I think your assessment is very accurate. In the world of startups you print shares to reward early employees. You print shares for multiple rounds of funding. You give free samples at a loss to grow network effect.
In the long run all platforms need to be sound: producing more value than they consume.
Where I was wrong is assuming value is a 0-sum game. The reality is that 1+1=4 when it comes to network effect and value creation.
I think it'd be more accurate to say that the network effect is superlinear with respect to value.
That is to say that N(x + by) \neq N(x) + bN(y) -- which is what is required for any functional to be linear, by definition.
In essence, N(x) > O(x). Depending on what the actual value creation is, i.e. features, functionality, etc., will determine the growth rate of network effect for a particular blockchain.