---
language: en
license: mit
task_categories:
- text-classification
tags:
- control-geometry
- control-surface
- reasoning
- clarus
- sios
size_categories:
- n<1K
pretty_name: Control Surface Classification v0.1
---

# 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:

```csv
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