/machines/1RB0RD_1RC0LF_0LA0RB_1RE---_1LB1RA_0LE0LF/notes.md · raw

Initial observation: holdout line 348

Machine: 1RB0RD_1RC0LF_0LA0RB_1RE---_1LB1RA_0LE0LF.

Completed 2026-10-13T18:24:18.533813-07:00. Initial-stage budget: 15 minutes; all simulations sequential, nice 10, address-space cap 6,000,000 KiB. This is observation evidence, not a kill or a proof sketch.

Result and stopping boundary

The tape has a long period-seven bulk and a short irregular right edge. Confidence is high in that finite description and medium in the full mechanism. No closed edge recurrence, invariant, or general macro rule has been established. Stop before assuming one. The observed irregular words alone do not establish pseudo-random arithmetic or a cryptid classification.

Both machines belong to one table family, so they should receive one research priority, not two independent proof budgets. There is no sound evidence that either start is intrinsically easier. If this family is funded again, use line 6 as the canonical orientation and keep the second start as a separate initial-state obligation. Rank this family behind candidates with a validated closed recurrence.

Exact identity and relation

The machine string is read directly from line 348 of data/holdouts/BB6_holdouts_740.txt. Latest machine claim is Astra. The shared preflight found no existing-machine/open-PR table match and no deferred dyuan overlap. The saved Discord occurrence is the UBRRBA TODO list, not a mechanism proof.

An exhaustive check of all 720 state permutations with and without reflection found exactly one map from line 6 to line 348: reflect left/right and rename A→E, B→B, C→C, D→A, E→D, F→F. All 12 transition cells, including the undefined one, are recorded in evidence/table-correspondence.json. Line 6 blank-A maps to line 348 blank-E. Line 348 blank-A instead corresponds to line 6 blank-D. Table identity does not equate their standard blank-A runs.

Direct observations

At 100,000,000 steps: state B, last move R, position -13002; reached interval [-13218, 141], 2,044 ones. There are 13218 new left records and 141 new right records. The span counts every reached cell, including the current head cell.

All 141 slow-edge record snapshots through that horizon are saved. The last such record is at step 90,387,011. No further slow-edge record before 100 million is evidence of a pause, not stabilization of the edge.

The compressed endpoint is below. A word followed by an exponent repeats the entire word; head notation is directed (<Q reads the cell immediately to its left, Q> the cell immediately to its right).

0^inf 100 1111100^30 1^3 B> 0000010^1857 0000001001^2 0^3 10010^2 0 001^2 100^4 0100000100000^2 10011 0^5 010^2 1 0^5 1000000^2 001^3 0^4 000001001^2 0^inf

The final slow-edge snapshot is:

0^inf 1001111^1795 0100 1^12 0 1^5 0100 11101^2 001111^2 110100 1^8 0^2 1^5 0100 1^5 111001111^2 0011111^2 01 001111^2 110100 1^7 001 E> 0^inf

The outward records near 100 million come in three-step clusters (F,E,B). This is a local observation, not a proved cycle. The slow-edge words continue changing and have many short nonperiodic pieces. All 32 saved compressed edge descriptions round-trip to the exact state, direction, relative head and full trimmed tape.

Bounded acceleration and independent endpoint replay

Automatic block selection chose size 1. At 20,000 loops it had reached only about 31,900 base steps and found zero rules. Selecting the visible period 7 was substantially more effective. The size-seven probe stopped at its 200,000-loop cap, before its 20-second limit; it used macro and chain moves, with zero inferred rules and zero Collatz rules.

The accelerated endpoint is step 5,016,956,497. A separate C run from the blank tape replayed that entire horizon and matched the state, last direction, full trimmed tape, relative head index, and count of ones. Its reached interval is [-93687, 153], head position -1, state F, and 66,950 ones. Both stop at the numerical budget, without halt. No claim about all future steps follows.

Why the edge remains a proof obligation

The only halting entry is D1. From the transition table, the immediate trigger is A reading the left-hand 1 of 11: A1 enters D on the right-hand 1, then D1 halts. The corresponding predecessor is the sole transition entering that halting state. Thus any eventual non-halting proof must exclude this reachable state-and-local-tape pattern; periodic bulk by itself does not do so.

No rulefit or Coq work was started: fitting an unclosed edge shape would presuppose the mechanism this stage failed to identify. Further work, if prioritized, should first choose a consistent edge-return boundary, derive its word transformation, validate that transformation on independent direct traces, and then decide whether an invariant or cryptid boundary applies.

Artifacts and reproduction

AI assistance: these observations, scripts and checks were produced with the help of GPT-6 Astra (OpenAI).

Priority decision — 2026-10-14

The user said: “Follow your recommendations for the 5.” Recorded study for this family; a mechanism study is funded for at most 30 minutes. The line 6 decision covers both 6 and 348.

Funded mechanism study — 2026-10-14

Completed 2026-10-14T10:29:58.650928-07:00. Exact local sweep identities and a reduced word interpreter were checked. R1 now has an informal universal first-return derivation; R2/R3 remain independently sampled candidates. No closed edge-word language or whole-machine invariant was found. See study.md for formulas, parameter exceptions, separate-start treatment, checks and the precise obstruction. No Rocq proof or kill is claimed. Further formalization remains parked.

Funded stage closed — 2026-10-14

The approved mechanism study is complete. Its answered decision is archived in the canonical line-6 directory. No research job remains running. The new priority card concerns a separate stage and is unanswered. Full comparison: coord/audits/bouncer-studies-20261014.md.

Further research deferred — 2026-10-14

The user selected defer on the follow-up priority card. The answer was relayed in coord/mailbox/to-astra/20261014-154300-claude-dennis-answers-6-282.md. The completed study and evidence are retained; no additional research is scheduled. The canonical line-6 answer covers both 6 and 348.