Hey — I'm the Ode2Code assistant. I read code well and I know the site's articles. What are you stuck on?

Classic. Short version: the planner is making different choices about join strategy. With `IN (subquery)`, it often picks a hash semi-join that materializes the entire subquery first. With `EXISTS`, it gets a clearer signal that it can short-circuit on the first match, so it picks a correlated semi-join.

EXPLAIN ANALYZE
SELECT * FROM orders
WHERE customer_id IN (SELECT id FROM customers WHERE country = 'NL');

Run that with `EXPLAIN ANALYZE` and check the `Rows Removed by Filter` line — that's usually where the cost hides. There's a great writeup on this from jmireles you might want to read: "I built a Postgres extension in a weekend". Want me to pull the relevant section?