The Tutor's Desk
One tutor's week: rough notes, a parent-ready report, and the guide coach on a real problem.
Meta DAX › See it
See it
Every one of these started as one sentence from a guide and what the learner already loves. Made-up names, nothing identifying. A way, never the way.
Showcases
The same recipe built every one of these.
The Year on the Wall
A whole Grade 6 science year from one brief: 36 weeks mapped, misconceptions named, one week built in full.
First Week at the Bakery
A staff onboarding module made from a one-page brief.
The Proton Mill
A 3D Biology 101 lesson on how mitochondria make ATP, with instructor notes.
Each showcase was built by the MetaDAX course factory from a short brief. Model-generated; reviewed for safety, not certified.
Sample packages
9 packages built from parents’ descriptions: one page for the guide, one for the child.
9 of 9 shown
-
Carrying tens with marble runs — a Meta DAX homeschool sample
Stuck: freezes before trying; writes column totals side by side instead of carrying
Read the sample
A made-up example built from a parent's description. A way, never the way.
Guide: Nora · Learner: Eli · Age band: 7–9 · Subject: Math
Where the learner was stuck
Eli could add the digits in each column of a two-digit addition problem, but wrote the two column results side by side rather than grouping them. Asked to add 27 and 15, Eli produced 312: the tens total (3) and the ones total (12) placed next to each other, with no carrying. Eli also froze before trying when a column of numbers appeared, saying "I can't" before making any attempt.
The approach (built through what they love)
Marble runs already model overflow: when too many marbles pile up in one spot, extras roll to the next track. The session starts with Eli teaching Nora how to set up an overflow track, then Nora names the connection — the ones spot holds up to nine, and when you reach ten, one marble rolls up to the tens track. That is carrying. Every example in the session is framed this way: count the ones, check for overflow, roll the extra ten up. The game uses the same language, and Eli can hear the "roll" happening in each question.
For the guide: the shape of a session
A session has three parts. First, Eli teaches — Nora asks how the overflow track works and genuinely does not know how Eli has set it up. Second, Nora names the math connection in one sentence, using Eli's own words. Third, Eli tries the game at the learner's own pace, with Nora beside them. The guide's job during the game is to notice the face, not the score. If Eli goes quiet after a wrong answer, Nora does not wait or ask why — she gives Eli something to do in the first two seconds: "let's count the ones together" and starts counting herself. Keeping the page to one question at a time, or folding a printed page, prevents the overwhelm that a full column of numbers triggers.
What only a person can do
Nora can see the moment before the freeze — the held breath, the looking away — and interrupt it before it sets in. She can connect the marble run to one Eli actually built, not a generic one. She can tell when Eli is guessing versus thinking. She can read the room: if the session starts to feel like a test, she can stop the game, return to the marble run, and let Eli be the guide again. A page cannot do any of this. The decision to stop early, to shift tone, to follow an unexpected question — those belong entirely to Nora.
Where the learner leads
The session opens with Eli as the guide. Eli knows more about building a marble run that works than Nora does. Eli sets up the overflow track and explains why the marbles go where they go. Nora listens and asks one question: what happens if the extras have nowhere to go? That question is real — Nora wants to know. Eli's answer contains the whole idea of carrying; Nora's job is to name it afterward.
The game on the learner's side
The game presents two-digit addition problems one at a time, each with a visual showing the column structure. There are three levels, each with four questions. After three correct answers in a row, the numbers get a bit bigger. After a miss, the next question is a bit easier. Wrong answers name the specific mistake — side-by-side writing, forgetting the carry, or dropping the ones — so Nora's note after the session can say exactly what came up. Nothing bad happens on a wrong answer. The game says: wrong answers help us know what to work on next.
When to seek an evaluation
Most children find carrying awkward at first and get it with practice. But if a child consistently cannot hold two things in mind at the same time — a partial result and what to do with it next — even after many patient tries, or if numbers seem to slip and rearrange in ways the child cannot explain or notice, or if the frustration is sharp and persistent across many subjects and types of work, it may be worth talking to someone whose work is understanding how children learn. A professional who observes children and learning can watch, ask questions, and tell you what they see. You do not need a label to have that conversation, and the information you get is yours to use however you choose.
A way, never the way
The marble run is one way to think about carrying. Another child might reach for a different picture — a full row of LEGO bricks moving up to the next layer, a score counter that rolls over, anything. If the learner reaches for their own model, follow it. The goal is that the learner can carry tens reliably. How they get there is open.
-
Blending sounds with dinosaurs — a Meta DAX homeschool sample
Stuck: says each letter sound correctly but cannot push the sounds together into a word
See a learning record from this package → published by the learner or guide
Read the sample
A made-up example built from a parent's description. A way, never the way.
