Why Technical Books Still Arrive as PDFs

Many computer book PDFs are scans. That part is true. The stronger explanation sits in the page itself and in the publishing chain. Publishers can make EPUB. They still put PDF first for technical books.

The page has to stay fixed

EPUB reflows text to fit the screen. Line length changes. Font size changes. Paragraphs move. For a novel, that helps. For a technical book, it can break the page.

Technical books carry code, formulas, diagrams, pseudocode, and structural figures. Code indentation is syntax. Python uses indentation to define blocks. In C, brace levels define logic. Algorithm pseudocode depends on alignment. When an EPUB reader handles pre and code tags, tabs and spaces often disappear. Once indentation goes, the code fragment loses structure, and the reader loses the thread.

PDF keeps the original indentation and line breaks on every device. Formulas work the same way. Integral limits, fraction alignment, matrices, and equation systems carry meaning through position. Reflow can make a formula look awkward. It can also make it ambiguous. Technical books trade flexible layout for stable layout. That is a choice between easy reading and exact reading.

The workflow already ends in PDF

Many technical authors write in LaTeX. LaTeX is a markup-based typesetting system. Plain-text source files compile into a precisely typeset PDF. If the workflow already produces PDF, a separate EPUB becomes extra work.

Publishing workflows also end in PDF. Adobe InDesign exports a finished product as PDF. EPUB adds a processing chain, validation, and cost. Publishing professionals in Italy have noted that textbook publishers still mainly offer PDF e-books because the production line outputs PDF. Scanned PDFs are not the whole story. Native digital editions often start with PDF too.

Copyright and control push toward PDF

Technical books have narrow audiences and high production costs. Publishers watch electronic editions closely. PDF describes the final page, not the content structure. Its text stream follows page coordinates. Change one word, and a paragraph may collapse. Extract text, and the copy may come out garbled or out of order. Change font size or line spacing, and you are rebuilding the page.

EPUB is ZIP plus HTML and CSS. Its structure is clear. It is easier to edit, extract, and copy. It sits closer to a Word document.

PDF also makes DRM easier. Publishers can control copying, printing, and expiration. Even when EPUB has DRM, removed restrictions leave text that is easier to copy and distribute. Springer Nature’s e-book licensing policy shows this split: some versions carry no DRM, while others carry strict usage limits.

O’Reilly is the counterexample. It sells PDF, EPUB, and MOBI in DRM-free formats. One purchase gets all formats. O’Reilly, Springer, and Packt began selling e-books early, and their online memberships allow direct reading. Their e-books are well made, with linked tables of contents. “Publishers only make PDF” is false. PDF is the safe default, not the only format.

Chinese and foreign PDF sources also differ. One shared comment says foreign PDFs come from publishers, while Chinese PDFs come from readers. The line is exaggerated, but it marks a gap in origin and quality.

Converting scans to EPUB still needs human proofreading

The main obstacle in scanned PDF to EPUB conversion is OCR, or optical character recognition. Traditional OCR fails often on code and formulas. Recent AI OCR tools have improved.

Calibre is the established e-book manager. It includes PDF to EPUB conversion. It mainly handles text-based PDFs. Its OCR on scanned pages is limited, and code and formulas scramble easily. MinerU is a more convenient option, often paired with ABBYY. ABBYY has high OCR accuracy. In some cases where MinerU fails, ABBYY performs better. In most cases, MinerU wins on simplicity, and AI handles much of the preprocessing.

Conversion is not the finish line. Large amounts of algorithm pseudocode and complex formulas still need review. Two-column text, such as English on the left and German on the right, defeats ABBYY too. Manual processing is the only route. Technical books converted to EPUB usually still need a human proofreader.

PDF is not the only answer. In technical books, it is still the safest default.