The file knows where, not what
A word processor file describes a document: this is a heading, this is a paragraph, this belongs to that list. Change something and everything after it moves, because the file understands what the pieces are.
A PDF describes a page. It is closer to a set of instructions for a printer: put this character at this position, in this font, at this size. Then the next one.
Once you have seen that, most of the frustrating behaviour of PDFs stops being mysterious. There is no paragraph to edit because no paragraph was ever recorded. Nothing reflows because nothing knows what a line is. Delete a word and the gap stays, because the remaining characters are at fixed coordinates and have no reason to move.
Which is why editors and converters guess
A PDF editor works by reconstructing structure that was never in the file. It looks at the positions and infers that these marks are probably a line, those lines are probably a paragraph, that gap is probably a column.
That is inference rather than information, and it explains exactly where these tools succeed and fail. Clean single-column text reconstructs well. Columns, tables, footnotes and anything laid out carefully by hand reconstruct badly, because the arrangement was held in place by coordinates and there is nothing underneath it to recover.
Converting to Word is the same problem with an extra step: the invented structure then has to be expressed in a format that reflows, which is the precise thing the original was built to prevent.
A scan is a different problem entirely
Everything above assumes there is text in the file. A scanned document usually has none at all: it is a photograph of a page, wrapped in a PDF, containing one large image and zero characters.
So search finds nothing, selection selects nothing, and a converter returns an empty document. Nothing is broken. There was never anything there to find, and the fix is OCR, which reads the picture and writes a layer of real text behind it.
And why a black box hides nothing
This is the version of the misunderstanding that does actual damage, and it follows directly from the model above.
Drawing a black rectangle over something adds an instruction: put a filled shape at these coordinates. It does not remove the instruction that puts the text there. That one is still in the file, underneath the shape, completely intact, and it can be selected, copied or pulled straight out with a text extractor.
Governments, law firms and large companies have all published documents this way and been read within hours. Covering is not removing, and the file does not care what is on top of what.
The complaint is the feature
It is worth ending here, because the difficulty is not a defect that somebody failed to fix.
A PDF looks the same on every machine, in every application and on every printer, and it achieves that by specifying positions instead of leaving the layout to be recalculated at the other end. A word processor file, opened on a different computer with different fonts, genuinely can move.
The format bought that certainty by giving up structure. Hard to edit is the bill for looks the same everywhere, and for a contract, a drawing or a certificate that is a trade almost everybody would make.
So what do you actually do with it
If you need the words, PDF to text pulls them out without pretending to keep the layout, and PDF to Word attempts the reconstruction and is honest about what does not survive it. If the file is a scan, OCR is the step that has to happen first.
If you only need to add something rather than change what is there, sign and fill puts text and a signature onto a document nobody made a form out of, which is a far easier job than editing. And if the aim is to remove information, redaction is the page to read before doing anything, because a black box is not one.
Common questions
Why is a PDF so hard to edit?
Because it is not storing a document, it is storing a page. A word processor file holds structure: this is a paragraph, this is a heading, this text is in that list. A PDF holds instructions for putting marks in positions, so there is no paragraph to edit and nothing to reflow when something changes. Editing means finding the individual pieces and moving them, which is why every PDF editor feels like fighting the file rather than working with it.
So how do editors manage it at all?
By reconstructing structure that was never recorded. An editor looks at the positions of the marks and infers that these ones probably form a line, those lines probably form a paragraph, that gap is probably a column break. It is inference rather than information, which is why the result is good on a clean single-column page and falls apart on anything with columns, tables, footnotes or unusual spacing. Nothing is being read wrongly, the answer simply was not in the file.
Why does converting to Word lose the layout?
Same reason, one step further. The conversion has to invent a document structure out of positioned marks and then express it in a format that reflows, and reflowing is exactly what the original was designed to prevent. Anything that was held in place by coordinates rather than by structure, which is most careful layout, has nothing to hold it once it is in a word processor. It is a reconstruction rather than a translation.
Why does search find nothing in some PDFs?
Because there is no text in them at all. A scanned document is a photograph of a page wrapped in a PDF, so it contains one large image and no characters whatsoever. Nothing is missing or broken, there is simply nothing to search. That is a different problem from the one on this page, and the fix is OCR, which reads the picture and adds a layer of real text behind it.
If I draw a black box over something, is it hidden?
No. Drawing a rectangle adds a new instruction to the page saying put a black shape at these coordinates. It is the most damaging version of the misunderstanding. The instruction that puts the original text there is still in the file, underneath, entirely intact. Anybody can select it, copy it, or pull it out with a text extractor, and this has embarrassed governments and law firms repeatedly.
Then why does anybody use PDFs?
Because of exactly the thing being complained about. A PDF looks the same on every machine, in every application, and on every printer, because it specifies positions rather than leaving the layout to be worked out again at the far end. A word processor file can and does move when it is opened somewhere else. The format bought that reliability by giving up structure, and the difficulty of editing is the price of the guarantee.