Test, build, review: how we teach coding and chess
Every session runs in three stages. Test — we find out what your child already believes before we teach anything. Build — they make something real from a blank screen, never from a tutorial. Review — they replay the mistake until they can explain it themselves.
Most children's coding classes look like progress
Hand a child a tutorial and they will produce a working game this afternoon. The screen fills, the parent is pleased, everyone agrees it is working. Then the tutorial is removed, and the ability goes with it.
A child who copied a game has learned to follow instructions. A child who built one has learned to recover when it breaks.
First we loosen the ground
You cannot plant into hard soil. A child who is afraid of being wrong — or who has already decided they are "not a computer person" — will not take in a concept however clearly it is explained.
So the first job is not teaching. It is loosening: small wins, permission to break things, questions with no marks attached. Then we plant, and only then do we move in order, one idea resting on the one before it.
Once the ground is loose: three stages
Test — before we teach anything
We ask your child to predict what a piece of code will do, explain a game they have just played, and describe the move they would make next. Then we watch where they guess instead of reason. About ten minutes, and the most useful ten minutes of the term.
A wrong belief is harder to fix than a missing one. A child who thinks a variable is a kind of button will build on that for weeks. Better to find it in the first session, while it is one sentence to correct.
Build — from a blank screen
Your child makes something small enough to finish in a session and deep enough to break: a character that moves, a score that changes, a rule that stops the game. No tutorial, no copy-and-paste, no template with the answer in it.
Seymour Papert's constructionism holds that children build knowledge by making something shareable and then thinking about what they made — not by being told. MIT's Scratch team turned that into a spiral of imagine, create, play, share, reflect. The finished game is not the reward for the learning; it is the learning.
See Resnick, Sowing the Seeds for a More Creative Society and Scratch's creative learning approach.
Review — the mistake, out loud
When the game breaks, we stop. Your child replays what they did, finds the moment it turned, and explains why they thought that would work. The teacher's question is not "what is the answer" but "what did you expect to happen?"
In chess it is the same move every time: we replay the loss and find where the game turned. A child who can say I thought my knight was defending that square has learned something no explanation could have installed.
Little minds, handled carefully
A seven-year-old is not a small adult. Sessions are short, mistakes are expected, and nothing is ranked in front of other children — a child who feels embarrassed stops thinking and starts guessing.
Staying with one problem long enough to see its shape. A chess puzzle can take ten minutes; a bug in a game can take longer.
Asking "is that actually true?" before acting — of their own code, their own move, and eventually their own homework.
If this, then that. Order, conditions and consequence, practised until the reasoning is something they can show you.
What the research actually says
Chess is the most studied "thinking game" for children, and the result is narrower than the marketing. The apparent benefits shrank as the studies got stricter, and the randomised trial found no effect on school marks at all. Chess makes children better at chess.
Coding builds the habit of testing your own work, and the evidence that it lifts general reasoning by itself is thinner than most providers imply. So we do not rely on a transfer that may not happen: we teach the reasoning explicitly, and let the game be the practice ground.
Who is actually in the room with your child
Sammy Subu
Founder and programmes director
I teach every session myself right now — no rota, no assistant. I designed this curriculum and I teach every class.
A bachelor's degree is the floor, with a strong preference for BSc. A teacher from an education background must also hold a specialisation in science or mathematics — being good with children is not enough for this curriculum.
What this looks like over a term
The first two sessions look slow. By week four your child has a game they built from nothing; by week eight they can tell you why it broke. The full fee structure is on the fees page.
What we don't publish. The sequence itself — the order ideas arrive in, and which ones we hold back until later — is the part that took longest to get right. It changes what a child can do, and you will see it in the work rather than on a page.
What parents ask about the teaching
Is this just tutoring?
No. A tutor follows the school syllabus and helps with homework; we do not teach the syllabus. Sessions run in three stages — what the child already believes, something built from a blank screen, and the mistake reviewed out loud.
Why does the method need one-to-one sessions?
Because the first stage only works if the teacher can hear how one child reasons. In a room of twenty, a child who has guessed correctly sounds the same as a child who understands. One-to-one, the guessing is audible. Groups return as we add teachers.
Does my child need to be good at maths to start?
No. Children from age six start with blocks and the rules of chess. The arithmetic in a game is coordinates, counting and conditions, and it arrives because the game needs it, not as a separate lesson.
What if my child already codes?
The first session finds out what they actually understand rather than what they have copied. Many children who have followed tutorials can build a game but cannot explain why it breaks. That gap is where we start.
See the method in a session
Tell me your child's age and what they already do, and I will say honestly where they would start — including if they are not ready yet.