Session two

Session one is a start. Session two is where a family’s effort begins to add up, and it adds up because of a few lines written down at the end of the first. When a session ends, whoever is a step ahead on this subject, which today is you, writes a next-time note. The next package is built from that note instead of from scratch. And each session leaves one dated entry in a record your family keeps, readable by you and by a program. What compounds is not the cleverness of any one lesson. It is the record of what was tried.

What session two looks like

Open the hockey session-two sample. It says at the top that it is an illustrative example made with a simulated family, not a real family’s record. A fourteen-year-old can argue for ten minutes about which player belongs on a line, and freezes when asked to write one paragraph of the same argument. Session one got the spoken case onto paper in fragments. The sample then shows the note the learner page left at the end of that session, in the learner’s words: “the start is fine now, but the middle is a pile — and I don’t know how to say the other guy is wrong without just saying he’s wrong.”

The rest of the page is built from that one sentence. The guide opens with a callback, not a reset: the first claim sentence stays, because it works. The whole session is the middle. Pick the two strongest reasons, give each its own sentence, then add one sentence that names what the other side would say and answers it. The note also said the learner tires once it feels like an English class, so the session is short, stays on hockey, and stops while it is still going well. Further down, it lists what only the guide can do, such as telling when “this is dumb” means too long rather than too hard. It closes by saying that session two went to the middle only because that is where this learner’s note pointed. Another note would send the next session somewhere else entirely.

The next-session guide states the same habits plainly. Paste the note into the skill or the AI you already have, or point it at the file you saved, and ask for the next session. It keeps what worked and opens with a callback, moves one step on from the stuck spot (never a leap), never repeats a thing the note says already failed, changes the theme only if your child tired of it, and paces the length to how the last session ended.

Why a record, and why the second session

The record is deliberately plain. By its format, each entry is one dated session, and it is append-only: a correction, or the next session, is a new entry, never an edit of an old one. It is private by default, a file on the guide’s machine with nothing uploaded, and publishing is a separate choice made later. There is deliberately no name field, because a program cannot detect a real name; the learner appears only as a package reference and a nickname you chose. Slips are written as behaviour, never as a diagnosis, and a check blocks a named condition, an email or a link in free text. The learner record template is a folder you can copy, and its page states what a stranger would see if you ever chose to publish. The longer argument that a kept record is what compounds is in our piece on how knowledge compounds under a constraint.

There is also a reason to want session two to come back to session one. In a laboratory study, Roediger and Karpicke (2006) had undergraduates aged 18 to 24 read a short passage once, then either read it again or try to write down what they remembered. Five minutes later, re-reading was ahead, 81% of the passage to 75%. A week later the order flipped: the group that had tried to recall it remembered 56%, the group that had re-read it 42%. A broad review by Dunlosky and colleagues (2013) rated practice testing and distributed practice, which means spreading study out over time, as having “high utility assessments because they benefit learners of different ages and abilities.”

Those are adults in a laboratory, not your kitchen table. We have not measured whether our sessions do this for a family; nothing has been measured yet, and the first measurement is the first real families. What the studies give us is a reason to build session two as a return. The record is what makes a return possible.

This week

After the next session, write three lines: what landed, what slipped, what they loved. Say what happened, not what the learner is. Use a nickname, never a real name. Be specific about the third line, because it is what the next session will be built through; in the sample, hockey is the way in only because it is the thing the learner argues about without being asked.

If you use the player, the learner page leaves a note for you and offers to save the session to a learning record, a file that downloads to your device and goes nowhere else. Without an AI, the no-AI guide ends the same way: tell us how it went, and the next one comes back shaped by it.

the previous piece explains why the game has no AI in it, and the sample shows what that game does do, naming each slip as it happens so the note has something exact to say. The next piece is addressed to anyone who is the tutor in the group.

What did you write in your next-time note, and where did it send session two?

Sources

• Roediger, H. L., III, & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255. https://journals.sagepub.com/doi/10.1111/j.1467-9280.2006.01693.x (percentages from Experiment 1: five-minute test 81% re-read vs 75% recall; one-week test 56% recall vs 42% re-read; participants described as undergraduates aged 18 to 24.)
• Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving Students’ Learning With Effective Learning Techniques: Promising Directions From Cognitive and Educational Psychology. Psychological Science in the Public Interest, 14(1), 4–58. https://journals.sagepub.com/doi/abs/10.1177/1529100612453266