Metadata Classes ​

For Realith, several classes of metadata must be distinguished.

1. Metadata without which the network cannot work ​

These are data needed for:

• routing;
• causal continuity;
• verification of admissibility;
• linking transitions to the canon.

Typical examples at this level:

• identifiers and references linking a transition to the object line;
• contour markers without which the reading regime cannot be determined;
• metadata linking a candidate step to the canonical outcome.

Without them, the network will not be able to retain object canon.

2. Metadata admissible for limited service ​

These are data that may be needed for:

• diagnostics;
• replication;
• refusal logging;
• publication of outcomes;
• servicing of the environment.

Typical examples here:

• log data about intake, rejection, or replication;
• limited diagnostic features;
• metadata needed to publish an outcome without disclosing the full content.

They must not automatically turn into a right to see all content.

3. Metadata that must be hidden or minimized ​

These are data the expansion of whose visibility destroys the boundaries of:

• the hidden contour;
• the closed contour;
• the compact publication regime.

Typical examples here:

• metadata through which hidden content can be reconstructed;
• excessive correlation features between closed objects;
• service traces that silently turn limited observability into a broad indexable surface.

If metadata classes are not separated, different implementations will begin to expand the observable surface in different ways.
Then:

• privacy boundaries will blur;
• contours will begin to sound stronger than they are actually implemented;
• operator observability may quietly expand.