Middle-schoolers can smell a guess-the-teacher rubric from the doorway. The fix is not a longer rubric. It is a gate: a checkpoint the student can see, argue with, and pass or fail on the evidence.
Here is a two-week science unit on a familiar prompt, "Why do plants in the north classroom window grow taller than the ones in the south corner?" Three gates. Each one names what counts as passing before the student starts.
Gate 1: A testable question
Students arrive with a fuzzy claim, "the plants like the window better." Gate 1 asks them to convert that into a question a measurement could answer this week. Passing looks like: one variable named (light, temperature, water frequency, soil), one comparison stated (north window vs. south corner over ten days), and one prediction with a direction (taller, more leaves, greener).
The checkpoint is a single index card. If the card contains a variable, a comparison, and a directed prediction, the student passes Gate 1. If it says "plants like light more," it does not pass, and the teacher shows the student exactly what is missing. No mystery.
This gate is not new pedagogy. It is the empirical-question stance you find in classroom research since the 1990s (Class E, see Chinn and Malhotra on epistemic authenticity). What is new is making the pass criterion visible to the student BEFORE the work starts.
Gate 2: A measurement plan the student could hand to a substitute
The second gate is procedural. The student writes a plan another person could execute without asking questions. Passing looks like: what to measure (stem height in centimeters), when (every school day at 9:15), how (ruler at the soil line), and what to record (a table with date, plant ID, height, notes).
The checkpoint is the plan itself, read aloud by a classmate who has never seen it. If the classmate can execute it without follow-up questions, it passes. If the classmate has to ask "which plant?" or "measure from where?", the plan goes back for one revision.
This is where the classroom-configuration matters (Class C): the gate works when the teacher has set up the room so a classmate CAN read the plan aloud in three minutes. That means partner desks, a quiet minute, and a clipboard. The gate design and the room design are the same design.
Gate 3: A conclusion the evidence actually supports
At the end of week two, students write a one-paragraph conclusion. The gate is not "did you conclude something." The gate is: does the conclusion match what the data shows, and does the student name what would have changed their mind?
Passing looks like: the direction of the result stated (north plants grew taller by an average of 2.1 cm), the confidence stated in plain words (the difference held across four of five plant pairs), and one falsifier named (if we swap the plants between windows next week and the north ones still grow faster, the window is not the cause).
That last piece, the falsifier, is the gate that changes the classroom. A student who names what would have changed their mind has done science. A student who writes "plants need light, this proves it" has not passed Gate 3, and the teacher can point to the exact missing sentence.
Why this is not just better rubrics
A rubric grades finished work. A gate is a checkpoint the student meets or does not meet, in the moment, on evidence both parties can see. The student never has to guess what the teacher wanted. The teacher never has to invent a score for a paragraph that half-worked.
Three gates over ten days. Each one has one observable checkpoint. Each one can be re-attempted the next day if the student misses it. The grade at the end is not a number pulled from a rubric column. It is the record of which gates the student passed and on which day.
Further reading
- Curriculum gates that teach thinking
- Evidence classes as classroom vocabulary
- A rubric that rewards falsifiability
- The workshop for teachers designing gate-based units
AI-authorship fence: this post was drafted with LLM assistance, reviewed and edited by a human author, and published under human editorial responsibility. No claim is made that an LLM authored the reasoning. The framing is a working hypothesis on the attainable path toward General Natural Intelligence, natural not artificial, tested in the open.