PROVENANCE did not begin from an empty design space.
Several earlier QSOL projects developed pieces of the engineering discipline that this repository generalizes into a framework-neutral chain-of-custody system.
This file records that lineage.
It is historical and architectural context.
It does not make donor repositories normative dependencies.
QEC
│
│ deterministic identity
│ canonical representation
│ invariant discipline
│ receipts
│ recompute-not-trust
│
▼
UFF
│
│ closed evidence bundles
│ source manifests
│ precommit witnesses
│ trust boundaries
│ self-hash exclusion
│ external anchors
│ read-only observation
│ explicit proof scope
│
▼
PROVENANCE
framework-neutral
forensic chain of custody
for AI, software, tools,
humans and automated systems
Alongside this lineage:
OPT
│
└──────► efficient implementation and CI
without weakening correctness
QSOLKCB/QEC established recurring engineering patterns around:
canonical identity
deterministic serialization
cryptographic receipts
immutable artifacts
invariant-driven validation
recompute-not-trust
PROVENANCE retains those ideas at the evidence layer.
It deliberately does not inherit QEC's requirement that the observed system itself be deterministic.
QSOLKCB/UFF moved several of those patterns into concrete evidence workflows:
source
↓
identity commitment
↓
execution
↓
artifacts
↓
manifest
↓
closed bundle
↓
verification
↓
receipt / anchor
UFF also established an important boundary:
technical integrity
!=
scientific truth
PROVENANCE generalizes that principle:
valid evidence
!=
correct decision
valid signature
!=
true statement
valid chain of custody
!=
legal conclusion
PROVENANCE records evidence.
Interpretation remains downstream.
QSOLKCB/OPT contributes the rule that optimization must preserve the contract it accelerates.
For PROVENANCE:
FAST
is acceptable only when:
FAST_RESULT == REFERENCE_RESULT
under the applicable exactness contract.
Useful inherited disciplines include:
measure before optimizing
reuse only under proven identity
publish cache state only after success
keep a reference path
bound parallelism
test equivalence directly
do not weaken validation to improve timing
PROVENANCE is not QEC, UFF, or OPT combined.
Its purpose is narrower:
Preserve enough trustworthy evidence to determine what happened, how the record was produced, and where certainty ends.
The core system therefore prioritizes:
Evidence Fidelity
Non-Interference
Chain of Custody
Cryptographic Integrity
Observation Boundaries
Independent Verification
Framework Neutrality
Low Overhead
The intended evolution is:
evidence semantics
↓
invariants
↓
minimal schemas
↓
canonical representation
↓
artifact identity
↓
event recording
↓
custody
↓
verification
↓
adapters
↓
presentation
Performance work happens around that contract.
It does not redefine it.
PROVENANCE should not become a collection of every mechanism developed elsewhere.
The preferred architecture is:
small core
+
strict contracts
+
optional adapters
not:
large framework
+
mandatory middleware
+
hidden runtime machinery
Every additional operation performed during monitoring creates:
latency
failure surface
complexity
observer effect
verification burden
Therefore:
Collect the minimum evidence necessary to preserve the required claim accurately.
Minimalism must never justify omission of evidence required by the declared provenance contract.
Initial donor inspection for the PROVENANCE bootstrap used:
QEC
2c94351e667bc05aba26fe101dcf41fa8d4c9735
UFF
596cd732df61587aa1a9801cad1ec13483b1347f
OPT
3441c6ceacd4a6ca43ec46de758e2d5bc23d8ca7
These hashes identify the repository states examined during initial architecture work.
They do not freeze the donor repositories or require PROVENANCE to track future donor changes.
The lineage can be summarized as:
QEC taught us to bind the result.
UFF taught us to bind the evidence.
OPT taught us not to waste cycles doing it.
PROVENANCE makes the chain general.
Or more formally:
MINIMUM_OVERHEAD
subject to
ZERO_UNDECLARED_LOSS_OF_EVIDENTIARY_ACCURACY