I'd think the question is really where the trade-off point is: how many more users can you gain if you're willing to accept the loss-leader of coding those extra amounts? ...because, yes, it's true that "nonpaid users" are a drain on the system in terms of computational and financial issues, but non-paying characters are always going to be the largest group of any online group -- and thus it's likely that they're also contributing the bulk of the content. And that content, in turn, is what prompts other users to want to pay for the nifty aspects of the service as part of getting all that content regularly.

So... I'd say the extra bit of coding is worth the trade-off, if it can be shown that it's a service that would be a substantial boost to users, pay and nonpay alike. That, though, I'm not so certain of. No way to get the stats to gauge it, I mean, which leaves me as undetermined.