Permanent project lineage
52 repositories. One readable development history.
Source repositories may be deleted after their useful code, results, and ideas are preserved. This page records what each branch attempted, what survived, and where the work moved next.
Archive rule
Deleting a repository is allowed. Deleting the lesson is not.
“Historical” does not mean false or worthless. It means the code no longer carries the current public architecture.
“Superseded” means the useful mechanism has a clearer successor. “Scratch” means no enduring public claim needs to survive.
External links appear only where a repository still has an active or supporting role. The archive itself remains meaningful after a GitHub purge.
2026 — present
Current programmes
The repositories carrying the current scientific and engineering questions.
Ten-SON-LM
Original idea
Recurrent semantic workspace with learned tension controlling revision, sparse routing, relaxation, and readout.
What survives
Workspace architecture, causal tension interventions, and matched learned-gate baselines.
Successor
Current model-science line
TensionForge
Original idea
Verifier-first OpenCL training runtime for legacy AMD hardware.
What survives
Forward/backward/optimizer parity, persistent device state, and honest performance diagnosis.
Successor
Current compute/runtime line
tsq
Original idea
Verifier-gated adaptive precision that escalates compute when tension justifies it.
What survives
Precision-routing semantics, repair loops, and cognitive/compute receipts.
Successor
Current adaptive-inference line
ts-chat-language
Original idea
Deterministic text-to-MeaningGraph compilation and a bounded verifier-first symbolic agent loop.
What survives
Stable semantic identities, explicit goals/topology, transactional execution, replanning, replay, and a non-LLM language boundary.
Successor
Current language/interface line
TS-Reasoner-v0
Original idea
Typed verifier authority over conclusions, goals, plans, actions, effects, lessons, memory, and rejection.
What survives
The verifier-first operation firewall, transactional symbolic environment, and receipt-backed agent transitions.
Successor
Current verifier/reasoning line
TS-Benchmarks
Original idea
Audit-first falsification harness for graph, reasoning, model, and compute experiments.
What survives
Fixed generators, matched baselines, replay, and visible failure categories.
Successor
Shared evidence layer
bogbin
Original idea
Verified storage and computing substrate with canonical state, capabilities, workspaces, journals, and rollback.
What survives
Exact round trips, state roots, deterministic transitions, and runtime receipts.
Successor
Current verified-computing line
TensionLM
Original idea
Earlier sigmoid pairwise-tension attention and model telemetry experiments.
What survives
Runnable baselines, public artifacts, matched comparisons, and negative results.
Successor
Ten-SON-LM
Reusable or specialist branches
Supporting infrastructure
Useful machinery and bounded demonstrations that should not compete for the word flagship.
TS-Core
Original idea
Compact graph/tension kernel for propagation, relaxation, contradiction, and revision.
What survives
The reusable substrate vocabulary and deterministic mechanics.
Successor
Shared library
BoggersTheCIG
Original idea
Large provenance-aware claim, evidence, confidence, contradiction, and revision graph.
What survives
The durable knowledge-graph vision and provenance discipline.
Successor
cig-ts-engine and this website
cig-ts-engine
Original idea
Smaller deterministic CIG engine for graph state, tension, derivatives, and Break/Evolve proposals.
What survives
A compact implementation that is easier to inspect and test.
Successor
Canonical compact CIG implementation
TS-Codex-OS
Original idea
Project graph, tension ledger, planner, and receipts for local coding workflows.
What survives
Engineering state and planned transitions should be explicit and reviewable.
Successor
Verifier-first development tooling
3b_solution
Original idea
Tension-adaptive integration applied to the three-body problem.
What survives
A concrete application and an honest loss against DOP853 in the tested setup.
Successor
Application evidence
BogOS-Native
Original idea
Native-oriented exploration of lower-level BogOS execution.
What survives
The boundary between user-space verified prototypes and future native work.
Successor
bogbin
Navigation and identity
Public entrypoints
Repositories that explain, route, or present the work rather than add another mechanism.
TS-Start-Here
Original idea
Credibility-first ecosystem map.
What survives
The need for one canonical explanation.
Successor
This website
BoggersTheFish
Original idea
GitHub profile index.
What survives
A minimal route from GitHub to current work.
Successor
Reduced profile plus this website
boggersthefish-site
Original idea
Permanent field archive for current systems, evidence, failures, and deleted-repository ideas.
What survives
The whole public narrative and lineage.
Successor
Canonical public record
Prototype waves
Earlier reasoning and model lineage
Earlier attempts that developed the vocabulary and boundaries of TS. Their code may disappear; their useful ideas remain here.
GOAT-TS
Original idea
Early broad implementation of TS as graph-based reasoning.
What survives
Constraint graphs, propagation, tension, relaxation, and explicit reasoning state.
Successor
TS-Core and TS-Reasoner-v0
GOAT-SIMPLE
Original idea
Stripped-down presentation of early GOAT mechanics.
What survives
Small inspectable demonstrations before ecosystem claims.
Successor
TS-Core
BOG-TS
Original idea
Alternate TS packaging during architectural consolidation.
What survives
The move from broad branding toward modular components.
Successor
TS-Core and TS-Reasoner-v0
GOAT-TS-DEVELOPMENT
Original idea
Fast-moving development branch for GOAT experiments.
What survives
Intermediate experiments absorbed by dedicated repositories.
Successor
Specialised TS repositories
GOAT-TS-LITE
Original idea
Reduced GOAT runtime.
