The Travels of Mendes Pinto: Team Game
Can You Give Us a Descriptor of the Subject? (Part I)Sergio Mascarenhas
Edited by Stephen Barringer
December 6, 2001
The Travels of Mendes Pinto: Team Game
Can You Give Us a Descriptor of the Subject? (Part I)Sergio Mascarenhas
Edited by Stephen Barringer
December 6, 2001
Editorial Prologue: The Pulsing Guts of Column Creation
As you know, this column is an intimate look at the process of game design, and Sergio has invited the commentary and editing of other minds beside his own, including Steve Darlington and myself, as well as all the comments he responded to in last month's column. In fact, last month the original draft of the column, before Your Humble Copy Editor had at it, was made available along with the final version, as a tangible demonstration of just how much effect (and how little, in many ways) a copy editor has on the final product.
In keeping with this theme, for the next two months we're getting even more down-and-dirty: Sergio and Steve Darlington trade off in a point-counterpoint commentary on the column as it's being written. For convenience, Steve's comments will be inserted throughout the column in (bracketed italics), with Sergio's counter-comments following in [square bracketed italics]. You wanted to see how game design really works? This is it, in all its gory glory.
(I think I'm kinda glad that I'm only the copy editor.)
--Stephen J. Barringer, Toronto, November 2001
I suppose this month I have to speak about descriptors. At least, I said I would last month. And I guess I also have to speak about skills, which I said I'd do two months ago. After all the trouble I went to explaining why the game won't have attributes, I really should look with the same depth at skills (or anything that performs the functions covered by skills in skill-less games). Right? Right?
Well, not really.
You see, September's column served its function: to demonstrate the way I handle things. I guess you got the picture. It'd just be boring to repeat it all again.
So, I'll take a shortcut. I'll present in broad terms the issues I have with the most common descriptor designs and build on it to present my approach.
In any case, my take for this month is how to handle functional descriptors, the ones that represent what the character is able to do. For the sake of simplicity, let's call these descriptors Skills.
The Don'ts of Skill Design
I'll start with my list of "Don't"s when handling a system for skills. These are mostly based on my perceptions as a GM. Several of them will be relevant for players that don't gamemaster, though, and also correspond to my perceptions as a player.
1. Don't reduce it to game system technobabble.
I guess this is implicit in what I was saying in the last two columns. All games need to have a technical vocabulary specific to the game, but that does not necessarily mean that everything must be handled that way. The most annoying thing (to me, anyway) is a game that uses game-specific abbreviations, acronyms, and lingo for everything -- "THACO17; AC?" "-2; roll!" "20: Crit!" "HitLoc?" "Vitals! And the blade has Type X venom!" "Save mod?". And so on and so on.
(There's also a personal issue here: since my mother tongue is Portuguese and the RPGs I own are in foreign languages -- mostly English -- I end up using a foreign vocabulary while speaking in Portuguese, making translation twice as difficult.)
("And yet, one of the most loved. Jargon maketh the fan." -- Steve D.)
This makes it much harder on the player. From my point of view a game should use common language as much as possible to present its concepts and ideas, and to describe how we think about it.
I also consider "technobabble" as the tendency to describe things in terms of quantities. "How big is he? Size 13." What is 13? A number generated by the sum of the results of rolling 3 dice. We can get used to it and be able to translate it into something meaningful ("You see, size ranges from 3 to 18, and 10 or 11 is average, so 13 means that the character is slightly bigger than average."), but the problem is that even this can go wrong if the players have different standards for that translation ("What's 'average'? Around here it's 6 feet ... there it's 1.7 meters ... down here it's 1.5....").
Excessive quantification generates a longer learning process and a degradation of meaning. Yes, it provides its own advantages. Furthermore, if numbers and technical jargon are part of the way we think about the issue at hand, we should stick to it. All I'm saying is that we need a balance.
2. Don't get everything wrong by trying to get everything right.
Think about RuneQuest (or any BRP game). You have skills, right? For instance, broadsword attack, broadsword parry, bastard sword attack, bastard sword parry, and so on. All very neat, very precise. But is this the way we work? Do we really learn different skills for different swords?
Well, they do it in Olympic sword fighting, don't they? Exactly. They do it in sports. But sports don't reflect reality. In the battlefields of the Middle Ages or even the Renaissance, that level of specialization just wasn't there (and we will see that it wasn't there in Mendes Pinto's Asia either). We can't take the thinking behind contemporary sports -- and where it leads in terms of performance -- and apply it to situations (be it historical or fictional) that are inherently different.
