September 25, 2012

Toaster repairman: the strange (and unfinished) evolution of RPG character creation

When Wasteland 2 reached full funding on Kickstarter, a discussion began to brew in the RPG community about ways to improve on the original game. Among the many subjects of debate was whether (and if so, how) to improve on the skill-based character creation system employed by Wasteland (and due to reappear in Wasteland 2 in one form or another).

This struck me as a good time to write about skill-based character creation systems: how they came about, and what they mean for cRPG design today. I’ve spent the intervening time building this article, bit by bit. Let’s talk about how character creation first came about, how the skill-based approach evolved, and where that leaves us today.

A brief history of character creation

Character creation, like all core elements of role-playing games, is best understood against the background of the historical wargames that preceded them.

Baron von Reisswitz is credited with creating the first true wargame (i.e. a game meant to simulate battles with a certain degree of fidelity, and not merely a chess derivative). Created in the early 1810s, this game went by the name Kriegsspiel. It featured units actually in use by the militaries of the day, and was meant to simulate battles. Character creation was therefore a matter of faithfully emulating the real-world characteristics of the units those pieces represented, then using die rolls to simulate unforeseen factors in resolving combat.

Von Reisswitz’s son created a revised version of the game in 1824. The revised Kriegsspiel paid such close attention to accuracy that the Chief of Prussian General Staff was moved to recommend the game as a military exercise, with the king of Prussia actually ordering that every regiment of the Prussian Army be supplied a copy. (No, really.)

A third version of Kriegsspiel was put out by Colonel Julius Adrian Friedrich Wilhelm von Verdy du Vernois in 1876. Verdy du Vernois was suspicious of the idea that military outcomes could be predetermined according to fixed rules, and replaced die rolls with the mediation of impartial “umpires” who would determine the outcomes of various engagements based on their knowledge and experience. Yes, that’s right: the first Dungeon Masters were Prussian military men from the 1800s.

“No, you can’t cast Magic Missile across the Rhone.”

The American military began putting out its own wargames around this time, with Jane’s Fighting Ships following suit across the Atlantic in 1898. Like Kriegsspiel, Jane’s Fighting Ships spelled out the characteristics of the game’s numerous units in astonishing detail. Google Books has a digitized copy of the rulebook online, so you can see for yourself just how intricate this got:

Even H.G. Wells got in on the action, producing Little Wars in 1913. The rules of Little Wars were far simpler than those of other wargames, but it generally followed the practice of simulating large-scale battles, with the characteristics of different unit types decided rigidly according to the type of troops each unit represented.

It wasn’t until the early 1970s that wargames started delving into the idea of individual men and women as units. The games that did this eventually became known as “man-to-man wargames” (not to be confused with Steve Jackson’s ruleset of the same name). It may seem obvious to us now, but this focus on individual men and women was such a radical departure from wargaming tradition that it wouldn’t even appear in the rules for Gary Gygax’s Chainmail until 1971, three years after Chainmail’s initial publication. Even then, it seems the man-to-man rules in Chainmail were largely an afterthought, relegated to a mere 2 pages out of the entire 44-page book. There, too, character creation remained a matter of looking up prefabricated unit values in a table. Even Dungeons & Dragons, published in 1974, retained many of the Chainmail rules. Character creation consisted of nothing more than allowing players to select from one of three prefabricated character classes.

Things changed dramatically with the publication of Advanced Dungeons & Dragons in 1979. The AD&D First Edition character creation system required each player to roll three six-sided dice to determine each of the following statistics for any given character: Strength, Intelligence, Wisdom, Dexterity, Constitution, Comeliness(!) and Charisma. Character class was not chosen until later. This wholly upended the method of character creation that had prevailed up until that point. Statistics were no longer determined by class: instead, characters got statistics, and only then chose a class based on which roles the statistics made available to them.

RPGs continued to diverge from war games as the genre developed, and so too did their character creation systems. With increased focus on unique, individual characters came an increased focus on the abilities and limitations of each individual character. At their peak, these considerations would come to supplant the notion of character class entirely. Published in 1986, Steve Jackson’s GURPS represented a coming of age for skill-based RPG systems. GURPS characters have no classes at all–rather, they have four primary attributes and a huge variety of skills that can be leveled independently of one another.

In a way, this represented the zenith of the individual-focused approach to character creation. All vestiges of the old system were gone: in the skill-based paradigm, characters became unique, fully realized individuals rather than mere instances of a uniform military unit to be used in battle simulations. This approach became quite popular among pen-and-paper role-players, not just with GURPS, but later with White Wolf RPGs such as Mage: The Ascension and Vampire: The Masquerade.