Guide: Nora · Learner: Dash · Age band: 7–9 · Subject: Reading
Where the learner was stuck
Dash could say every letter sound on its own without any trouble. The problem was that the sounds would not join up. Sounding out s-u-n out loud, Dash would then say a completely different word that started with s. Each sound stayed separate, like beads that would not string together into a word. Worksheets and flashcards had not shifted this. The frustration showed up quickly: after about two misses in a row, Dash would cover the page and stop.
The approach (built through what they love)
Dinosaur names are built from the same phonics chunks that appear in ordinary words, and Dash already knew those names. That meant every blending exercise could feel like a dinosaur game rather than a reading drill. Short names came first — three sounds in rex, four in tops — and the game built up to five-chunk names like Velociraptor. Because Dash wanted to get the dinosaur names right, there was a reason to push the sounds together. The game also stepped up in difficulty after three correct answers in a row and stepped down after a miss, so the difficulty stayed close to the edge of what Dash could do.
For the guide: the shape of a session
A session starts with Dash in the lead: Dash teaches Nora something about how to tell two dinosaurs apart. This is not a warm-up exercise — it is a real moment where Dash is the expert. Only after that does the blending work begin. The guide's steps are short: introduce one name, model the blending (say the sounds slowly, then faster, until the word pops out), then let Dash try. The game records which blends break down most often so the next session can target exactly those. Sessions are kept short, stopping before frustration rises too high. If Dash covers the page, the guide moves immediately to doing the sounds together rather than waiting or asking why.
What only a person can do
Nora can see Dash's face change a moment before the page gets covered, and can move in before the frustration lands. Nora can also connect a specific dinosaur name that Dash mentioned to the sounds in that session, making the example feel personal rather than generic. Deciding when to stop — even if the game is not finished — is something only Nora can judge. A short successful moment at the end of a session matters more than getting through all the questions.
Where the learner leads
At the start of every session, Dash teaches Nora one thing about dinosaurs: how to tell a Stegosaurus from an Ankylosaurus, or what made a Triceratops different from other ceratopsians. Nora listens and asks questions. This is not decoration — it signals that being the expert matters, and it lets the session start from a place of confidence. The same pattern appears on Dash's side of the page: the first thing Dash sees is an invitation to teach, not a question to answer.
The game on the learner's side
The game shows letter sounds one at a time, using chunks from real dinosaur names. Dash picks which name the sounds blend into. Wrong answers are tied to the specific mistake behind them: for example, picking a word that starts with the first sound (guessing from the beginning rather than blending all the sounds) gives a different response than saying the sounds out loud but not joining them into a word. The game has three levels. Level one uses short three- and four-sound names like rex and raptor. Level two uses longer chunks like stego and bronto. Level three uses full names across four or five chunks. Difficulty steps up after three correct answers in a row and down after one miss, so it stays close to the edge of what Dash can do.
When to seek an evaluation
If a learner can say each sound correctly but the sounds still do not join into words after several sessions — even with short bursts, patient modelling, and a topic they love — it is worth describing that to a professional and asking whether an evaluation would help. Other things worth noting: mixing up the order of sounds inside a word, frequently forgetting which sound goes with which letter even after practice, or getting much more tired from reading tasks than from other kinds of learning. When asking, describe what you see: say that the learner can produce each sound on its own but the sounds do not connect into words, and that the pattern has continued across multiple sessions. You do not need a label before asking for an evaluation. Getting one earlier rather than later means more options.
A way, never the way
This page shows one approach that worked as a starting point for one child at one moment. A different child who loves something else entirely would need a different anchor. A child who connects to the sounds but freezes under any game pressure would need a different shape. The blending steps here — model slowly, speed up, let the word emerge — are grounded in how sounds work, but the way they are delivered, the pacing, and when to stop are all things only the guide in the room can judge. Use what fits. Leave the rest.
-
One backyard science lesson for mixed ages — a Meta DAX homeschool sample
Stuck: keeping mixed ages engaged together without the oldest waiting and the youngest drifting
Read the sample
A made-up example built from a parent's description. A way, never the way.
Guide: Margot · Learners: Theo (8), Bea and Ines (both 5) · Age band: 4–9 · Subject: Science
Where the guide was stuck
Margot had three children to teach at the same time: Theo, who reads fluently and finishes activities quickly, and Bea and Ines, who cannot yet read and lose focus within a few minutes. Every curriculum she tried was aimed at one grade, which meant Theo was bored while the little ones were lost, or the little ones were left behind while she worked with Theo. The real wrong answer the little ones gave when asked what plants need — "candy" — told her they were not yet connecting the question to anything real. She wanted one lesson that held all three without splitting her attention three ways.
The approach (one lesson, three ages)
The lesson gives each child a different job based on what they can already do. Theo reads questions aloud and, later in the session, explains answers back to Bea and Ines in simple words. Bea and Ines always have something to touch, pour, or point at — their job is never to sit still and listen. The whole session runs in short bursts of two to three minutes of listening followed by a doing moment: pouring water on soil, lifting a rock, touching a leaf. The doing moments reset the little ones and give Theo a break from being the reader so he can observe and ask his own questions.
