Gaming Origins to Fiction

Introduction

The Malazan Book of the Fallen is one of the very few works of contemporary epic fantasy whose entire cast, cosmology, and pre-novelistic history were developed as a shared tabletop role-playing campaign before a word of the first novel was written. Steven Erikson and Ian C. Esslemont met on an archaeological dig in the early 1980s, began running games together shortly afterward, and spent roughly a decade co-creating what would eventually become the ten-volume Malazan series and Esslemont's parallel novels. The trajectory from gaming session to finished fiction is not an incidental anecdote about the series' origins. It is a load-bearing fact about how the books work — and Erikson has, across numerous recorded conversations, articulated its consequences in unusually specific terms.

This essay reconstructs the gaming-to-fiction pathway under six headings: the anthropological pact that shaped the early sessions; the deliberate "violence to the mechanics" that placed narrative above game balance; the marathon character-driven sessions that produced the cast; the alternating and convergent GM structure that gave the two authors overlapping claims on a single world; the Karsa Orlong campaign as the cleanest documented case of a player's unwitnessed contribution to the series; and the famous natural-20 dice roll that determined the outcome of a pivotal duel in Toll the Hounds. Each of these structural facts has a recoverable textual consequence in the finished novels, and understanding the consequence requires understanding the gaming practice it grew out of.

Origins at the Dig: An Anthropological Pact

Erikson and Esslemont met as working archaeologists, and the first gaming session between them occurred, literally, at a field camp. Erikson has described this origin moment with some candour:

"It came from gaming. Myself and Ian C. Esslemont were both archaeologists and we met on a dig, and I knew nothing about D&D but he did — he'd spent most of his university playing D&D instead of going to class. But he talked me into it, and the first game I ever played was on the site. So we were in a camp on the side of a lake, and unfortunately we smoked way too much dope, the game was a disaster." (An Evening with Steven Erikson transcript)

The first session's failure is less important than the second-order commitment it produced. When Esslemont subsequently joined Erikson in Victoria — where Erikson was enrolled in a writing programme — the two shared a flat and began gaming in earnest, and from the outset they imposed a professional constraint on their play that distinguished the practice from conventional RPG campaigns:

"We shared a flat, so we started gaming, and both being archaeologists as opposed to being an anthropologist, we wanted to make sure that our fantasy world was internally consistent in terms of cultural evolution, civilizations, all that kind of thing. And so that's what the gaming basically involved." (An Evening with Steven Erikson transcript)

What is being named here is the importation of a disciplinary standard from archaeology into the worldbuilding of a fantasy campaign. Conventional RPG worlds are designed primarily as combat environments; their cultural elements function as colour. Erikson and Esslemont were, by contrast, applying the same internal-consistency criteria that an archaeologist applies to a reconstructed material culture: if the society has metal tools, what is the metallurgical infrastructure that supports them? If the society is mobile, what is the subsistence base that allows mobility? If two cultures are in contact, what is the trade network, the linguistic interface, the history of conflict? The Malazan world's distinctive "lived-in" density — the sense that every named city has a hinterland, every hinterland a prehistory, and every prehistory a set of extinct or absorbed predecessor cultures — is the direct consequence of this pact. It was, before it was a novel, the documentation of a world built to archaeological tolerances.

"Violence to the Mechanics": Story Over Balance

From very early in the campaign, Erikson and Esslemont treated the rule systems they were using — first D&D, then GURPS — as instruments to be reshaped whenever the rules and the story diverged. Erikson's own phrase for this practice is pointed:

"The Origins are well known — we start with the D&D system and then went on to GURPS when we were playing, but all the time we were sort of messing with the mechanics because we were looking for those limitations... I keep talking about how the mechanics — the violence that we did to the mechanics while we were playing — because all considerations of balance, game balance, all the recipients out the window. We were just trying to serve the story." (Community Malazan Questions with Esslemont and Erikson transcript)

