splitting the chain has never been an option by those wanting a to upgrade bitcoins blockchain capacity.
the option is >95% mutual consensus to activate, meaning >95% stay on a single chain and the other <5% are left no longer being a node if they choose not to be part of the network. if 95% choose to go for it, it activates meaning by the time its actually running more than 95% of nodes are happily ready. and by more, i mean the other 5% have had time to make their decision too
it is however core that refused to be part of a consensus. and they have been very public to suggest other implementations should not use consensus and split off if they dont want to follow core. (not the other way round).
What you are describing is what
I and others call a bilaterial hardfork-- where both sides reject the other.
I tried to convince the authors of BIP101 to make their proposal bilateral by requiring the sign bit be set in the version in their blocks (existing nodes require it to be unset). Sadly, the proposals authors were aggressively against this.
The ethereum hardfork was bilateral, probably the only thing they did right--
ethereum was an intentional split "--oopose-dao-fork" which gmaxwell (loves inventing buzzwords) calls a bilateral fork.
again no consensus upgrade intends to bilaterally fork.. apart from gmaxwell & his colleagues that dont believe in diverse open consensus.
core have been hard headed and trying to set themselves as the centralized leader of the bitcoin network. trying to push anything not core away from the bitcoin network and into the land of altcoins..
totally going against bitcoins consensus ethos.
they have used fake rhetoric such as
Meanwhile hardforks are almost certain to split the chain even if just accidentally. They force users to upgrade at a particular time, and there is no easy upgrade:
the stupidity of these rhetoric's is that there is no initial timescale for when 95% should be achieved when using consensus. AND it only upgrades IF 95% is achieved.
meaning people choose the consensus option in their own time, even years before and happily run their implementations before activation to do their tests and just use bitcoin happily. knowing it only activates when 95% of other people also agree.
and after activation only 5% need to decide if they want to be a full node.
rather than the other way around that cores softfork are proposing, where 20 pools make a decision and to activate first then, the nodes move over AFTERWARDS.
cores very own features DONT use network consensus. their features bypass the requirement of nodes to consent or to veto the change, where the change is blindly passed to the nodes. making them less than a full node. thus causing nodes to move to core features if they want to become a full node AFTER THE FACT.(no choice given)
laughingly other implementations who want a consensus upgrade have had implementations running live for over a year.
yet cores own features are not even at the point of flagging desire, but want it activated by christmas, which goes against the same core devs fake rhetoric of
They force users to upgrade at a particular time, and there is no easy upgrade:
all consensus upgrades are options to upgrade a single network of nodes (single chain of data). yet cores features are 3 different sets of data.
think of it like triplets. same DNA (blockheaders) but different lives
pruned nodes(not entire history)
legacy(entire history/nowitness)
segwit(entire historywitness)
which actually causes the 5500 nodes to not be less intercommunicatable with each other.
EG
a new user wanting to sync the entire history cannot connect to a prunned node.
a new user wanting to fully validate cannot connect to a legacy node because legacy wont have the signatures.
thus we will not have 5500 'full nodes' all with the same data and all helping each other but 3 clusters of mismatched nodes.
also there is a 4th level of node, branded fibre. that just relays data uncheckedin short when segwit activates there will be less than 1400 full history, full validation, fully interconnect-able nodes on the network
but hey.. ignore security, ignore decentralization, ignore consensus, ignore bug risks.. and "trust" just a few men and their 90 unpaid spell checking interns,
who want to separate the nodes into clusters.
who want to make legacy nodes not be full nodes in under 2 months.
the hypocrisy of core has been too loud yet many people ignore the technicals and prefer blind trust.