Why do you say that? In the games I've been observing, most City States are conquered after they've already built their walls. As far as I can tell, starting the City States with walls would delay the timing of the conquest of some of them until the AI has Swordsmen or Archers, but most would still go under. It would also have knock on effects in terms of what the AI civs do with their armies during that time (attack each other? nothing?)
I'm genuinely curious about how those who argue for starting the City States with walls think that would improve gameplay. I don't think it would have a significant impact on the number of City States conquered, but it could impact the timing. Would that timing make a big difference?
Pushing the timing back means that you're probably going to more often get AI vs AI battles early, or AI vs human ones. Maybe it has enough time for AI civs to get a couple envoys in a city-state and thus change their mood against it. Or it gives the city-state enough time to get some units online and better defend themselves. It at least gives them a fighting chance, and pushes back when people might want to attack them to a point that you can also maybe do something about it. If you push it back long enough for someone to suzerain them as well, that could have another impact in terms of possible CB too.