1. Account and Login
Create an account with Google, then log in. If it is your first visit, complete onboarding (display name + school).
Signup
Use your school Google account
name@school.edu
Continue with Google
After signup: redirected to onboarding.
Login
Return users sign in directly
Continue with Google
Forgot access? Contact school admin
2. Create Group
Groups define who can manage tasks and judge submissions. Add teachers/admins to the same group.
Groups
Year 7 Science - AC Generator Unit12 members
Year 8 Humanities - Essay Moderation8 members
3. Create Task
Create a grading task, pick owner group, set confidence target, and open for grading.
New Task Modal
Task: AC Generator Explanation
Group: Year 7 Science
Open for grading: ON
Confidence target: 90%
Use ad-hoc names: Optional
Create task
4. Add Data from Google Sheets
In Manage submissions, paste rows copied from Sheets. Typical format:Student Name, Submission Text.
Paste Import
Fatima Al-Rashidi, The AC generator works when...
Amelia Thornton, An AC generator converts...
James Oduya, AC generator produce alternating...
Amelia Thornton, An AC generator converts...
James Oduya, AC generator produce alternating...
Tip: Keep one student per line.
Mapped Rows
Fatima Al-Rashidi matched
Amelia Thornton matched
James Oduya review needed
5. Judge with Anchored Runs
FairForm keeps one anchor fixed while you compare challengers, reducing fatigue. Progress segments show completed vs remaining comparisons for that anchor.
Judging Screen
Anchor
2 comparisons remaining
“An AC generator converts rotational energy into electrical energy by electromagnetic induction...”
Challenger
“When a coil rotates in a magnetic field, flux changes and current alternates...”
6. Compare Judge Results
Use results to compare global ranking and per-judge ranking side by side. Focus the same submission across all judges to quickly inspect disagreement.
Global Ranking
Rank 1 - Amelia Thornton 93% conf
Rank 2 - Fatima Al-Rashidi 78% conf
Rank 3 - James Oduya 61% conf
Per-Judge Columns
Judge AJudge BJudge C
Focused submission highlighted across all columns
Key Terms
Anchor: The fixed submission in a short run of comparisons.
Confidence: How certain the model is about an item’s position.
Confidence Band: Likely rank range (for example, rank 3-5).
Judge Agreement: How often a judge aligns with global model expectations.
Upset/Bias Rate: How often a judge picks against expected outcome.
Surprisal: Average “unexpectedness” of a judge’s picks.