My first thought is to provide some means of just putting the file into the repository, and damn the consequences. For example a web upload, the script/program

My first thought is to provide some means of just putting the file into the repository, and damn the consequences. For example a web upload, the script/program on the server uses the current HEAD, puts the file in there, git add, git commit etc. In that case, git won't detect conflicts, but humans might, so some git-savy user can remedy that. A slightly more sophisticated mechanism would assign a branch to each user, and then merge as you see fit.