[FOUNDATION] Canonical Systems vs. Interpretive Systems
There are two kinds of systems in the world.
Canonical systems preserve what is.
Interpretive systems produce what it means.Most modern platforms pretend these are the same thing. They aren’t. And most systems do not explicitly preserve the distinction — it’s a foundational fault line in modern information systems.
A canonical system is the place where the record exists. It’s the place where the record lives, in a form that can be independently verified. A canonical system does not need to be smart. In fact, it should avoid being smart.
An interpretive system is a meaning engine. It filters, summarizes, scores, ranks, and narrates. It may be useful. It may even be honest. But it is not the record. It is a lens.
When you collapse these two into one, you get the worst of both worlds:
interpretation disguised as record
narrative disguised as evidence
“helpful” changes that cannot be audited
summaries that quietly replace originals
systems that can rewrite history without detection
The simplest definition that holds up under pressure
A canonical system can say:
This existed.
In this form.
At this time.
And that claim can be independently verified.
That’s it.
Anything beyond that is interpretation.
And interpretation is allowed — but it must never be confused with the canon.
Why this boundary matters (more than any feature)
Interpretation is inherently lossy. Even when done by the best people with the best intentions, it must select:
what to include
what to exclude
what to emphasize
what to minimize
what to frame as relevant
what to frame as noise
That selection is not evil. It’s unavoidable.
But selection becomes dangerous when the system does it silently — and then presents the output as if it were the record itself.
Canonical systems don’t assert truth
This is the part most people miss.
A canonical record is not a truth claim. It’s an existence claim.
A canonical system can record an event without asserting its correctness or truth.
An audit log can function as a witness, not a validator.Meaning is determined by humans and institutions outside the canonical system.
That restraint is not weakness. It’s the whole point.
Truth lives downstream — in institutions, courts, journalism, scholarship, and human judgment. The system should not compete with those. It should serve them.
The practical rule
If a system:
edits the record to “improve” it
merges records into a “single clean view”
replaces primary artifacts with summaries
makes interpretation the default posture
prevents export or independent verification
A canonical system does the opposite: it preserves, exposes, and allows independent verification of the original state.
Why I’m writing this now
We’ve built an entire civilization on interpretive systems pretending to be canonical.
That’s why we argue about “misinformation” endlessly while the actual problem — provenance — remains optional.
We don’t need smarter interpretations first.
We need a canon whose state can be independently proven.
In the next post: why provenance beats summaries every time.
--- CRAIGS Public License v1.0