Steering Portfolio

Track your intellectual journey. Build positions, test predictions, trace consequences, and create something new.

Demo Students

How the Steering Portfolio Works

A semester-long system that measures what tests can't - whether your thinking is actually developing, not just producing.

The Student Journey

1

Pick a Topic

Choose something you care about. You'll live with this question for a whole semester.

2

Build Your Position

Write journal entries, make predictions, and engage with evidence. Use AI freely - the tool isn't the test.

3

Let It Affect Reality

Share your work. Get challenged. Let your thinking change someone else's - and let theirs change yours.

4

Create Something New

Identify a gap nobody else has named. Build something - an argument, a tool, an analysis - that fills it.

The Four Components

Position Journal

Measures: Coherence Over Time

Every two weeks, write where you stand on your topic and what changed since last time. Each entry must reference the previous one - you can't dodge what you said before.

What the teacher sees: Did your position evolve with real evidence, or did you write the same thing eight times? Did you change your mind and explain why, or did you lock in after week two?

Entry chain Prior references Reversals Evolution dots

Prediction Log

Measures: Judgment Under Uncertainty

Make specific, falsifiable claims with a check date. "Social media use will increase" is vague. "Our school survey will show 60%+ students spend 3hrs/day on social media by March" is testable. When the check date arrives, record what actually happened.

Calibration: Are you right as often as you think you are? If you mark "high confidence" on everything and you're wrong half the time, you're overconfident. Good calibration means your confidence matches your accuracy - that's judgment.

Falsifiable claims Confidence levels Calibration score Reflections

Consequence Record

Measures: Skin in the Game

Document when your thinking affected something real. Another student used your research. A teacher changed their approach based on your analysis. You published something and got challenged. If your thinking never leaves the page, it's an exercise, not a portfolio.

Verification: Consequences must be confirmed by someone else - the person who used your work, the teacher who saw it, the public record. You can't self-certify impact.

Real-world impact External verification Feedback loops

Origination

Measures: Declaration

Once per semester: identify a gap - something that should exist but doesn't, or a question nobody is asking. Build an artifact - an argument, a checklist, an analysis, a proposal. Then defend it to your peers in a 5-minute presentation.

What matters: Not whether you're right. Whether you identified something real that nobody else named, and whether you can defend why it matters when challenged. AI can help build the artifact. It can't identify the gap.

Gap identification Artifact Live defense Novelty

How Teachers Grade

Each component is scored 1-3 at the end of the semester. The teacher reads the trajectory, not any single entry. A student who writes beautifully but never changes position scores lower than one who writes roughly but genuinely evolves.

1
Static

Position doesn't evolve. Entries don't reference each other. No engagement with challenges.

2
Responsive

Engages with evidence. Position shifts when warranted. References prior entries. Beginning to show intellectual movement.

3
Adversarial

Actively challenges own position. Seeks disconfirming evidence. Changes mind for good reasons and explains why. References personal experiences, not just articles.

Pattern Detection

The system automatically detects thinking patterns that help teachers focus their attention:

Adversarial Genuine intellectual development
Responsive Engaging but not self-challenging
Performed Looks like development but no personal grounding
Static Position unchanged, no evolution

Key Terms

Steering

The ability to push AI past its default answer. Not prompt engineering - it's taste, domain depth, and the courage to ask "where am I wrong?"

Calibration

How well your confidence matches your accuracy. If you say "high confidence" and you're right 80% of the time, you're well-calibrated. If you're right 40%, you're overconfident.

Trajectory

The path your thinking takes over a semester. Not where you end up - how you got there. The reversals, the dead ends, the moments your mind changed.

Reversal

When you genuinely change your position based on evidence. Not weakness - intellectual courage. The system rewards reversals that are grounded in real experience.

Skin in the Game

When your thinking has real consequences. Someone used your work. You published something and got challenged. Your predictions were tested against reality.

Origination

Seeing a gap that nobody else has named and declaring "this should exist." The one skill AI can't replicate - because AI doesn't know what's missing from your specific context.