Increasing the block size does not lead to increased mining centralization.
Interesting. Do you have any proof for this conjecture?
I have explained it in more detail here in the article that I written on the subject:
https://bitcointalk.org/index.php?topic=1164464.0It is not conjecture since my theories are based on the factual observations of how the Bitcoin network functions today. To simplify the argument for you however it is based on the fact that the vast majority of miners do not run full nodes for the purpose of mining. The pools run the full nodes for mining instead, that is why miners are not effected by the increased difficulty of running a full node, because miners do not run full nodes for the purpose of mining. This is why increasing the blocksize does not lead to increased mining centralization.
Not sure I get it. Either way, miners, be it pools or solo miners, have to run full nodes (if they are actually doing their job properly). The centralization pressure here mainly comes not from validation costs, but from propagation costs. And a lot here depends on the network topology (e.g. The Great Firewall), i.e. the limiting factor is latency, not bandwidth. It's been discussed many times here.