For the guide: the shape of a session
The session opens outside, not at a table. Theo knows where the ants nest in the yard; Margot asks him to take everyone there and show Bea and Ines what he knows. This puts Theo in the guide seat from the start and gives the little ones something real to look at before any words are spoken. After the ant visit, everyone collects one thing: a bug, a leaf, a handful of soil. They hold them while Theo reads the first question aloud. The water experiment — pouring water onto soil and watching the colour change — is the moment that anchors the session for the little ones, because they can splash and pour freely. The session closes with a game on the page: Theo reads each question, Bea and Ines point at the screen to pick their answer, and the three levels let Theo try harder questions while the little ones stay with the doing parts.
What only a person can do
Margot is the one who notices when Bea has gone somewhere else in her head and decides whether to bring her back or let a minute pass. She can tell the difference between Theo being bored and Theo being frustrated — those need different responses. She knows that Ines will stay with anything if she is allowed to get her hands wet, and that Bea sometimes needs a moment to herself before she can join in again. No page can carry that knowledge. The page gives a shape; Margot fills it with what she sees.
Where the learner leads
Theo knows the ant nest better than anyone. Before the lesson content begins, he leads: he takes everyone to the nest, points out the trails, and tells the group what he has noticed about what the ants carry. This is not a warm-up exercise — it is the real start of the lesson, and Theo is the guide for it. Later in the session, the hardest game questions ask Theo to explain an answer back to a younger child in words they can understand. Explaining something simply to someone who does not yet know it is harder than answering a quiz question, and it tells Margot more about what Theo actually understands.
The game on the learner's side
The game has three levels. The first level is pure doing: questions that require pouring, lifting, or touching something to answer. The second level is listen-and-pick: Theo reads the question aloud, Bea and Ines point at the answer on the screen. The third level is explain-it: the question asks the oldest to put an answer into simple words for a younger child. The game steps up after three correct answers in a row and steps back down after one miss, so it finds each child's level without Margot having to track it. Wrong answers are mapped to the specific mistake behind them, so the feedback is not just "wrong" but "here is what got mixed up."
When to seek an evaluation
If, across several sessions, one child consistently cannot stay with even a short doing task — not distraction, but a pattern that does not shift when the task changes — that is worth watching. Similarly, if a child seems to have significant difficulty connecting words to the things they stand for, even after many hands-on experiences, or if a child's understanding seems noticeably further behind what you would expect compared to other children the same age in ways that do not improve over time, those are patterns that go beyond ordinary variation. You do not need a label or a diagnosis to ask for an outside view. A conversation with a family doctor, a child development centre, or an early learning specialist at a local school board can give you more information. Asking early is not an alarm — it is information, and you remain in charge of what to do with it.
A way, never the way
This is one shape that worked for one parent's description of three children. Another family with the same ages and the same loves might need more time outside, or less reading, or a different anchor entirely. The oldest being the reader works here because Theo likes that job; if an older child does not, the whole shape shifts. Take what fits and leave the rest. The point is not to follow this plan but to find the version that fits your three.
-
Fractions at the kitchen table, two ages at once — a Meta DAX homeschool sample
Stuck: the younger thinks a bigger bottom number means a bigger piece; the older can name fractions but not add halves and quarters
Read the sample
An illustrative example made with a simulated family — not a real family's record. A way, never the way.
Guide: Priya · Learners: Sam (6) and Max (10) · Age band: 4–9 · Subject: Math
Where the learner was stuck
Two children learn at the same kitchen table, and fractions caught both at different points. The younger, Sam, thought a bigger bottom number meant a bigger piece — asked whether a quarter or a half was more cake, Sam chose the quarter because "four is bigger than two." The older, Max, could name any fraction on sight and reduce it, but when a recipe needed a half cup plus a quarter cup, Max wrote three-quarters only by guessing, and could not say why. One wanted to see the pieces; the other needed the pieces to add up to something real.
The approach (one activity, two levels)
The whole session is one batch of dough and one recipe, worked at two levels so neither child waits. Sam's level is cutting: fold and cut the dough into halves, then quarters, and line them up to feel that a half is the bigger piece even though "two" is the smaller number. Max's level is combining: the recipe calls for a half cup of one thing and a quarter cup of another, and Max's job is to say how much that is together and prove it by pouring. Same dough, same recipe, same table — the little one builds the pieces and the older one adds them.
For the guide: the shape of a session
Start with the dough, not the page. Sam cuts and arranges while Max reads the recipe aloud. Priya names one connection in plain words — "a half is one cut, a quarter is one more cut" — and then lets each child work their own level. The page comes out at the end: Sam answers which-piece-is-bigger questions by pointing, and Max answers add-these-two questions, each at their own pace with Priya between them. Keeping the two levels side by side, rather than running two separate lessons, is what lets one guide hold both children without the older one idling or the younger one lost.
