I would say that the GNS criteria are still worth observing, if not a be-all, end-all.

One problem with the model has been a confusion over what "simulation" vs. "narrative" means. Are we simulating a story, or are we narrating a story?

For example, many OSR games had a morale mechanic -- when 25%, 50% or more of the bad guys get taken out, there's a random roll to see if the bad guys flee. Fair enough. But the question is, is this morale rule a SIMULATION (that people who take heavy losses would be inclined to run away) or a NARRATIVE (that it makes sense in the story that Bulk & Skull would run away once it's clear they're out-matched).

Many early games were obsessed over "realism." If you read D&D 2e, there's long screeds in it about what the Olympic weightlifters could lift, therefore we made our encumbrance rules this way ... or once somebody survived falling 17,000 feet, therefore our falling damage rules make sense, etc. Many game design decisions were justified about how they were supposed to simulate reality.

Reality-simulation didn't last very long. The World of Darkness games preferred narrative over everything, with players encouraged to make their own backstories, motivations, and histories, going above and beyond any numbers. D&D3 introduced "extra-ordinary" feats -- abilities blatantly labeled as not even remotely realistic. D&D4-5 don't even bother labeling them as such.

For what it's worth, video games also wandered from realism. Grand Theft Auto characters regularly take 20 bullets to the face, get hit by a car, and then wander off to a hot-dog cart to regenerate all their HP. Richard Hillman of EA used the term authenticity instead of realism, to describe the simulation that he thought the users would be expecting, not the simulation that was realistic.* You see this often in video games: of course you just shove another clip in there. Of course, these guns make these precise sounds, whether they're the correct ones or not. Of course rolling on the ground briefly makes you invincible to all harm, etc.

When it comes to authenticity, then are we talking about something that is authentic to a fun GAME, or something that's authentic to a fun SIMULATION? Or are we talking about something that's true to the kinds of stories we would tell... something authentic to the NARRATIVE? Yowsa.

So let's back up a bit. What if we use GNS to refer to how do you resolve a situation in the game. Do you:

• Use a gamble, double-blind, or some other formal mechanic that has risks, reward, and strategy? If you are, then it's a Gamist solution
• Have a meta-solution that resolves everything, using fiat, editorial considerations, and anything else because you want it to go that way? That's a Narrative solution.
• Look at what's already been going on, and use a detailed analysis of the world's fiction to decide what happens next? That's a Simulationist solution

A discussion of GNS is better if there's some definitions of what GNS is not.

• If it doesn't have some kind of strategy, wager, or random element, then it's not a gaming solution. (For example, if you convince your MC that since it's foggy today, your opponent can't see you, that's a simulationist decision. If you spend a hold or story-point to suddenly say it's foggy when nothing in the fiction had said it foggy before, that's a strategic use of a resource, and now it's gamist.)
• If there's no story going on, it's not narrative. (For example, if you have a generic mob of NPCs show up because someone failed a roll to sneak around, that's a gaming thing. If you go through a precise set of calculations to determine how many gang members can show up based on shifts, transport, and other logistics, that simulation. If you have a mob of NPCs who are here because they've been tracking you and they're angry that you stole their last spark plug, and the MC thinks this is the right time to have them show, that's narrativist.)

Naturally, there's some wiggle room here. For example, using story points to change the environment or to produce new elements could be said to be "high-gamist, low-narrative.".

The GNS model is very much keeping in mind when writing PbtA games. PbtA games tend to be gamist first, narrative second, and simulationist last.

• Pbta games are high in gamist qualities. Moves can be described with barely any information, entitling players to actions because they rolled high. ("on 10+, you find an escape route"). Story factors are opened by high rolling. PbtA games are renowned for being "fail forward" -- that is,  you're going to do it anyway, we're just rolling to see what price you have to pay to do it. Relationships between characters will be formally tracked, they will rise and fall based on strategic decisions and random die rolls, and (in many games) they're the major way to increase your character's power, moreso than life experience, better gear, etc.
• PbtA games are medium in narrative qualities. The players make all rolls -- the MC never rolls anything in secret. If the MC wants to have some bad event show up, everyone's aware that it's either the MC's fiat or the result of a bad (gamist) roll ... never the result of some wandering-encounter table or other (simulationist) random event. For a game about a horrid landscape of grim survival, Apocalypse World will let players declare some NPCs to be immune to any badness what-so-ever, no matter what -- a narrative immunity, beyond any dice, tactics, or simulation.
• PbtA games are low in simulation. In Apocalypse World, the actual disaster isn't specified -- players are encouraged to "barf forth" whatever apocalyptica they think is appropriate to the narrative, and not to worry about simulating nuclear strikes, viral plagues, or Hadron colliders. Hunger, thirst, and other resources are abstract game decisions, when they show up at all. (AW2 encourages players to ignore gasoline evaporation, because having gas is more fun.)
  

