Marathon and the Thrill of Losing

2026-03-19T04:57:44Z

I wasn't planning to play Marathon.

Christa, a Bungie lover of the old school^[1] and particular aficionada of the original Marathon trilogy, talked it up to me non-stop since the announcement. Undeterred by the decidedly underwhelming closed technical test, the delay that that test prompted, or the plagiarism scandal of the visual design^[2], she kept excitedly sending me updates and videos. I read along interested enough, but it did little to make me want to pick up the game myself.

I'm not much for shooters. That's not to say I haven't played or enjoyed them; I played through a couple James Bond games^[4] and the Halo 2 campaign as a child, I played Splatoons 1 and 2 for a respectable number of hours, and I even cleared the original Destiny single-player content while recovering from surgery. But these games washed over me like waves; none of them inspired any particular affection for the genre or desire to play the latest thing. Certainly they are far outnumbered by the big-name shooters I've touched barely or not at all—Doom, Quake, Half-Life, Call of Duty, Team Fortress, Fortnite, Overwatch, or indeed the original Marathon games.

When the open server slam came, I didn't play it, even as more friends beyond Christa were starting to admit it might have the juice. When the game launched, I didn't get it, although I quite enjoyed watching friends stream over Discord. It was those streams, I think, that did it. Being in the moment with someone, feeling the ebb and flow of tension and release, and seeing how much of the game wasn't shooting, the thought started wriggling its way into my brain.

"What if I did play Marathon?"

As I continued watching my friends, as Christa continued sending me videos of feats of meticulous planning as well as derring-do, as I learned more about what the structure of an "extraction shooter"^[5] meant in practice, this thought grew. I found my modding work in a lull, waiting for upstream changes and code reviews, with no particular video game on deck^[6]. I decided to give it a try.

What really started singing to my soul and got me to spend the $40 USD wasn't even necessarily the prospect of hand-fun from playing the game, but rather mind-fun^[13] from engaging its design with a critical eye. Christa is fond of repeating the idea that extraction shooters are a "game designer's genre", but it wasn't until picking it up with my own two hands that I really began understanding why. The last time a game has given me this much insight into the relationship between mechanical design and player experience was Resident Evil GCN. Everything in this game is part of the texture of interactivity in a way that's just not true of other genres; every sound design choice has repercussions on how other players might hear you, every piece of level geometry is a place to hide or a vantage point to look for other players. Small changes have massive ramifications: tightly limited inventory space means that high-level players have to drop their good items to pick up great items, which means that low-level players can expect to scavenge good items after a fight is over and the enemy has left, which means that mid-level^[14] players can prey on low-level scavengers to get their loot they've accumulated. The design choices create an ecology of players with different goals, approaches, and reactions that's far more varied than any NPC AI could hope to achieve, and it does so as a natural outgrowth of the system rather than a structure imposed from on high.

But I don't have room in this post^[18] to go on at length about every aspect of this game's design I find fascinating. I want to focus instead on one particular aspect: the way the game handles loss.

The Sweat

I personally have never struggled much with the despair of losing my equipment in this game. Maybe it was the fires of Magic: the Gathering that forged my emotional fortitude, where the difference between a win and a loss was at the mercy of the top card of my deck^[19]. Maybe it was seeing other players at those tournaments tilting to the point of public meltdown and resolving early to learn to bend with the good and bad but never snap.

Bending with it is crucial, though: I gotta feel the emotions even though I try not to let them overwhelm me. The point of a game, after all, is the emotions^[23]. I don't just want to go through the motions when I play a game, I want to care about the outcome. I want to feel the sweat.

The sweat is the sense of anxiety that comes from a combination of caring about the outcome of a game and knowing that outcome to be uncertain. Like the sense of fear from watching a horror movie (or more directly, the tension from watching a thriller), it's a way to play-act uncomfortable emotions in a space whose unreality shields the player from real consequences. Unlike a film, though, the outcome depends in part on the player's own actions. The sweat is the feeling that you hold something fragile and precious in your hands and it's up to you to keep it safe. The sweat can end in one of two ways: you can succeed and make actual the potential of your treasure, or you can fail and smash it to bits.

The possibility of failure is critical. Without it, there can be no sweat, no tension. And without tension, there can be no release.

Extraction Shooters and Roguelikes

In the first few days of Marathon's release, when a few of my friends were beginning to pick it up, I heard the same advice bandied about from a few places: think of it like a roguelike. You know the core conceit of a roguelike, right? Perhaps not. More likely, you don't know which of the constellation of traits that cluster around this genre-concept I might specifically be referring to. Here, I mean it in the broad sense, what might be called a "roguelite"^[24] by some: a game organized into individual runs whose progress is largely divorced from one another and lost forever if a run fails.