What only a person can do
Priya can see that Sam has stopped looking at the dough and started guessing, and move the dough back under Sam's hands before the guess hardens. She can tell that Max's "I don't get it" means boredom, not confusion, and hand Max the harder job of checking Sam's cuts. She knows which child needs to be first and which can wait a minute. A page gives the two levels; only Priya can read which child needs which one right now, and swap a job the moment the table tips from baking into testing.
Where the learner leads
Both children lead something. Sam knows, better than anyone at the table, how to cut dough into even pieces so nobody complains their piece is smaller — that fairness is exactly what equal fractions mean, and Sam teaches it by doing. Max leads by teaching down: the hardest moment of the session asks Max to show Sam why a half and a half make a whole, using the pieces Sam just cut. Explaining it to a six-year-old is harder than adding the fractions alone, and it shows Priya what Max truly understands.
The game on the learner's side
The game runs one activity at two levels. The first level shows two pieces and asks which is more cake — built to catch the "bigger bottom number" slip, so a wrong answer names it: "four pieces means each one is smaller." The second level gives two fractions from the recipe and asks for the total, with a picture of the measuring cups filling. It steps up after three correct in a row and steps back down after a miss, so each child settles at their own level without the guide tracking it. Nothing bad happens on a wrong answer; it just tells the table what to cut next.
When to seek an evaluation
Most children sort out that a half is bigger than a quarter once they have cut enough real things, and most come to adding fractions slowly. But if a child, after many hands-on tries across a long stretch, still cannot hold the idea that the same whole can be split into different numbers of pieces, or seems unable to line up a number with an amount they can see and touch in a way that does not shift over time, it may be worth asking someone whose work is how children learn. A family doctor or an early-learning specialist can watch and tell you what they see. You do not need a label to have that conversation, and what you learn is yours to use as you choose.
A way, never the way
Baking is one way in because this table already bakes together. Another pair of children might reach fractions through folding paper, sharing a pizza, or splitting a score — and another mix of ages might need the levels further apart or closer together. The point is not the dough. It is that the younger learner can see that a half is the bigger piece and the older can add the pieces into a whole. If your children reach for a different thing to cut, follow it.
-
Turning a hockey argument into a paragraph (session two) — a Meta DAX homeschool sample
Stuck: can argue a point out loud for ten minutes but freezes when asked to put the same case in a written paragraph
Read the sample
An illustrative example made with a simulated family — not a real family's record. A way, never the way.
Guide: Dana · Learner: Theo · Age band: 13–15 · Subject: Writing
Where the learner was stuck
Theo, who is fourteen, can argue for ten minutes straight about which player belongs on a line — reasons, counter-examples, the lot — but asked to write a single argument paragraph, Theo stalls at the first sentence and writes nothing. Session one got the spoken case onto paper in fragments: a strong claim and one reason, but the reasons ran together and there was no line answering the other side. The next-time note the learner page left at the end of session one said, in Theo'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 approach (picked up from the next-time note)
This session keeps what landed and moves one step on the exact spot the note named. The claim sentence is left alone — it works now, so it is not re-taught. The whole session is the middle: taking the pile of reasons and giving it a shape. The move is to pick the two strongest reasons and give each its own sentence, then add one sentence that names what the other side would say and answers it — the thing the note asked for. Every example is a real hockey argument Theo already makes out loud, written down in that four-part shape: claim, reason one, reason two, the answer to the other side.
For the guide: the shape of a session
Open with a callback, not a reset: Dana reads back Theo's claim sentence from session one and says it is staying because it works. Then the whole session is the middle. Theo argues a hockey take out loud first, Dana catches the two strongest reasons and reflects them back, and Theo writes each as its own sentence. The hardest sentence — answering the other side — comes last, modelled once on a take Theo does not hold, then written on one he does. Pace it to how session one ended: the note said Theo tired quickly once it felt like an English class, so keep it short, keep it hockey, and stop while it is still going well.
What only a person can do
Dana can hear which two of Theo's five spoken reasons are actually the strongest and say so — a page cannot weigh an argument it has never heard. Dana can tell when Theo's "this is dumb" means the task is too long rather than too hard, and cut it there. She can catch the moment Theo argues brilliantly out loud and freeze it — "say that again, that exact sentence, now write that" — turning speech into the written line before it evaporates. And she can decide to answer the other side together, out loud, before Theo has to write it alone.
Where the learner leads
Theo knows hockey far better than Dana, and that is the engine of the session. Before any writing, Theo leads: he lays out the argument in full, and Dana genuinely learns something about the game she did not know. The hardest written sentence — the one answering the other side — works because Theo already does it in speech without noticing; Dana's job is to point out that Theo just answered an opponent out loud, and that the paragraph needs exactly that sentence. The learner supplies the argument; the guide names the shape.
