Both sides - "big blockers" and "segwitters" - have valid arguments in my opinion. It's all about a tradeoff between decentralization and centralization in two different areas:
- the Core plan would mean a more centralized payment system (becaues it realies on off-chain solutions), but a more decentralized full-node structure (because everyone could run a full node without even have to prune)
- while the Unlimited plan would maintain the decentralized payment system (on-chain payments for almost all use cases), but with a more centralized full-node structure (most nodes will have to be pruned/light nodes and full nodes will be limited to industrialized areas with good connections and datacenter users).

It's well possible that in well-connected areas of industrialized countries and urban areas upstream bandwith already is high enough for, let's say, 15-20 MB blocks (~1 TB/year). Now even in a highly developed country like Germany there are still "black holes" without >1 Mbit internet connections. In the developing world, naturally these black holes are much larger (almost the entire rural areas/small cities).

So the goal should be to find an agreement (Satoshi Roundtable Consensus 2?) on block size that makes possible for nearly everyone with a full-sized computer to run a full node, but not on a cheap smartphone, that would be too limited even today. And for the "payment system" my proposal would be to find a block size that should be able to handle all mid-to-large payments (let's say over 20-30 USD) on-chain but also facilitate off-chain payments with solutions like LN for real microtransactions (coffee, mp3's ...).

Edit: And for the future, the technologic progress - data like bandwidth growth and storage capacity - should be used to determine block size.