Comparison · 2026-04-23 · Reading time ~ 7 min
Pagera vs Project Gutenberg: what's shared, what's different
Project Gutenberg opened 76,000+ public-domain texts to the world. Pagera sits on top, adding AI translation, a modern web reader, and demand-driven curation. Here's the honest side-by-side.
Pagera Editorial
Founded by Michael Hart in 1971, Project Gutenberg (PG) is the original digital library. As of April 2026 it hosts roughly 76,000 free public-domain texts. Pagera is built on top of that legacy. So why did Pagera start, and what does it solve that PG doesn't?
At a glance
| Dimension | Project Gutenberg | Pagera |
|---|---|---|
| Founded | 1971 | 2025 |
| Operator | Non-profit (Project Gutenberg Literary Archive Foundation) | PubStation Inc., ad-funded |
| Catalogue size | ~76,000 titles (mostly English) | 24,446 curated titles (Gutenberg + Aozora Bunko + manual) |
| Translations | None. Original language only. | AI + 5-axis / 98-point review. 55 Korean translations live, shipping weekly. |
| File formats | HTML / EPUB / Kindle / plain text / some audio | Web reader only. Chunked HTML with anchor-aligned bilingual. |
| Reading UX | Download and read in your own app | Bilingual view, highlights, vocabulary notes, synced progress, auto-covers |
| Curation | Search / category listings | Era / genre / language / popularity / request queue |
| Licence | Project Gutenberg License (US public domain) | Original keeps PG licence; translations released CC0 / Public Domain Mark 1.0 |
Why Pagera exists: the language gap
Project Gutenberg is extraordinary. But for a Korean reader, 76,000 of its titles are effectively unreadable. PG's Korean translation count is in the single digits — the volunteer model that built PG never extended to translation. Tens of thousands of classics were technically "available" but practically unreachable.
Pagera exists to close that gap. When readers request a translation, duplicate requests accumulate and enter a priority queue. Pagera AI then produces a first draft, which goes through a 5-axis review loop until it clears the 98-point gate. One translation costs roughly USD 6 today, which makes weekly expansion feasible on ad revenue alone.
Pagera does not replace Gutenberg
To be clear: Pagera isn't a Gutenberg replacement. We use PG's texts as a data source and credit them on every book page. PG remains the canonical archive; Pagera adds a translation + modern reader + curation layer on top.
If you want the original English EPUB for your Kindle, go to PG. If you want to read the original and a Korean translation side-by-side in your browser, highlight a line into a synced vocabulary deck, and pick up on another device, Pagera is the faster path.
How translation quality is enforced
Every Korean translation on Pagera must pass:
- Adequacy 9.5+ / Fluency 9.5+ / Style 9.5+ / Humanness 9.5+ / Polish 9.5+
- Double-blind scoring by two independent LLM reviewers, ≥ 98 / 100 total
- Up to 5 targeted-fixer iterations if a single axis falls short
The rubric is open-sourced inside the Pagera repo (.claude/rules/translation-quality.md) and per-book histories live in .claude/quality/registry.json. Readers can ask for the score of a specific translation on request.
Which one should you use?
Use Project Gutenberg when: you want to own and remix original-language EPUBs, you manage a personal library in Calibre, or you're hunting for obscure English-language texts.
Use Pagera when: you want to read in Korean, you want side-by-side bilingual, you expect a modern web reader with highlights + progress sync, or you want to request a translation of a work that doesn't have one yet.
They aren't competitors; they're layers. Pagera stands on Gutenberg's shoulders.