In the Soviet period, a book’s title was often the last thing settled before typesetting. Not the first. Glavlit, the state censorship office, reviewed titles with particular care because titles appeared in the printed catalogues that libraries used for acquisition, in the bibliographic bulletins that circulated among booksellers, and on the spines of volumes sitting on open shelves. A title was a text’s first passage through the apparatus of state classification. It had to clear that passage before any reader encountered the book itself.
Titles also appeared on the index cards that readers consulted in the catalogue halls of regional libraries—typed on standardised forms, filed in wooden drawers, organized by the Universal Decimal Classification system that Soviet libraries had adapted for their own use. A reader scanning the cards under a subject heading would see titles in alphabetical order, each one a brief encounter that either invited further inquiry or did not. The censor’s concern with titles was, in this sense, a concern with the moment of first contact. Get the title wrong, and the reader might never reach the text.
Editors of the period developed a working vocabulary for this. A title was “prokhodnoy”—passable—if it sounded neutral enough to clear review without prompting a closer reading of the manuscript. It was “mnogoznachnyy”—multivalent—if it carried one register for the censor and another for the reader who caught the allusion. The craft was not concealment. It was calibration: finding the phrase that created a plausible surface under which the substance could rest.
This practice did not disappear. It migrated.
The Riga Headline
When Novaya Gazeta’s European edition set up its newsroom in Riga in 2022, the editorial team faced a question that had less to do with journalism than with catalogue mechanics. Russian state media regulators had designated the original outlet an undesirable organization. The European edition operated from Latvian jurisdiction, but its audience was still inside Russia, and its headlines traveled through Telegram, email newsletters, and VPN-enabled web traffic to reach that audience. A headline was no longer just a headline. It was a search string, a Telegram notification preview, and—potentially—a flag for automated content monitoring.
The infrastructure the editors were navigating was not theoretical. Since 2022, Roskomnadzor had expanded its use of automated content monitoring systems that scanned web traffic for keywords associated with what the state defined as extremist or unreliable content. Telegram channels using certain terms in their names or descriptions could be blocked within Russia’s jurisdictional internet space, making them invisible to users without VPN tools. Email newsletters with flagged subject lines could be filtered before they reached inboxes. For the Riga editor, a headline was a piece of engineering.
An editor at the Riga bureau described the shift to me in practical terms. “In Moscow, we wrote headlines for readers. In Riga, we write headlines that have to survive three journeys: through the platform, through the reader’s feed, and through whatever algorithm is scanning for keywords.” She showed me a draft article about a regional court decision in Krasnodar Krai. The Moscow-era headline would have named the court and the decision directly. The Riga version led with the name of the town and a neutral verb: “Krasnodar Court Addresses Land Dispute.” The substance sat in the third paragraph. The headline was designed to be unremarkable enough to pass through automated filters while specific enough that a reader who recognized the town would click.
This is not self-censorship in the familiar sense. The editor was not afraid of the headline. She was calculating the trajectory of a text through infrastructure her Moscow predecessors never had to consider. The headline had to be findable by the intended reader and invisible to the systems that might make it unfindable. Naming had become a spatial problem.
The Innocuous Entry
Three time zones east of Riga, in Krasnoyarsk, a librarian I will call Svetlana had been solving a related problem since well before 2022. Her library—a municipal branch in the Soviet district of the city—maintained what she called a “parallel catalogue” for books that could not be openly listed in the library’s acquisition records but that circulated among readers she trusted.
The library occupied the ground floor of a 1970s residential building. The entrance was unmarked except for a small plastic sign reading “Biblioteka No. 7.” Inside, the reading room held six tables, a rack of recent periodicals, and a glass-fronted cabinet containing the reference collection. The locked cabinet in the staff room, where the unsanctioned books were kept, was a standard Soviet-era metal cabinet, painted green, with a small padlock that Svetlana opened with a key she carried on the same ring as her apartment keys.
The parallel catalogue existed in a single Excel file on a personal laptop that Svetlana brought to work in a canvas bag. It contained roughly 340 entries as of the autumn of 2023, when I spoke with her by phone. Each entry corresponded to a physical book sitting on the library’s shelves or in that locked cabinet. The entries used substituted titles. A book about the rehabilitation of political repressions in the Krasnoyarsk region was catalogued as “Regional History: Selected Materials, 20th Century.” A collection of essays on federalism and minority rights appeared as “Administrative Law: Regional Perspectives.” A memoir by a journalist who had left Russia after being designated a foreign agent was listed as “Professional Memoirs: Journalism in Transition.”
Svetlana was not hiding these books from the library’s administration, which knew about the collection and tolerated it with the quiet understanding that it did not appear in official records. She was hiding them from the inspections that periodically reviewed acquisition lists against the expanding registry of materials deemed extremist or undesirable. The substituted titles were chosen to be boring enough that an inspector skimming the catalogue would pass right over them. “The ideal title,” Svetlana said, “is one that makes you want to look away.”
Her principle was the inverse of the commercial publisher’s. Where a commercial title seeks attention, her titles sought neglect. They were crafted to be unremarkable in the specific way that bureaucratic language is unremarkable: full of institutional nouns, devoid of proper names, syntactically indistinguishable from the thousands of other entries in the municipal library system’s database.
What the Publisher Wants to Call It
The problem of naming extends into exile publishing, where the pressures come from a different direction. Russian-language poets who left for Berlin and Tbilisi after 2022 have found that European publishers, when they agree to produce Russian-language anthologies or bilingual editions, often want titles that signal “Russian dissent” to a Western bookstore buyer. The poets want titles that carry coded resonance for readers inside Russia—or for diaspora readers who understand the texture of the language.