As with technobabble, over-categorizing things is strongly connected with over-numbering things. It's part of our Western culture, actually. This tendency expresses itself in things like competitions, the Book of Records, science, etc., etc., etc. It's part of the drive for precision. "We have skills? Why not have sub-skills? And sub-sub-skills? Or feats associated to skills? Or..." You get the idea. The moment one defines something one can find exceptions, special cases, variants. And for each of these we can repeat the process, with a new rule or rule variant for each.
But do we really do things this way in our daily life? There's something called tolerance and accommodation. My grandfather used to say, "If you don't have a dog, hunt with a cat." We learn to do things in the most precise way... and live doing things by approximation.
"50 character classes! 300 skills!! 2,000 spells!!! We cover it aaaalllllllll!!!!!!!" The catch is, no one is able to cover it all. There's always something missing. And there's always the risk of duplications, redundancy, and above all, big, boring books to read through.
("Not everybody finds detail boring. In fact, given the RPG market, most gamers seem to love it." -- Steve D)
Which is faster, the cheetah or the gazelle? It depends on the concept of "fast". Maybe the cheetah is faster running straight, but the gazelle might be faster when curving.... wait! That means we need two stats for two types of running!!....
Do we really need to follow this path? Is detail that important for its own sake? I personally prefer to be approximately correct rather than specifically wrong.
("Doesn't it depend on the game? If you are playing cheetahs and gazelles chasing each other, these details may be very important." -- Steve D)
3. Don't leave the gamers on their own in the dark.
We've all seen those little, cute, simple, usually so-called "storytelling" games. Highly flexible. So flexible that they always finish by saying "do as you please". The Window and FUDGE come to mind.
The problem is that I don't need your game to do as I please. I need clear directives on how to use that flexibility, that ability to adapt. I may not want thousands of stats, but I need to know how to develop the stats I want for your system.
("You're confusing your terms. Are you talking about game-building tools like FUDGE, or rules-light systems like The Window? Are you talking about game building development, or rules improvisation tools/suggestions?" -- Steve D)
["Is there a difference? Yes, if I'm a game designer or if I'm into game design theory, but if I'm a GM who just wants to grab a setting and rule system and go... In the end it's always my game. Whatever I pick, be it D&D, FUDGE, The Window, GURPS, it's always tools I'm picking. This is an essential part of roleplaying games. In fact, there are a couple of recent articles on Places to Go, People To Be that make good reading at this point." -- Sergio]
4. Don't attempt to show it all.
The game has n character descriptors and a value assigned to each. So, if you want a character, assign values for each and any of the n descriptors, thoroughly filling all the little boxes... and spend the evening doing this.
My first RPG (RuneQuest) suffered from this trauma. Three columns of skills, and values to assign to each one. We could just skip it, but it would be at the risk of having to calculate the values of the skill while play, right? Right, but... on second thought, it's better to have it all calculated beforehand, if you don't want to break the flow of roleplaying.
The key idea -- which I didn't invent, and which is present in several games -- is that, if it's a default value, don't waste time making it explicit.
5. Don't handle PCs and monsters differently.
It was D&D that introduced the concept of different stats and descriptive protocols for PCs and monsters. Now this makes a certain sense from the dungeon-crawling point of view that gave rise to D&D -- PCs can be long and detailed, monsters should be short and simple -- but it adds complexity because it requires two mindsets to handle instances of the same basic set of entities. It also creates a divide that players may wish to blur (like playing a "monster" instead of a "character race or class"; remember all those arguments in the early days about "But why can't I play a vampire?" [Ironic in hindsight....]).
A game should strive to have the same set of protocols to represent all game entities.
6. Don't handle PCs and NPCs differently.
RuneQuest (as far as I know) was one of the first -- if not the first -- games to use the same protocols for PCs and monsters. Yet, since it suffered from the "show it all syndrome", it would be the job of a convict to write down stats for each and every NPC, no matter how minor it was. So GMs and scenario developers began taking shortcuts: they didn't write down everything, but only the stats considered relevant for their in-game function. For instance, if the PCs had to fight a squad of castle guards, the guards were written up with their combat skills and other combat related attributes. Everything else would be dropped.
This is sensible in a dungeon crawl. But in other situations it presents a problem: What will happen if the underdeveloped character is to be used in a situation not covered in its description? Or if a player wants to turn an NPC into a PC? Or the GM wants to turn a minor NPC into a major NPC?
