Right, but in the example they were talking about block sizes over 1MB, which as I understand it is a hard limit.
The paper is specifically talking about raising the block size limit with a hard fork. They are going with the argument that miners might be opposed to such a hard fork and users not. So the solution to that is for miners to implement a soft limit after such a hard fork goes through.
Also, is there anything to stop everyone from connecting to just one "master-node" on the network, thus making all transactions just one hop to anyone else? I presume this master-node would need enough BTC to handle the liquidity of everyone, but that is theoretically possible.
No, nothing stops people from doing that. But likewise, nothing is stopping people from bypassing that "master-node" and directly connecting to the person they wish to transact with.
Do you know if there's a more recent version of the Lightning Network white paper? The latest I've found is version 0.5.9.2 dated January 14, 2016.
There are not versions of the paper itself, but there is a much more detailed specification of the lightning network available here:
https://github.com/lightningnetwork/lightning-rfc/.That specification is what implementors are using to actually implement software for the Lightning Network.