Compare two texts

How to read the comparison result

Lines on a red background, marked with a − sign, appear only in the first text: they were deleted or edited. Lines on a green background, marked with a + sign, exist only in the second: they are additions, or the new version of a line that changed. Lines with no highlight are identical in both texts. The comparison uses the "longest common subsequence" alignment, the same principle behind the diff tools programmers rely on. That means if you insert a line halfway through a document, everything after it isn't flagged as different: the remaining lines are re-aligned correctly. It is computed with Myers' algorithm, the same one git uses: once the identical beginning and end are set aside, the work grows with the differences rather than with the length, so two long and similar versions are compared in a moment. If instead the texts are long and very different (for example the same lines in a completely different order), the comparison settles for a good alignment instead of the best one so you don't have to wait, and it tells you so.

When comparing two texts comes in handy

Classic use cases: two revisions of a contract or a thesis, the draft and the edited version of an article, two lists you need to reconcile (guest lists, inventories, email lists), or two configuration files. The comparison works line by line, so it shines with text that has frequent line breaks. Each box takes up to 100,000 lines, more than enough for most documents. If the second text is a chat's answer that rewrote the first, look at What the AI changed.