I am not sure exactly how github is set up but there is a simple way of limiting the impact of duplicate files: use hardlinks. `git clone` uses hardlinks when

I am not sure exactly how github is set up but there is a simple way of limiting the impact of duplicate files: use hardlinks. `git clone` uses hardlinks when possible so if the fork is on the same filesystem, it should limit duplication automatically. The caveat here is that `git gc` has to be handled with more care (all references have to be updated). `git clone` also provides a `--reference` for more explicit caching of duplicate objects. This also requires some care during cleanup but otherwise can work pretty well.