Compare Text

The comparison works on lines, using an algorithm called the longest common subsequence. Rather than comparing line 1 to line 1 and line 2 to line 2 — which falls apart the moment a single line is inserted — it finds the longest sequence of lines that appears in both texts in the same order, and treats everything else as an insertion or a deletion.

That is why adding one paragraph at the top does not mark the entire rest of the document as changed. The matching lines are still recognised as matching, just at different positions, which is what the two line-number columns show you.

A changed line appears as a removal followed by an addition. That is not a limitation to work around. It is the honest representation. “This line was edited” is a judgement about intent; “this line is gone and this one is new” is what actually happened to the text.

The algorithm builds a table the size of one text multiplied by the other, so the input is capped at 2,000 lines per side. Beyond that the memory cost in a browser tab stops being reasonable, and the tool says so instead of freezing.

A supplier returns your contract “with a couple of small tweaks”. You want to know exactly what moved.

Press Load an example to see the shape of it. The result shows four kinds of row. The payment terms line appears twice — struck in red at 30 days, added in green at 14 days. That is the tweak: your cash sits with them for two weeks longer.

The termination and governing law lines are unmarked, so they are untouched. The confidentiality line is red-then-green, gaining “for three years”. And there is a green line with no red counterpart at the end: an entirely new arbitration clause that was not in your version at all.

That last one is the reason to run a diff rather than skim. A new final clause is easy to miss when you are reading for changes to text you remember — your eye is checking what you know, not looking for what you do not.

Press Changes only to hide the unchanged lines. On a long document that turns forty pages into a dozen rows. And the whole comparison ran in your browser: a contract under negotiation is precisely the kind of document that should not be pasted into a site that uploads it.

If the two versions came out of PDFs, extract the text from each first, then repair the line breaks — otherwise the diff will report differences that are only line-wrapping.