Vincent's post on design theory** follows this model. If a player has an ability, they should be allowed to use it, in this moment. What's going on right now, the decisions the player wants to make (the game) is considered top priority, moreso than a long, drawn-out discussion over whether it's better for how the history of events would permit this (the narrative) or how something might actually be possible given our logistics (the simulation).

When designing your own PbtA game, it's a good idea to remember this priority list of "game first, narrative second, simulation third". The purpose of GNS is to give vocabulary to these decisions, and to guide you in the right direction ... not as a precise taxonomy or a check-list. If we can keep in mind how GNS applies to design theory, then it's still very useful.

* http://famousaspect.com/what-is-game-design-with-rich-hilleman-part-1/
** http://lumpley.com/index.php/anyway/thread/466

When we first came across the PbtA engine, we were intrigued with the possibilities of the core mechanics. The playbooks-and-moves structure gets players into the hands of wild and crazy characters quickly and easily. We knew it would be a good fit for our "over-powered sci-fi after dark" game.

The key design issue was "over-powered". It's one thing to play a game about real people having real problems, or natural people with some unusual reach in time and space. But what do you do when some characters can lay waste to entire civilizations, and others can barely open a bag of chips? How do you make a game that has one player with Rick and another with Morty? Or Bender and Frye? Or KOS-MOS and Shion? You might be lucky enough to get a group of players who intuitively understand that different characters will have different levels of power... but will the game know that? Will the players work together by working with the game, or working together despite the game?  We knew that this would require some thinking.


__________________________
Fishes out of Water: The Genre

FARFLUNG is inspired by the high-concept sci-fi genre. These stories have a lot of ASYMMETRY — some characters are far more powerful than others. In HITCHHIKER’S GUIDE TO THE GALAXY, Arthur Dent is just a normal human who wanders around in a bathrobe — he’s not a three-armed ski-boxing playboy, a polymath engineer, or an immortal robot. In DOCTOR WHO, Rose Tyler is a day-to-day temp worker — like many companions to a time lord, she doesn’t have unusual skills, psychic powers, or durability. And in the TV series LEXX, Stanley Tweedle may be the only person to command a planet-destroying starship, but he himself is just this human — he’s certainly no undead assassin or seductive death-robot, like his other companions are. And so on.

Arthur, Rose, and Stanley are asymmetrical characters because the other characters CLEARLY have more influence. It’s not merely that they have less combat prowess… these other characters have LESS power overall. Even on the few occasions when they show off their technical skills, their diplomacy, or their knowledge, other characters — even in their own party! — are clearly more capable.

There’s a problem when adapting these stories to the tabletop. If threats are rendered as contests where “you do something to it” by making rolls, any group of players is going to suggest the logical thing: that the player with the best rolls hammer on the thing every turn. And when some characters are more capable than others, that means their players are front and center all the time, making all the rolls, while everyone else watches.*

How do we empower non-confrontational players with weaker characters to participate in group challenges, even when their characters might not have the agency to participate? Let’s lay out the problem, and then, lay out our solution.

_______________________
A Wargame, Interrupted: Conflict in Tabletop Gaming

PbtA games tend towards the confrontational: You declare that you do something against something else, then roll to see if you get your way. High rollers are rewarded: They seize the thing by force, they seduce the thing. While middling rolls may require the player to make some kind of compromise, they still force another character to submit.**

PbtA games also usually focus on single-player actions. Group activities, such as “we all have to pitch in to fix the water purifier” or “we need to make the Cultists of the Rusted Root happy” are largely absent. Some games imply that players should work together; for example, a group event could be assigned a countdown that everyone could work to roll back. At best, a group-countdown has each player pitching their solo actions in to fight one big fire.

PbtA games (like most tabletop games) often have more rules for combat than anything else. Weapons will be assigned distinct, varied harm levels (1, 2, 3, 4, 5, etc.) and, curiously enough, a descriptor (reload, ranged, ap, etc.) This descriptor rule usually isn’t usually used anywhere except combat.  Because combat gets so much space in the book, and has so many rules and resources to manipulate, many players get the message that combat is INTERESTING and therefore must be a LARGE PART of the game.