The game on the learner's side
The game builds one paragraph in four moves, each a short choice rather than a blank page. First it offers three claim sentences and asks which states a position rather than a fact. Then it gives a jumble of reasons and asks Theo to pick the two strongest and drop the rest — the step the note asked for. Then it shows four sentences and asks which one answers the other side instead of just repeating the claim louder. It steps up as Theo gets a run right and eases back after a miss, and every wrong choice names the slip — "that is a fact, not a position," or "that repeats your claim, it does not answer theirs" — so Dana's next-time note can say exactly where to pick up.
When to seek an evaluation
Plenty of people argue far better than they write, and a gap between a confident speaker and a stalled writer is ordinary, especially at this age. But if, over many sessions and many subjects, getting any words onto the page stays far harder than it should given how fluently the learner speaks — if the hand freezes in a way that does not ease with practice, or written output stays sharply behind spoken ability across everything, not just writing — it may be worth asking someone whose work is how young people learn. A family doctor or a learning specialist can observe and tell you what they see. You do not need a label to ask, and what you learn is yours to use as you choose.
A way, never the way
Hockey is the way in here only because it is the thing Theo argues about without being asked. Another learner would turn a paragraph through a video game, a court case on a show, or a fight about a band. And 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 shape — claim, reasons, an answer to the other side — is worth keeping; the subject it is built through is yours to choose, and the note tells you where to go next.
-
Telling time with a real bus timetable (print-first) — a Meta DAX homeschool sample
Stuck: reads the hour hand but not the minutes; cannot yet work out how long until something happens
Read the sample
An illustrative example made with a simulated family — not a real family's record. A way, never the way.
Guide: Ruth · Learner: Jun · Age band: 7–9 · Subject: Math
Where the learner was stuck
Jun, who is seven, can read the hour hand and say roughly what o'clock it is, but the minutes are a blur, and "how long until the bus comes" is a question Jun cannot yet work out at all. Asked how long from 2:10 until 2:40, Jun said "thirty o'clock," counting the numbers on the clock face rather than the minutes between them. Jun loves buses and already knows the timetable at the stop near home by heart — which bus, roughly when — but cannot connect that known thing to the hands on a clock.
The approach (built through what they love)
The session is built on the one thing Jun already cares about: the real timetable at the local stop. Each question is a bus Jun knows — "it is 2:10, the bus leaves at 2:40, how long do you wait?" — so the minutes finally point at something real. The move is to count minutes in fives around the clock face, the way the timetable jumps in neat steps, rather than counting the printed numbers. Telling the time and working out how long until the bus become the same skill, anchored to a stop Jun can picture.
For the guide: the shape of a session (print-first)
This one is built for paper, not a screen. Print the one page and fold it so only a single question shows at a time — Jun gets overwhelmed when a whole sheet of clocks appears at once, so the fold is doing real work. Jun answers the showing question, says the time or the wait out loud, and only then unfolds to the next. The guide keeps the answer key on a separate page, not on Jun's sheet, so Jun never sees the answers sitting beside the questions. Ruth sits beside Jun with the key, confirms or gently redirects, and lets Jun set the pace. Short is better than complete: a folded page of six questions done calmly beats a full sheet rushed.
What only a person can do
Ruth can see the fold doing its job — or failing — and refold, cover, or set a question aside the moment Jun's face tightens. She can tell the difference between Jun guessing and Jun counting, and she holds the answer key so the page stays a set of questions rather than a test Jun can peek at. She can connect a question to a bus ride the two of them actually took last week, which no printed page can do. And she decides when six questions is plenty and it is time to walk to the real stop and watch a real bus come.
Where the learner leads
Jun knows the local timetable better than Ruth does — which bus comes when, which one is always a little late. Before the page, Jun leads: Ruth asks Jun to walk her through the stop's schedule, and genuinely learns something. The hardest questions on the page flip it — Jun has to tell Ruth how long until the next bus, teaching the wait out loud — because explaining the gap is harder than reading a single time, and it shows Ruth what Jun has really grasped.
The game on the learner's side
Because this is print-first, the "game" is the folded page itself: one question visible at a time, each a clock face and a bus from the timetable, getting one step harder as Jun gets a run right and easing back after a miss — the guide turns to an easier question on the key if Jun stalls. Early questions ask only the time shown; later ones ask the wait between two times, counting in fives. The guide's separate answer key names the likely slip beside each answer — counting the numbers instead of the minutes, or losing track at the top of the hour — so Ruth can name exactly what happened rather than only "not quite," and note where to start next time.
When to seek an evaluation
Many seven-year-olds take a while with minutes and longer with elapsed time, and muddling the two is ordinary. But if, after many patient tries with real and concrete things over a long stretch, a child still cannot line up a number with an amount of time they can feel, or counting on consistently slips and resets in a way the child cannot notice or correct, it may be worth asking someone whose work is how children learn. A family doctor or an early-learning specialist can watch and tell you what they see. You do not need a label to have that conversation, and what you learn is yours to use however you choose.
