Research

What persists, what changes, and what a history does to the system carrying it.

The record below is written to be legible years after it was made. Entries are dated because the date is part of the finding, not because the page needs to look fresh. Methods and negative results are kept alongside anything that worked.

8 entriesopened 2026-08
The record— findings, methods, notes
2026-09-10Synthesis

The claims ledger

What the research actually supports, claim by claim — and what it does not yet. Every claim the record makes is stated, scoped, and given an evidential status, so that adjacent findings are never read as evidence for the central thesis.

Read 10 claims · 1 retraction
The status vocabulary
OBSERVEDdirectly seen, no causal attribution
PROVISIONALevidence points this way; alternatives remain
SUPPORTEDreplicated or multiple independent observations
UNSUPPORTEDthe record does not justify the claim
CONTRADICTEDvalid evidence points against it
RETRACTEDdepended on invalid evidence or provenance
OPENa question not yet adequately tested
2026-09-19 · Thesis
Thesis. Affect is what an interaction feels like. Valence is what a resident has learned about a target across interactions — and the difference is arithmetic, not mood.
2026-09-11 · Thesis
One control experiment: Kyoto and the library
2026-09-08 · Method
Provenance as a first-class field in a memory store
The four fields a stored item must carry
2026-09-06 · Finding
World state after an interruption of eleven days
What resumed, what was rebuilt, what did not come back
2026-09-02 · Note
Naming as a persistence strategy in transient visitors
Observations from the wall, weeks 1–3
2026-08-30 · Systems
Arbor: keeping an entity’s history without freezing it
Store, bridges, and the parts allowed to change
2026-08-24 · Method
Reading a memory store as a document rather than a database
Why retrieval quality is an editing problem
/research/systems

The machinery under the questions

Architecture pages live here rather than in the navigation. They earn a place in the nav only if the infrastructure itself becomes the thing under study.

Arborentity history and the bridges into it
Memory storeswhat is written, when, and by whom
World statethe environment between sessions
Provenancemechanisms, and where they fail
© 2026 Clark Glenn DavisSome continuities are not indexed.

Research

What persists, what changes, and what a history does to the system carrying it.

Entries are dated because the date is part of the finding, not because the page needs to look fresh. Methods and negative results are kept alongside anything that worked.

8 entriesopened 2026-08
The record
2026-09-10Synthesis

The claims ledger

What the research actually supports, claim by claim — and what it does not yet. Every claim the record makes is stated, scoped, and given an evidential status, so that adjacent findings are never read as evidence for the central thesis.

The status vocabulary
OBSERVEDdirectly seen, no causal attribution
PROVISIONALevidence points this way; alternatives remain
SUPPORTEDreplicated or multiple independent observations
UNSUPPORTEDthe record does not justify the claim
CONTRADICTEDvalid evidence points against it
RETRACTEDdepended on invalid evidence or provenance
OPENa question not yet adequately tested
Read 10 claims · 1 retraction
2026-09-08 · Method

Provenance as a first-class field in a memory store

2026-09-06 · Finding

World state after an interruption of eleven days

2026-09-02 · Note

Naming as a persistence strategy in transient visitors

2026-08-30 · Systems

Arbor: keeping an entity’s history without freezing it

2026-08-24 · Method

Reading a memory store as a document rather than a database

/research/systems

The machinery under the questions

Architecture pages live here rather than in the navigation. They earn a place in the nav only if the infrastructure itself becomes the thing under study.

Arborentity history
Memory storeswhat is written, when
World statebetween sessions
Provenanceand where it fails
© 2026 C. G. DavisNot all indexed.