Awesome! I did have a few things I was hoping I'd get more information on. I watched the PWL talk on Zanzibar and that gave me a better idea of the available options. 1. For a new product with zero current users, hosting `ory/keto`with a database that supports strong read+write consistency would also make sense, correct? Scalability isn't the biggest factor at the moment. I'm curious what you guys think about the current state of the project? I recall the PWL talk mentioned something about the package developers making puzzling decisions. 2. I saw a paper mention "attribute aware ReBAC". Out of curiosity, is there something inherent about Zanzibar's design that would prevent this or is this more an implementation decision. I agree with @User (from our conversation on reddit) that constraining the operations does force a certain consistency and simplicity in design. The reason I'm curious is because I've read about the difficulties in scaling issues faced with graph databases (think Neo4j). So would Zanzibar + attributes essentially make a highly scalable graph database?