Oh interesting… I have the entire SEC database to index (only indexing the text in a sliver of it at the moment) and use tsvector and I could use something faster.