Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> syncing what page you've read up to in a book (a feature many people find crucial!), cannot really be divorced from having the raw data to create server-side metrics about people's reading habits.

Sure it can. You encrypt the data so the client can read it and the server can't. When you add a new device, you e.g. scan a QR code on your old device so that it can add the decryption key to the new device, and the server never knows what it is.



That scenario is addressed in the next few paragraphs.


The next few paragraphs were not originally in the post. And you obviously get less information from knowing the user has updated some encrypted data than having the specific page of the specific book. It is also not inherently necessary for the server to have even that information; it could be sent directly from one device to the other(s) or via an independent third party relay (e.g. Tor).


This. Distributed consensus is easy and cheap to implement when all participants can be trusted (as opposed to the Byzantine problem that I.e. blockchain tries to solve). It’s even easier if you’re ok with approximately eventual consistency with a reasonably high probability of convergence. Page state seems like it fits into this category




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: