"Roberts, Jon" <Jon(dot)Roberts(at)asurion(dot)com> writes:
> What if the new query has a significantly lower cost compared to the older
> one?
Much as I'd like the planner to be infallible, it ain't; estimated costs
are no proof of any real-world performance difference. Better show
EXPLAIN ANALYZE numbers if you want to be taken seriously --- and *not*
ones from a Greenplum-modified PG.
In any case "I can make this particular query faster" seems a rather
different argument from "you guys should eliminate all use of correlated
subqueries".
regards, tom lane