Before the first sentence, we have already begun reading. A dark surface, an isolated object or a very large title suggests a way in. We may expect a mystery, an intimate story or an explanation. The cover starts working while the text remains closed.
To discuss this anticipation I prepared three proposals for a book that does not exist. Its title is The Last Box. Its narrative material is only the short passage below, written for the comparison: no published edition, no author or reader opinions to attribute to the proposals.
Keeping one text fixed
Here is the passage: “When the warehouse was empty, one unnumbered box remained. The caretaker left it beside the door. The next morning it was still there, but someone had moved the chair to make room for it.”
It does not tell us what the box contains. It does leave a relationship between an object, an absent presence and a small decision. Three visual directions follow: a slit in darkness, an ordered arrangement of volumes and an everyday object in a warm field. The title stays the same in the typographic comparison.
The same title, three entrances
Imaginary book: title and fragment written for this comparison.
The first proposal may encourage a search for a threat. The second may suggest an inventory, perhaps a text about space and order. The third may allow a domestic closeness to emerge. These are design interpretations, rather than findings collected in a reader test.
The expectation that returns to the words
Rereading the passage after each proposal may give some details a different weight. The unnumbered box may become an unsettling anomaly, an administrative gap or a small object that has been given room. The text does not change, but the question with which we enter it may.
A cover proposes a hierarchy before the page has time to build its own. I therefore consider its design part of reading, rather than a wrapping to settle at the end. The visual direction needs to return to the text, asking which passages it helps us encounter and which it makes harder to hear.
Exaggeration may be intentional. A cover can create friction, contradict a first impression or open up irony. But the friction needs a relationship with the work. If I promise danger the text never intends to approach, I may have built only an effective invitation to a different room.

The title between mark and sentence
The sample’s title is real text set in the document, readable without interpreting lettering generated inside an image. The two AI illustrations instead accompany the relationship between a closed object and expectation. They depict imaginary books, rather than photographs of editions or reproductions of existing covers.
The relationship between title and image changes the promise too. An explicit title can restrain an ambiguous figure; a strongly narrative image can prevent a few words from remaining open. If both tell everything, the reader may no longer find room to formulate a question of their own.
The comparison concerns the complete proposals: colour, composition and figure all change. It cannot attribute a possible preference to one factor. That would require another trial, with narrower variations and a disclosed method.
A promise the text can keep
A cover also has to work in the form in which it will be encountered. A thumbnail may preserve a patch of colour and lose a figure; a printed sheet may show a detail that disappears on a digital shelf. The choice therefore requires returning to actual sizes and supports, beyond the attractive large view of a proposal.
For the imaginary book I would choose the direction that keeps the chair’s small displacement recognisable without immediately turning it into an omen. This is an editorial position on this passage, rather than a rule about genre or a sales forecast.
A successful cover leaves the text free to become more complex than its promise. It should offer an entrance while allowing the reader to find something in the pages that the surface could not already contain.
