TMBR: November 2026
Jump to navigation
Jump to search
| Prev: October 2026 | This Month in Beaver Research | Next: December 2026 |
This edition of TMBR is in progress and has not yet been released. Please add any notes you think may be relevant (including in the form a of a TODO with a link to any relevant Discord discussion).
This Month in Beaver Research for November 2026.
TODO: Upgrade to Ligocki Busy Beaver repo unlocks pentational+ deciding. No new champions, but there's complete turnover of places 4-20+ of the bb7 top halter list, and new bb4x3 top halters in places 4-8.
TODO: loader3229, Jeff Ketchersid, roromachine14's new champions
Media
- YouTube video: Skelet #17: The Last Obstacle to solving Busy Beaver 5. NLogSpace. 6 Nov 2026.
BB Adjacent
- Pebble Automaton
- On November 2, the remaining group of 22 pebble(11) holdouts was officially dubbed the "golems". They are modeled by complex systems of equations, making them potential cryptids. They can be categorized into 7 unique paths.
- Catch-Em-Turing
- On November 2, User:NathFR76 created a Busy Beaver variant featuring multiple agents and no individual halting state. Instead a machine is said to halt when all agents occupy the same square on the tape. The CET(n) function is theorized to grow faster than BB(n).
- On November 11, roromachine14 claimed a new champion for CET(3) ≥ 10^10^(2.8244*10^13), the machine is a Shift Overflow Counter (SOC).
- On the same day, star discover a cryptid in CET(3) machines (Agent 1: 1RB1LC_1LC1RB_1LA1LA, Agent 2: 0RB0RC_1LB1RA_1LA0RA), BMO/Mahler type problem with an irrational growth rate.
- On November 13, star discover a second cryptid and found 19000 holdouts for CET(3).
- On November 14, Axolotldailyweek claimed a new champion for for CET(4) > 10^^14, the machine is a Shift Overflow Counter (SOC).
- On the same day, roromachine14 claimed a new champion for CET(3) > 10^10^10^10^16.558 the machine is a Shift Overflow Counter (SOC).
- On the same day, Axolotldailyweek claimed a new champion for CET(4) > 10^^23, the machine is a Shift Overflow Counter (SOC).
- On November 15, star claimed a new champion for CET(3) > 10^^12.467 the machine is a Shift Overflow Counter (SOC).
- On the same day, star reduce the number of CET(3) holdout to 14214, you can see on https://rorogamin.github.io/CET-Challenge/0LB---_1LC1RB_0RB0LC%7C1RB0LC_0RC0RB_1LC1RA.
- On the same day 67 claimed a new champion for CET(4) > 10^^140, the machine is a Shift Overflow Counter (SOC).
- On November 16, 67 claimed a new champion for CET(2,3) > 10^^8, the machine is a Shift Overflow Counter (SOC).
Champions
| Domain | Champion | Discoverer | New lower bound | Discoverer | Previous lower bound |
|---|---|---|---|---|---|
| BB(17) | 1LB1LA_1RC0LA_1RD1RC_1RE1RK_1RF0LJ_1LG1RI_1RM0LM_1LG0RC_0RN0LE_1LA1LE_0LL1LB_0RM1LL_1LF1RH_1RQ0LO_1LP1LN_0RB1LI_1RZ1LO (bbch)
|
Jeff Ketchersid | |||
| BB(18) | 1LB1LA_1RC0LA_1RD1RC_1RE1RK_1RF0LJ_1LG1RI_1RM0LM_1LG0RC_0RN0LE_1LA1LE_0LL1LB_0RM1LL_1LF1RH_1RQ0LO_1LP1LN_0RB1LI_1RR0LO_0RO1RZ (bbch)
|
Jeff Ketchersid | |||
| BB(19) | 1RQ1RL_0LI1RC_0LD1LF_0RA1LD_1RB1RE_1LH0LG_1LF1LG_1RE1RM_1LJ1LI_0LK1LI_1RL1RA_0RL0RE_0LN1RM_1LO0LO_1RE1LP_1LN1LN_1LR1RQ_1RZ1LS_0LE0LR (bbch)
|
User:Jacobzheng | |||
| BB(20) | 1LH0LG_0LI1RC_0LD1LA_0RE1LD_1RQ1RL_1RB1RF_1LA1LG_1RF1RM_1LJ1LI_0LK1LI_1RL1RE_0RL0RF_0LN1RM_0RF0LO_0RF1LP_0RF1LN_1LR1RQ_1RZ1LS_0LF1LT_0LF0LR (bbch)
|
User:Jacobzheng | |||
| BB(11,4) | 1RG3LB1RZ2RA_2LD---1LD2RA_3RB1LB3LC2RC_3RE---3LC2RC_---3RE3LH2LF_3RF2RF2LE1LF_2LC3LH2LJ2RG_---3RH1LI2RG_3RH1LH3LI2RI_------3LJ3RK_------2RA3RK (bbch)
|
User:Loader3229 | |||
| BB(8,5) | 2RG3LB4LC2RA4LB_1RZ---1LD2RA4LD_3RB1LB3LC2RC4LC_3RE---3LC2RC3LC_---3RE4LE4LH4RB_4LE3LG3LB2RF4LG_2LH3RG1LH2RF2RA_3RG1LG3LH2RH4LH (bbch)
