3. Afaik the new versioning capability that will be introduced with SegWit, basically hands the power to Blockstream to drive forks any time they want to. Again this is another game theory discussion which I don't wish to elaborate on at this time. Note Monero's periodic forks as a protocol also seems to a form of centralized control. One can argue that miners can vote, but it is fairly impossible to stop the political clout of the lead developers especially when that control is hardwired into the protocol. Satoshi's design was that it should be nearly impossible to hard fork, because that is the only way you truly have decentralization.
Regarding [3], if miners/exchanges/merchants don't agree with the lead developers they can just remain on an old(er) version.
I said it is a complex game theory discussion. Seems you want to force me to have that discussion when I said I am busy on other more urgent priorities. Because I can't just leave your afaics incorrect assumption there unrefuted.
You can't retain an older version, because soon the new features will become intertwined in everything and soon you won't be able to function in the network. The complexity of failure modes with an older version are beyond the scope of the discussion I want to have now. That will be a technical discussion. Just remember what happened the last time I got into a technical discussion with gmaxwell on Ogg streaming format indexing, a topic that he thought he was more expert on (and he should have been since he was the co-inventor of one of the Ogg codecs).