝ 1/L or L²-scaled — gaps form a converging series
• If allowed windings are harmonic modes (n turns on the internal circuit), transition energies follow mode differences
• Measure first, pick later: several level structures get tried and the spectrum output decides — same falsification style as link3

Tick phases (inherited order, extended)

1. junction collide — mass interactions (inherited)
2. mass stream — lUP transport (inherited)
3. photon stream — offers propagate, ν-transit decay ∝ ν (inherited)
4. transaction pass — offer/confirmation matching (NEW)
5. session strength — duplex EMA (inherited)
6. emission — level-drop emits at gap ν (replaces iE decay)

Honesty rules (carried forward)

• Ratios and shapes, not absolute values; confounds logged
• Every mechanic falsifiable; regressions re-run after changes
• No virtual photons — every exchange is a real photon on a real channel

Build-out completed (the audit queue)

• 9 duplex edges — structure is per-edge (bonds, confirms, sessions); transit stays per-channel (18 dirs)
• Two-beat heartbeat + lottery — offers arm INCIPIENT transactions held on the edge; next tick a weighted lottery (countnu(1+bond)) actualizes ONE per site; losers rejoin
• Circulation object — absorption splits deposits across edge pairs (closed loops); level = matched-loop content
• Quantized emission — one quantum per transition, site emits the ACTUAL gap n->n-1, paid by siteUnwind; charge is bound energy that THERMALIZES into winding
• Attractor drain — quantization enforced dynamically: only complete winding bands persist (sub-quantum = forming)
• E_n family — gap_mode 1-4: {E0/n^2, E0/n, E0n, E0n^2} falsifiable candidates sharing one transition table
• RTT handshake — offers carry birth ticks; actualization spawns an ACK that walks back ttl=tau hops delivering RTT=2*tau as qBondDist — retarded confirm, real latency, dual-end bond credit
• Conservation — uint32 registers, sealed ledger, exact nu=128 kinematic fusion wall, fusion launches pairs apart
• Phase structure measured — cold vacuum / stochastic nucleation band (P(ignite) sigmoid) / saturated hot phase