Why PDFs read so badly on a Kindle

Load a PDF onto a Kindle and the result is bad enough that it feels like a bug. It isn’t. It falls straight out of what a PDF is.

A PDF is a description of a printed page

The format exists so a document looks identical everywhere: same layout, same line breaks, same page breaks, on any screen and any printer. It manages that by recording exactly where every single character sits on the page.

Brilliant for print. It also means a PDF has no concept of a paragraph, a chapter, or even a word in the sense you’d recognise. Just characters at coordinates. Any structure you see when you look at one is something your eyes are supplying.

A Kindle wants the opposite

An ebook is reflowable. It stores “this is a paragraph, this is a heading, this is a chapter” and leaves the device to work out where lines and pages fall. That’s what lets you bump the font size, switch to a bigger screen or rotate the thing and have the text rearrange itself sensibly.

Give a Kindle a fixed page and it has exactly one move: shrink the page until it fits. On a six-inch display an A4 page drops to roughly 40% of its intended size, which puts the type somewhere around four points. Hence the pinching, the panning, and the landscape mode that shows you two paragraphs at a time.

What conversion actually does

Converting a PDF to EPUB isn’t a translation between two similar things. It’s reconstruction: taking the flattened output and working out the structure that got thrown away when the PDF was made.

Which means deciding that these characters form a line, that those lines form a paragraph, that this line is bigger and shorter than its neighbours so it’s probably a heading, that the text at the top of every page is a running head rather than prose, and that this word ending in a hyphen carries on below.

None of that is in the file. Every bit of it is a guess.

Which is why results vary

On ordinary prose the guessing is reliable. Novels, reports, papers and manuals have consistent, unfussy layouts, and the converted EPUB reads much like a real ebook.

On documents where the layout carries meaning, it gets much harder and sometimes pointless. Tables, multi-column magazines, mathematics, sheet music: there’s no correct reflowable version of a page whose arrangement is the information.

A rule of thumb: if you could read the thing aloud start to finish and it would still make sense, it’ll convert well. If it depends on where things sit on the page, keep the PDF and read it in landscape.

The best conversion is the one you avoid

If the document exists in any other format, start there. EPUB, HTML, Word and LaTeX all record structure explicitly, so a converter reading those has nothing to guess at.

Convert from PDF when the PDF is all you’ve got. Which, let’s be honest, it usually is.