And Kafka ends up with the infinite-data / sync issue anyways: if an event stream grows unbounded, eventually it will be too big and you'll run into problems, s

And Kafka ends up with the infinite-data / sync issue anyways: if an event stream grows unbounded, eventually it will be too big and you'll run into problems, so you start needing to cull events, which then means that you need to do some kind of batch sync to start new consumers. Or when you need to store state, have the event stream be your entire source of truth and the applications be stateless... but there be dragons.