It's a shame that there doesn't exist a tool where you can say "I want to seed X gigabytes of this dataset" and it automatically selects and seeds the X rarest GB". It would be fantastic for distributing storage for archivists, while the data set could still be edited and added to and downloaded and automatically balanced across the seeders.

Does such a thing exist? Why isn't it more widely used, and everyone still relies on immutable torrents?