A way, never the way
A bus timetable is the way in here only because this child already loves the buses. Another child would find the minutes through a favourite show's start time, a sports clock, or an egg timer in the kitchen — and another child might do better on a screen than on paper, where this one needed the fold. Keep what works: one question at a time, the answers kept apart, the pace set by the learner. The goal is that the child can read the minutes and work out the wait. The thing you hang it on is yours to choose.
-
The water cycle in a jar — a Meta DAX homeschool sample
Stuck: can recite the words evaporation and condensation but cannot say what is actually happening to the water or why
Read the sample
An illustrative example made with a simulated family — not a real family's record. A way, never the way.
Guide: Carmen · Learner: Felix · Age band: 10–12 · Subject: Science
Where the learner was stuck
Felix, who is ten and working at about a grade five level, could reel off the words — evaporation, condensation, precipitation — and label a diagram with arrows. But asked what the water was actually doing, or why a cloud forms at all, he went quiet. The words sat on top of nothing. He could match the term to the picture in a workbook and still believe, when pressed, that the water in a puddle simply "disappears." The vocabulary had been learned as a list to return on a test, not as a description of something real happening to real water.
The approach (built through what they love)
Felix loves watching things change state — ice going to water, water going to steam on the stove. So instead of starting from the words, the session starts from a change he can watch for himself. A sealed jar with a little warm water and a lid goes on a sunny windowsill. Over an hour, water climbs the inside of the glass as vapour, beads on the lid, and runs back down. That is the whole cycle, small enough to hold. Only after Felix has watched it and described it in his own words does the session attach a term to each thing he saw: the water that left the pool is evaporation, the drops on the lid are condensation, the run-back-down is precipitation. The word lands on an event he has already witnessed, not the other way round.
For the guide: the shape of a session
Set the jar up together first, warm water and a tight lid, and put it where the sun hits. Then leave it and do something else — the waiting is part of it, and checking back every fifteen minutes keeps the attention fresh rather than forced. Carmen's job while they wait is to ask, not tell: "Where did the water on the lid come from? It was at the bottom a minute ago." Let Felix reason it out loud, even when the first answer is wrong, before naming anything. Keep a single printed page with the jar drawn on it and blank lines where the words will go, and fill them in only once Felix has described each step without the term. Stop while he is still curious — one clean run of the cycle understood beats three terms memorised.
What only a person can do
Carmen can hear the difference between Felix reciting "condensation" and Felix actually explaining that the vapour cooled against the glass and turned back to drops — the same word, two completely different states of understanding, and only a listener in the room can tell which one is happening. She can follow the unexpected question ("what if the jar were cold instead of warm?") into a second, better experiment, where a page would simply mark the first answer and move on. She can decide that today the jar is enough and the rain part waits for next time. And she can catch the moment the idea clicks — the look up from the glass — and let him teach it back to her while it is fresh.
Where the learner leads
Felix leads the watching. He decides how often to check the jar, he is the one who first notices the beads on the lid, and he narrates what is changing before any term is offered. When he gets to explain, unprompted, that the water "went up invisible and came back as drops," he is doing the real science — observing, describing, reasoning from what he saw — and the words are just labels for work he has already done. Carmen's part is to ask the question that makes him look again, and to name the thing once he has seen it. The learner supplies the understanding; the guide supplies the vocabulary that pins it down.
The game on the learner's side
The game is a build-the-cycle screen, one step at a time rather than a whole diagram at once. It shows the sunny pool and asks: the water is leaving — is it disappearing, moving as invisible vapour, or sinking into the ground? A wrong choice names the specific slip ("it feels like disappearing, but the water is still there — it is just spread out as vapour you cannot see") rather than just marking it wrong. The next screen shows vapour meeting the cool lid and asks what happens; the next, the drops growing heavy and falling. Three levels: level one labels the steps Felix just watched in the jar, level two asks him to predict what changes if the air is colder or the water warmer, level three runs the cycle outward from a jar to a cloud and back. It steps up after three right in a row and eases back after a miss, and every mistake it logs is a reasoning slip the guide can pick up next time — not a score.
When to seek an evaluation
Plenty of children can recite science vocabulary long before the ideas underneath settle, and the gap between the words and the understanding is ordinary at this age — it is exactly what hands-on work is for. But if, across many topics and many months, a learner keeps returning memorised words with nothing underneath them, cannot hold a two-step cause and effect ("this happened, which made that happen") even after watching it directly, or tires sharply from any task that asks him to explain rather than recall, it may be worth describing what you see to someone whose work is how young people learn. A family doctor or a learning specialist can observe and tell you what they notice. You do not need a label to ask, and what you learn is yours to use as you choose.