Compared to pen-and-paper games, however, skill-based systems in cRPGs have been comparatively rare; and not for no reason, as we will see below.

Why use skill-based character creation?

GURPS employs skill-based character creation because of the game’s central conceit. “GURPS” stands for “Generic Universal Role-Playing System.” As the words “generic” and “universal” suggest, GURPS is designed to work independent of any particular setting. Thus, in theory, GURPS is as well-suited to a fantasy campaign as a science fiction campaign, a steampunk campaign or a campaign in the mythical Wild West.

GURPS has this flexibility because skills are far more transferable than class roles. Class is largely a function of setting: it is a statement on the character’s position within a particular social or military context. You can’t easily transpose, say, a Barbarian from a medieval setting to a contemporary setting; it simply wouldn’t make any sense. But a physically strong character with Acrobatic and Melee Combat skills? That could certainly work. The skill-based system is therefore convenient for running a wide variety of different pen-and-paper role-playing campaigns with friends.

Skill-based character creation systems offer unique benefits to cRPGs as well. To my mind, there are two fundamental advantages to using skill-based systems in cRPGs: (1) they provide interesting alternatives in the game, and (2) they provide role-playing opportunities. Those might sound like the same thing at first, but they’re actually quite different.

Skills offer flexibility in character creation and development. Characters are not all just minor variations on a theme; rather, their abilities represent one entirely unique combination of trade-offs among thousands–or even millions–of possibilities. The player isn’t limited to a predefined set of complementary abilities: she can build her character in as odd and incongruous a manner as she likes.

To analogize: class-based character creation is the Apple of character creation systems, founded upon the use of a few pre-built constructs that share the same core components and come in a limited variety of flavors. By contrast, skills-based systems are the PCs of the character creation universe, cobbled together from a dizzying array of components that can combine in interesting (though sometimes deeply flawed and incongruous) ways.

Skills also force a certain flexibility into the game’s design. Because the developer can never count on your character having one skill or another at any given point in the game, the developer is forced to provide numerous alternative means of bypassing obstacles and resolving quests, or else create a wide variety of content with the understanding that only a limited selection of possible characters will be able to see it on any given playthrough. Branching design can (and often does) appear in class-based RPGs as well, of course, but the possibilities tend to be far more limited, as they typically rely more upon factional and alignment choices than they do the player’s actual abilities, which are generally rather restricted.

There is also the matter of role-playing. Giving the player the flexibility to shape their characters using a wide variety of different skills gives the player greater flexibility to play the character he or she wants to play. Rather than “a priest,” the player might be–for instance–a highly persuasive character with healing skills and a knowledge of ancient languages, a sort of smooth-talking archaeologist medic. The level of specificity with which one defines a character in a skill-based system can open the door to much more detailed (and satisfying) role-playing than the limited array of options offered in class-based systems.

The problems with skill-based character creation in a cRPG

Lest you think that I advocate skill-based systems without reservation, I will now proceed to talk about the rather substantial downsides of using a system like this in a cRPG.

As a general proposition, most of the design problems inherent in computer RPGs stem from the fact that there is no Dungeon Master to appeal to. Because there is no Dungeon Master, the wide variety of choices available to the player in a skill-based system presents a huge array of possibilities that require baked-in responses. The player can improvise, but the world can seldom improvise a response in kind. Every instance in which a skill may be useful therefore has to be built into the game ahead of time.

RPGs with skills-based character systems are more susceptible to this issue than most, simply by virtue of the number of possible character builds. It’s easy to balance a class-based system, since there are a relatively small number of possible character builds. A skill-based system, by contrast, involves so many possible combinations of abilities that balance becomes extremely difficult.

To revisit the computer analogy from earlier: it is easy to find a good Apple computer because it only comes in a few configurations, and it is easy to ensure that that handful of available configurations works flawlessly. By contrast, the sheer variety of possible PC builds means that while there are far more possibilities, many of those builds are bound to be deeply flawed. So it is with skill-based character creation systems: any cRPG with more than a small handful of skills is going to provide a near-infinite variety of possible character builds.

This presents a serious design challenge for any developer. Each new skill requires a new system, or–at the very least–a new branch or skill check in the game’s conversation trees. In turn, the more systems and skill checks one adds to the game, the harder it becomes for the developer to ensure that each skill is more-or-less equally important to the game.

This last concern becomes especially problematic because skills used in emergent systems are mixed with those that can be used only at specifically scripted points in the game. Consider combat for a moment. Combat is generally an RPG’s most highly emergent system. Combat is built upon a number of actors operating according to simple rules, which then produces unexpected and complex situations. There are dozens of different factors in play at any given time in battle: a single scenario involves dozens of player decisions, and is unlikely to play out exactly the same way twice. This emergent quality makes combat a tempting choice to form the backbone of a cRPG, since it offers a much higher ratio of entertaining possibilities relative to development time spent than a more static system (such as, say, dialogue trees) would. Thus, combat ends up accounting for a huge chunk of playtime, which in turn makes combat-related skills the most consistently valuable skills in the game.

