I just read this article on Lost Garden. I'm sort of in agreement, except that I'm mostly not.
While I like the overall idea of ditching some of our arbitrary constraints and crufty assumptions, I don't think that the proposition he actually writes is a good one. If you haven't read it, the article can be summarized as being a giant endorsement of A) multiplayer, B) rules to let players express themselves, and C) integrating games into daily life.
I agree with most of these in almost all situations (only (C) is somewhat iffy), but I disagree with his examples. For example, he specifically says that we can "dodge" the hard problem of highly interactive/adaptive NPCs by having other players instead. This only works in certain situations. It is inherently an extremely low-immersion solution, and is almost always implemented in a way that damages any one player's ability to express himself in the game. It very often (although not absolutely always) produces hierarchical player gradients, with certain players being extremely powerful and dominant and others being unable to accomplish anything aside from following in their shadows.
There are theoretically ways to mitigate these problems, but that produces other problems. It is just as easy to dodge the "NPC problem" in other ways: they're all hard problems whose "solutions" have complex repercussions.
To make it clear what I mean, look at any MMORPG. While a MMORPG is perhaps the multiplayer game type, you'll find the majority of player-player interactions are deeply non-immersive. Some of the talk is about game rules (rather than character interaction), and the rest of the talk is blather about the player's personal life and preferences. On the other side of the spectrum, hardcore RPers might be much more immersive, but the high-end game requires you to go meta in order to succeed. There's too many rules and details that need to be accounted for in order to win a raid. Most of those details would be invisible to the character, and the rest don't contribute much to the character's personality or experience, unless you consider "mind bogglingly tedious bookkeeping" to be the experience. In which case you should work and at least get paid for it.
An NPC will be much less adaptable, but there are many advantages to relying on them even so. They have no problem staying in character, they can be scripted to further the story/mood/arc without deviating, and they're always available to act specifically when the player wants to. The limits can be disguised somewhat through clever UI design and scripting.
Both approaches have their issues, both approaches have very wide repercussions, brutal tradeoffs. More than that, I don't mean to sound like a cheerleader for NPCs. Simply put, it is important to realize that there are no easy solutions. As a designer, you must always realize what the issues are about your approach, and what sort of design it necessitates and allows.
NPCs vs players are only one facet. I use them because they're first in the essay. I could write more on the various other topics - such as integrating into daily life - in the same way. But I won't. This essay is long enough.
Please keep in mind that there are no easy design solutions, no perfect template that can be modified. Everything you do will have repercussions, and everything you do will have hard problems associated with it. Your design needs to mesh and tick along well, which often dictates which approaches work well together: the various aspects need to mask each other's issues, support each other's strengths.
For example, his three big suggestions mesh very well together. Each aspect plays to the strength of another aspect. Which is presumably why he focuses on them. But you could make an equally interesting game out of the very opposite assumptions: it all depends on what your design goals are.