Simplify Diff for Substantial Changes
For some changes in title and abstract, there are spurious similarities that make the diffs difficult to understand (for example https://twitter.com/nyt_diff/status/1353771665017221120).
Would a pull request that detected if, say, 30% or less of the original text was unchanged and replaced it with "-old-headline- new headline" be accepted?
Hey, thanks for asking and the suggestion. My fear with that would be that with some long titles and specially with long abstracts the image would be even higher than what they currently are and make them harder to follow. But the main issue I have is that I don't want two different kind of outputs for the bot. Currently the bot has alt-text for the images that was added to address #3, that has the
Before: xyz After: xyx
format. But I know that most people don't see the alt-text with the default browser configs. Maybe an alternative could be a companion bot that answers to each diff with the text versions of old and new.
I'm not sure what you mean by "two different kinds of outputs for the bot." Wouldn't it be the same format, but just have the entire old headline removed and the entire new one added?
Also I understand how making the images taller might cut them off in previews, but I'm not sure I understand why that would make it "harder to follow."
I don't mean to be rude or insistent; apologies if I come off that way.
By "two different kinds of outputs" I mean two styles of diffs, one with the changes inline as we currently have for the cases where most of the original text is unchanged and another with the complete old title and the new title when the texts are very different but still have some word in common. I played around with the script using the headline in the link you sent and with the abstract of that same thread to test how they would look:




I agree that they are easier to read and the difference in height isn't considerable, but I'm still not convinced of having the two styles. And the inline format is important to notice the small changes, that was the original purpose of this bots. I used to have a bot that just posted the main headline of a news site, sometimes they would make a small change (typos most of the time) and I would have to read the tweets several times until I would find the difference, eventually I remembered about the newsdiff project and decided to create this diff bots.