Cold write exercise

# Cold write exercise

An assignment type called **"In-Class Essay"** — a timed, unassisted writing mode where the tutor is disabled and the screen is locked to the writing surface. Students write without AI help, mirroring standardized test conditions. Produces a baseline measurement of unassisted writing ability.

## The pedagogical idea

Throughout the year, students use the tutor and grading assistant on their regular essays ("warm writes"). Periodically, the teacher assigns an In-Class Essay — same writing task, no tutor, no AI help. The point is to measure whether the tutoring is actually making students better writers on their own, not just better writers with a crutch.

Teachers should see improvement over time in both warm writes (tutor-assisted) and cold writes (unassisted). If warm write scores go up but cold write scores stay flat, the tutor is doing the work, not the student. If both trend up, the student is internalizing the skills. This is the signal the cold vs hot write analytics dashboard would surface.

Scores will naturally be lower on cold writes — that's expected and fine. Teachers can make these worth fewer points. The lower scores aren't a punishment; they're honest signal. And knowing they'll face periodic unassisted writing incentivizes students to think about *how* the tutor makes them better — to actively try to internalize what it teaches, not just lean on it. That's transfer.

The teacher-facing description for this assignment type should explain this framing — why periodic cold writes matter and how to read the results alongside warm writes.

## What it looks like

A single open writing surface for the full duration — no section flow (intro → body → conclusion), no tutor, just the prompt and the editor. Think blank page + countdown.

- **No tutor.** Where the tutor panel normally lives, show a placeholder — the prompt, or a simple "This is an in-class essay. No tutor available." Not an empty void; something intentional that signals this is by design.
- **Single open editor.** No section-by-section flow. The student writes the full essay in one surface, like a real in-class essay or standardized test. They decide their own structure and pacing within the time.
- **Teacher-set timer.** Free-form input — the teacher types in the number of minutes at assignment creation. A countdown is visible to the student. **Hard auto-submit when time's up** — no grace period, no extension on the student's end.
- **Teacher timer controls.** The timer is continuous for students (no pause button), but teachers can pause, reset, or add time from their end. Covers bathroom breaks, fire drills, or "I gave them too little time" without giving students the ability to game it.
- **Full-screen mode.** The exercise launches in full-screen. Not truly enforceable in a browser (students can exit full-screen — this isn't Chromebook kiosk mode), but it creates the right environment and deters casual distraction.
- **Copy-paste: blocked and flagged.** Paste is actively blocked during the exercise (paste events are intercepted and prevented). If a student somehow gets around the block, it's flagged for the teacher the same way the existing paste alert works. Belt and suspenders.
- **Teacher monitoring encouraged.** The spec should be honest with teachers: this creates test-*like* conditions, not a proctored exam. Full-screen and paste blocking deter casual cheating, but a determined student on a non-managed device can work around browser-level restrictions. Active teacher monitoring during the exercise is part of getting good data. The teacher-facing description should say this plainly.

## Grading

GA grades In-Class Essays with the same rubric as regular thesis-driven essays. No special leniency. The scores will be lower — that's the point. The gap between warm write scores and cold write scores is the data. A different rubric would pollute the comparison.

Teachers can set a lower point value on In-Class Essays if they want to reduce the grade impact while preserving the signal.

## Resolved decisions

- [x] Locked screen: full-screen mode + paste blocking. Not truly enforceable in a browser; teachers should actively monitor. Honest about the limitation.
- [x] Timer expiration: hard auto-submit, no grace period.
- [x] GA grading: same rubric as regular essays. Lower scores are expected and informative, not punitive.
- [x] Timer: continuous for students. Teachers can pause/reset/add time from their end.
- [x] Timer input: free-form (teacher types minutes).
- [x] Copy-paste: both blocked and flagged.
- [x] Writing surface: single open editor, full duration. No section flow, no tutor.

## Open questions

- [ ] Should there be a "1 minute remaining" warning before auto-submit, or is the visible countdown sufficient?
- [ ] What does the teacher's monitoring view look like during an active In-Class Essay? Can they see who's writing, who's idle, word counts in real time?
- [ ] Can a teacher launch an In-Class Essay for the whole class at the same time (synchronized start), or does each student's timer start when they open it?
- [ ] Should the full-screen exit be logged/flagged for the teacher (i.e., "this student left full-screen 3 times")?
- [ ] Does the essay get GA feedback in addition to the grade, or just the grade? Feedback could be valuable for learning but might change the pedagogical dynamic if students expect coaching even on cold writes.