Actually this might prove the opposite. The intention is of course that his removing of 2000 lines of code while improving behavior is a good thing. Therefore a change in number of lines of code is significant. Therefore you should track lines of code. I track lines of code to get a sense for when there's more lines of code than seem necessary so I can go chopping it away until it becomes manageable.
I do a similar thing on a file by file level. Once a file or single function grows too long, it is time for a refactor.