問題の核心──創造者は、なぜ報われないのか

書籍が書かれた瞬間に、著作権は発生する。絵が描かれた瞬間に、映像が撮られた瞬間に。日本が採用する「無方式主義」は、登録不要でその権利を認める。

だが現実はどうか。

その権利を「誰が、いつ、何を創ったか」と証明する手段がない。デジタル空間ではコンテンツが瞬時に複製・流通し、創造者への対価還元は著作権管理団体を通じた複雑な手続きに依存する。絶版・品切れ作品の著作者遺族には、そもそも対価が届かない。外国著作物への対応は、さらに困難だ。

この提言は、その仕組みの設計図だ。ブロックチェーン技術とスマートコントラクトを組み合わせることで、「誰もが自由に使えて、意識せずに作者に対価が届く」世界を技術的に実現できる。

三本柱──新著作権保護インフラの設計

価格決定の逆転発想

著作物の価値を事前に決めることは不可能に近い。無名の写真が突然バイラルになることもある。だからこそ、事前に価格をつけるのではなく「使われた量が価格を決める」という設計が有効だ。

誰も使わなければゼロ。億回使われれば高額になる。作者は価格を決めなくていい。使われた実績が、そのまま報酬になる。1回0.001円でも、億回使われれば100万円だ。

関連する法体系と必要な整備

国立国会図書館への具体的提案

本提言はすでに国立国会図書館・電子情報部へ送付している。法改正を要しないフェーズ1として以下を提案している。

量子コンピュータ時代への備え

現行のSHA-256ハッシュは、今日の計算機では解読に事実上無限の時間を要する。だが量子コンピュータの本格稼働が現実となる2030年代以降、この前提が崩れる可能性がある。

世界への発信──日本発・新知財秩序の構築

まず概念に名前をつける

世界を変えた概念には、必ず名前があった。「インターネット」「ブロックチェーン」「オープンソース」──名前がつくことで、概念は伝播する。

この提言が目指す世界を、一言で表す名前が必要だ。提案する。

発信先と役割

RAI国家戦略との連携

本提言は、RAI（Refined Autonomous Intelligence）国家戦略と深く連携する。RAIが「日本の精神性をロボットに宿す」ならば、UCAは「日本の知的創造物に魂を宿す」試みだ。どちらも、中国・米国とは異なる「日本型の知的インフラ」の構築という一点に向かっている。

全体ロードマップ

実際に動くプロトタイプを公開中

The problem — why creators are not paid

Copyright arises the moment a book is written. The moment a painting is made, the moment a film is shot. Japan, like every Berne signatory, grants the right without registration or formality of any kind.

But what happens in practice?

There is no means of proving who created what, and when. In digital space content is copied and circulated instantly, and remuneration for creators depends on elaborate procedures routed through collecting societies. For the families of authors whose works are out of print, nothing arrives at all. For foreign works the difficulty is greater still.

This proposal is a blueprint for that mechanism. Combining blockchain and smart contracts, a world in which anyone may use a work freely while payment reaches its author unnoticed is technically achievable.

Three pillars

Inverting how price is set

Determining the value of a work in advance is close to impossible. An unknown photograph can go viral overnight. So rather than pricing it beforehand, let the volume of use set the price.

Nothing used, nothing paid. Used a hundred million times, paid accordingly. The author never has to name a figure. The record of use is the remuneration.

The legal framework, and what it would need

A concrete proposal to the National Diet Library

This proposal has been sent to the Electronic Information Department of Japan's National Diet Library. Phase 1 requires no amendment to any statute.

Preparing for the quantum era

SHA-256 would today take effectively infinite time to break. But once quantum computers enter genuine operation — from the 2030s onward — that premise may fall away.

Reaching outward — a new order for intellectual property

First, give the concept a name

Concepts that changed the world all had names. "The internet." "Blockchain." "Open source." A name is how a concept travels.

Roadmap

What has changed since this was written

The centre moved to AI training

As written above, the design centred on digital archives and out-of-print works. It has since narrowed onto a sharper question: was this work used to train this model? A proposal on that basis was submitted to Japan's Cabinet Office and Agency for Cultural Affairs in August 2026, under the heading from disclosure to measurement.

The reasoning is that the transparency debate has stalled on a single axis — how much of a training corpus must be published — and that axis has no equilibrium. But a rightsholder does not want the corpus; they want an answer about one work they already hold. That is a closed question, answerable without disclosure.

Identification is delegated to an existing standard

The design no longer defines its own identifier. ISO 24138:2024 (ISCC), published in May 2024, specifies a similarity-preserving content code carrying both an exact component (Instance-Code) and a perceptual one (Content-Code). The latter matters practically: collection pipelines resize images, re-encode audio and normalise text, so exact matching alone fails systematically. UCA now implements that standard rather than competing with it.

The substantive contribution is on the operator side

Existing ISCC work already constrains the party that declares a work, through a transparency log. Nothing constrains the party that ingests it. UCA now specifies a corpus commitment log applying the same RFC 6962 construction to the operator, with one structural change: the membership tree is sorted rather than append-ordered, which permits non-membership proofs.

That change matters more than it appears. Without it, "present" is provable and "absent" is not — leaving unconstrained precisely the verdict an operator has a financial incentive to falsify.

Limitations are now stated rather than left to be found

A commitment binds a claim, not completeness: an operator can commit to a corpus that omits what it trained on, and no cryptographic mechanism detects this. Similarity-preserving codes collide, so an attestation asserts that a member with that code was present, not that a specific work was. And non-membership proofs reveal bracketing neighbours, permitting partial enumeration under repeated query — whether they can be blinded without losing verifiability, we do not know.

A working prototype