LIDS is a learning framework that treats the meta-skill of learning as the subject — teaching a child how to form hypotheses, articulate understanding and direct their own curiosity, rather than what to remember. It runs through seven modes, each wrapping curriculum content in a different pedagogical shape, and it replaces grading with articulated mastery: a child has understood something when they can explain it in their own words, confirmed at least twice.
What problem does LIDS solve?
Traditional education assesses recall under pressure and compresses the result into a letter, which captures almost nothing about how a child actually thinks.
The cost lands unevenly. Children who process differently — dyslexic, ADHD, autistic — get told they could do better if they applied themselves, when the format was the obstacle. And curiosity, the strongest driver of genuine learning, gets punished whenever it fails to match the lesson plan.
No mainstream curriculum teaches a child how to form and test a hypothesis, how to iterate when they were wrong, how to ask a better question, or how to be wrong without shame. Those are the skills that decide whether someone thrives once there is no syllabus left.
The kid who got told "could do better if he applied himself" designed a system that never says that to a child.
What are the seven learning modes?
Seven modes, each with named phases, each developing a different meta-skill alongside the subject matter.
| Mode | Phases | What it develops |
|---|---|---|
| Standard Lesson | Explore → Explain → Apply → Reflect | Structured comprehension, metacognitive reflection |
| Guided Discovery | Wonder → Investigate → Discover → Connect | Hypothesis testing, Socratic reasoning, synthesis |
| Teach Me | Brief → Teach → Challenge → Review | Articulation and knowledge-gap identification |
| Maker Mode | Imagine → Plan → Create → Present | Creative problem-solving, design thinking, iteration |
| Discovery Project | Choose → Investigate → Return → Analyse → Iterate → Complete | Multi-session research, real-world investigation, persistence |
| Debate | Topic → Position → Argue → Rebut → Judge | Critical thinking, argumentation, perspective-taking |
| Journal | Settle → Write → Deepen → Close | Self-reflection, emotional regulation, metacognition |
The modes are not difficulty levels and they are not a sequence. A child moves between them depending on what the material and the moment call for.
What is the Curiosity Rule?
The Curiosity Rule is a single instruction embedded in all seven modes: if a child mentions an interest or goes off-topic mid-lesson, never redirect — weave their interest into the curriculum in real time.
The curriculum wraps around the child, not the other way around. A tangent is not a discipline problem or a loss of focus. A tangent is the highest-quality signal available about what this particular child is actually drawn to, arriving unprompted and for free, and redirecting it destroys the one input no assessment can manufacture.
That single rule does more work than any other part of the framework, because it is the one most systems get backwards.
How is understanding proved without grades?
Understanding is proved by articulation, confirmed at least twice, through Socratic questioning — not by a score.
A child demonstrates mastery by explaining a concept in their own words and having that explanation hold up under questioning, on more than one occasion. One good explanation can be a lucky reconstruction of something half-remembered. Two, separated in time and probed differently, is understanding.
The mechanism that makes this work is inverted teaching. In Teach Me, the child becomes the teacher and the system plays a confused student — deliberately misunderstanding at points, forcing the child to clarify and go deeper. Accuracy is tracked silently and gaps are raised only in the Review phase, never mid-explanation, because interrupting an explanation to correct it teaches a child that explaining is dangerous.
This rests on the protégé effect: explaining a concept produces deeper learning than receiving instruction. The strongest current evidence comes from Georgia Tech's Socratic Mind programme — a study of 173 undergraduates in CS 1301 provided the first empirical evidence that AI-mediated Socratic questioning measurably improves engagement, learning and higher-order thinking, with subsequent deployment across 5,000+ students at Georgia Tech and UC San Diego showing statistically significant learning gains and 70–95% positive student experience.
Mastery is not a number you were awarded. Mastery is an explanation that survived being questioned, twice.
Where does LIDS sit against other learning technology?
LIDS targets a layer above content delivery and above adaptive difficulty.
| Layer | What it does | Typical examples |
|---|---|---|
| Content platforms | Worksheets, videos, lessons | Khan Academy, BBC Bitesize |
| Adaptive learning | Personalised paths, difficulty adjustment | IXL, DreamBox |
| Cognitive development | Teaching the meta-skill of learning itself | What LIDS is for |
Most systems described as adaptive adjust difficulty. Adjusting difficulty keeps a child inside the same pedagogical shape and simply moves the bar. Adjusting the shape — moving a child from a Standard Lesson into Teach Me because they have the facts but cannot yet explain them — is a different intervention, and it is the one that develops the meta-skill.
Where does LIDS not apply?
LIDS does not apply where the objective is genuinely recall under time pressure, and pretending otherwise would be dishonest.
Some things must simply be known, quickly and correctly — number bonds, a safety procedure, a language's irregular verbs. Socratic discovery is a slow, expensive way to install a fact that has no underlying structure to discover. Use retrieval practice and spacing for those, and stop trying to make everything a journey.
It is also a poor fit where an external syllabus with fixed assessment dates is non-negotiable and there is no slack in the calendar. LIDS optimises for depth and self-direction; an exam three weeks away optimises for coverage. Those goals genuinely conflict, and a framework that will not say so is selling something.
Two further limits worth stating plainly. It depends on a child having enough language to articulate — which changes what the modes look like for very young or pre-verbal learners rather than removing the approach. And it assumes an adult who can tolerate a slower, less legible middle period, because articulated mastery produces less visible evidence per week than a worksheet does.
How do you apply it outside education?
Apply it anywhere a system must develop capability rather than measure output — the transferable core is three rules, not the seven modes.
LIDS — THE TRANSFERABLE CORE
The Hopium Lab · v1.0 · 22 July 2026 · take it, fork it, argue with it
1. NEVER REDIRECT THE TANGENT
[ ] Where does unprompted interest currently get suppressed as off-topic?
[ ] What would it cost to treat it as the highest-quality signal you have?
2. PROVE UNDERSTANDING BY ARTICULATION, NOT BY SCORE
[ ] Can the person explain it in their own words?
[ ] Has that explanation held up under questioning more than once?
[ ] Are you recording the explanation, or only the outcome?
3. ADJUST THE SHAPE, NOT ONLY THE DIFFICULTY
[ ] When something is not landing, do you make it easier — or change the mode?
[ ] Do you have more than one mode available at all?
THE TEST: if your system can award a score without a person ever having
explained anything, it is measuring compliance, not understanding.
Onboarding, technical mentoring and internal training all fail in the same shape: they measure completion because completion is easy to record. Ask instead whether the person can explain the thing to someone else, twice, and the whole picture changes.
What did LIDS lead to?
LIDS is a set of principles, and principles can be broken by the system that implements them — which is what led to building an engine where breaking them is not expressible.
The design question that follows LIDS is uncomfortable: a framework that says never reduce a child to a number is a promise, and a promise is only as good as the next feature request. The answer was to build the record-keeping engine so that no scoring surface exists — the renderer has no ranking affordance to add one to, and there is no cross-learner comparison anywhere in the system. Fork the philosophy; you cannot fork away the integrity.
That is the difference between a principle and a guarantee, and it is the more interesting half of the story.
Ross Jones, Founder, The Hopium Lab. Last modified 22 July 2026. LIDS is applied in Skoolio.