There is an argument to use --numeric_ids to prevent accidental mapping of IDs. This would be the case for a general purpose backup system, like in the cloud. You would never know the IDs on the source machine, and any mapping would be purely accidental.
On the other hand, if you're in your own world with a user ID directory that spans the whole network, mapping might be helpful. Except it probably wouldn't matter, because everyone should have the same ID everywhere anyway.
There is an argument to use
--numeric_idsto prevent accidental mapping of IDs. This would be the case for a general purpose backup system, like in the cloud. You would never know the IDs on the source machine, and any mapping would be purely accidental.On the other hand, if you're in your own world with a user ID directory that spans the whole network, mapping might be helpful. Except it probably wouldn't matter, because everyone should have the same ID everywhere anyway.
So I think I should use
--numeric_ids.