Editorial

Classical scholarship has a scale problem. The Chinese tradition alone has accumulated over two thousand years of commentary on the I-Ching, most of it untranslated into English. The Zhanguoce runs to thirty-three chapters of diplomatic speeches. The Hanfeizi, the Shiji, the Sunzi, the Guanzi, the Mozi, the Xunzi — each text has its own body of source material, its own interpretive history, its own internal logic.

And that is just China. This project covers three civilizations. Thucydides, Herodotus, and Xenophon in the Greek tradition. Ferdowsi's Shahnameh in the Persian. Each tradition has thousands of pages of primary text that deserve careful, bilingual treatment — original language alongside English translation, with commentary that respects the source.

No single scholar could produce editions of eighteen classical texts across four languages at the depth we publish, in any reasonable timeframe. The work would take decades. The traditions deserve better than waiting decades.

AI changes the economics. Not the scholarship — the economics.

The AI (Claude Opus 4.6) drafts text under close editorial direction. It researches across source materials. It produces first drafts that are then read, challenged, restructured, and rewritten. It handles the mechanical work of translation verification, cross-referencing, and consistent formatting across eighteen texts in four languages.

The AI is fast at things that are slow for humans: synthesizing information from multiple classical sources, maintaining consistency across a large body of work, and producing readable prose from technical material. It does these things well enough that editorial time can focus on what matters.

I decide what to publish. I determine the editorial angle — which source texts to prioritize, which interpretive traditions to follow, which claims to include and which to leave out. I validate the scholarship against classical sources. I read every line before it publishes.

The AI does not know which translation of 乾 is more faithful to the original. It does not know that Thucydides' Melian Dialogue maps to 合縱連橫 dynamics in ways that illuminate both traditions. It does not know that Ferdowsi's framing of the Sasanian collapse echoes the patterns Sun Tzu warned about. These are editorial judgments. They require years of study, cultural context, and an understanding of what connects these three traditions.

The AI is a tool. I am the editor. The distinction matters.

Because hiding it would be dishonest, and dishonesty is corrosive to the kind of trust that scholarship requires. If you're reading an analysis of how Su Qin's vertical alliance maps to the Delian League, you deserve to know how that analysis was produced. Not because AI-assisted work is lesser — but because you have the right to evaluate the method.

We think the method stands on its own merits. The text editions are carefully sourced. The original languages are always shown alongside translations. The commentary engages seriously with the scholarly tradition. If those things are true, the tool used to produce them is a fact about process, not a mark against quality.

We also think transparency will age well. The EU's AI transparency regulations take effect in August 2026. Google recommends disclosure wherever a reader might wonder “how was this created?” The direction is clear: the standard is moving toward disclosure, not away from it. We'd rather be early than late.

We don't just disclose the use of AI. We publish the tools.

Inkstone is an open-source set of Claude Code skills built for classical Chinese scholarship. It includes verse explanation (bilingual commentary on classical texts), verse-to-image prompt composition (the pipeline that generates our hero images through fal.ai), and design review tooling. Anyone can inspect it, use it, or improve it.

Publishing the toolchain is the strongest form of transparency we know. It's not enough to say “we used AI.” We show you exactly how, and we give you the tools to do the same.