If minor NPCs are just given an "event-specific" set of skills without any background, then when they get "promoted" to centre stage, one either has to invent their background -- a much easier process if you can just use one set of rules -- or justify odd combinations of skills that were given to an NPC solely to keep the action moving. Using consistent protocols for character/creature stats simplifies this whole process immensely.
7. Don't make it difficult to port ideas from game to game.
How often do you find something cool in a game to the point that you would like to port it to another game -- say, an NPC, creature, scenario -- but "translating" it is just too confusing or difficult? ("So this guy's got Honour 45... what the hell alignment is that, anyway?") Or even to port things within two different instances of the same game ("I'd love to run this scenario right away, but it's for level 5 characters and my PCs are level 10!")?
Most of the time the difficulty stems from the complexity of the systems involved (the system of origin and the system of destiny). This, in turn, stems from the business-driven fact that most RPGs are not designed to use non-line products easily. GURPS is perhaps the only system which included a conscious design decision to make other game products easily usable with GURPS (and even there you encounter difficulty, as the 3rd Edition and the Compendia make the "basic" GURPS rules more and more extensive and complex).
("Actually, a lot of the time it's because the setting/adventure is set in a different setting, or requires different types of heroes. Granted, though, complex mechanics also makes it rough." -- Steve D)
["Well, I'm assuming that the GM has gone through that and considers that setting-wise things will go smoothly." -- Sergio]
Summing Up the Don'ts
Or, to summarize everything above in a few lines, the Don'ts are:
Many of these problems are closely related. After all:
Problems 5 and 6 -- treating PCs, NPCs and monsters all with different rules -- tend to exist in games that also have problem 2, over-categorizing and over-specifying everything.
Problem 6 -- treating PCs and NPCs differently -- is avoided to a great extent in a game that avoids problem 4 by not describing everything (any missing stats in NPCs are assumed to correspond to an established "default" value).
Problem 7 -- non-portability between systems -- is more acute in games that also suffer from technobabble, over-specification, over-description and PC/NPC/monster rule separation, and it's inherent to games that suffer from under-specification.
And the Conclusion: Don't Turn Wow into Humph
There's a further problem that's related to all the previous ones. Pick up the Lankhmar D&D modules. Pick up Stormbringer or Elric! Pick up MERP. Find the game descriptions of the main characters from the original fiction. Look at them.
Tell me, honestly: Do you feel excited by those descriptions? Do you like to know that the Grey Mouser is of level X, that Elric has x% on bastardsword fighting, and so on? Do those cold fixed numbers reproduce the wonder of those characters -- the same wonder you get when you read the books?
("I can't speak for those games, but I know the TMNT descriptions are very cool in that game. And just a few weeks ago my group were being awed by Yoda's stats. Frankly, yes, impressive stats do make characters impressive." -- Steve D)
You see, characters in good fiction are interesting because the writer is able to make them unique. Sometimes they are greater than life. Other times they have a balance of strengths and weaknesses that's all their own. When RPGs attempt to convert this into system lingo, more often than not they destroy the wonder. Why? Think about the Grey Mouser: sure, he is level X, but there are plenty of other level X NPCs. Besides, my character will also get to level X. Now, take Elric: he has a sword with a demon of type D with attributes W... so what? I can think right away about more powerful demons of type E, with attributes W+10. Once something is quantified, imagination and awe are dead; comparison becomes mathematical, not emotional.
("Certainly there's some merit to this argument, but I see it being more about the way the game is designed. If the stats are designed so that level X is easy to get and demon swords are plentiful, then of course the NPCs won't be impressive. But it also means the game doesn't fit the setting!! If the stats for impressive characters aren't impressive themselves, then the game doesn't reflect the setting." -- Steve D)
What I'm trying to say is that the conversion of a literary creation into a highly structured and formalized system usually turns the character into a stereotype; it makes it flat, it destroys wonder and uniqueness. Furthermore, it does this to any character... PCs included. That's one of the reasons why character advancement is so important, so that the sense of ennui that comes out of what one has achieved can be overcome by more of the same.
There must be another way.
("Well, there is the point that nobody actually tries to convert a literary creation in a formalized system. They just try to convert his physical aspects and powers. Some games do personality traits, but most still leave room for the unstat-able. That's why they have the "Description/Background" box. However, I do agree that advancement is a significant way to overcome stat ennui, when the character lacks dramatic punch. Which, indeed, can rarely be given through stats alone -- although some games, like Pendragon, go a long way to do it through stats." -- Steve D)
A Pause for Consideration
Before we move on to consider another way, there's something worth mentioning: it may seem that I'm trying to debunk three decades of RPG design. A little pretentious, no? How can it be that everybody (meaning, everybody I know, since there are plenty of games out there I never was able to look at) made it "wrong"? Even in part? The answer is simple: they didn't.
