[nrubin]

Quick question about one of the algorithms this article proposes:

>Instead, we compute a seed by combining a server_seed (which is provided in hashed form to the client before the deal - in the same way as the initial_shuffle is) and a client_seed.

What if the server picked a large value and this appending caused an overflow? Don't some languages/runtimes turn an overflow into 0? Then the server could always seed the RNG the same way.

I'm probably very wrong, can someone tell me why?