HCI / 07 / Initial protocol
World-Model Interaction.
How should a requestor inspect incomplete, conflicting, and changing world state?
Interactive prototype / synthetic state
Inspect the uncertainty.
Illustrative envelope within corridor95% radius · 7.3 m
- Accuracy
- Not measured
- Integrity
- Not evaluated
- Confidence
- 95% assumed model
- Latency / age
- 0 s simulated age
Model, provenance, and limits
This is a deterministic interface test bench, not a sensor feed. For an isotropic 2D Gaussian, the assumed radial coverage is r95 = √(−2 ln 0.05) × σ. The toy scenario uses σ = base uncertainty + age × an illustrative source coefficient. Coefficients are authored test controls, not measured GNSS, visual, or inertial performance. A ±20 m corridor is illustrative. A positional confidence envelope is not an integrity protection level. No safety authority or hardware actuation is granted.
Research record
From question
to evidence.
- 01 / Question
- How should a requestor inspect incomplete, conflicting, and changing world state?
- 02 / Hypothesis
- Task-oriented questions with inspectable evidence may be easier to use than exposing every raw layer by default.
- 03 / Method & prototype
- Compare task-first questions with layer-first exploration against the same bounded environment. Include contradictory observations and unavailable data. The shared synthetic test bench is an interaction prototype; completing the proposed study requires task design, participant recruitment where relevant, and recorded evaluation.
- 04 / Observation
- Evaluation pending. No participant-study results or field-performance findings are reported in this initial protocol.
- 05 / Measures
- Task completion, provenance comprehension, missed conflicts, and unnecessary interaction steps.
- 06 / Systems implication
- Candidate: return the required state with its uncertainty and lineage, while keeping deeper evidence inspectable. Promotion depends on evidence of improvement against the baseline, with failure cases and limitations recorded.