Editorial · Who edits Storgy
I edit Storgy. Every page, by hand.
I’m Nikola Gulevski. I read every poem before it goes on this site, write the editorial intros on each poet hub myself, and review the Sonnet-assisted line-by-line analyses one by one. When the AI drafts something, I’m the one who decides whether it stays on the page.
Storgy is built as a one-person reading library on purpose. The point of writing my name on the door is so a reader can hold someone accountable for what shows up here. If a line break is wrong, an attribution is off, or a poet you care about is missing, you know who to email.
What I publish
Three things, all reviewed line by line.
Storgy isn’t a magazine. It’s a reading library. What I put on these pages falls into three categories, all of which I read carefully before they ship.
Public-domain texts
Every poem on Storgy is sourced from Project Gutenberg or another public-domain corpus, with the original line breaks preserved. Visitors in the US (and search crawlers) see full bodies; visitors elsewhere see authorised excerpts where US copyright still differs from theirs.
Drafted by AI, edited by me
Claude Sonnet 4.6 produces the structured analysis against a strict schema, then a second-pass humanizer strips the AI tells. I read each one and rewrite the parts that drift. The provenance band on every analysis page tells you when AI helped.
Signed, two-paragraph reads
Each poet hub opens with a two-paragraph intro that says, plainly, what makes this poet worth reading and where to start. The byline is mine. The Article schema on the page names me as author so credit isn’t laundered through “the Storgy team”.
What I don’t publish
The things I leave off the site.
- Full texts by living poets. Their pages on Storgy carry a biography, a curated list of representative works, and direct links to the poets’ own pages on Poetry Foundation, poets.org, and their publishers. We don’t reproduce the poems.
- AI “lost poems” or speculation. The site never publishes a poem “in the style of” a real poet or claims a piece is by someone it isn’t. The Poem Generator tool exists and labels its output as generated; that is the only AI-authored poetry on Storgy.
- Translations without a named translator. A translation has two rights-holders, not one. Until I can verify the translator and confirm the translation itself is public-domain, the poem doesn’t go up.
- Anything I haven’t read. The whole point of an editor with a name on the page is that the bar is human. If a page exists on this site, I’ve read it.
Methodology
How a poet gets onto Storgy.
- 01Sourcing. Project Gutenberg supplies the text. Wikidata cross-checks the metadata (birth, death, publication years, language). Poets.org fills small gaps. The bibliography is recorded with the poem.
- 02Public-domain check. Each work is classified globally (pd_status) and US-only (pd_us). The render gate honours the visitor’s country so EU/UK readers see excerpts on works that are public-domain only in the US.
- 03AI analysis pass. Sonnet 4.6 returns a strict JSON shape with seven sections — summary, themes, line-by-line, tone, symbols, historical context, FAQ. The schema constrains what the model can produce so every poem page has the same anatomy.
- 04Humanizer pass. A second model rewrites the AI draft to strip em-dashes, hedged adverbs, and the AI vocabulary this site won’t print (“iconic”, “timeless”, “masterpiece”, “in the world of poetry”).
- 05Editorial review. I read the page, fix what drifted, and sign off. Pages that don’t survive review get re-drafted or dropped.
Per-poem provenance and verification dates live on the sources page.
Reader questions
Things people email me about.
Because the site is small enough that one person can hold the whole thing in their head, and that consistency is the brand. Every poet intro, every comparison essay, every public-domain check is run by the same reader against the same bar. When that stops being possible, I'll write here about who else has joined. Not before.
Claude Sonnet 4.6 drafts the line-by-line analyses, the editorial intros, and the comparison essays against a strict JSON schema. Every AI-drafted block then runs through a second-pass humanizer that strips em-dashes, hedged adverbs, and the words 'iconic', 'timeless', and 'masterpiece'. Nothing is published before I read it. The provenance band at the bottom of analysis pages tells you when AI helped.
Project Gutenberg is the primary text source. Wikidata cross-checks author birth/death years and original publication dates. The pipeline also tracks pd_status (global) and pd_us (US-only) per poem — visitors from outside the US see authorized excerpts on works that are public-domain in the US but not yet in their region. The verification record lives on the /sources/ page.
Full texts by living poets. AI-generated 'lost poems' or speculation that something a poet might have written. Translations without a named translator and a public-domain check on the translation, not just the original. Anything I haven't read carefully enough to defend.
Email hello@storgy.com. Misattributions, wrong line breaks, missing poets, broken sources — all welcome. If you're a working poet or scholar and want a poet/work added or corrected, mention that and I'll prioritise.
Write in
The fastest way to reach me.
If a poem is misattributed, a translator is missing, or a poet you read should be here, write in. Suggestions for new themes, comparison pages, or tools are welcome too.
hello@storgy.com