Furthermore, credits can come across as a bit 'sleazy' to the consumer. Every person I know assumes that, in general, credit-based systems are implemented to make more money (again, festivals: I just want another beer, not buy another ten credits if I just need one more to get my fucking drink!).
Though, it seems to me that even in the current model what is missing are very strong network effects. It would be interesting to see if adding additional incentives increases/improves sharing. For instance, what if an article becomes free after N friends read it after sharing? It might cause people to share articles that their social circles are very likely to read.
Another interesting feature would be a share-once link. Sometimes you read an article that you want to share with someone specifically. You don't want to be an *ss pointing them to a paywall, so you pay the Euro 0.25 for them :). It's not only a convenience, but also a potential way to bring in new customers.
In addition, if you have tiers of credits (so you have 3 distinct credits, rather than just replacing euro sign with a credit sign), you lose plenty of flexibility.
So yeah, I can see why you rejected this option :)
I think a solution might be buying credits, e.g. you have bronze/silver/gold credits, and each article could be priced at 1 bronze/silver/gold credit (because the price being e.g. 3 credits would keep the overhead). This layer of abstraction removes the user from thinking in terms of money (e.g. this article cost 0.25 euro), a bit.
In terms of monthly subscription, I would assume pricing articles individually makes it easy to deal with the publishers? X people read it at price of $Y, so you get X*$Y minus some fee, can't get any easier. It would get much more tricky if trying to split the monthly cost.