The problem in question is really a problem with the subscription model that charges you a flat rate regardless of usage. The ostensible selling point is that what you pay does not depend on how much you use, and that you can use as much of a service as you want, such that the price is significantly below what you'd pay if you were using the service maximally. This is similar to dial-up internet in the old days, where you used to be charged per hour, before flat rates became common.
But here's the rub. While it is true that any given person in practice might benefit from such a pricing scheme, it isn't true that everyone could in principle benefit from such a pricing scheme. The model depends intrinsically on uneven and non-maximal usage of the service, which is what the aforementioned subsidizing is doing. For it to work, it requires that a large share of people overpay for the service, where overpaying means paying more than the value of what you receive (by definition, if you don't use a service, then you are overpaying). If everyone were using the service maximally, there would be no difference between paying for how much you use and paying a flat rate, because you can be sure that the service provider would raise prices to cover costs and reap profits.
So no one is doing you any favors here. Poor utilization is not just an "oops" on the part of the subscriber, but an essential feature of the business model. If everyone was being "personally responsible", the business model simply wouldn't work. And because this isn't a charity, the idea of being happy about subsidizing others users is kind of weird.
So two natural questions to ask are:
1. What could a pay-as-you-go model look like in such cases? Could it cover the expenses of the service? Arguably, no. Those who don't use much would probably continue not to use much. Those who do might reduce consumption, because now you must pay for what you use.
2. Is there a morally sound justification for paying at least a base rate for an unused or underutilized utility to keep it afloat (perhaps charging additionally according to usage)? Putting aside all utilitarian arguments, which I take to be unacceptable, we can find a number of cases that seem to operate similarly that we do not appear to object to. Take the salary, for example. One is not payed strictly according to the value one provides, though you could argue that salary is, ideally, a method of paying for the value provided in a diffuse way (value provided previous year reflected in the following year, esp. bonuses, or upfront payment with expectation of value). So salary doesn't seem quite the same. What about the fact that you can use the service on demand? This is like having a driver who gets payed for being on-call. It seems like this may be a good analogue to begin with to try to grasp the ethical and economic reality of the subscription model in question.