About this idea, you won't usually think about it until you really need it. Here is an example where you need it:
Assume you have a `table` with primary key `id` and you have `id` value 1 to 10000 stored in that table. When you are going up adds every `id` by 1, you may write 

update `table` set `id`=`id`+1

and this SQL will usually results in SQL error. The reason is simple: `id` is the primay key, and while updating `id`=1 to `id`=2, the duplication happened.

And the solusion is 

update `table` set `id`=`id`+1 order by `id` desc
which means the order of updating can be defined to avoid relative unique-duplication problem.