Inside my seeders I often use this trick to prevent from re-seeding already present values. For example, if you are creating a base seed of production data -- say a pre-defined list of values -- and add a new item but do not want to delete all seed data and re-seed, you can try this.
Note that, there are some flaws to it. Such as: It will not delete missing values if your new list has changed, or if you set the attributes too specific, you may end up creating a duplicate (for instance, if you have the same item with different `order` values due to rearranging in the list).