the RPG-ish board game from the 80s called HeroQuest) is the purest example of a narratively-focused tabletop RPG out there that still plays more-or-less like a traditional tabletop RPG (so, excluding GM-less games like Microscope or Fiasco). To be clear, that's not the same thing as the best narratively-focused ttrpg; though I think HQG is very good, what the "best" game is will always depend on circumstances and preferences of the people playing. No, my claim is that HQG does the best job of focusing in on, and equally importantly making explicit, the basic assumptions and concepts of narratively-focused play. HQG is very clear about what it is doing and why it is doing what it is doing, in a way that other games that have similar goals often obscure.
But what do I mean by "narratively-focused"? The core mechanics of HQG were designed by Robin Laws, who I think has to be considered one of the top five or so living ttrpg designers. In addition to HQG and its predecessors (HQG is the 3rd edition of Heroquest, which in turn was derived from the earlier HeroWars, all by Laws), Laws is best known for the core engine running the GUMSHOE series of games, as well as Feng Shui. Both GUMSHOE and Feng Shui are all about genre emulation--mysteries and Hong Kong-style action movies, respectively. HQG, though, goes deeper than that, and explicitly tries to emulate the structure of stories themselves. Key game components function off of "story logic," as opposed to "world logic" or "game logic."
Mechanically, characters in HQG have a list of abilities, with an associated rating. Unlike most other games, there is no set list of abilities, but instead are entirely open-ended and player defined. For example, the sample HQG character "Vargast the Thunderer" has the ability "Lunars Killed my Wife." Much like with the Backgrounds in 13th Age, any time an ability is relevant to the story, it can be used to accomplish some action. So, when faced with a Lunar soldier, Vargast might use his "Lunars Killed my Wife" ability as the basis for attacking the soldier, as his rage and hatred of the Lunars powers his blows. Abilities can be "break outs" of other abilities, acting like a specialization. So, returning to Vargast, he can use the Air Rune to do Air Rune things (fly, fight, be boastful, etc.), but when he uses the "Lightning Spear" break-out of the Air Rune, he gets additional bonuses.
Abilities are rated from 1 to 20, with the additional concept of "masteries" (signified by the "Mastery Rune," which looks like a flat W). If an ability goes over 20, it is converted to a "mastery" and 20 is subtracted from the ability. So, an ability of 27 is expressed as 7W (i.e. "7 and one mastery"); an ability of 48 is expressed as 8W2 ("8 and two masteries). It's a little bit of a code, but you pick it up rather quickly. To use an ability, the GM assigns a difficulty (more on that in a bit) that is rated on the same scale as abilities. Next, masteries are compared and cancel out--if the player ability is 8W2 and the difficulty is 7W, then the player ability becomes 8W and the difficulty becomes 7. Both the GM and the player roll a d20 and try to roll under the numerical component of the ability rating. Rolling under the number is a success, rolling a 1 is a critical; rolling over the number is a failure, rolling a 20 is a fumble. Now, the side with masteries remaining can "bump" their result up by one level per mastery (fumble to failure to success to critical); any remaining "bumps" bump down the opposing roll. You can also spend Hero Points to bump rolls in an emergency. The player and GM results are then read on a simple chart to determine the outcome--anything from Complete Victory to Complete Defeat.
For more dramatic situations, these rules can be expanded into extended contests. Extended contests are a series of rolls where the outcome of a particular roll is converted into points, and the goal for the player is to inflict five points on an opponent before he or she suffers five points. There are also rules for using one ability to augment another prior to a conflict, and some more noodly rules for using Rune abilities, having companions or followers, and a couple of other smaller things. But, basically, that's the whole game.
If it were just that, then HQG would be "rules light" and free form-ish, but otherwise pretty conventional. The secret sauce, though, is in the way the game instructs the GM to set difficulties. Rather than setting difficulties based on some sense of how tough something would be in the "real world," HQG encourages GMs to set the difficulty in a way that makes sense from, and advances, the story and the story-goals the GM has. There are base difficulties that are tied into an approximation of character power, but otherwise there are no fixed difficulties for any particular challenge or opponent. If the encounter is the climax of the story, then it should be harder; if it is early on and the players are just getting into things, you can make it easier.
