/dev/disk/by-uuid/, i.e. managing disks by UUID, is a very neat idea.
Until your hard disk begin to fail (without SMART warnings) on friday afternoon.
Then you decide to copy its complete content onto one spare disk.
Then you begin to have multiple partitions with the same UUID and real problems begin (do you mount root by UUID?).
Note that you can manage a GPT disk where the partition description has space to store the UUID partition, but anyway ext2/3/4fs is also storing the UUID in the superblock and I do not want to guess which UUID (internal to the FS or in the partition table) is being used by each tool, when I try for the third time to emergency save the partition content of the failing disk to the new disk.
At least with filesystem labels you can easily change them.