The difference is that the altcoin is visible from the start as a clearly separated ecosystem, while in the case of an chainsplit that is not so clear. In every chain split there will be services that were available to the users of the existing chain that limit themselves to only one chain (because of ideology or practical considerations).
I don't understand that argument. After all, the original chain is still functioning.
I think here lies our misunderstanding - in my chain split scenario (the most likely if BIP148 would compete with "Barrycoin") there would be no agreement which one is the original chain. Both the BIP148 client and the "Barrycoin" client have code in it that will eventually become incompatible to unupgraded non-Segwit clients.
How would services "disappear" from the original chain ? [...] Of course, it could be that the service provider makes up his mind to CHANGE his service and not offer it any more to the original chain - but that's an active decision on his part.
If there is no agreement which chain is the "original" one, and there is a new client A (for chain A) and a new client B (for Chain B), then there will be services that will upgrade to client A and others to client B. It is an active decision in both cases, but it's not too different from a regular upgrade of the client. It isn't important if one of the chains is "hard-forking" or only "soft-forking", both are forking.
That is exactly one of the problems of "collectibles" which aren't collectibles, because, exactly, they can be copied over. If a chain had a value-regulation system, then after the storm is over, the value would go back to whatever it was supposed to be regulated to.
My fear is that a "regularly splitting" Bitcoin will become usable only for people with a technically advanced understanding of the ecosystem. Now, with all the volatility, Bitcoin is still usable as a saving/speculation vehicle for common people (a sort of "poor man's stock"). The user has to follow one price feed and can take decisions based on it. Having to follow two feeds is already far too complex for most people.
Also in non-speculative use cases like remittances, then in a "regularly forking" scenario both parties must pay attention to the danger of a chain split all the time. That would mean, de facto, that it couldn't be used for this purpose in an independent/decentralized way by common people but only by specialized companies that hide the hassle for you. That is also a centralization risk.