Hm, the word “likely” is doing a lot of work there. If anything local storage of this stuff is encouraging, it suggests at least the
possibility that this isn’t all living in the cloud. But it’s being interpreted as a negative with an unsubstantiated assumption about how the data is being used.
I’d also like to think we could have a better discussion on HN than “big number scary”. 122 tables sounds like a lot, sure. They could denormalise the whole dataset and keep it in one table, key/value store style. Would that be better? It’s a photo app with facial recognition. Stands to reason that it needs to store facial recognition data.