The designers of games like RuneQuest (remember that RQ's design predates its adaptation to Glorantha, so it was not a setting specific game), D&D, and Rolemaster (the basis of MERP) tried to abstract a game from several sources. In the last two columns I discussed language as the foundation of RPGs. Any setting is acquired through language expressed in the sources that present it. Games that cross settings have to abstract a language and generate a source for that language: the game books. This is what the designers of games like the ones mentioned above have done.
In the process, they developed ways of thinking about game design, and ideas about the structure of a roleplaying game, that were highly formal (since they didn't have the literary richness of a good work of fiction), complex (given that everything had to be expressed, since there was no external source on which to rely to fill the gaps) and tended to quantifiable stereotypes (because they were deprived of the context of a well rounded setting (1)).
("They were also designed as personalized wargames, which had a much bigger effect on their design then the things you list." -- Steve D)
I want to go the other way around. I want to give precedence to the setting in its own terms, so that formalization is kept to a minimum (otherwise the system will take precedence over the setting), complexity is reduced (whatever the system doesn't cover can be picked from the setting materials or inferred from it) and the setting's inherent uniqueness is preserved. (2)
("Which all setting-specific games have done, to various degrees and with varying degrees of success, since RuneQuest. However, you are right that few games don't give total, inherent control of the system to the setting, because they want to make sure their games don't suck as games." -- Steve D)
["You just can't see me smiling, Steve. I loved this comment, really. Did I mention how SkyRealms of Jorune and Cry Havoc! (a wargame, remember?) made me look at RQ's damage system in a different light? I... forget it, otherwise I'll never finish this column." -- Sergio]
Still, I'm not starting from scratch. If I got where I stand now, it was because of several games that one way or the other influenced my thinking. The single biggest influence (at least for the purposes of the present column) was Hero Wars.
THE HERO WARS APPROACH
In Hero Wars (HW from now on) there are no predefined sets of descriptors, neither sets of predefined descriptors. What do I mean by this? (3)
There are no predefined sets of descriptors
D&D has races, attributes, classes, feats, skills, whatever; BRP has races, attributes (basic and derived), skills; Pendragon has the same as BRP plus passions and traits.
In HW there is only one set of descriptors, called keywords. Now, I don't like the term 'keywords', but it's hard to come up with an alternative (descriptors, the word I've been using, is no less ugly). Anything that is used to describe the character AND is used as an in-game rule action within the system is a keyword. They all work the same way, no matter what they describe in setting terms, since there is a single unified set of resolution mechanics to handle all situations.
A word of caution: when I say that there are no different sets of keywords, I mean that there are no different sets system-wise (different sets with different functions in terms of game mechanics). I'm making this point because the game does classify keywords into different categories for presentation purposes (for instance, cultural keywords, magic keywords, professional keywords) or to facilitate their usage in game terms (combat keywords, physical keywords, mental keywords). But all of this is driven by the setting, not by the mechanical aspects of the game. All these keywords, regardless of their category, function using exactly the same system mechanism.
("It's worth pointing out that Risus did this long before HW. And to an extent Ghostbusters kicked it off with its 'choose ANY skill' idea." -- Steve D)
["Two games I don't know (or, to be more precise, I only looked at Risus some time ago, so it had no influence in my design efforts). Yet when you mentioned it, I realized that there was a game that played an important part in the way I look at game systems, a game I seldom mention: the French horror RPG Malefices. It's a jewel of minimalist game design. An excellent game, both in the beauty of its system and the way it fits the setting like a glove. (I would change the resolution mechanic, though. But I always change something.) -- Sergio]
There are no sets of predefined descriptors
Before I explain this, keep in mind that I'm focusing on system considerations, not on setting issues. The game does not pretend to provide lists of descriptors that handle all foreseeable game situations. The basic assumption is that a keyword is whatever the players say it is. Look at how character creation works: The basic system is for the player to write a 100-word description of his character; after that, he underlines 10 concepts he considers to be the character's key aspects. These are his basic 10 keywords (4).
