lithoSpore — Targeted GuideStone Deployment

USB-deployable, self-verifying scientific validation artifacts. The first instance reproduces the Long-Term Evolution Experiment (LTEE) targeting the Barrick Lab at UT Austin.

What This Is

lithoSpore is the ecosystem’s deployment system for self-verifying scientific artifacts. A lithoSpore is a USB drive (or network-deployable archive) that carries validated science, its data, its tools, and its provenance chain — everything needed to independently reproduce results on any machine, with zero dependencies and no internet.

The first instance targets the Barrick Lab at UT Austin — the continuation of Richard Lenski’s Long-Term Evolution Experiment (LTEE), the longest-running evolutionary biology experiment in history.

Repository: sporeGarden/lithoSpore

Spore Taxonomy

The ecoPrimals ecosystem uses a biological metaphor for deployment artifacts. Each class adds capability:

Class	Self-sufficient?	Size	What it carries
coldSpore	No	< 1 KB	Static marker + frozen data snapshot
liveSpore	Partial	~KB	+ `liveSpore.json` journal + `./refresh` update mechanism
pseudoSpore	No	KB–MB	+ Braids, receipts, derivation configs, provenance — proves the mountain was climbed
lithoSpore	Yes	MB–GB	+ Python runtime + Rust binaries + full data — carries everything to reproduce on its own

The spore can’t carry the mountain, but it proves the mountain was climbed. A pseudoSpore provides the proof. A lithoSpore provides the proof and the tools to re-climb it.

Current Status

75/75 checks across 7 science modules — all PASS at Tier 2 (Rust)
6 published papers reproduced: Wiser 2013, Barrick 2009, Good 2017, Blount 2008/2012, Burden 2024, Tenaillon 2016
Anderson disorder framework applied to LTEE fitness data — GOE/Poisson eigenvalue statistics
Single 5.1 MB binary (musl-static) — no runtime dependencies
Cross-platform: Ubuntu, Alpine, Fedora, Debian, read-only FS, Windows (7.9 MB)
108 unit + 16 integration + 15 chaos/fault-injection tests

Three Operating Modes

lithoSpore discovers its environment at runtime and adapts:

Mode	Network	Discovery	Validation Tier
Standalone	None — airgapped USB	No primals	1–2
LAN	Local network	env vars / UDS socket	2
Geo-delocalized	Remote / WAN	songBird TURN relay via cellMembrane	2–3

probe_operating_mode() records the mode in liveSpore.json, the append-only provenance journal that tracks every validation run (BLAKE3-hashed hostnames, no PII).

Three-Tier Validation

Tier	Runtime	What runs
1	Python notebooks	numpy/scipy baselines — reference implementations
2	musl-static Rust	In-process `run_validation()` — 5.1 MB binary, zero dependencies
3	NUCLEUS primals	Tier 2 + provenance trio (rhizoCrypt DAG, loamSpine attestation, sweetGrass braid)

Tier 2 proves the science. Tier 3 proves the provenance chain is sovereign.

pseudoSpore Lifecycle

pseudoSpores are lightweight proof artifacts emitted by springs and consumed by lithoSpore:

Spring validates → litho emit-pseudospore → pseudoSpore archive
                                              ↓
                         litho ingest-pseudospore → registry.toml
                                              ↓
                         litho promote → lithoSpore candidate

Command	Purpose
`litho emit-pseudospore`	Create archive from validated module state
`litho ingest-pseudospore`	Validate + register in `pseudospores/registry.toml`
`litho fetch-pseudospore`	Remote download + validate
`litho audit`	10-check pre-handoff validation
`litho promote`	pseudoSpore → lithoSpore candidate

Browse available pseudoSpores in the pseudoSpore Gallery.

Deployment Vision

The target architecture makes lithoSpore artifacts accessible through the mesh:

USB handoff — plug in, run ./validate, done. Zero install, zero internet.
Pepti depot — pre-built ecobins distributed via the sovereign depot (plasmidBin)
Mesh-accessible — sporePrint hosts the gallery; songBird routes capability calls to the serving gate
Historical provenance — liveSpore.json journals accumulate across validations, building a chain of independent reproductions

The chassis is domain-agnostic. New guideStones require only a scope.toml, data.toml, and domain-specific module crates — the deployment infrastructure, validation harness, and provenance machinery are reused.

Seven Science Modules (LTEE Instance)

#	Module	Paper	Checks
1	ltee-fitness	Wiser 2013 (Science)	8/8
2	ltee-mutations	Barrick 2009 (Nature)	7/7
3	ltee-alleles	Good 2017 (Nature)	20/20
4	ltee-citrate	Blount 2008/2012 (PNAS/Nature)	11/11
5	ltee-biobricks	Burden 2024 (ACS SynBio)	15/15
6	ltee-breseq	Tenaillon 2016 (Nature)	7/7
7	ltee-anderson	Anderson localization framework	7/7