In the early years of the MMO industry, from roughly 1997 to 2001, there were a few big MMOs that had active player populations. By the time we started ArenaNet in the summer of 2000, we knew of at least eighty MMOs that were in development. Based on the success of UO and EQ, publishers were reviewing their portfolios and planning to migrate their existing game franchises to the online world, where they believed they could adopt a subscription model and "make bank". Clearly, it did not work out that way. As more MMOs came into the market, two things changed. First, players now had a choice about which game they would play, and as a result their expectations for polish, content quantity, and service increased substantially. Second, and perhaps more telling for the future of the industry, it became clear that the subscription model forced players to choose a single game, rather than playing many different games.
Gamers will no longer buy the argument that every MMO requires a subscription fee to offset server and bandwidth costs. It's not true – you know it, and they know it. Gamers may buy the argument that your MMO requires a subscription fee, if you can tell them what they are getting for their money. This is the legacy of games like Guild Wars, Maple Story, and Silkroad Online, all of which introduced new business models into the MMO genre and were quite successful. The subscription model is still perfectly viable, but the pain threshold is very low now. It's no secret that gamers don't want to pay a subscription fee. If you can convince them that your game offers enough value to justify it, more power to you! But be prepared to defend your decision, often and loudly, and back it up over the lifetime of your game.
Be very aware of the choice you are asking players to make, and the frequency of that choice. In a subscription model you are asking players to make a choice every month, and it is a fairly drastic choice: Stay married, or get divorced? It is certainly the case that if every player decides to stay married every month, you can make more money from each player in the subscription model. But that will rarely be the case, and not something that you should count on. Every month, some percentage of your player base will decide on divorce, and as with marriage in the real word, once you are divorced you rarely get married to the same person again. If you go the subscription route, you'll need to have the confidence that your marriage rate will exceed your divorce rate.
With Guild Wars we ask players to make a choice only one time, and that choice is whether to buy the game, or not to buy the game. While we don't enjoy a recurring revenue stream each month, we do benefit from the fact that most Guild Wars players come back to the game when we release new content, so we are less concerned about players putting the game down for a few months. Players don't have to decide whether to stay married or get divorced, they just have to decide whether they want to play today or not. Beyond the benefit of a lower pain threshold to get into the game, this is the core strength of the Guild Wars business model, and one of the reasons it continues to thrive when many other subscription-based MMOs are struggling.
Innovate with your game play, and innovate with your business model! The two go hand in hand, and are mutually dependent on each other. Decide on your business model first, and then build your game around it. Guild Wars can be successful with its business model because we decided that we would not charge a subscription fee before we wrote the first line of code, and every design and technology decision we made served that purpose. We could never turn Guild Wars into a subscription-based game, just as Turbine could not suddenly decide to eliminate the subscription model for Lord of the Rings Online. If you decide to require players to subscribe to your game, be prepared to build a game that thoroughly justifies it.