eduKate Sengkang Education Runtime | How the Learning System Works

eduKate Sengkang is the compiled education runtime: a connected way to locate a learner, understand what is happening, repair the first consequential break and return increasing control to the student.

This page is the canonical parent for the learning system used across Which Student Are You?, Voyage, Darwin, English, Mathematics, Primary Science, Examination Craft and the Educational Field.

The current service state

eduKateSengkang.com is the Sengkang-facing education world. The current tuition service operates in small groups of up to three students at 83 Punggol Central. The academic routes are Primary and Secondary English, Primary and Secondary Mathematics including Secondary 3–4 Additional Mathematics, and Primary 3–6 Science.

The research laboratory upstream is eduKateSG. eduKate Sengkang does not need the reader to traverse that laboratory to understand the live education system here: the validated parts are compiled into this local graph.

The learner representation tube

Subject world → curriculum → representation → learner receives → learner reconstructs → learner answers → tutor observes → diagnosis → repair → retest → transfer → learner state changes.

A wrong answer is therefore not yet a diagnosis. The visible error may come from missing knowledge, vocabulary, representation, retrieval, method selection, working-memory load, expression, transfer or examination deployment. The job is to find where information first stopped travelling correctly.

Three student models

• Internal model: what the learner actually understands.
• Expressed model: what the learner can explain, write or represent.
• Performance model: what the learner can execute under the conditions that matter.

A learner may understand but fail to express; explain correctly but fail under time pressure; or reproduce a procedure without reconstructing the underlying idea. Those are different failures and require different repairs.

Find the break at the right depth

• First observed break: where failure first becomes visible.
• First consequential break: the earliest weakness materially reducing current capability.
• Root causal break: the underlying capability failure generating later errors.
• Lock-in gate: the point after which new learning increasingly depends on the weak capability and repair becomes more expensive.

We do not automatically repair the earliest historical weakness. We compile the smallest justified repair that releases the greatest useful capability for the learner’s present state and available time.

Correction has to land

E0 error observed → E1 feedback heard → E2 feedback understood → E3 internal model changes → E4 method changes → E5 next attempt changes → E6 unfamiliar transfer succeeds → E7 future error rate falls.

“I explained it” is not the same as “the learner learned it.” A correction becomes useful only when it survives retrieval, appears in independent work and transfers into a changed context.

Five learner clocks

• Developmental time: how long this learner needs to build the capability.
• Curriculum time: where the syllabus is moving.
• School time: what school is teaching and testing now.
• Examination time: how long remains before independent performance is required.
• Repair time: how long the missing capability is likely to take to stabilise.

These clocks do not always agree. That is why the correct route may be full repair, parallel repair, temporary containment or examination-priority work rather than simply “do more questions.”

Two graphs operate at the same time

The persistent education graph is the long-lived map: levels, subjects, Voyage, Darwin, learning, examination and the routes between them. The temporary learner compilation is built for one learner at one time: current state → target → missing capability → selected repair → expected evidence → retest → recompile.

Humans and eduKateAI use the same runtime

A tutor can perform this reasoning directly. eduKateAI performs the same bounded sequence computationally: identify the receiver and task, keep several plausible lost states alive, eliminate what the evidence does not support, ask the smallest discriminating question when important state is missing, and only then compile HELP. The website supplies the stable education world; the AI supplies live routing through it. The AI should therefore become more useful as the public graph becomes cleaner, not by inventing a second education system beside it.

The connected runtime

• Which Student Are You? locates the current learning state.
• Voyage maps development and movement across levels and subjects.
• Darwin asks why a formerly successful learning method has reached its limit and what must adapt.
• English, Mathematics and Science are the subject-world implementations.
• Examination Craft takes over when installed capability must survive time, marks, uncertainty and independent execution.
• The Educational Field shows the wider network, clocks, constraints and feedback around the learner.
• Start Here remains the human-facing entrance to the live system.

Reason for existence

Turn complex education research into a coherent, correctable, navigable and runnable education world that helps a parent, student, tutor or AI locate the learner, find the first consequential break, compile the next useful repair, observe whether correction landed and return increasing control to the learner.

The acceptance test is simple: a fresh reader or AI should be able to reconstruct how eduKate Sengkang works from this public graph without having to reinvent the architecture.