Comment by grep_it

> More complex examples, like IN (1, 2, 3) or id != 25 can be handled as well. For the former, we can hash all values and route the query to the matching shards. For the latter, we can do the opposite.

This makes it sound like the query would be routed to all shards except the one matching HASH(25). But wouldn't id != 25 need to go to all shards, since there could be values in the id set that hash to the same shard as 25?