If I use both random sampling (via Sort & Limit) and deduplication, what will be executed first?

Let's say two duplicates exist in the Target Data Extension. If random sampling was to run first, then some duplicates could be removed by this process without using deduplication rules. 

When using DESelect this is not the case. To make sure the results of a selection are deduplicated properly, DESelect first runs deduplication and then performs random sampling.

Note: to apply random sampling, click the gear icon on the Target Definition screen and choose Sort & Limit.