This is an intriguing suggestion, but the issue you raised about the quantification by line number is also not negligible.

Its certainly worth discussing, so thanks for opening this task as a forum...

On the other side of the spectrum to your nice example, are changes that are simply done over many files/lines with a simple `grep' (on Emacs), like commit bcd59d5b (`%zu instead of %lu for printing size_t'), implementing it didn't take more than a few minutes, or require much thought. But only counting line numbers, it would be considered as a major change (57 files changed, 465 insertions, 465 deletions).

In terms of credit, I would give much more credit to that one-byte example you gave than this commit. Vladimir has made many such great corrections for example, (like this commit, or this one).

Counting commits, we would at least give both these two examples equal credit and also encourage developers to break up the commits into managable fragments as much as possible. But when counting lines, the developers might make a single commit with a large number of changes (sometimes on contextually different things).