A spreadsheet (with undo) is pretty much exactly a bitemporal database but it only shows you the "latest" version. You can undo to go back in time but it is not

A spreadsheet (with undo) is pretty much exactly a bitemporal database but it only shows you the "latest" version. You can undo to go back in time but it is not very convenient. THere are also DAGs and the code is data. Building this at scale is what a lot of large orgs end up trying to do. Tricky bit is probably the trade of between lazy eval (pull?) and streaming push through the calcs. I think push is needed for low latency but am not 100%.