The inversion here is structurally important. In most tabletop play, the rules constitute the world: they determine what is and is not possible for characters, and the story emerges from their operation. In the Erikson-Esslemont sessions, the world was primary and the rules were subordinate. When a rule produced an outcome that contradicted the world the authors had built, the rule was rewritten. The resulting practice is closer to collaborative fiction with dice than to conventional RPG play — and crucially, it trained both authors to think of the mechanical apparatus of magic and combat as servant to character and plot, not the reverse. This is the root of the famous difficulty readers have in producing a "hard" taxonomy of the Malazan magic system: there never was one. The system was always a set of negotiable constraints whose purpose was to generate dramatic situations, and when a situation required a break from the constraints, the break was taken.

Erikson has extended the principle to a meta-observation about the ludic-versus-narrative tension:

"The ludic experience and the narrative experience are not necessarily the same, so this idea of violence where you're balancing it more and towards a story and a world that you're experiencing... that's actually quite interesting." (Community Malazan Questions transcript)

The distinction between ludic and narrative experience is itself an advanced piece of craft thinking. Most RPG designers working in the 1980s and 1990s were explicitly optimising for ludic balance; Erikson and Esslemont, without any apparent programmatic theory, were optimising for narrative resonance. The result was a decade's worth of shared imaginative practice directed at a set of aesthetic goals that conventional game design would have classified as side-effects rather than objectives.

The Six-Hour Conversation: Character as Prior Commitment

Perhaps the most striking detail about the Erikson-Esslemont gaming practice — and the one most directly responsible for the distinctive depth of the Malazan cast — is the length and character-orientation of the sessions. Erikson has noted that a single session might run six hours and contain no combat at all, because the session was devoted entirely to dialogue and character interaction:

"There was times that we gamed entire sessions — how many people here playing role-playing games? All right, you all know. We would play an entire session, so it'd be six hours where..." (An Evening with Steven Erikson transcript)

Extrapolated across the years of the campaign, this practice produced something unusual: a cast of characters whose voices and relational histories had been rehearsed — in real time, between two writers — before any of them appeared in print. By the time Gardens of the Moon was drafted, Whiskeyjack, Kalam, Quick Ben, Fiddler, Anomander Rake and the rest had been improvised in conversation for hundreds of hours. Their dialogue styles were not invented for the novel; they were recovered from memory of play.

This accounts for a quality of the finished prose that has often been remarked upon by readers and critics: the sense that the characters are older than the text — that they exist prior to their appearances, that their silences and asides refer to events outside the current frame, that they know each other in ways the narrative does not need to explain. The quality is not a stylistic trick. It is a residue of the gaming practice: the characters had in fact been interacting with each other in a shared imaginative space long before the novel began, and Erikson was writing them as remembered friends rather than as newly invented constructs. When the Bridgeburners speak in shorthand because they have too much history for explication, the shorthand is real shorthand — it refers to a history Erikson actually rehearsed with Esslemont in play.

Alternating GMs and Convergent Campaigns

A further feature of the Erikson-Esslemont practice was the alternation of game-master duties and the eventual merging of multiple parallel campaigns into a single narrative timeline. Erikson's own description of the structure is worth quoting at length:

"We would alternate who ran the game that night. But sometimes we play both sides... we would actually split the squad. So some games he was running, campaigning two other times, and I think my characters were Quick Ben, Kalam, Fiddler and — he has the other ones. And then my first character ever rolled up was Anomander Rake, and then we just blended him into the whole storyline, and we played out the fete at the end of Gardens of the Moon — that was actually a game goat." (An Evening with Steven Erikson transcript)

Three points are worth extracting here. First, Erikson's very first rolled-up character was Anomander Rake — a fact that retrospectively explains the character's anomalous centrality in Gardens of the Moon. Rake is not a character Erikson built to fit the novel; he is a character whose existence predated the novel and whose presence shaped the story around him. Second, the "splitting of the squad" — running separate but temporally simultaneous campaigns for different subsets of the Bridgeburners — produced a narrative architecture in which characters have independently witnessed events, hold independent knowledge, and can meet later with asymmetric information. The structure of Gardens of the Moon, in which the Bridgeburners operate in split squads with partial views of the overall plot, is the fictional descendant of this practice.