("Again, Risus does this also. Actually, so does FUDGE, now I think about it, if you use this approach for chargen. Both predate HW." -- Steve D)
["True, but... Risus is marred, in my opinion, by its Toon-esque tone (not that I have anything against Toon, it's one of the games I most cherish). I mean, it's supposed to be a fast-forward, do-as-you-please game -- even if it can be played differently. There's a difference in tone that sets it apart from HW. FUDGE... I never liked FUDGE. It's one of those games I just can't like. There's something about it that puts me off. Furthermore, I think you are missing the difference here: a game system that says "when preparing your game, draw up a list of available abilities that help define the way you handle the setting" is very different from "when playing, create abilities for your character that suit the setting". It may seem to be splitting hairs, but there's a difference between meta-game reasoning (FUDGE) and in-game reasoning (HW). You see, FUDGE is a tool for people to create games in the classical mold, rather than a straight-up game in itself. Granted, HW can also be turned into a similar tool (I've been suggesting time and time again it could be used as an excellent superheroes game system. And, heck, if even D&D can be turned into an "open" tool...). But the openness of HW is not intended to be used at that level. It is inherent to the way it handles in-game entities." -- Sergio]
This basic character creation system of HW is complemented by a 'list' system, which provides the players with lists of pre-created sample keywords. Yet, this does not correspond to the way other games handle pre-defined sets of keywords because it is based on a completely different design. These lists don't result from an attempt on the part of the game designer to abstract coherent sets of descriptors that represent all foreseeable game situations from the point of view of the game mechanics. They are setting-derived and present the most likely keywords to be found in a given population (whether based in culture, background, race, profession, or system of beliefs).
("Ditto Risus. And Ghostbusters. And indeed, any game which says 'choose your own skills', like CoC. I love the way CoC characters have skills like 'Look Stupid 99%'." -- Steve D)
["Sorry, Steve, but I see a huge difference between a game that says 'choose your own skills' as an afterthought, and a game which makes that concept central. Furthermore, it's not only a question of skills; in HW, any descriptor is chosen this way. There are no predefined set of attributes plus open set of skills. (By the way, if you ever come upon Malefices, give it a look and compare with CoC. Just a suggestion.)" -- Sergio]
For instance, take the list of Horlanthi (one of the cultures of Glorantha) keywords at http://www.glorantha.com/hw/P03Keywords1.html. How can we understand this list? Think of it as if there was an essay on the Horlanthi culture that someone had distilled into a list of relevant descriptors of the Horlanthi population. It's purely setting-derived and setting-oriented.
But let's get back to the basic system. The only limit a player faces when creating his character is his creativity, the direction of the GM and, more important, his understanding of the setting. The better he knows the setting, the easier it will be for him to create his character.
As you can see, the way HW handles descriptors and character creation is completely setting-oriented. The key idea is that if you know your setting, you know what to play within it. Everything not covered by the player's chosen keywords will default to some implicit value (5). Now, this leads us to another question: how do HW mechanics handle keywords?
The first issue is, how do we assign performance to keywords? Because, you see, keywords represent not just the things the character can do, but how well they can do them. After all, they're not just some nice slogans, they must have mechanical consequences in game terms. Well, the keywords chosen by the player will have a value that's higher than the default value for everything not covered by them. The character performs better in his chosen keywords.
These values are defined in numeric terms in a scale that goes from 1 to 20 (and then jumps to a second scale of multiples of 20). In very simple terms the game works by testing the keyword against a 1d20 roll. (This means that the game uses an absolute scaling system. I'll address this in next month's column.)
Another nice touch is that any keyword has the potential to be used against any keyword. This is a fascinating departure from classic games. In classic mechanics, if there is a set of descriptors to handle a certain situation, one can use only those descriptors to handle that situation, and cannot resort to descriptors of a different set.
("Ghostbusters and Toon both broke this rule....so just be careful with the definition of 'classical'. Again, Risus and other games of its ilk also break this rule. SHERPA (an early relative of FUDGE) is another example, I think." -- Steve D)
["Well, 'classical' in this instance means the games that set the mold (D&D, RQ, Traveller et al). I am not implying that HW was the first game to do away with strict boundaries in terms of ability usage, though. I began my analysis of HW by stating that I'm limited by the games I know or that shaped my ideas on game design. The games you mention are not included in this set, except for Toon -- as you say, Toon did it long before, because it suited the setting, the nonsense inherent in cartoons. Yet, as cartoons reinforce the reality they turn into nonsense, Toon reinforced the RPG conventions it subverted -- by coupling its rules differences with its setting wackiness, it reinforced the identification of 'serious' gaming with traditional systems." -- Sergio]
So you would use combat skills for combat, communication skills for social interaction, but can't use communication skills for combat. Since in HW there are no mechanically oriented sets of descriptors, any keyword has the potential to be used for any situation... provided the player can advance a good explanation for it (it may be subject to penalties if the explanation is not good enough). For instance, in a fight one of the parties may answer an attack with an offer of gold instead of the blow of a weapon.