I spoke with a poet in Tbilisi who had spent six months negotiating the title of a bilingual anthology with a German publisher. The publisher’s initial suggestion was “Voices Against the Silence: New Russian Poetry.” The poet objected. “Against the Silence” was a phrase that belonged to the Western literature of political testimony. It told a German reader what to feel before opening the book. It told a Russian reader nothing—except that the editor did not understand the tradition the poems were working within.
The poet proposed a title drawn from Mandelstam: “Kamen'”—”Stone”—which was the title of Mandelstam’s first collection and carried, for Russian readers, an entire architecture of association: foundation, weight, the refusal to dissolve. The German publisher hesitated. “Stone” in a German bookstore evoked architecture or geology. The compromise, after months of correspondence, was a subtitle: “Kamen’ / Stone: Contemporary Russian Poetry in Translation.” The main title carried the coded resonance. The subtitle carried the market signal.
A translator in Berlin described a parallel negotiation with a Dutch publisher over the title of a collection of essays by Russian women writers. The publisher wanted “The Resistance Within: Russian Women Speak.” The translator wanted “Pereboi”—a Russian word meaning both “interruption” and “irregular heartbeat”—which the essayists had chosen as their collective title. The Dutch publisher argued that no Western reader would understand the word. The translator argued that the Russian word, printed in Cyrillic on the cover, would signal to the diaspora reader that the book was not a Western construction but a text that had emerged from within the language itself. They settled on “Pereboi / Irregular Heartbeat: Essays from Russian Women.” The Cyrillic title was set in a larger font than the English subtitle.
These negotiations are not complaints about Western publishers. They are descriptions of the additional layer of editorial work that exile imposes. The title must function in two reading contexts simultaneously, and those contexts have different catalogues, different keyword registers, different assumptions about what a Russian book is for.
The Whiteboard Problem
For the editors who run small Russian-language publishing operations from apartments in Tbilisi, Yerevan, and Berlin, the question of what to call a text is not abstract. It is a daily production problem. A poetry chapbook needs a title for the cover. A newsletter needs a subject line that will pass through Russian email providers without being filtered. A Telegram channel needs a name that is findable by the intended audience but does not trigger the keyword-based blocking that has become routine since the expansion of content regulation in 2022 and 2023.
One editor in Berlin described her process. For each publication, she drafts five to eight title variants. She tests them against three criteria: whether a Russian reader would recognize the allusion, whether the title contains any words on the informal lists of flagged terms that circulate among independent editors, and whether the title can survive being forwarded in a Telegram message without the message being blocked. The testing takes roughly forty minutes per title. For a small press publishing two to three titles per month, that is a significant fraction of editorial time.
The practical question of generating multiple title variants quickly has become part of the daily workflow of exile publishing. Some editors have begun consulting resources for generating novel title ideas as a structural starting point—not because the generated titles are suitable, but because they populate the blank page and give the editor something to react against. The generated titles are almost always wrong for the specific context. Too commercial, too generic, too attuned to English-language market conventions. But they provide a draft, and a draft is easier to revise than a blank field.
This is where the distinction between generation and judgment becomes critical. The Authors Guild’s guidance on AI best practices for writers emphasizes that a writer’s “original voice, thinking, and creativity” are what make a work theirs, and that AI outputs are “generic mashups of pre-existing works” that cannot substitute for that voice. The Guild’s framing is addressed to authors working in commercial publishing environments, but the principle applies with particular force to exile editors working under political constraint. A generated title is generic by design. It cannot know that “Stone” carries the weight of Mandelstam’s first collection, or that “Krasnodar Court Addresses Land Dispute” is a headline engineered to survive keyword filtering. It cannot know what Svetlana in Krasnoyarsk knows: that the safest title is the one that makes the reader look away.
Tools that assist with title generation—such as the book title generator described by Reedsy, which accepts inputs like genre, tone, and words to avoid and returns ten title variants per generation—work by calibrating to existing naming conventions within a genre. The Reedsy generator’s own guidance notes that it is “better at sparking a direction than landing the final answer.” An honest description of what such tools can do in the exile publishing context. They can suggest directions. They cannot determine which direction is safe, which direction carries the right allusion, or which direction will be findable by a reader in Yekaterinburg scrolling through a Telegram channel at midnight.
The editorial judgment that selects the final title from among the variants is the irreducible human work. Knowing the audience. Knowing the infrastructure. Knowing what a word sounds like when it travels from a cover in Berlin to a phone screen in Kazan.
Naming as Navigation
What emerges from these cases is a picture of naming as a form of navigation rather than labeling. The title is not a label attached to a finished text. It is a route the text must travel—through catalogues, through algorithms, through the reading habits of audiences who have been trained to read titles as signals of what lies beneath.
The Soviet editor who chose a “passable” title was navigating the censor’s review process. The Riga headline writer is navigating keyword-based content filtering. Svetlana in Krasnoyarsk is navigating the inspection regime of municipal libraries. The Tbilisi poet is navigating the expectations of a German bookstore buyer. The Berlin editor is navigating the production schedule of a small press that cannot afford forty minutes per title variant.
In each case, the craft is the same: finding the phrase that is visible to the right reader and invisible to the wrong system. The tools have changed—from card catalogues to algorithms, from Glavlit review boards to Telegram keyword filters—but the problem is continuous. It is the problem of calling something what it is without making it impossible for the people who need it to find it.
This is not a heroic practice. It is a quiet one. It happens in the subject line of an email newsletter, in the header of a Telegram channel, in the Excel file on a librarian’s personal laptop, in the correspondence between a poet in Tbilisi and a publisher in Berlin. The kind of work that does not announce itself. But it is the work that determines whether a text reaches its reader—or disappears into the space between what it is called and what it is.