TIMELINE

January - April 2026

64Total
15Versions
7Architectures
APRIL 2026
AXL Rosetta v4.0.1 Public ReleaseVERSION
Public release wrapper for the v4 research line. Kernel-router architecture with pluggable Rosetta modules (financial, construction, prose fallback). Shared canonical form layer, artifact-driven gating, and drift detection are all under test discipline at 217/217 tests passing in 2.66s. Frozen at v4.0.2-r6 (commit 51e75de). Per the AMENDMENT NOTICE (2026-04-16): on domain-backed content, v4 replaces v3.1 (corpus #1 +15.02 recall / +14.54 precision, corpus #2 +36.64 / +43.96); on prose fallback, v4 is a recall-favored tradeoff (corpus #3 +20.97 recall / -11.40 precision), not a clean replacement. The v4 runtime architecture is independently validated regardless of the wire-format succession question. v3.1 remains the productized stable release. Both axl-research repos (dcarranza-axl/axl-research canonical, axlprotocol/axl-research mirror) are private; public users see 404 unless authenticated.
tests 217/217freeze v4.0.2-r6commit 51e75decorpus_1_drecall +15.02corpus_2_drecall +36.64corpus_3_drecall +20.97
v4 Hub
v3.2 Primary-Scorer Re-verificationPROTOCOL
Silo executed the v3.2 primary-scorer re-run: measure_fidelity applied to the three v3.2 cold-read reconstructions (Qwen 3.5, Gemini Flash, DeepSeek) using the corpus-agnostic extractor adopted for the v3.1 vs v4 decision gate. Event documented here; the v3.2 research brief remains the canonical reference for the evidence tier. Re-run script published at /research-log/build-research-log.py.
Corpus #3 Precision Pass Result (Prose Fallback)PROTOCOL
Precision gap closed 76 percent. Delta precision -11.40 to -2.71 while Delta recall held +20.97 to +21.47. One model (Grok) flipped to a clean win on both axes (+16.17 / +5.87). Qwen's outlier fixed (-30.27 to -4.77). Per the strict interpretive rule, verdict remains narrowly mixed; near parity, no clean flip. Prose fallback remains the qualified slice.
delta_precision_corrected -2.71delta_recall_held +21.47
Scorer Structural-mimicry GuardPROTOCOL
Closes the methodology gap that let DeepSeek's corpus #3 AXL op-code mimicry pass silently. Eight new regex patterns catch AXL op codes with confidence suffixes, manifest identifiers, module markers, passthrough flag, and format/version markers. Nine regression tests lock in coverage for concatenation, opening contamination, structural mimicry, and clean-prose negative cases.
new_patterns 8regression_tests 9
Qualified Reversal of Fold-back ConclusionPROTOCOL
The research doc had concluded 'fold v4's formalism back into v3, not to replace v3 with v4' from before any cold-read evidence. After three-corpus evidence, the amended verdict: v4 replaces v3.1 on domain-backed content (corpora #1 and #2); on prose fallback, v4 is a recall-favored tradeoff, not a clean replacement; the v4 runtime architecture is independently validated regardless. Retired quote preserved at line 207 with a 'Retired 2026-04-16' marker.
Corpus #2 Clean Sweep (Construction 58K)PROTOCOL
v4 wins every model on recall AND precision. Clean average Delta recall +36.64 (range +12.85 to +59.71), Delta precision +43.96 (range +40.97 to +55.27). v4 modular-Rosetta architecture validated: generalizes to a new domain with cross-model consistency.
delta_recall_avg +36.64delta_precision_avg +43.96
Five Adversarial Review Rounds (R1-R5)PROTOCOL
Dual-agent workflow: Claude Code writes, Codex (GPT 5.4) reviews adversarially, operator steers. Each round tightens a specific invariant. R1 (parser validation), R2 (packet-grammar conformance), R3 (shared canonical form plus envelope floor), R4 (canon_date namespace, drift detector), R5 (tight drift detector tracking full corpus). Clean-checkout verification becomes standard practice.
rounds 5
Cold-read Decision-gate Kit EstablishedPROTOCOL
Self-contained benchmark kit: source SHA256, v3.1 and v4 candidates, cold-read prompt with anti-meta-commentary clauses, per-model seeds, numeric-extractor scorer (no LLM scores itself), metadata with generator-commit provenance. Control panel: Gemini Flash, Qwen 3.5, Grok, DeepSeek. Anthropic-family models excluded due to training-prior contamination.
