한국어 English 日本語 中文
A LEARNING FRAMEWORK

QuestLoop
Question-Driven Learning

Not a framework for problem-solving but a framework for learning. Questions, not challenges, become the engine of learning, and failure is not permitted — it is required.

Question → Hypothesis → Investigate → Verify → Question again
QUESTLOOP CYCLE Question Hypothesis Investigate Verify
The Starting Point — One Nagging Question

The belief that "solve a valuable problem,
and learning will naturally follow"

This framework began with a question we kept meeting while running challenge-based learning (CBL) in the field for years. While the team races toward a problem in the world, is individual learning really happening?

CONCERN 01

Overconfidence in accidental learning

A team challenge optimizes for problem-solving, not for individual learning. Learning gets pushed aside — not an intended goal but a byproduct picked up along the way.

CONCERN 02

The team's ZPD ≠ the individual's ZPD

A difficulty that suits the team as a whole is too easy for some members and too hard for others. One challenge cannot satisfy the learning conditions of N people at once.

CONCERN 03

Role entrenchment

The more rationally a team operates, the more each member takes on what they are already good at. Output quality rises, but individual learning curves flatten.

QuestLoop is not an improved version of CBL. It inherits the motivational energy of challenges — real problems, personal interest and relevance — but it is an independent framework redesigned from the ground up so that learning becomes intention rather than accident. The reason for putting learning first is plain — solve only what your current skills can handle and your skills stay put; and if your skills stay put, the ultimate hard problem can never be solved. Climbing by solving small challenges while learning, growing the very ability to solve, is the only path to the real problem.
How It Was Derived

Eight decisions
built the framework

Starting from those concerns, the principles, structure, and devices were settled one after another. Each step is an answer to the question the previous step left behind.

DECISION 01 — IDENTITY

This is a framework for learning

Team output is a means to learning, not the goal. When team efficiency and individual learning collide, individual learning wins. This one declaration unlocks role entrenchment — an expert taking an unfamiliar part is no longer an imposition but the correct way to use the framework.

DECISION 02 — DUAL GOALS

Nest the individual challenge inside the team challenge

The team challenge aims at a problem in the world; the individual challenge aims at one's own learning. The two are not separated but nested — "while solving this team problem, I will deliberately learn X." Accidental learning becomes intentional learning.

DECISION 03 — ENVIRONMENT

Failure is not permitted — it is required

Psychological safety is only a precondition. QuestLoop stands on productive failure — failure is not something to endure but the core mechanism of learning. A drop in deliverable quality is officially accepted.

DECISION 04 — ENGINE

The question opens learning; the hypothesis is its unit

You first ask a real question, then form a tentative hypothesis for it from your current understanding alone, learn while chasing the answer, and evaluate the change against the hypothesis. Because a hypothesis is a prediction, the moment of verification is the exam itself — the exam becomes the engine of learning, not its endpoint.

DECISION 05 — STRUCTURE

A fractal where two layers run the same cycle

The team, too, forms and verifies the hypothesis "this solution will solve the problem." Each individual forms and verifies their own learning hypothesis. Two layers, one cycle.

DECISION 06 — DEVICES

Structure holds it up, not a person

The problem that learners struggle to set good challenges and hypotheses on their own is solved with structure, not facilitator talent — constrained choice, the hypothesis sentence template, the peer-verification ritual. The organizer steps back from intervener to infrastructure manager (Loop Keeper).

DECISION 07 — EXPERTS

Experts learn by the same rule

For a ten-year veteran, "what I can't do" is the adjacent area that amplifies their expertise — design, domain, market. The single rule "pick what you can't do" covers the entire range from beginner to expert.

DECISION 08 — AUTHENTICITY

Visible connection beats cynicism

The ultimate challenge is decomposed hierarchically down to an attemptable size. Even the smallest challenge is connected by lineage to the real problem. What decides toy versus stepping stone is not the challenge's size but the visibility of its connection.

The Engine — The Learning Cycle

Start with a question,
return to a question

Click each phase to explore. The cycle turns identically on the team layer and the individual layer.

Question QUESTION Hypothesis HYPOTHESIZE Investigate INVESTIGATE Verify VERIFY
PHASE 1 · QUESTION

Question — Start from real curiosity

Every cycle begins with something you genuinely wonder about — "What don't I understand yet, and what am I truly curious about?" Before searching for an answer, you first sharpen the question into a verifiable form.

A good question narrows the direction of inquiry. Only with it does the next hypothesis escape vagueness.

↻ When verification ends, the updated understanding raises the next question — the loop never stops

Structure — Fractal Nesting

Two layers, one cycle

TEAM LAYER Team challenge — a hypothesis aimed at the world Person A Learning hypothesis Person B Learning hypothesis Person C Learning hypothesis

The team challenge aims at the world;
the individual challenge aims at yourself

An individual challenge is not a separate assignment split off from the team challenge but a learning hypothesis nested inside it. The very process of solving the team problem becomes the individual's testing ground.

The moment the two layers collide — "if I take this part in order to learn, the team's output slows down" — the framework's answer is unequivocal.

On collision, individual learning wins
Same Cycle, Different Use

How the cycle turns differently
on the team layer and the individual layer

One engine, different fuel. The team layer verifies a problem in the world; the individual layer verifies one's own understanding. Compare how each phase works on the two layers.