Further, because combat skills impact an emergent system, their use offers a wide variety of possible effects. Gunplay, for instance, can deal direct damage; it can injure particular parts of an opponent’s (or player character’s) body; it can detonate explosives; it can break parts of the scenery; and so on. It offers variety and flexibility because the system it impacts responds organically to its use. This, in turn, makes these skills carry more water in terms of offering satisfying, unpredictable results across multiple playthroughs.

By contrast, skills that are useful only in scripted systems do not possess these benefits. A skill check or variable check in a scripted event will always produce one of two (or, in a best-case scenario, one of a few) pre-generated results. Because each result has to be hand-crafted rather than arising emergently, such skills are much less efficient at generating interesting variations throughout the game, and tend to feature less heavily in wRPGs than those skills which directly impact the game’s emergent systems.

All of these factors converge to cause significant problems for an unwary player. Providing skills that are less often useful can have nasty effects on an RPG playthrough, including: (1) consuming skill points that could have gone toward other, more helpful skills; (2) offering fewer opportunities to gain experience points and improve the character (which, in turn, further cripples the player’s ability to put points into more useful skills); and (3) failing to actively help the player character survive the most common situations in the game, leading to frequent player deaths.

Even worse, these effects emanate primarily from choices made before the game even begins (and thus, before the player has any way of knowing how often various skills will be useful). These initial, blind choices can be overcome later only at great cost, if at all; and as explained above, a bad initial choice will actually make correction more difficult because of decreased character survival rate and decreased opportunities to gather experience points and level up. It has the potential, in short, to create a highly undesirable positive feedback loop.

Some games are nice enough to suggest the player use pre-generated characters on a first playthrough, but that is essentially just offering the player a choice of character classes. It only begs the question: why give the player the option to create a broken character in the first place?

What about role-playing?

Now you might say, “Who cares? I’m playing this game to role play, not to create a super character.” That’s a good attitude to have, but why don’t we really test it out? Role play someone who is not skilled in combat or movement/stealth skills. Someone who is skilled only in areas that don’t play into the game’s emergent systems. It almost doesn’t matter which game we pick to perform this experiment in. Try to successfully complete–oh, say, Fallout 2–by role-playing a scientist. Go ahead. Roll up a new character and put all of her points into the Science skill. No save scumming, now. Or how about a scientist-doctor? Maybe one who loves hiking. Stick all your points in Science, Doctor and Outdoorsman. See how well that works out.

Better yet, don’t. I’ll save you some time: you are going to die. Repeatedly. With few opportunities to level up, your character is going to stagnate, and you are going to struggle to get much of anywhere in the game unless you are already intimately familiar with its details. (Of course, there’s no point in having a discussion about game balance using someone who knows how to speed run the game in under 30 minutes as our reference point.)

The point is, cRPGs aren’t pen-and-paper role-playing games. There’s no Dungeon Master to appeal to with creative uses for your characters’ various skills. Every last skill check that applies to a non-emergent game system has to be incorporated into the game in advance. The average player’s skills are useful in direct proportion to the number of times the game checks for them. Skills which are rarely used may serve a role-playing purpose, but they can also completely undercut a player’s enjoyment of the game by making survival and progression extremely difficult based on front-loaded choices the player is forced to make blindly.

Three potential solutions

After all of that, you might think that I’m going to suggest eliminating so-called useless skills from wRPGs. I’m not, though that is certainly one way to avoid the problems of skill-based character creation. The original X-Com, for instance, features characters with individualized aptitudes spanning a wide variety of skills, each highly useful in relation to an emergent system. (X-Com also has its share of balance issues, but those issues stem more from the design of certain facets of the combat system than they do the use of individualized character skills.)

The problem with the X-Com approach, however, is that every single skill in the game plays into only one system: combat. There’s very little in the way of non-combat approaches to different situations, which undermines much of the value of using a skill-based character creation system in the first place. There’s character diversity, but no diversity in terms of broader approaches to problem-solving.

There is a second approach. Brian Fargo has said that he approaches the balance issue by giving the player a party of characters so the player can cover a lot of bases between the characters. This is smart design, but it only works well in party-based games where you have direct control over your entire party. In party-based games where you have direct control over a single main character and no one else, certain skills (e.g. dialogue or stealth) provide no advantage at all when present in an AI-controlled party member.