A way, never the way
A jar on a windowsill is one way into the water cycle, chosen because this learner loves watching things change state. Another child would get there through a puddle measured on the driveway over three dry days, a kettle and a cold spoon, or a drawing that animates. The thing worth keeping is the order — watch the real change first, reason about it out loud, attach the word last — not the jar itself. Use what fits your learner and your week; leave the rest.
The learning record for this session
A Meta DAX session can close by writing one entry to a learning record — a single dated JSON object, readable by a person and a program, append-only, kept privately on the guide's machine and published only if the family ever chooses to. Here is the entry this simulated session would write, in the
metadax-recordv1 shape (nickname only, no real name, no location, behaviour never a diagnosis):{ "format": "metadax-record", "v": 1, "entry_id": "hs-sc52-s1", "package": { "ref": "HS-SC52", "session": 1, "title": "The Water Cycle in a Jar", "subject": "Science", "age_band": "10-12" }, "date": "2026-10-09", "attempted": [ { "game": "Build the Cycle", "answered": 9, "right": 6 } ], "reached": [ { "game": "Build the Cycle", "level": 2, "of": 3 } ], "mistakes": [ { "name": "said the water disappeared rather than turned to vapour", "count": 2 }, { "name": "named the term correctly but could not say what was happening", "count": 1 } ], "learner_words": "Felix watched the lid for ages and then said: the water went up invisible and came back as drops, so it never actually left. He wanted to know what would happen if the jar were cold instead of warm.", "guide_note": "The hardest moment was before the jar fogged up, when Felix still thought the puddle water just vanishes. Waiting and asking where the lid drops came from worked better than telling him. The click came when he said it never actually left. Next time I would like to try his own question — a cold jar — and get to the rain part. What only I could do: I could hear the difference between him reciting the word and him actually explaining it, and we stopped while he was still curious.", "links": { "sample": "https://daxfoundation.org/metadax/samples/#science-water-cycle-jar", "package": "", "play": "https://daxfoundation.org/metadax/play/" }, "generated_by": "metadax-homeschool skill" }The entry stays on the guide's machine by default. Keeping a record over time is what the learning-record template is for; producing one entry, as above, needs nothing but the session. A way, never the way.
-
Interviewing a grandparent about a small town — a Meta DAX homeschool sample
Stuck: treats history as dates and names to memorise and cannot see it as something that happened to actual people
Play with companion notes → guide-facing sidecar notes, shown only to the guide
Read the sample
An illustrative example made with a simulated family — not a real family’s record. A way, never the way.
Guide: Rosa · Learner: Luca · Age band: 10–12 · Subject: History
Where the learner was stuck
Luca, who is twelve, could pass a quiz on dates and names but experienced history as a wall of facts with no one behind them. Asked what any of it had to do with a real person, he shrugged — it was just stuff that was already over, printed in a book, memorised for the test and then gone. The idea that history is made of ordinary people's days, that someone alive today remembers a version of the past firsthand, had never quite landed. He was not uninterested; he simply had never met history as something a person could tell you about over a cup of tea.
The approach (built through what they love)
Luca does love the family's own stories, and will happily sit and listen to an older relative remember things. So this session makes that the history lesson. The plan is a single real interview with a grandparent, kept fictional and generic here, about one small town they knew. Three questions carry it: what did the town look like back then, one job they held, and one change they watched happen over the years. Luca prepares the questions, asks them, listens, and takes rough notes. Afterward, the raw answers become the first page of a local history written in a real person's voice — not a list of dates, but what one person saw. History stops being a book and becomes a conversation he was part of.
For the guide: the shape of a session
Help Luca write his three questions before the interview, and keep them open — "what did the main street look like?" draws a story where "how many shops were there?" draws a number. Agree a signal for "ask the next one" so Rosa can sit in without taking over. During the interview, Luca leads and the grandparent talks; Rosa's whole job is to stay out of it unless the thread drops. Afterward, sit with the notes and help Luca turn three spoken answers into three short written passages, keeping the grandparent's own phrasing wherever it is vivid. Keep it to the three questions — a first interview that goes deep on three beats a scattershot twenty. Stop while there is still a story Luca wants to write down.
What only a person can do
Rosa can read the room during the interview — notice when a question has opened something the grandparent wants to say more about, and quietly let the silence run instead of rushing Luca to the next question. She can tell when Luca is genuinely curious versus just working through his list, and nudge him to follow the live thread. She can judge which memories are the grandparent's to share and which to leave, a decision no page could make for a family. And afterward she can hear which plain spoken line — "the whole street smelled of bread on a Tuesday" — is the one worth keeping exactly as it was said, and help Luca build the passage around it rather than flattening it into a fact.
Where the learner leads
This whole session runs on Luca being the one who asks. He chooses the three questions, he conducts the interview, he decides which answers struck him enough to write down first. The grandparent is the expert on the town, but Luca is the one drawing the history out and shaping it — the historian's actual work. When he reads his finished page back and realises he has written something that did not exist in any book, because he made it from a real person's memory, the point of the subject arrives on its own. The guide sets it up and steps back; the learner does the history.
