This is the join. The steps a child sees on the board at the start of the lesson are the exact things the teacher confirms at the end, and the exact things that appear in the tracker. Not a separate assessment. Not a second job. The same four sentences, all the way through.
The hinge told us something. 9 of your 30 children (30%) voted "hammer" on the Moon question. That is above the 25% line where SEQUENS raises it.
Suggested for step 3: A few. This is a suggestion from 30 votes, not a verdict. You can overrule it. You were in the room and SEQUENS was not.
9 children (30%) still think heavier falls faster. That misconception does not fade on its own - it hardens.
Next lesson's retrieval: re-run the two-object drop as a starter, and ask the Moon question again before you teach anything new.
The four steps are in the tracker against every child. The misconception is flagged. Next week's starter is aimed at it. You did not fill in a mark sheet, and nobody typed a child's name.
Why steps to success are written this way. They are thinking steps, not task steps. "Write the date. Underline the title. Answer three questions." is a to-do list, and a child can complete every item on it having learned nothing. These four say what the child can do with their head at the end that they could not do at the start.
Why this closes the loop. The steps on the board at the start are the steps the teacher confirms at the end, and the steps that appear in the tracker. One set of sentences, doing three jobs. No separate assessment task. No mark sheet. No re-typing. The lesson assesses itself because it always knew what it was for.
And SEQUENS suggests, the teacher confirms. The hinge already told it that nine children think heavier falls faster - so it offers a judgement on step 3, with its evidence, and the teacher overrules it or accepts it. A vote is a signal, not a verdict. The teacher was in the room.