Now, GMs do this sort of thing all the time. If a GM builds an adventure session around a progressively tougher series of encounters, culminating in a "boss fight," then he or she is already incorporating story-based thinking into the design of the session. But that story-based thinking is camouflaged by, or at least channeled through, a set of "objective" mechanics. To implement the story-based thinking of progressively more challenging encounters, in most games the GM will select monsters or other antagonists based on their pre-determined stats, "slotting" them into the adventure in a way that progressively ratchets up the difficulty. So, you might start with some weak monsters like goblins (or, to be Gloranthan, broo), then follow with ogres, then trolls, culminating with a dragon fight at the end. The GM needs to go in that order to maintain the difficulty curve--if you flipped the dragon and the goblins, you wouldn't have an steadily increasing difficulty progression.
In HQG, by contrast, opponents have no fixed stats, only the difficulties the GM assigns according to whatever narrative arc he or she is trying to establish. In a way, it's just "cutting out the middle man"--you are functionally doing the same thing GMs do in other games, but without the constraints of messing with predetermined stats. If for some reason you wanted the dragon to be the easiest antagonist and the broo the most difficult, in HQG you set difficulties a higher difficulty for the broo and a lower one for the dragon.
If you are the sort of person who thinks that tabletop RPGs are or should be about trying to simulate on some level the way the fictional world would "really work," then all of this probably sounds like dangerous nonsense. Broo are "objectively" less powerful than dragons, and thus should always have lower stats, according to this view. But HQG explicitly disclaims simulating Glorantha according to any sort of "objective" criteria. By not having any fixed stats for anything, including antagonists, you would have a hard time running the game according to some sort of simulationist logic. Removing fixed stats for opponents is to some extent taking away the net that can catch the GM if he or she is tempted to go down a simulationist road.
Instead of fixed stats as a net, HQG gives you the "pass/fail cycle." The basic idea is that the GM should track whether the PCs are succeeding and failing at whatever it is they are doing. If they are on a hot streak of success, then the GM should jack up the difficulty level; if they have failed several times in a row, then the difficulty should be reduced. The end goal is to create a balance of successes and failures to make a compelling story--you don't want your heroes breezing their way through challenges nor constantly falling on their faces.
Again, many GMs do this in an informal way all the time. And HQG goes out of its way to present the pass/fail cycle as truly a net for GMs, something to fall back on if the GM isn't sure what difficulty to set (I seem to remember the pass/fail cycle being more prominent and without the qualifiers in the previous edition of Heroquest, but I don't have it in front of me to check). But there are going to be people that are going to have big problems with this sort of thing. If getting rid of the fixed difficulties tied to some "objective" criteria severely limits the simulationist dimension of HQG, extensive use of the pass/fail cycle arguably undercuts the "gamist" dimension. If a run of success is going to cause the GM to crank up the difficulty until you fail, and a run of failure will reward you with an easy one, then people who like the idea that they are accomplishing things and overcoming challenges through their own moxie are going to feel a little cheated.
But, again, this is a narrative game, and it sells out along that dimension. The fun of the game is in the story that the GM and the players collaboratively create. HQG strips out elements that support other kinds of play. This is why I think HQG is the most pure narrative game, because you basically have no choice as a GM to run it according to anything other than story-logic. If you don't want that sort of play-style, then obviously this game is not for you. It's very clear about what it is doing.
I like narrative-style play, so I have no problem with this. My concern is a little more specific, and it relates to something I've noticed about narratively-focused games generally--they tend to have greater amounts of "genre drift" than the more mechanically-oriented games. In a mechanically-oriented game (like, say, D&D, or Runequest), the game is defined by the mechanical pieces that are pre-made as part of the design of the game. GMs are of course free, and often do, come up with their own classes, monsters, etc., but that is usually an appreciable amount of work and so most of the time you are going to be using the pre-made stuff. And the designers are going to tend to pre-make stuff that fits into the themes and tropes of the setting, so the majority of the stuff that gets used in games is consistent with a particular feel. If you run a game with characters using D&D classes fighting D&D monsters, you are going to end up with a game that feels like the particular high-ish medieval fantasy that is D&D. In a narrative game, there isn't that barrier to making up a bunch of stuff, so there aren't those constraints that keep things within the genre boundaries. One of the things I've noticed about Dungeon World streams is that most of them very quickly leave the confines of Tolkien-esque fantasy and spiral off into some sort of "weird fantasy" direction.