TEAM LAYERTeam layer — a hypothesis aimed at the world
INDIVIDUAL LAYERIndividual layer — a hypothesis aimed at yourself
A real problem the world hasn't yet solved is the challenge. Declare the ultimate challenge and decompose it into a hierarchy tree. The problem's authenticity becomes the authenticity of the whole tree.
ChallengeCHALLENGE
Your own learning is the challenge. Pick "what you can't do right now" from the tree's leaves and nest it inside the team challenge. Beginners pick foundational skills; experts pick the adjacent areas that will amplify their expertise.
Sharpen the challenge into an answerable question — do users really face this problem, what counts as solved, why now. "What is the real
problem to solve?"
QuestionQUESTION
"What can't I do yet, and what am I genuinely curious about?" You frame the real question that becomes the starting point of learning, in your own words. "I really
don't get this"
Form a prediction for that question. It is one hypothesis the whole team writes together. "This solution
will solve this problem"
HypothesisHYPOTHESIZE
Written before learning, with no information, from current understanding alone. There are N hypotheses — as many as there are members. A peer asks — "When this hypothesis breaks, what will you learn?" "Right now I understand X
like this"
Build, collide, meet users. The team's investigation is the work of exposing the solution to reality.
InvestigateINVESTIGATE
The team's investigation site doubles as the individual's laboratory. Not in separate study time but in the very process of solving the team problem, you verify your hypothesis and record the moments it breaks.
Check "how much of the problem was solved," but the result is not a grade — it is material for the next team hypothesis. Deliverable quality is not what gets graded.
VerifyVERIFY
Measure the learning delta — the distance between the starting hypothesis and current understanding. Because each person's hypothesis is their own evaluation criterion, growth is measured even when teams produce different results.
Rejecting a solution hypothesis is information for changing direction. You have gained grounds for a pivot — the project hasn't failed.
FailureFAILURE
Rejecting a learning hypothesis is the point of maximum learning. Understanding updates by exactly the gap between expectation and reality. A hypothesis that never breaks is the real warning sign.
WHEN THE TWO LAYERS COLLIDE

"If I take this part in order to learn, the team layer's investigation slows down" — at the moment the two layers' interests clash, the individual layer wins. The team layer's output is a means for the individual layer's learning. This priority is QuestLoop's identity.

Authenticity — The Challenge Hierarchy Tree

The meaning of a small challenge
is proven by its lineage

The ultimate challenge is hierarchically decomposed down to an attemptable size. Click a leaf node — its lineage lights up, showing how that small challenge connects to the real problem.

← Swipe sideways to see the whole tree →

Ultimate challenge Keep blind people from getting lost indoors Perceive the space LiDAR indoor mapping Convey without sight Non-visual interface Know the way Indoor map & route data Measure the error limits of ARKit anchors firsthand Try simplifying a point cloud into flat planes Use VoiceOver gestures for a day with eyes closed Model a floor plan as a graph data structure Try translating pathfinding into haptic rhythms LEAF = the challenge an individual picks · "pick what you can't do"
Select a leaf to reveal its lineage A small task that looks like a toy to one person becomes, for another, a stepping stone toward the result. What makes that difference is this tree.

What decides toy versus stepping stone
is not the size of the challenge but the visibility of its connection

Build your own team's tree →

A tool to declare the team challenge, assign individual challenges, and visualize the lineage

Devices — Structure Holds It Up

Five devices drive it,
not the organizer's talent

The difficulty of setting good challenges and good hypotheses on your own — QuestLoop solves it with structure, not with people.

DEVICE 01

Constrained choice

Not "set your own challenge" from a blank page. You pick what you can't do right now, from the leaf nodes of the challenge tree. Because the options derive from the team's task, relevance is guaranteed — and a single rule guarantees difficulty.

Rule: pick what you can't do
DEVICE 02

The hypothesis sentence template

The form is the filter. An unverifiable hypothesis can't even complete the sentence.

Right now I understand [X].
Trying [Y] will reveal whether it holds.
DEVICE 03

The peer-verification ritual

Not the hypothesis's author but a teammate asks. Verification runs without a Loop Keeper, and the asking peer trains their own metacognition in the process.

"If this hypothesis breaks, what will you learn?"
DEVICE 04

The Loop Keeper as infrastructure manager

The Loop Keeper is not someone who inspects challenges but someone who watches whether the structure runs well — the caretaker of tree decomposition, the templates, and the peer ritual. Not intervention — infrastructure management.

DEVICE 05

Learning-delta evaluation

The criterion for success is not "was the problem solved" but "what changed relative to the hypothesis." The problem must be real and the attempt must be real, but what gets evaluated is the amount of learning change, not the deliverable. Because each person's hypothesis is their own evaluation criterion, the ambiguity of assessment disappears. Here, too, lies the grounds for officially accepting a drop in deliverable quality.

THE BEGINNER'S LEAF

Foundational skills

Pick, from the tree's leaves, a foundational skill you can't handle yet. Things to learn lie scattered all over the map.

THE EXPERT'S LEAF

Adjacent areas that amplify expertise

For a ten-year developer, "what I can't do" is the design sense or the market's domain knowledge needed for a better result. The surrounding knowledge that makes your strength even stronger becomes the challenge.

Same rule, different leaf — "pick what you can't do" covers every level of skill
Summary — Five Principles

QuestLoop on one page

01

Learning is the goal

Team output is a means. On collision, individual learning wins, and a drop in deliverable quality is officially accepted.

02

The hypothesis is the engine

Question → Hypothesis → Investigate → Verify. The moment a hypothesis breaks is the point of maximum learning, and the moment of verification is the exam itself.

03

The structure is fractal

The team layer and the individual layer turn on the same cycle. Individual challenges nest inside the team challenge.

04

Failure is required

Failure is an object of design, not of permission. Productive failure and psychological safety are the floor of the environment.

05

Lineage is authenticity

Every small challenge connects, through the hierarchy tree, to the ultimate real problem. Visible connection beats cynicism.