I see the LOC numbers come up in fish eye and can't help comparing my productivity against the others. I'm quite familiar with the issues of
LOC, but it's still interesting. There are a few issues that make it
of much more questionable value in our group however. The metric counts too much and blows the numbers out of the water for some of us.

• adding a way to limit LOC to just .java files would help eliminate noisy and very larget data files would help keep checkins of that type of file from making it look like someone is on performance enhancing drugs checking in 100,000 lines over a weekend. I might implement this as a filter that could by directory or file type. You never know when someone is doing something similar, but doing it in java files.

• In a way that may be more valuable to people that don't check in large amounts of test or resource data, having the LOC counter not count the creation of a branch would help as well. If you happen to be the guy who created a branch, you get "points" for all 100k LOC or so. It's a trivial operation that throws metrics way off.