`SELECT *` fetches more data than you need (more I/O and memory) and couples code to schema changes (adding a column can change results, payload size, or expose sensitive fields). Selecting only needed columns is clearer and can enable better index-only plans.