Combat is often the only group activity where everyone can participate. It’s one of the few times progress is measured in numbers. Enemies have this many HP, and when that HP reaches zero, we win.*** High numbers are the way to win. High numbers with powerful powers, even better.
So let’s tie this back to ASYMMETRY. When all conflict is confrontational,quieter players with non-confrontational characters can get lost. They aren’t invited to roll to defeat challenges – if their numbers are too low, they can’t beat it. In fact, by putting themselves at risk, they may be “feeding the other team” by forcing their team-mates to spend effort pulling them out of trouble.

_______
None of Us is as Strong as All of Us

FARFLUNG had some serious design challenges. Many of the characters are people who are victims of circumstance. Their only major ability is determination and luck, not skill or power. Obviously, seeing those characters overcome adversity is part of what makes the genre interesting. But in the context of having agency in a GAME, it’s hard to argue that death-blasts at +4 aren’t way cooler than throwing rocks at -1.

FARLUNG’s playtest encouraged these solutions:

• Harm has not one kind, but three: DOING (physical exhaustion), FEELING (emotional anxiety) and THINKING (mental stress). A character who is strong against one kind of harm will be weak to other kinds of harm. Even if a character has the best ADDS to deal with a problem, they may stand down and let another player’s character deal with it, if they have better RESISTANCE to the harm that problem causes. (Brutes will do things physically, talkers will handle the feels, and smarties will handle the thinking.)
• Characters begin the game with “Time Points”, which can be used to enable powers and to boost rolls. It’s important that the characters BEGIN with these points, rather that forcing them to earn them through play – especially confrontational play. (You must win rolls to get points to boost rolls… so you can win rolls? That’s not going to work.)
• These time points aren’t destroyed or lost, but cycled. Fx points (for super-powers) and Px points (for extra-ordinary abilities) become Hx points. And Hx points are used to boost OTHER PLAYERS, not yourself. An active character who frequently hogs the spotlight becomes exhausted… and then asks their friends for help, bringing them into the action.
• It’s very important that time points are cycled. If the points were merely spent, players would be encouraged to spend all their points to deal with a problem… and then rest. By cycling points into the Hx category – abilities that only boost other PCs – players must rely on each other.
• The key Hx move is INSPIRATION. One player calls on another for inspiration, and the two players must briefly describe how the inspiration works, possibly in a flashback. The players must describe the moment when they were inspired within the story itself, bringing the fiction and mechanics together.
  The inspiration move doesn’t have to be pre-planned. When there’s too many “carry +1 forward” moves, the players might spend all their time buffing each other before pressing on. Instead, players say they’ll act and then act, knowing they can call  upon help when they need it.
• The inspiration moment doesn’t require the two characters to be in the same place, only that they had a connection at some past point – that is, a flashback to a previous moment. Players will often get separated. FARFLUNG is a sci-fi game, so there’re high-speed vehicles to split them up over great distances, and radios or telepathy so they can stay in contact despite that.Even though the character may be physically absent – and it might make sense that they are, since they might not survive being on the surface of the sun or whatnot – they can still help another character out.

_____
Rules Endorse Behavior

The most important decision of FARFLUNG was transparency. The players shouldn’t have to think about “meta” factors, or how to manipulate the rules for best advantage. Time spent meta-thinking is time spent away from being immersed in the game’s fiction. When all a player has is Hx points, they will act to exchange those Hx points. When a player’s character has high resistance to harm to their Feelings, but low resistance to harm to their Doing, they will choose moves where emotions are at risk – because they don’t care – and they will avoid moves where physical harm is a danger – because they do care. The player doesn’t have to know all those bullet points are in effect… they will have a playbook in front of them, and they will know when to shine, and when to help their fellow players shine, almost by accident.


===================================
* DOCTOR WHO has long been plagued with the problem of passive companions, with a variety of responses.  (http://nerdist.com/doctor-who-for-newbies-the-companions-part-two/) (http://www.warpedfactor.com/2015/01/doctor-who-companion-pieces-ace.html)
** There’s been progress on this front. APOCALYPSE WORLD 1e (2010) has the move, "When you try to seduce or manipulate someone, tell them what you want and roll+hot." AW 2e (2016) has updated the move to read, "When you try to seduce, manipulate, bluff, fast-talk, or lie to someone, tell them what you want them to do, GIVE THEM A REASON [emphasis mine], and roll+hot." Still confrontational, but now it acknowledges the target won’t submit to any request just because you rolled high.
*** Some events might have a countdown, instead… but the only mechanical difference between “everyone takes turns rolling to reduce the enemy’s HP” and “everyone rolls to reduce the countdown” is that you might be using a different number than +fight.