Third, the fete at the end of Gardens of the Moon — the party at Lady Simtal's estate, which culminates in multiple convergent crises — was "actually a game goat": actually played as a session. The scene's distinctive quality, which has often been remarked upon, is the feeling that a dozen agendas converge in a single location and produce consequences no individual agent had planned. That feeling is generated by the fact that the scene was originally produced by a dozen separately-motivated players colliding in a shared space with each agent pursuing their own goals. The author did not design the scene's chaos; he transcribed it from memory of a session in which the chaos had been real.

Karsa Orlong: The Player Who Didn't Know

One of the most revealing pieces of Erikson's gaming-to-fiction testimony concerns the origin of Karsa Orlong, the Teblor warrior whose arc dominates the opening chapters of House of Chains. Karsa was originally produced through a one-on-one campaign between Erikson (as GM) and a player named Mark Paxton-McRae:

"Later on we moved from sort of one-on-one gaming — which I know sounds strange — but where I had a whole group of people and then I ran them through... in a linear fashion. The poor guy, I did a one-on-one game with a guy named Mark Paxton-McRae, and he rolled up his character, he gave the name to me as Karsa Orlong, and then I sat down and put together a campaign for him. Okay, you're living in a tribe in the mountains, and your friends come up to you... Most of our gaming was just to actually create a moral quandary in the players and really mess with their heads." (An Evening with Steven Erikson transcript)

Two features of this origin story are worth emphasising. First, the character's name was supplied by the player, not the writer — which is why the name "Karsa Orlong" has the phonetic distinctness of a player's invention rather than the derivation from Erikson's usual linguistic patterns. Second, and more significantly, the campaign itself was constructed around a player who did not know the shape of the world he was entering. Paxton-McRae was playing a mountain tribesman from within the tribesman's worldview, with no knowledge of the larger Malazan cosmology into which the character would eventually emerge. The resulting psychological texture — Karsa's ferocious provincialism, his difficulty in understanding the concerns of the "civilised" world, the slow expansion of his moral vocabulary as the books proceed — is the authentic residue of a player who was himself moving from a limited to an expanded perspective as the campaign developed.

The Karsa chapters of House of Chains are famous for their opening ferocity and for the slow moral education they impose on both character and reader. What the gaming origin reveals is that the education was not a device invented for the novel; it was transcribed from a real process of a player's gradual discovery of the world he was embedded in. The character's cognitive development is not a retrospective authorial construction but a replay of an actual historical fact about how Paxton-McRae experienced his own character at the table.

The Natural 20: Dice as Narrative Authority

A second famous origin story concerns a combat encounter in book 8 (Toll the Hounds) whose outcome was determined not by narrative design but by a literal dice roll:

"One of the famous things I've talked about before in terms of the Malazan stuff in book eight — the battle between two major characters, I won't spoil it too much, was decided on a 20d natural roll. So we were just rolling to see who was going to win, because they were both characters we had rolled up and played, and they were now facing each other, and the first guy who got a 20 decided the story. And so, I mean, that's kind of rare, but at the same time it is kind of how we viewed it: whatever comes, we just sort of take it and run with it." (Steven Erikson Talks Building Malazan transcript)

The principle on display here is the inverse of the "violence to the mechanics" principle examined above. When the story required an outcome, the mechanics were violated to achieve it; but when two characters were of equal authorial importance and no story-derived preference existed between them, the dice were accepted as legitimate arbiter. The philosophical position is consistent: the rules are subordinate to story, but in the absence of a story-reason to override them, the rules produce outcomes that the author will accept. The consequence is that at least one major plot event in the later volumes of the Malazan Book of the Fallen was determined by chance and then incorporated as fact — a method closer to surrealist aleatory composition than to conventional novel-writing.