Now, there is nothing inherently wrong with weird fantasy that goes outside of genre boundaries--I love Friends at the Table's Hieron stuff, and it can definitely be described as "weird fantasy." But if the goal is to have a game that emulates a particular genre or captures the feel of a particular setting--and the explicit goal of HQG is to create a Glorantha game--drifting off into some other space should be considered at least a partial failure. HQG requires that players and the GM be, for lack of a better term, "disciplined" about telling Glorantha stories, making Glorantha characters, and playing those characters in a way that is consistent with the feel of Glorantha. And that requires the players and GM to bring to the table a bunch of pre-existing knowledge of Glorantha. For example, the character creation process is really just "pick a clan, pick an occupation, pick three Runes, add some other relevant abilities, a few flaws, and then assign ratings." That's great if the players and the GM know Glorantha and what sort of campaign they want to play, but if you are coming to this fresh there is not much in the way of hand-holds for newbies.
[On the other hand, you could just run Heroquest rules without Glorantha, in whatever setting your group wanted. The 2nd edition of Heroquest was pitched as a generic game system, and I understand Chaosium will soon be releasing a d20-style System Reference Document for the underlying system. But since Heroquest Glorantha is explicitly about Glorantha, I think evaluating it in terms of Glorantha alone is fair.]
By contrast, in the Runequest review I mentioned that the systems and the way they interlock do a great job of creating that Glorantha experience, even if the GM and players don't really have deep familiarity with the world. If you follow along with the family generation system to create your character, that process will give you a bunch of Glorantha hooks and elements for your character, teaching you bits of the setting that you need to know as you go along. The game helps you stay within the lines of the tone of Glorantha. HQG requires the players and GM to do that work themselves.
To be fair, HQG does have a ton of Glorantha information--I would say at least 75% of the book is either pure setting information or a hybrid rules/setting material. But where I wouldn't run Runequest without being confident that the players were going to dig in and learn the rules, I wouldn't run HQG without a similar confidence that the players were going to dig into the setting of Glorantha. If you ran HQG for people with no prior knowledge of Glorantha, I think the game will tend to drift off into something that might very well be very fun, but not really recognizably Glorantha, or otherwise flop around like a fish out of water.
But, if you had a group that was all-in on doing the Glorantha thing and telling Gloranthan stories, and everyone was either was up on Gloranthan lore or willing to put in some reading time to get up to speed, then HQG is a great vehicle for telling those stories. If you are looking for some pointers in that direction, I would be remiss not to mention the "Colymar Campaign" adventure path found in the HQG supplement Sartar: Kingdom of Heroes (reading S:KH, by the way, was when Glorantha first "clicked" for me) and the Eleven Lights Campaign in Eleven Lights I haven't gotten my hands on Eleven Lights yet, but the Colymar Campaign is great and weird in all of the best ways, most notably the big heroquest that is a centerpiece of the story. Or, really, you can repurpose any published Glorantha material--one of the sneaky good features of HQG is that you can run any published module from any system without doing any conversion work by just using the text and ignoring all of the stats. The system is so simple that even the most crunch-phobic group will pick it up quickly, and it brings just as much drama to tense negotiations or interactions as it does to combat or chases (or, more accurately, the same drama, since they use the same system).
I feel like I am becoming predictable by saying this in every review, but not everyone is going to be into the narrative play model that you get from HQG. In many ways, it's the opposite of the play experience that you get from Runequest, and it is very different from the vast majority of games out there, especially outside the indie world. Picking up a tabletop RPG that gives you no stats for any antagonists is weird--the first time I encountered it, I flipped through the book to find the stats that I thought had to be there and couldn't find. But I think it nails a very particular way of approaching tabletop RPGs, and more specifically a tabletop RPG about Glorantha. Heroquest Glorantha is my favorite narratively-focused game, and it is something that I think anyone interested in that style of play should check out.