マルチバース攻略本 (Multiverse Guidebook) is a 12-chapter guide to the multiverse, string theory and the many-worlds interpretation — written for readers without a physics background, with no equations from start to finish.
You don't just read it. You run the experiments. A double-slit simulator, a branching-worlds demo, and quizzes along the way; XP accumulates as you go. It reads like a game, and the whole thing is free.
The main text is written in Japanese. This page is a complete English introduction to what the guidebook covers and how it keeps itself honest — not a machine translation of it.
The one rule: every claim carries a label
This field mixes established physics with informed speculation, and most popular treatments blur the line — which is the same as lying. So the guidebook labels every claim with one of four levels of confidence:
確立 · ESTABLISHEDRepeatedly confirmed by observation and experiment. Textbook material.
有力な仮説 · LEADING HYPOTHESISWidely supported by researchers with indirect evidence — but unsettled.
検証困難 · HARD TO TESTThe theory works on paper, but current technology cannot check it in practice — sometimes in principle.
解釈 · INTERPRETATIONExperiment can never decide it. A question of how to read the same mathematics.
The labels do real work. By the final chapter, "what do we actually know?" has a precise answer instead of a shrug.
The twelve chapters
How much of the universe can we see? — the observable universe, and why its radius is 46.5 billion light-years in a 13.8-billion-year-old cosmos. Established.
Levels 1 and 2: beyond the horizon, and the bubble universes — eternal inflation's consequence for a flat, infinite space.
Try it: one particle, two slits — an interactive double-slit experiment you operate yourself.
Level 3: many-worlds as "subtraction" — Everett's reading, presented as the option that removes the collapse postulate rather than adding worlds by hand.
The branching you — lunch choices, survival probabilities, and what "probability" can still mean if every branch happens.
String theory: why extra dimensions are needed — and what the demand costs.
The landscape: a universe with a settings file — the string landscape's ~10⁵⁰⁰ vacua and why it tempted physicists toward anthropic reasoning.
Summary: the four levels of the multiverse — Tegmark's hierarchy, sorted with the honesty labels.
What the latest observations say (2025–2026) — CMB anomalies and what would, or wouldn't, count as evidence of a collision with another bubble.
Is there free will, or is everything determined? — where many-worlds genuinely touches philosophy.
The infinite computer — computation without purpose or direction at the bottom of it all.
So what is consciousness for? — the closing chapter, kept carefully inside the interpretation label.
New today (auto-updated)
Six newest arXiv preprints each in cosmology (astro-ph.CO) and quantum physics (quant-ph). You do not have to read them; the titles alone show the field is moving.
Cosmology (needs JavaScript)
Quantum physics (needs JavaScript)
Eight things people get wrong
Mistakes that are easy to make when explaining this to someone else. Each has a tempting shape, so they are listed up front.
Many worlds are not parallel worlds you can visit. Branching is one-way; branches can only interfere while the system is tiny and isolated from its environment.
Observation does not need a mind. The apparent collapse happens when a system entangles with its environment (decoherence) — a single air molecule will do it.
Entanglement cannot send messages faster than light. Correlation without control of outcomes carries no information (the no-communication theorem).
Inflation is not "before the Big Bang". In the standard picture, reheating at the end of inflation is what produces the hot Big Bang.
"The universe is infinite" is not settled. The observable region is finite (radius about 46.5 billion light-years); near-zero curvature is consistent with infinite, not proof of it.
The multiverse is not one theory. The four levels rest on different evidence with different confidence; "is the multiverse science?" has no answer until you say which level.
String theory is neither dead nor proven. It has not yet produced a testable prediction; as mathematics it is used across physics.
"If everything is determined, nothing matters" does not follow. Determinism and "does your choice change the outcome" are different questions; a choice is part of the causal chain, not outside it.
Who writes it, and how to read it
The guidebook is part of toriumis.com, a one-person studio's collection of free learning sites and tools — a quantum mechanics course for non-scientists lives on the same domain, and the two are designed to be read together.
The working language of the guidebook is Japanese. If you read Japanese even slowly, the interactive experiments need no language at all — start with Chapter 1 and see whether the format works for you. If demand from English readers ever justifies it, a full English edition is the natural next step; this page exists so that at least the shape of the thing is visible now.
A note on honesty. Nothing on the guidebook's pages claims the multiverse is real. What the labels allow you to see, chapter by chapter, is exactly where observation ends and where interpretation begins — which is a rarer thing to find than another confident summary.