That's exactly the point: a hard fork does not just affect miners, and cannot just get decided by miners. All full nodes must have accepted the new rules, or they will be forked off when the hashrate percentage triggers.
Furthermore, 75% is pretty terrible as a switchover point, as it guarantees that old nodes will still see a 25% forked off chain temporarily.
My opinion is that the role of Bitcoin Core maintainers is judging whether consensus for a hard fork exists, and is technically necessary and safe. We don't need a hashpower vote to decide whether a hardfork is accepted or not, we need to be sure that full noded will accept it, and adopt it in time. A hashpower vote can still be used to be sure that miners _also_ agree.
Currently, a large amount of developers and others believe that the treshhold for a hardfork is not reached, especially given the fact that we scale in the short term, as well as make many changes that long-term benefit scalability, with just a softfork (which does not require forking off nodes that don't adopt the new rules, for whatever reason).
--
Pieter
So it seems that everyone is in agreement then, since a hard fork to 2M is orthogonal to SW, and also that core should not be involved in dictating what the consensus rules should be, then why not put BIP102 into play with a 75% mining majority activation and let the market decide.
As it’s activation only involves the miners, and its implementation doesn’t affect users or exchanges, then it poses no complication to SW which can do done in parallel.