i continue to think you're focusing on the wrong metric: nodes.
if tx fees go to $100 with 1MB, how many 3rd world ppl can afford that? answer: none. and when fees go that high consistently, they won't even bother with Bitcoin.
what good is it then to run nodes when Metcalfe's Law squares with user growth?
Users is the wrong metric too. Paypal has an enormous number of users. Conventional banking has even more.
Bitcoin has to remain (or arguably return to a state of being) decentralized in a meaningful way and
also get users, or it is just a waste of time. The metric has to be both.
Peter-R posted a while back an correlated graph between transactions (used to infer number of users as end point as that information is not avalanche). The correlation mapped to Metcalfe's law well.
So I think it fair to infer it's actually number of transaction that is the metric and that's a function of users and nodes respectively.