We have to understand that our problems with advancing manned spaceflight are mainly organizational. And because of that, it's beneficial to focus efforts not on a grand goal, which may not survive the next budgetary fight in Congress, but instead on capabilities which can mostly be used for whatever goal we'll have in mind at any particular moment.
We're well on the way along this idea already. Whatever payload can be used for on LEO, we already have means to get some 20 metric tons there reliably. This, as Heinlein put, is already half-way to anywhere in the Solar system. We also work on manned spacecrafts, which, while useful for getting to ISS and back, can also be used - mostly - for much more distant trips, to near-lunar space and more. We're growing a network of automatic stations, which can provide early reconnaissance for various sky bodies, navigational beacons and relays. A bunch of companies have growing experience with controlled landing and launching of one-stage rocket devices.
Gateway has a good property in that it's another component on the way to the Moon - and elsewhere - which has higher chances to survive another organizational shakeup because it's easier to build Gateway than to mount a sizable manned program to the Moon surface. Gateway is an ISS on steroids - in terms of getting to a much higher point - but not much more in terms of complexity, and we became pretty good with building and maintaining ISS. A smaller ISS in Moon vicinity can be less risky and less expensive than many other projects. As for usefulness - given that Gateway is in place, it's much easier to manage risks for more bold manned and unmanned projects on the Moon. Here I disagree with Griffin - he doesn't seem to recognize organizational challenges which Gateway allows to solve.
Gateway has benefits in easier construction (you don't need to land heavy modules on the Moon), better understanding of international cooperation (e.g., ESA will get experience launching modules towards the Moon), uses other than Moon (for later stages, when we might have some flow of materials from the Moo surface), easier return to Earth (so less risks), better coverage of Moon surface in terms of accessibility etc. IMO, not bad for something which actually has a better chance to get funded to completion than existing alternatives.