Ownership and maintenance of ETL rules, as they are documented in e.g. Excel or Airtables, and as they are actually implemented as well as understanding the results they give when applied on actual, often imperfect, data – is a challenging task.

Hard learned lesson: Always go to the data - never trust yours, or others, assumptions.

@jervenbolleman Good point 🙂When it comes to the core entitors in the domain I am working in: Sharing and reuse of clinical trial data, it is me who developed the logic to publish the flat datasets centerns around Study, Study Consents, Data Requests etc.

3NF tables, key value pairs, triples / graphs are all powerful models. However, when consuming data I must admit I ❤️ flat datasets with denormalised, pivoted and subsetted data centered around domain/master entities with URIs.