I favor the subscription model because it changes the paradigm from an RPG as a product to an RPG as a service - I much prefer the latter, and I think most people would if they understood how a fully realized tabletop RPG-as-a-service would look. A product is static. There is little (some, but not much) incentive for a publisher to continually refine or improve a product, but a service model requires the publisher to maintain relevancy over the course of the game's projected lifecycle. It also - eventually - is the only reasonable solution to the edition treadmill issue: a game as a service can go through iterative updates that allow it to stay competitive and compelling without ever having to invalidate any part of itself. For an example of how this works (keeping in mind that the fact that these examples are computer games is largely unimportant; what's important is that they are games as a service, and it just so happens that currently the only way to do that is with a digital product), see games like EVE Online, or World of Warcraft. An RPG as a service also provides the publisher with a much more stable stream of revenue, which allows them the flexibility to try things that would not be considered conservative, smart moves in a product-based environment. Paizo benefits from this model - their subscription base is vital to the way they like to do business.