The only skills that AI-controlled party members typically use with any proficiency in such games are combat skills, which gives the player a bit more of a buffer to focus on non-combat abilities, but otherwise accomplishes little. Meanwhile, it renders certain of your own abilities, like Stealth, borderline-useless unless the game offers you the ability to consistently leave behind most of your party. And of course, in non-party-based RPGs, giving the player a party simply isn’t an option in the first place.

Recently (so recently, in fact, that I had to edit this article to include it prior to publication), Tim Cain announced a third approach for the upcoming cRPG Project Eternity. His preferred method is to separate combat and non-combat skills:

Non-combat skills are gained separately from combat skills. You shouldn’t have to choose between Magic Missile and Herbalism. They should be separate types of abilities, and you should spend different points to get each one.

This is interesting (and in many ways, quite similar to what I suggest below). However, it does have the disadvantage of removing a major mode of specialization from the player. No matter what, the player is always going to be good at combat, and is always going to have non-combat alternatives to fall back on. It’s a smart hedge against the eventuality that the developers can’t successfully balance the usefulness of all abilities, but it comes at the cost of making character specialization less consequential.

A fourth way: the Survival/Elective Tiered Approach

Back in March, I casually suggested that Wasteland 2 segregate useless (but fun) skills from the ones that are most consistently used in the game. I’d like to take this opportunity to elaborate on that.

The thing is, so-called “useless skills” actually do have a legitimate place in RPGs: they add color to the world and can provide more role-playing opportunities for characters. Ostensibly, that’s the sort of thing we want to encourage, not strip away. The problem with these “useless skills” isn’t that they exist; it’s that they are put into the same pool as skills that are orders of magnitude more helpful. In a zero sum character creation system, where putting points into one skill means giving up competency in another, placing skills that are used only sporadically alongside skills that are used constantly is a recipe for a lot of ruined play experiences.

So here is my humble suggestion to developers who want to create a game with skill-based character creation: don’t force players into making a choice between skills of high value and skills of more dubious benefit. Instead, create two tiers of skills: survival skills and elective skills.

The survival skills tier should contain only those skills so consistently useful that at least one of them is probably necessary to complete the game in a reasonable playthrough (i.e. completing most main quests, and not exploiting special knowledge of where things are hidden in the game world). These will necessarily include skills that play into the game’s core emergent systems (e.g. combat; stealth; and physical manipulation abilities suited to the setting, which might include pick-pocketing or hacking). Ideally, persuasion should feature as well, assuming that the developers take sufficient care to make the game both challenging and beatable primarily through the use of dialogue options.

The elective skills tier, by contrast, should consist of what I choose to call “flavor skills”–those which are useful only in specialized, uncommon scenarios (e.g. Science from Fallout 1-2 or the infamous Toaster Repair skill from Wasteland), or which provide benefits that do not directly impact the player’s ability to survive most in-game challenges (e.g. Outdoorsman from Fallout 1-2 or Cartography from Eschalon 1-2).

The player should get two pools of points to spend: one pool of points that may only be spent on survival tier skills, and a second pool of points that may be spent on either survival tier or elective tier skills. This ensures that the player cannot create a character incapable of surviving a normal playthrough while still giving the player the flexibility to pursue unique avenues of play and opportunities for role-playing.

Superficially, this approach might sound like the approach now under consideration in Project Eternity, a tiered system with similar gameplay benefits. However, it differs in one crucial respect: it does not force all characters to be combat specialists. Using the survival/elective tiered approach, the player has the option of playing someone who is not skilled at combat and must rely on other major skills (such as Stealth or Persuasion) to survive. It is, in short, a way of removing most of the risk of “useless” characters while at the same time preserving character variety.


The notion of what it means to control an individual unit in a game has evolved tremendously in the past 200 years. Where once we used identical battalions straight out of a war game’s unit tables, now we can expect to command highly individualized, deeply customizable characters. Skill-based character creation systems represent the best we have to offer in terms of individualized cRPG character creation; and yet, the enormous difficulties involved in balancing them have nonetheless rendered them exceedingly scarce among cRPGs.

A good skill-based cRPG must accommodate a large diversity of game systems. Separating necessary survival skills from elective skills that lend flavor to the role-playing experience provides an elegant solution to the thorny balance issues this diversity raises. The survival/elective tiered approach isn’t the only possible solution, but I do happen to think that it’s the one which holds the most promise.

My modest hope is that this approach leads to more fruitful and widespread use of deep, skills-based character customization in our cRPGs going forward. We’ve come a long way since Kriegsspiel; and this land is far, far from fully explored.

Craig Stern is an indie developer currently working on the multiplayer turn-based tactics game Telepath Tactics. He is the founder of, and can often be found rambling in short, 140-character bursts on Twitter.