The question is about the use case and how much technical people are.
Excel format is bad because it’s not clear simple text in the raw text format. Also for simple local database with not that much data.
CSV with quirks could be edited with any text editor, put into LLM, shared, GIT history edits. That’s also the benefits of local notes in filesystem in MD format (oobsidian).
Problem with plain CSV is that I can’t do any formatting even the simplest or make simple formulas. I could make calculations using DuckDB or something but it have friction.