What survives
Pressure toward smaller reproducible kernels.
Successor
TS-Core
GOAT-TS-SUPERLITE
Original idea
Minimal GOAT derivative.
What survives
Preference for falsifiable mechanisms over all-in-one systems.
Successor
TS-Core and microbenchmarks
GOAT-PUBLIC_TEST
Original idea
Public testing surface for early GOAT work.
What survives
Repeatable entrypoints and evidence packs.
Successor
TS-Benchmarks and this website
BoggersTheSystem
Original idea
Minimal synthesis of memory, reasoning, and response.
What survives
System functions later separated into explicit boundaries.
Successor
ts-chat-language and TS-Reasoner-v0
BoggersTheAI
Original idea
Broader TS-native assistant/runtime synthesis.
What survives
Usability and reasoning substrate should be developed separately.
Successor
ts-chat-language and TS-Reasoner-v0
BoggersTheAI-Dev
Original idea
Development branch of BoggersTheAI.
What survives
Intermediate experiments absorbed elsewhere.
Successor
Current language and verifier lines
BoggersTheMind
Original idea
Local assistant and personal cognitive-system branch.
What survives
Persistent state, inspectable memory, and a human-facing interface.
Successor
ts-chat-language and CIG
BoggersTheLLM
Original idea
Early custom language-model direction around TS concepts.
What survives
A TS-native language system need not be an external-LLM wrapper.
Successor
TensionLM, Ten-SON-LM, and ts-chat-language
ts-llm
Original idea
Attractor-oriented TS language-model experiment.
What survives
Stable semantic configurations and structured state as model targets.
Successor
Ten-SON-LM
woke-baby-llm
Original idea
Tiny baseline exposing basic trainable language-model mechanics.
What survives
A simple reference against more complex mechanisms.
Successor
TinyLLM and TensionLM
TinyLLM
Original idea
Small educational model experiment.
What survives
Hands-on training experience and a compact baseline.
Successor
TensionLM and Ten-SON-LM
BAGI
Original idea
Broad sketch of how TS components might form a more general system.
What survives
System-decomposition questions, not an AGI claim.
Successor
Modular current programmes
BLM
Original idea
Alternative custom language-model branch.
What survives
Experimental model-building experience.
Successor
TensionLM and Ten-SON-LM
BoggersBrain
Original idea
Early cognitive-architecture container.
What survives
Cognitive functions should be explicit modules.
Successor
TS-Core, CIG, and TS-Reasoner-v0
BoggersThePulse
Original idea
Pulse, activation, and evolving-state dynamics.
What survives
Dynamic activation and propagation as first-class state.
Successor
TS-Core and tension-based models
BoggersTheEngine
Original idea
General engine/runtime synthesis for multiple TS mechanisms.
What survives
Clear runtime boundaries and reusable components.
Successor
TS-Core, TS-Reasoner-v0, and bogbin
Discarded scratch branches
Original idea
Two repositories contained no enduring public research idea worth carrying forward.
What survives
Their existence is counted for an honest inventory; the names and code do not need a permanent public role.
Successor
None
Represents 2 repositories
Parallel explorations
OS, CIG, wave, and convergence branches
Interfaces, operating-system metaphors, wave dynamics, multi-agent coordination, and attempts to converge the ecosystem.
GOAT-OS
Original idea
Early TS-as-operating-system substrate.
What survives
Runtime, capability, transition, and receipt framing.
Successor
bogbin
BoggersTheOS-Alpha
Original idea
Alpha-stage OS branch combining TS and low-level runtime ambitions.
What survives
The distinction between simulation, user-space substrate, and native execution.
Successor
bogbin and BogOS-Native
BoggersTheOS-Beta
Original idea
Later staged OS branch.
What survives
Deterministic state, persistent storage, and verifiable transitions.
Successor
bogbin
CIG-APP-V1
Original idea
First UI surface for navigating CIG knowledge and provenance.
What survives
Human-readable interfaces over claims and revisions.
Successor
This website
CIG-APP-V2
Original idea
Second CIG application iteration.
What survives
Separation between graph engine and presentation.
Successor
cig-ts-engine and this website
BoggersTheCIG_v2
Original idea
Second full CIG implementation branch.
What survives
Provenance-aware knowledge state.
Successor
cig-ts-engine and this website
WaveLab branches
ts-wave-lab · TS-WaveLab
Original idea
Parallel laboratories for propagation, relaxation, and wave interaction.
What survives
Wave dynamics as an experiment surface rather than a flagship.
Successor
TS-Core and TS-Benchmarks
Represents 2 repositories
ts-wave-colony
Original idea
Colony-style coordination among interacting TS processes.
What survives
Distributed local/global tension questions.
Successor
Future bounded coordination experiments
ts-visualizer
Original idea
Visualisation of graph state, activation, waves, and tension.
What survives
Internal state must be visible enough to debug.
Successor
Website diagrams and project diagnostics
TS-MultiAgent
Original idea
Independent tension graphs coordinating through shared waves.
What survives
A bounded multi-agent coordination question.
Successor
Future verifier-gated coordination work
ts-os-unified-convergence
Original idea
Wave 17 attempt to converge many TS repositories into one system.
What survives
Convergence should happen through interfaces and a public map, not another giant repo.
Successor
Current modular stack and this website
boggers-the-dream
Original idea
Offline consolidation or dream-cycle restructuring of stored state.
What survives
Memory consolidation should be explicit, bounded, and reviewable.
Successor
CIG and verifier-gated memory