Readers familiar with the Toll the Hounds duel in question will recognise that its emotional register — the combination of inevitability and arbitrariness, the sense that either character's death was equally possible until the moment it occurred — is precisely the register of a real dice roll between two characters the writer loves equally. The feeling is not simulated. It is documented.

The Division of Labour: Marines and the Crimson Guard

A final structural consequence of the gaming practice concerns the division of narrative territory between the two authors. Erikson has explained that their respective novels grew out of which characters each had played most extensively:

"Can explored it... with Fingers in the Mines, and I'm trying to remember now how we sort of messed around with that. It's hard for me to separate the gaming from what he wrote, because I played Fingers and Blues as members of the Crimson Guard, as well as Skinner and Shimmer." (Deadhouse Gates: A Chat with Steven Erikson transcript)

The practical outcome was that Esslemont's novels tend to follow characters and institutions Esslemont had primarily GM'd or played, and Erikson's novels follow those Erikson had GM'd or played — with substantial overlap in the middle. The Malazan Empire's upper command structure (Kellanved, Dancer, Dassem Ultor, Laseen) belongs to both authors in varying degrees; the Bridgeburners and the Bonehunters are primarily Erikson's; the Crimson Guard is primarily Esslemont's. The friction readers sometimes note between Erikson's and Esslemont's voices at the level of minor characterisation is the residue of different GM-styles: the same character handled at the table by two different players over years of play will acquire subtly different traits, and when both authors later write that character in prose, the subtle differences surface.

Erikson has been explicit that the two authors have never tried to enforce strict consistency across their respective bibliographies, and has framed this as a principled choice rather than a practical compromise:

"We basically leave it quite wide open because we're not trying to sort of strictly adhere to what we gamed. What we're trying to capture in the novels is the sheer fun we had while gaming and the atmosphere that we created with our characters." (Steven Erikson Talks Building Malazan transcript)

The formulation is philosophically revealing. What the novels are attempting to transmit is not a canonical rendering of events but an atmosphere — the specific emotional quality of the decade of play from which the events were drawn. Strict consistency would preserve facts at the expense of feeling; looseness preserves feeling at the expense of facts. Erikson and Esslemont have consistently chosen the latter, and their readership has generally accepted the trade.

Conclusion: History as Lived Experience

The practical consequence of a decade of gaming prior to writing is that Erikson's fiction is grounded in a model of history that is genuinely lived, not researched. Most historical-fantasy writers construct their worlds by reading about real history and imaginatively reconstructing it; Erikson and Esslemont constructed theirs by inhabiting a shared imaginative space for years, with real player-agents pursuing real goals under rules that were deformed whenever story required it. The resulting prose treats history not as a list of events but, in Erikson's own phrase, as "an assembly of information" — the accumulated residue of countless decisions by individual agents whose motives were rarely legible to each other and whose consequences were rarely what any of them intended.

The relevance of this to the craft of the novels is direct. When readers encounter a scene whose outcome depends on three unrelated plot threads converging in one room, the convergence is often not an authorial device but a gaming memory. When characters reference events the reader has never seen, the events often occurred in a session rather than in a prior book. When the cosmology's mechanics are deliberately loose, the looseness reflects a decade in which the mechanics were always loose — always subordinate to what the story needed on a given evening. The Malazan series is, in a sense that is both literal and consequential, a transcription of a finished game whose participants are now reporting its history.

This origin does not explain everything about the novels' success — the prose itself, the philosophical commitments, the extraordinary emotional ambition of the compassion argument all belong to the writing rather than to the gaming. But it does explain several features of the work that have puzzled readers arriving from conventional fantasy. The novels' inherited density, their unexplained friendships, their feeling that events have already happened before we witness them, their tolerance for chance as a legitimate arbiter of narrative outcomes, their refusal to produce a tidy taxonomy of their magic system — all of these features are the legible trace of a practice in which the world preceded the text. The novels are not worldbuilt in the conventional sense; they are reported from a world that was already built.

Related Essays

Sources