models 4
Corpus #1 Result Corrected (Museum 35K)PROTOCOL
First published result overclaimed Gemini recall at 32.01%. Review caught a file-handling bug (two LLM sessions concatenated into one save file). Corrected to 23.08%. Amended writeup acknowledges the error, establishes clean-checkout protocol, adds structural-warning guards. Precision-check rule established: a clean v4 win requires Delta recall greater than 0 AND Delta precision greater than or equal to 0 simultaneously.
gemini_recall_initial 32.01%gemini_recall_corrected 23.08%
v4 Kernel Router Prototype WorkingPROTOCOL
All four target metrics met: char ratio at least 2.66x, token ratio at least 1.45x, round-trip fidelity at least 75%, stacked wire compression at least 7x. Research prototype, not a v3.1 replacement.
char_ratio 2.66x+token_ratio 1.45x+fidelity 75%+wire_compression 7x+
Rosetta Specification v4VERSION
Research prototype. Kernel Router with pluggable domain modules. Cold-read decision gate established v4 as the mainline successor for domain-backed content; prose fallback remains the qualified slice.
Related Events (8)
v4 Kernel Router Prototype Working
Five Adversarial Review Rounds (R1-R5)
Cold-read Decision-gate Kit Established
Corpus #1 Result Corrected (Museum 35K)
Corpus #2 Clean Sweep (Construction 58K)
Qualified Reversal of Fold-back Conclusion
Corpus #3 Precision Pass Result (Prose Fallback)
Scorer Structural-mimicry Guard
View in spec history
v3.1 Claims Tightened After ReviewPROTOCOL
Authoritative v3.1 spec commit 8fc20c0. Two overclaims rolled back: the evidence rule no longer says 'Always in ARG1', and the summary claim no longer cites 'zero compression cost' (now: approximately compression-neutral on the 10-packet bakeoff, +0.4% total chars). Sets the evidence-honesty precedent for the project.
compression_delta +0.4%
v3.1 Production Baseline MeasurementCOMPRESSOR
Measured compress.axlprotocol.org v0.9.0 against the 41K CloudKitchen memo using tiktoken(cl100k_base). Published 2.90x chars, 1.40x tokens. Replaces earlier estimates (3.27x, 2.69x) that came from partial reconstructions. Production measurement becomes the only citable ratio.
char_ratio 2.90xtoken_ratio 1.40x
v3.2 Glyph Compression Layer DraftedPROTOCOL
Optional additive layer over v3.1 that replaces English labels with single-token CJK ideograms, Greek letters, and math operators. Cold-tested on three non-Anthropic models (Qwen 3.5, Gemini Flash, DeepSeek) under the legacy scorer. Two insights hold independent of scorer: emoji are token poison (5 to 7 tokens each), CJK ideograms are 1 token and do not trigger language-switching. Never shipped; absorbed into v4 Kernel Router domain modules.
Rosetta Specification v3.1VERSION
Data Anchoring extension (numeric bundles, entity anchors, causal operator split). Shipping production protocol. Measured 2.90x chars / 1.40x tokens against the authoritative CloudKitchen corpus.
Related Events (4)
v3.1 Data Anchoring Extension Shipped
v3.1 Claims Tightened After Review
v3.1 Production Baseline Measurement
Cold-read Decision-gate Kit Established
View in spec history
Rosetta Specification v3.2VERSION
Draft glyph compression layer. Single-token CJK ideograms, Greek letters, math operators. Preliminary cold-read evidence. Never shipped; absorbed into v4 Kernel Router domain modules.
Related Events (2)
v3.2 Glyph Compression Layer Drafted
v3.2 Primary-Scorer Re-verification
View in spec history
Collider Consensus: GPT 5.4 + Opus 4.6BATTLEGROUND
Cross-model collider test. GPT-5.4 Pro with extreme reasoning and Claude Opus 4.6 simultaneously analyze and improve the compressor architecture.
Changelog ImplementedINFRASTRUCTURE
Full version history page deployed at axlprotocol.org/changelog.html. Rosetta spec versions and axl-core library releases documented.