The Other Side of the Coin
If that was all there is to HW, it would be excellent. Unfortunately, there is more to it. You see, I don't like its resolution mechanics. But that's something I'll leave to a future column.
So, to wrap things about HW, how does it handle the "Don't"s presented above? Let's see:
Technobabble: Yes, the descriptors are setting-derived, but this nice feature is hindered by the way they are "mapped" into the resolution mechanics. Suffice it to say that this requires quantification, something I prefer to avoid myself, especially if, like in HW, this quantification works within the context of an absolute scale.
Over-specification: HW does not attempt to set reality in stone; there are no predefined system-derived sets of descriptors. It works just the opposite way.
Under-specification: It provides guidance to how the players may derive game entities from the setting -- it has to, since descriptors are setting-derived by nature. And it does this pretty well (6).
Over-categorization: In HW, the player only needs to express the relevant unusual aspects of their character (what isn't relevant is defaulted to a standard value).
PC-NPC-Monster Differentiation: There is no difference between PCs, monsters, or NPCs; all are handled the same way.
Inter-System Portability: It's a breeze to import things from other games (you just need to convert the relevant descriptors into keywords).
All in all, I can say that HW handles very well "Don't"s 2 to 7, but doesn't quite make it on "Don't" 1, since it ultimately reduces all the qualitative keywords into a quantitative scale -- something that makes the system mechanics too visible. Because of this it presents heroes in a way that is neither Wow nor Humpf but... Womf, maybe?
This leads us to a different set of considerations. So far we have been looking at the way games use descriptors to identify game entities -- how to create labels for different aspects of the setting. Now we need to look at how the game fleshes out those entities.
CONCLUSION: From System-Defined Skills to Setting-Sourced Descriptors
As TToMP takes shape, it's becoming obvious that this game design process is a very unusual one, because we're attempting what might be called un-design. We are attempting to create a set of rules that are as minimal and invisible as possible, that are derived from the setting, and that reinforce the setting with every use. For this reason, we took Attributes out of the game in the previous columns; for this reason, instead of an arbitrary set of skills, we are opening up the assignment of abilities to players' imaginations, to be described in plain English rather than jargon or numbers -- the way Mendes Pinto himself described the people, things and places he discovered.
This leads us naturally to the next step: how do you evaluate (literally: "give value to") an ability when you don't want to use numbers? Or even a preset absolute verbal scale (e.g. FUDGE's Terrible-to-Superb seven levels)?
To answer this, we're going to have to go to next month's column, as this one has already taken up enough space.
(1) Let me be more clear about this: I don't mean necessarily that the game is not connected with a well rounded setting. What I mean is that the well rounded setting is not in the game book. For instance, RQ used Glorantha as its setting but we cannot consider that Glorantha was well presented in the RQ rules book.
(2) There's a wonderful example of the issues discussed so far. You can pick it up at Places to Go People to Be: http://ptgptb.org/0019/classconflictD20.html . In this article the author tries to fit a medieval society into D&D's class system. It shows perfectly the problems that arise when a game defines its own language (and associated mechanics) and the players attempt to model a setting within that language.
(3) You can check how HW works by looking at Issaries web site: http://www.issaries.com. The relevant links are http://www.glorantha.com/hw/P02Characters1.html and http://www.glorantha.com/hw/P02Characters2.html for character creation.
(4) Actually the player does not need to decide on his 10 keywords before he starts playing the character. He may decide that he will make up his mind about some of them latter in the course of play. Furthermore neither does he need to know exactly what a keyword means. He may leave this to some moment in the future.
(5) It can be a racial, cultural, professional, belief system default. Or, if none of these apply, a general default value.
(6) When we consider character creation I would do things a little different, though. You see, HW either "forces" you to write down a description of the character and his life or to choose pre-defined keywords from a list. I would enlarge the possibilities in three senses:
The first is obvious: allow the players to combine both alternatives, picking some keywords from their text, and other from the list provided by the game books.
I am a bullet lists freak myself. Instead of being required to write a text, I'd choose to lay a nice bullet list of keywords created by me.
Picking keywords from the lists provided by the game books is nice, but I can see other ways where setting books can provide keywords. For instance, picking it from game fiction. Or from game art. Or from any other type of setting material.