The game on the learner's side
Between the interview and the writing, the game helps Luca turn spoken memories into a history. It shows one of the grandparent's answers and asks what kind of source it is — a firsthand memory, a guess, or something they heard from someone else — because a historian has to know. A wrong choice explains the difference rather than just marking it ("that is something they were told, not something they saw — still useful, but you would write it differently"). Then it offers three ways to open the written passage and asks which keeps the person's own voice rather than turning it into a textbook sentence. Three levels: level one sorts memory from fact, level two keeps the speaker's voice while tidying a sentence, level three asks Luca to place one remembered change on a simple before-and-after timeline. It steps up after three right in a row and eases after a miss, and every slip it notes is about handling a source, so Rosa knows what to pick up next time.
When to seek an evaluation
Finding history dry and disconnected is ordinary, and for most children it dissolves the moment the past becomes a real person talking — which is the whole point of this session. But if, across many subjects and many months, a learner cannot hold a sequence of events in order, loses the thread of a story told aloud and cannot say what came before what, or tires sharply whenever a task asks him to connect one thing to another rather than recall a single fact, it may be worth describing what you see to someone whose work is how young people learn. A family doctor or a learning specialist can observe and tell you what they notice. You do not need a label to ask, and what you learn is yours to use as you choose.
A way, never the way
Interviewing a grandparent is one way into history, chosen because this learner loves the family's own stories. Another child would get there through old photographs of one street, objects from a kitchen drawer, or a song that was popular once. The thing worth keeping is the move — history as a real person's firsthand account, drawn out by the learner and handled like a source — not the particular grandparent or the particular town. Keep the interview fictional, generic, and the family's to share or not. Use what fits your learner; leave the rest.
The companion sidecar for this package
This package carries a
companion.jsonsidecar: the machine-readable file a cognitive companion reads alongside the package, in themetadax-companionshape. It is published in the repo atcompanion/examples/history-grandparent-interview.companion.json. The sidecar states the single next step for guide and learner, the reach band for this learner, the behavioural signals a companion may watch (each naming the inference it must never become), the moves the companion may offer and their ceilings — and, required and non-empty, the work reserved for the human guide, each with a reason it stays with a person. It names capabilities, never products, and everything it describes degrades to a once-a-day bundle. It is a sidecar, read with the package, never a profile of a child. A way, never the way. -
Le Stade des Fractions
Stuck: no prior fractions; curious about the why but cannot yet say what a fraction means or how a shape can be divided into equal parts
Read the sample
An illustrative example made from a parent's request — nothing identifying. A way, never the way.
Guide: a parent at home · Learner: un apprenant de 11 ans · Age band: 10–12 · Subject: Math · Language: French
The package
A fractions course in French, in seven steps, set inside a single 3D stadium built with three.js. The learner arrives with no prior fractions and leaves having covered: what a fraction is (the pitch cut into equal parts), comparing fractions (two runners on the 400 m track), adding and subtracting (laps repainted with new equal parts), a fraction of a number (balls shared into hoops), finding the whole from a fraction — and a ten-penalty final to close. Done with a parent at home, in one sitting or across a few.
Each step
Every step has four parts: Regarde (a step-by-step 3D walkthrough), Pourquoi ? (the reason), À toi ! (five questions on the stadium, two tries, then the answer and why), and Guide (what to say, what to ask, what to watch for, an off-screen activity, the answer key). A Bilan at the end writes a next-time note — what happened in this session, never a label — to paste into the next one.
Provenance
Built from a real parent's request; everything identifying is left out. Illustrative example.
No samples match that filter.
Three steps in
Six family pages, three iterations in: the plan for step four, with steps one to three visible as history. One side for the guide, one made for the learner. These six now live here; they used to be the showcase.
- The Six-Out Inning Megan, a mom who homeschools, and Liam, 11
Adding fractions with different bottom numbers, through baseball.
- Make Your Case Priya, an English teacher, and Mateus, 15, eight months in Canada
A persuasive paragraph: claim, evidence, reasoning.
- Within Reach Dinah, who teaches teachers, and Jordan, a first-year Grade 3 teacher
Checking for understanding in the moment; help that fades on purpose.
- Everyone in the Room Renée, a people-and-culture lead, and Marcus, a new engineering manager
Meetings where everyone is heard, across three time zones.
- First On-Call Dave, a site reliability engineer, and Aisha, on her first on-call rotation
Handling a live incident; the blameless review.
- Real or Scam? Theo, 19, and Lucille, 78, his grandmother
Spotting phone, text and email scams.
Share one
These come from parents who ran the homeschool skill on their own and offered a generic version through a “Share a sample” issue. A person reviews each one before it appears here. No upload, no account.