It is a great idea in theory, but there's more to it than you make out.
Firstly, as people have pointed out, you'd get inundated with reports for completely trivial things.
Your (valid) point of:
the OP changes it to add a bunch of stuff that I no longer think makes it a great answer.
Can also be reversed to what is likely the largest percentage of edit scenarios:
The op changes it to correct a typo, uppercase a word, fix a link, put something in a quote....
This would only be useful if it is a large feature with lots of code and work put in, not just simply on/off - notify me or not.
You'd need all sorts of parameters, such as:
- Total chars changed
- Total words added or removed (or combination etc)
- What was changed - ie wrapped text in quote or code
- Only one notification per edit unless subsequent edits were
substantial as per checks in 1, 2 and 3
- "User can turn on/off feature" - scripts checking: IF a user edits question/answer, who voted, do they have "edit notify" on or off, have they already been notified for this question/answer, if yes was it trivial, is this one trivial, is it worth it, if yes notify them.. many many more things
- Others, many
Someone edits a question that has 200 votes, all those users who voted have to be sourced in DB to check everything as per 5. and then some more too.
And that is only one question of millions with potential to be edited.
Sometimes people edit their question 3-4 times quickly in succession. I try not to, but it happens sometimes and I read a bit of it and spot something, then something else.
How would the scripts for this feature not tell you every time a user noticed a trivial typo?
That could be done with total chars or words changed, but what about when the scripts determine that:
- Edit 1 = pointless - don't notify voter(s);
- Edit 2 = pointless - don't notify voter(s);
- Edit 3 = pointless - don't notify voter(s);
- Edit 4 = pointless - don't notify voter(s);
However, all of those pointless edits collectively would make you vote or change your vote, up or down...
It might sound trivial checking if user edited to (eg) wrap some current text in code or quote only, but sometimes this can make or break an answer being good so +1, or just "ok" so not bother voting.
Even if you don't think so, the next person might...
You have many permutations for checking this and in between - ie is adding 3 blocks of text into code block or quote block worth telling me about? Maybe 10,000 users do, but 20,000 users don't care...
So then people want to be able to fine tune the functionality to suit their requirements, as this would be too much work to implement and maintain, so doesn't get implemented, users just turn the feature off.
For the infinite future people come and ask if they can have XYZ feature, or why they did not get notified for something, or notified for things they don't care about, can the algorithms be explained so they can understand it, no? Ok I'll just turn it off.
Just too messy and too many things to consider and code for, which makes it a large project - much larger than you think and much larger than I've outlined here.