That simply makes two currencies - before-fork, and after-fork. There's no rule that the one that "should" win does, especially in the face of large computing power making the former more attractive. See Bitcoin Cash, versus Bitcoin Classic.
IIRC it didn't work for long (the of promoters of these tokens were likely well aware of this), and ASICs for them appeared anyway, once an ROI probability threshold was reached.
The example you gave has nothing to do with changing the PoW function to counteract ASICs. Both Bcash and bitcoin classic were forked because of completely different reasons, and had people who believed in those reasons defend the new forks and run the node software etc. For example XMR has been doing ASIC-deterrent forks for years now with great success.
Also it seems clear that Monero’s purported “ASIC resistance” is very much a social contract — requiring constant vigilance to uphold — one which isn’t guaranteed to be strictly adhered to over the long run, particularly as the cost/benefit analysis gets murkier.
Just like the rest of the cryptos that require all the participants to use specific protocol and software and agree on specific rules.