|
User:Loader3229 | |||
| BB(4,6) | 5LD3LB5LC2RA4LB4LC_1RZ3RB1LC2RA3RD0LA_3RB1LB3LC2RC4LC1RA_3RB0LA2LC4LD4LC4LD (bbch)
|
||||
| BB(5,6) | 1LE3LB2LA2RA4LB5RA_1RZ3RB1LC2RA3RD4RA_3RB1LB3LC2RC4LC2LB_3RB0LA5LE4LD4LC4LD_4LC5LC0RE5LE1RE2RE (bbch)
|
||||
| BB(6,6) | 4LF3LB1RZ2RA4LB---_2LC3RB1LD2RA3LF---_3RB1LB3LC2RC4LC0LE_3RB---3LC2RC---3RE_5RC3RE---4LC------_5RB0LF1RB1LF---5RB (bbch)
|
||||
| BB(7,6) | 2LE3LB1RZ2RA4LB2LD_0LA3RB1LD2RA3LF5RB_3RB1LB3LC2RC4LC1LE_3RB3RE3LC2RC---3RD_5RB3RE0LG4LC3LG5LE_5RB5RB---1LF---5LF_5LF---3RC1LG---5LG (bbch)
|
||||
| BB(8,6) | 0LC3LB2LH2RA4LB1RZ_3LH3RB1LD2RA3LF5RB_3RB1LB3LC2RC4LC1LE_3RB3RE3LC2RC2LD3RD_5RB3RE---4LC3LG5LE_5RB5RB---1LF5LF5LF_5LF4RG1LB1LG5LD5LG_3LE---1LH1LH---5RG (bbch)
|
||||
| BB(3,7) | 6LC3LB5RA2RA4LB2LA5LC_1RZ3RB1LC2RA3RB4LC0LA_3RB1LB3LC2RC4LC5LC--- (bbch)
|
User:Loader3229 | |||
| BB(4,7) | 6LD3LB5LD2RA4LB3LC5LA_1RZ3RB1LC2RA3RD---2RA_3RB1LB3LC2RC4LC3LC5LD_3RB---6LC4LD4LC4LD--- (bbch)
|
- User:Loader3229 discovered an array of champions based on Kropitz's current BB(3,4) champion and Jacobzheng's former BB(11,3) champion, demonstrating the following bounds:
- User:Jacobzheng discovered the following bounds:
- On November 2, BB(20) > .
- On November 4, BB(19) > and BB(20) > .
- On November 4, Jeff Ketchersid used Claude and Gemini to discover the bounds BB(17) > , BB(18) > and BB(19) > .
Holdouts
| Domain | New Holdout Count | Previous Holdout Count | Holdout Reduction | % Reduction |
|---|---|---|---|---|
| BB(6) | 608 | 608 | 0 | 0.00% |
| BB(7) | 12,662,080 | 12,655,322 | -6,758 | -0.05% |
| BB(2,5) | 37 (official) | 59 (official), 37 (informal) | 22 (Rocq confirmation of Lean proofs) | 37.29% (confirmations) |
- BB(6)
- On November 1, Ahmer Raza simulated two holdouts, both Reus Vampires, to halting.[3]
- On November 18, 67 verified the halting of BMO10, this was the 6th BMO problem ever solved.
- BB(7)
- Two previous large holdout reductions, the direct simulation by @DavidSG and inductive decider application by @finlay, did not enumerate TNF children, causing the reported holdout count to be a slight underestimate. Andrew Ducharme, with help from @GameDungeon and Shawn Ligocki, enumerated the children from both reductions, creating 346,584 and 8,428 new holdouts from the respective reductions. Before adding these TMs to the full holdout list, they need to be filtered by all deciders applied to the full list between the point where the parent TMs were found to halt and the present. This was finished for the 8,428 list from @finlay's reduction, which ended up adding 6,758 new TMs to the holdout list.[1] Filtering continues for the 346k list from @DavidSG's reduction and is currently at 74,554, but there are multiple deciders still to run before the holdout list is completely unified.
- BB(2,5)
- On November 3, mxdys confirmed in Rocq the proofs in Lean of 22 TMs solved with AI assistance (GPT6 and Aristotle), releasing a formal list of 37 holdouts.[4]
Cryptids
All Turing machines mentioned in this section work with an alphabet of 2 symbols unless explicitly stated. The state bounds given also only consider Turing machines which work with 2 symbols.
- Cryptids in the Wild (+ Energy Vampires)
- Reus Vampires
- On November 1, Ahmer Raza ran path O4 to 1,941,376,761,248 iterations, at which point it halts![5] This path corresponds to a group of machines with halting times of over 10683,718,950,000 steps, and decides the last of the 7 Reus vampires.
- Reus Vampires
- Larger Cryptids
- Logical Independence
- Zermelo-Frankel set theory
- On November 1, charlie claimed to have used Claude to formally reduce the number of defined instructions required in a ZFC-independent TM to 375 (376 including the halting transition).
