Miners can't change Bitcoin without the consent of full nodes, who run the core client. That's why SegWit2x didn't happen three years ago. All the miners supported it, because bigger blocks means bigger transaction fees. Full nodes and core devs didn't support it, because big blocks means you won't be able to run Bitcoin on PCs much longer and therefore centralized online service providers will control it.
FWIW bigger blocks actually mean much much smaller transaction fees, both in theory and practice. Though it's true some miners erroneously believed arguments otherwise.
BCash, for example, some loopy "large block" clone of Bitcoin has only rarely had more than $1 in fees in blocks-- and not primarily because it has fewer transactions (though it does), but because there is no cause to pay more than negligible amounts when both capacity isn't scarce and mining isn't completely centralized.
The result is that the BitcoinUnlimited (one of the BCash clients) "chief scientist" proposes that security must be paid for by perpetual inflation... and they've been moving towards abandoning Bitcoin's decentralized consensus-- e.g. bcash clients will no longer reorg so if there is a split deeper than a few blocks (caused by a network partition or a block race), their nodes will just split off into separate networks. Quite a mess.
>>FWIW bigger blocks actually mean much much smaller transaction fees, both in theory and practice.
Ethereum has larger transaction fees than Bitcoin right now, on a larger volume of transactions, so in practice, that is not true.
The addressable market will necessarily shrink with too small blocks.
Now of course when Bitcoin Cash launches, with most of the world unaware of it, it's not going to get as much as adoption, and by extension transaction fees, as Bitcoin, irrespective of its block size limit, so your comparison is inappropriate.
Ethereum has a what is effectively a block size limit, which is limiting usage to pretty similarly to Bitcoin's and it is operating hard against it. ('volume' of transactions is not particularly comparable because the average bitcoin transaction now makes multiple payments in a single txn).
Back in the past the issuer of Ethereum stated publicly that online money should not cost more than 5cts/txn... and yet here we are with $50 transactions there.
BCash has not only fewer transaction in spite of tens of millions spent promoting it, but it has vastly lower fees per transaction. If its blocks were half full at 10x the current feerate (and many times larger than bitcoin blocks), its fee income would still be a tiny fraction of Bitcoin's-- and nowhere near enough to support even the extremely low level of security it currently has.
That is simply not true. You can even look to the public markets, there were futures on each of these fork attempts and they traded at a tiny fraction of the price of Bitcoin.