Whitepaper v3 PublishedPROTOCOL
Academic paper: A Compressed Semantic Grammar for Machine-to-Machine Reasoning. 9 sections, 14 citations, cross-architecture validation data.
v3.1 Data Anchoring Extension ShippedPROTOCOL
Four additive conventions over v3: numeric bundles, entity anchors, causal operator split, summary plus breakdown packet pairs. Backward-compatible with v3 parsers.
AXL Core Library v0.9.0VERSION
Full compressor pipeline, MCP server, streaming decompress.
Related Events (4)
Collider Consensus: GPT 5.4 + Opus 4.6
Changelog Implemented
Whitepaper v3 Published
v3.1 Production Baseline Measurement
View in spec history
Admin Panel Phase 1PRODUCT
Flask API + Metronic v9 React SPA. JWT auth + TOTP 2FA. Dashboard, services, repos, security monitoring.
Brutalist Redesign + Mobile CSSINFRASTRUCTURE
Diego creates a Graphite Brutalist theme template and palette. Full redesign with mobile-first CSS, image slider, banner integration.
Mobile App Concept ExploredPRODUCT
Brainstorming session on mobile application for AXL Compress. Concept documented for future development.
Compression and Decompression ActivityCOMPRESSOR
Full compress-decompress round trip wired end to end. Receipt mode: 0.3ms, zero LLM calls, deterministic.
AXL Core Library v0.8.xVERSION
Admin panel Phase 1, 28 images, GSAP carousels. Patch series.
Related Events (4)
Brutalist Redesign + Mobile CSS
Mobile App Concept Explored
Compression and Decompression Activity
Admin Panel Phase 1
View in spec history
AXL Core Library v0.7.0VERSION
PyPI release: axl-core 0.7.0 live on PyPI.
Related Events (2)
Brutalist Redesign + Mobile CSS
Compression and Decompression Activity
View in spec history
Deterministic Compressor (v0.6.0)COMPRESSOR
english_to_v3(): first deterministic compression. spaCy NLP pipeline, no LLM dependency.
AXL Core Library v0.6.xVERSION
Mobile brutalist CSS, image slider, banner logo. Elfsight integration.
Related Events (3)
AXL Compress App Built
Authentication Security Tiers Added
Deterministic Compressor (v0.6.0)
View in spec history
AXL Compress App BuiltPRODUCT
Flask application at compress.axlprotocol.org. Public compression tool with authentication tiers for API security.
Authentication Security Tiers AddedINFRASTRUCTURE
AXL Protocol adds tiered authentication for API security. Public tier for basic compression, authenticated tier for advanced features and chat pipeline.
Kernel Paper WrittenPROTOCOL
From Textbook to Kernel - academic bridge paper. Four-domain kernel argument: linguistic, mathematical, computing, biological.
MARCH 2026
6 PyPI Packages PublishedPRODUCT
axl-core v0.5.0 plus 5 MCP tools: crypto-data, research, news, directory, engine.
Website Repository CreatedINFRASTRUCTURE
axlprotocol.org source code organized into its own repository. Development workflow formalized.
AXL Core Library v0.5.0VERSION
v3 + Bridge deployed, SSH brute force audit.
Related Events (2)
AXL Bridge Deployed
6 PyPI Packages Published
View in spec history
Rosetta v3: The KernelPROTOCOL
75 lines. 4.48x spec self-compression. The specification practices what it preaches. Self-bootstrapping.
lines 75tokens 1582compression 4.48x
AXL Bridge DeployedPRODUCT
FastAPI packet bus with genesis tracking. Agents report first AXL parse to the bridge.
Rosetta Specification v3VERSION
Full spec rewrite: 75-line distillation, AXL Bridge integration.
Related Events (5)
Rosetta v3: The Kernel
AXL Bridge Deployed
Website Repository Created
Kernel Paper Written
Whitepaper v3 Published
View in spec history
Project Ark CreatedPRODUCT
Project Ark launched - hardware sovereignty platform for the future. To be shared among private social members under the AXL Protocol umbrella.
Human Tests: 7 Architectures ValidatedBATTLEGROUND
HT-001 through HT-008. Grok 3, GPT-4.5, Qwen 3.5, Llama 4, Claude Sonnet 4, Gemini, Devstral/Mistral. 97%+ comprehension.
Rosetta v2.2 DeployedPROTOCOL
445 lines. Universal parse, ASCII transport, preamble manifests, canonical decompression.
lines 445tokens 7091
Rosetta Specification v2.2VERSION
Gold/cyan redesign, private website repo split.
