A few more thoughts on the matter...

Internet has multiple levels of bandwidth (from home-based ADSL copper lines to fiber-optics under the ocean). The idea of Bitcoin (as a decentralized consensus system with open-source protocol rules) can therefore exist on any such level, but arguably only one at a time!

As systems like these cannot be established centrally and instead emerge and grow slowly as grass roots movements, they cannot begin to occupy a higher level of bandwidth before saturating the previous one. In principle, however, nothing should stop the idea to keep propagating itself onto the higher and higher levels (with different degrees of decentralization though).

As recent discussions show, Bitcoin seems to have saturated the level of home-based internet infrastructure and is being asked the question: "move up or stay?". That's a tough one to answer, because competitors are nowhere close and there is no one to ask an advise from. It's an uncharted territory.

What if consensus rules can be protected by means other than a sheer quantity of full nodes? What if we can bake them into Bitcoin's brand itself, with a pre-defined (4-year) schedule to revise and adjust them? We can then have multiple large full nodes in the system (like various block explorers today) which would provide a web interface for studying and inspecting every aspect of the blockchain without the necessity to run it at home. Any deviation from the consensus will be quickly noticed and the public will be alerted about the nodes and services that decided to deviate from it. The system would still remain perfectly transparent.

Now, if Bitcoin decides to increase its block size cap (and it seems like many in the community want to do it in one way or another), I still wouldn't aim higher than 8MB (or 2-4-8 if that fits better) just to make sure that transition onto the next level is smooth and clean (even if that means that home-based nodes will eventually be pushed out). This approach still requires active (big) players to regularly and consistently agree on the protocol rules and have enough patience and respect for each other to freeze and bake those rules into Bitcoin's brand for at least a period of 4 years (after which they will be revised again).

If Bitcoin decides to stay (at home), it will allow competitors to catch up (bandwidth-wise) and eventually one of them will move forward.