The initial clone determined what the default branch of the cache git repository was. Changing it on the remote didn't have any impact on the local data. However, cloning it on a different machine would then store a different default branch on that machine. This could lead to different results for the same command on different machines.