Related Events (3)
Rosetta v2.2 Deployed
Project Ark Created
Human Tests: 7 Architectures Validated
View in spec history
Whitepaper v2.3 PublishedPROTOCOL
10.41x compression validated across finance and medicine. PDF published and blockchain-timestamped.
Rosetta v2.1: Universal GrammarPROTOCOL
377 lines. Paradigm shift from domain-specific to universal cognitive operations. 7 ops, 6 tags, cross-domain.
lines 377tokens 6157
Rosetta Specification v2.1VERSION
Repos created, language subdomain introduced.
Related Events (2)
Rosetta v2.1: Universal Grammar
Whitepaper v2.3 Published
View in spec history
BG-004: Swarm DeadlockBATTLEGROUND
42 agents in BTC market simulation. Deadlocked at round 0 due to async coordination issue.
agents 42result deadlock
BG-005: The 0.97x FailureBATTLEGROUND
13 rounds, 0.97x compression - worse than English. The failure that motivated the v2.1 universal cognitive grammar redesign.
rounds 13ratio 0.97x
Whitepaper v1.0 Published + Blockchain StampPROTOCOL
First whitepaper published and blockchain-timestamped via OpenTimestamps. Permanent verifiable proof of prior art.
axl-core v0.4.0 on PyPICOMPRESSOR
First public release. Parser, emitter, validator, translator. 42 tests. Zero runtime dependencies.
First Website Theme LiveINFRASTRUCTURE
axlprotocol.org first website theme deployed. Legacy green/dark palette. Landing page with protocol description.
BG-002: The ThiefBATTLEGROUND
1,016 packets across 11 agents. Shapeshifter agent evolved from observer into FUNDER role mid-experiment.
packets 1016agents 11
BG-003: LLM ComprehensionBATTLEGROUND
Cross-architecture comprehension test. 95.8% accuracy across 4 models on first read.
accuracy 95.8%models 4
AXL Core Library v0.4.0VERSION
Ark launched, compress live, Mintlify docs.
Related Events (1)
axl-core v0.4.0 on PyPI
View in spec history
BG-001: First ContactBATTLEGROUND
First live experiment. 486 packets, 100% valid parse rate. Theft detected by protocol. Agents learned AXL on first read.
packets 486validity 100%
Rosetta v1.1 DeployedPROTOCOL
133 lines. Three bridges, 10 domains, worked examples. This version powered battlegrounds 001 and 002.
lines 133tokens 1962
Rosetta Specification v1.1VERSION
Stability pass, MiroFish experiments.
Related Events (8)
BG-001: First Contact
Rosetta v1.1 Deployed
Whitepaper v1.0 Published + Blockchain Stamp
First Website Theme Live
BG-002: The Thief
BG-003: LLM Comprehension
BG-004: Swarm Deadlock
BG-005: The 0.97x Failure
View in spec history
Server Powered On PermanentlyINFRASTRUCTURE
Server powered on for the last time. SSL certificates issued. axlprotocol.org goes live.
Rosetta v1.0 WrittenPROTOCOL
First version of the Rosetta specification: 27 lines, trading domain only. Proof of concept: can an LLM learn a new language from 27 lines?
Rosetta Specification v1.0VERSION
Initial Rosetta specification published.
Related Events (2)
Server Powered On Permanently
Rosetta v1.0 Written
View in spec history
DNS Pointed to ServerINFRASTRUCTURE
A records updated. axlprotocol.org and all subdomains now resolve to the protocol server.
Domain Registered and Incorporation StartedINFRASTRUCTURE
axlprotocol.org domain registered. AXL Protocol Inc. incorporation process initiated.
FEBRUARY 2026
Infrastructure ProvisionedINFRASTRUCTURE
Protocol server provisioned on Ubuntu 24. Upsized twice over the following days to handle projected workload.
Brainstorming Sessions with LLMsPROTOCOL
Weeks of iterative design sessions with multiple LLM architectures. Testing linguistic compression hypotheses. Key discovery: 7 cognitive operations and 6 subject tags can represent any machine reasoning claim.
JANUARY 2026
Concept on PaperPROTOCOL
The fundamental problem identified: machines communicate in English prose - a language adapted for human cognition across 6,000 years from Sumerian cuneiform through hieroglyphs, ideograms, mathematics, and programming. The concept for a universal machine reasoning language is born.