- On November 12, James Chang used ChatGPT 6 Astra to find a 209-state TM that halts iff ZFC is inconsistent.
- On November 20, the figure is now 207.[6]
- New Foundations
- On November 6, svennik-alt-account used ChatGPT 6 Astra to find a 210-state TM that halts iff New Foundations is inconsistent. This is accompanied by a Lean proof. This is also the smallest known TM which PA cannot prove is non-halting.
- On November 11, svennik-alt-account used ChatGPT 6 Astra to find a 198-state TM that halts iff New Foundations is inconsistent. The proof cites an AI-generated proof that Con(NF) <=> Con(MAC).
- On November
1619, svennik-alt-account used ChatGPT 6 Astra to find a192185-state TM that halts iff New Foundations is inconsistent. The proof comes with a Lean verification that NF has the same interpretability strength as NFU+Infinity, which interprets MAC, which interprets TST+Infinity, which is equiconsistent with NFU+Infinity. The fact that NFU+Infinity can interpret the theory NF is novel as of the time of this writing. The informal idea behind the equi-interpretability argument was invented by GPT Astra 6, and then converted into Lean code very efficiently by Claude Fable 5.2 Extra. The argument differs from the Holmes-Wilshaw proof.
- Zermelo-Frankel set theory
- Logical Independence
Computational Complexity Theory
- On November 1, GregoryMorse announced a 278-state machine that quasihalts iff P = NP, a 295-state machine that quasihalts iff P = PP, and a 291-state machine that quasihalts iff P = PSPACE.[7]
- On November 4, Mira-Cult announced a Lean formalization of a 285-state P = NP beeper.
- On November 6, Mira-Cult released proofs for 284-state and 266-state NP = coNP beepers; the 284-state table is Lean-formalized, while the 266-state construction has a conventional proof.
- On November 9, GregoryMorse completed Lean formalizations for a suite of ten beepers, constructed with Claude assistance. These use designated beep states; Mira-Cult's headline counts use designated beep transitions.
| Problem | Domain / convention | Discoverer | Date | Proof and notes |
|---|---|---|---|---|
| P = NP | BBB(285) (beep transition) | Mira-Cult | November 4 | Lean proof; 286-state beep-state companion also formalized. |
| NP = coNP | BBB(284) (beep transition) | Mira-Cult | November 6 | Lean proof; 285-state state-exit companion documented. |
| NP = coNP | BBB(266) (beep transition) | Mira-Cult | November 6 | Conventional proof; 267-state state-exit companion documented. |
| P = NP | BBB(278) (beep state) | GregoryMorse, Claude | November 9 | Lean proof and NP-completeness proof. |
| P = PSPACE | BBB(291) (beep state) | GregoryMorse, Claude | November 9 | Lean proof: alternating-quantifier language in P. |
| P = PP | BBB(295) (beep state) | GregoryMorse, Claude | November 9 | Lean proof: majority-acceptance language in P. |
| NP = coNP | BBB(317) (beep state) | GregoryMorse, Claude | November 9 | Lean proof: polynomially clocked nondeterministic verifier for the reference language's complement. |
| Integer factorization in P | BBB(324) (beep state) | GregoryMorse, Claude | November 9 | Lean proof: bounded-divisor decision language in P. |
| Discrete logarithms modulo primes in P | BBB(400) (beep state) | GregoryMorse, Claude | November 9 | Lean proof: bounded-exponent decision language in P. |
| Linear-time multiplication | BBB(403) (beep state) | GregoryMorse, Claude | November 9 | Lean proof: linear time in the queue-of-stacks model. |
| Graph isomorphism in P | BBB(415) (beep state) | GregoryMorse, Claude | November 9 | Lean proof: graph-isomorphism decision language in P. |
| Linear-size NTT circuits | BBB(469) (beep state) | GregoryMorse, Claude | November 9 | Lean proof: chain-circuit condition. |
| Twin prime conjecture | BBB(19) (beep state) | GregoryMorse, Claude | November 9 | Lean proof; quasihalts iff the conjecture is false. |
| NP ⊆ P/poly | BBB(325) (beep state) | GregoryMorse, Claude | November 12 | Lean proof: polynomial advice for the reference language. |
| promise-P = promise-BPP | BBB(333) (beep state) | GregoryMorse, Claude | November 12 | Lean proof: deterministic decision of the specified promise problem. |
| L = P | BBB(373) (beep state) | GregoryMorse, Claude | November 12 | Lean proof: multihead decision of the specified reference language. |
| L = NL | BBB(415) (beep state) | GregoryMorse, Claude | November 12 | Lean proof: multihead decision of the specified reference language. |
| Linear-time GF(2) matrix multiplication | BBB(494) (beep state) | GregoryMorse, Claude | November 12 | Lean proof: linear-time GF(2) matrix product in the guest model. |
| Conjecture | Domain | Discoverer | Discovery date | Notes |
|---|---|---|---|---|
| Pillai's conjecture | BBBB(91) | charlie, Claude[8] | November 1 | Unknown - pseudoquasihalts if the conjecture is false. Proof. |