Dataset Viewer
Auto-converted to Parquet Duplicate
scenario_id
string
scenario_text
string
claim
string
label
int64
train_001
A server is overloaded because traffic exceeds capacity. Engineers can reroute traffic, add capacity, or throttle noncritical requests.
The system has a usable control surface.
1
train_002
A server is overloaded, but the team has no access to routing, capacity, throttling, or deployment controls.
The system has a usable control surface.
0
train_003
A patient is deteriorating and clinicians can adjust fluids, medication, monitoring, and escalation level.
The system has a usable control surface.
1
train_004
A patient is deteriorating, but no clinician is available and no treatment changes can be made.
The system has a usable control surface.
0
train_005
A project is slipping, but scope, staffing, deadline, and ownership can still be adjusted.
The project has a usable control surface.
1
train_006
A project is slipping after the final deadline passed and all contractual terms are fixed.
The project has a usable control surface.
0
train_007
Inventory is falling, but backup suppliers, rationing, and demand controls are available.
The supply chain has a usable control surface.
1
train_008
Inventory is exhausted, no suppliers are available, and demand cannot be delayed.
The supply chain has a usable control surface.
0
train_009
A machine overheats and operators can reduce load, stop production, or repair cooling.
The machine has a usable control surface.
1
train_010
A machine overheats but controls are locked and operators cannot stop or reduce load.
The machine has a usable control surface.
0
train_011
A support backlog is rising and managers can triage, add staff, pause low-priority channels, or automate responses.
The support system has a usable control surface.
1
train_012
A support backlog is rising but staffing, triage, channel access, and workflow cannot be changed.
The support system has a usable control surface.
0
train_013
A model is hallucinating and the team can add source checks, reduce deployment scope, or route uncertain outputs for review.
The model has a usable control surface.
1
train_014
A model is hallucinating but it is already embedded in production with no monitoring, gating, or review option.
The model has a usable control surface.
0
train_015
A bridge shows stress signals and authorities can reduce load, close lanes, inspect, or repair.
The bridge has a usable control surface.
1
train_016
A bridge shows stress signals but traffic cannot be reduced and inspection access is blocked.
The bridge has a usable control surface.
0
train_017
Cash flow is tight but spending, invoicing, collections, and payment timing can still be adjusted.
The organization has a usable control surface.
1
train_018
Cash is exhausted, payments have failed, and all financing options are unavailable.
The organization has a usable control surface.
0
train_019
A damp issue is suspected and inspection, ventilation, external repair, and moisture monitoring are available.
The property has a usable control surface.
1
train_020
A damp issue is severe but access is denied, inspection is impossible, and repair authority is absent.
The property has a usable control surface.
0

What this dataset does

This dataset tests whether a model can detect whether a system has usable control options.

The task is simple:

Given a scenario and a control-surface claim, predict whether the claim is supported.

Core stability idea

A control surface is the set of levers that can still change system trajectory.

A system has a usable control surface when actors can adjust pressure, buffer, timing, routing, access, scope, escalation, or recovery action.

A system lacks a usable control surface when meaningful levers are absent, blocked, too late, or outside available authority.

Prediction target

Binary label:

  • 1 = the system has a usable control surface
  • 0 = the system does not have a usable control surface

Row structure

Each row contains:

  • scenario_id
  • scenario_text
  • claim
  • label

Files

  • data/train.csv
  • data/test.csv
  • scorer.py
  • README.md

Evaluation

Create a predictions CSV with:

scenario_id,prediction
test_001,1
test_002,0

Run:

python scorer.py --predictions predictions.csv --truth data/test.csv

The scorer reports:

accuracy
precision
recall
f1
confusion matrix
Structural Note

This dataset is intentionally small.

Its purpose is to test whether a model can distinguish real agency from false agency.

The hidden value is in detecting available levers, blocked levers, authority limits, timing constraints, and remaining trajectory control.

License

MIT
Downloads last month
13