Losing a run in a roguelike can be painful, but at the end of the day all you're losing is your shot at an abstract sense of victory. In games with meta-progression across runs, the bulk of that progression (especially early on) moves forward on failure as well as success. Even in games that start each run from absolutely the same state, every new run is a fresh chance to learn a little bit more about the game and get a little better for an eventual victory.

The extraction shooter structure is very similar to a roguelike from a sufficient level of abstraction^[25]. It's also organized into individual runs and losing a run also loses your entire build. "Think of it like a roguelike" helps frame that loss as something familiar, a learning experience, a chance to start again.

But there's a critical difference between the two structures. In a roguelike, nothing carries over from one run to another. In an extraction shooter, that carrying-over is the whole point^[27].

There are two ways to look at the difference between a run in a roguelike and an extraction shooter. In one sense, an extraction shooter is more punishing: if you want a good chance of success, you must put resources at risk. Runs in a roguelike are "free" in that you can fail without any consequence beyond the time you spent, whereas runs in an extraction shooter are not. In another sense, though, an extraction shooter is far more generous: no matter how fantastic your build, it's lost forever even when you win a roguelike run. An extraction shooter lets you keep that snowball rolling downhill for run after run, and even when you lose, any additional resources you've accumulated through your success will give you a leg up on your next few runs.

These perspectives are two sides of the same coin. The gear you take in only matters because it's drawing on the pool of resources you've extracted; extracting resources only matters because they're risked in future runs. From a strictly mechanistic standpoint, you can lose no more than you've already gained, which is definitionally more than you would have had in a roguelike, so each loss should be painless. But from an emotional standpoint, we know that's not true.

Soulslikes and Dungeon Delvers

A parallel difference exists between the soul mechanic pioneered in Demon's Souls (made popular by Dark Souls) and the traditional way dungeon delvers handle the loss of experience. In Demon's Souls, experience points are called "souls" and like most RPGs you accumulate them by killing enemies. Unlike other RPGs, though, you don't automatically level up when you accumulate enough souls; you can only spend them and level up once you reach the safety of a checkpoint, which may be somewhat hidden within the intricate and twisting maps. If you die, your souls are lost, but not necessarily for good: if you make it back to the point where you died without dying again, you can retrieve your lost souls and spend them at the next checkpoint^[28].

A common refrain among new players of these games is that losing their souls is painful. For some, it's too painful to bear, and leads them to bounce off the genre as a whole. But let's take a look at these mechanics from an abstract perspective as well, especially in contrast to the classic dungeon-delving RPGs from which they draw influence. The classic RPG structure also has checkpoints (in the form of save points) and experience. If you die, you go back to the last save, and everything you've done since then is erased.

We see here the same dichotomy that we did before between extraction shooters and roguelikes. From one perspective, soulslikes are more punishing than traditional RPGs, because they set you up to see a great many souls vanish because of your own failure. In another, they're gentler, because they give some hope of recovering those souls, even if that hope might be dashed. And again, what truly matters is the player's emotional experience.

Losing Rules

A lesson that dedicated soulslike players learn fairly quickly is that this dichotomy doesn't actually matter because the souls you accumulate don't actually matter. The amount of souls you need for a given level rise quickly over the course of the game, so just wait a couple regions and you'll be getting as many souls as you lost just by killing a couple mobs. People regularly go through these games never leveling up at all^[29]!

When I play, though, I try not to let myself fall into thinking of souls as unimportant, even though that's entirely accurate. I actively try to get back to that big pile of souls just past three knights who want my blood even if in reality they won't make a bit of difference to how effective I am in my next fight.

It sounds obvious when I write out: games are more fun if you care about them. I want to care about the texture of the world in a soulslike, I want to care whether I live or die, so I trick myself into thinking that little number in the corner of the screen has value.

The brilliance of an extraction shooter, just like the brilliance of the soul mechanic, is that by creating something just so that that thing can be put at risk it makes the player care. The souls you lost matter in a way the experience points you lost don't because the potential to get them back makes them your souls, even if they're ultimately lost anyway. The gear you bring in an extraction shooter is your gear because you chose to risk it, even if you stole it off someone else's corpse in the first place. You have a stake in the game, and that stake is the wellspring of the sweat.

None of this would be possible without loss. In fact, engineering situations where tougher, more impactful losses are possible is critical to making this dynamic exist in the first place. And not just possible in theory—loss has to be real, a specter haunting your every move, to truly make these games feel alive. You have to experience it, not just at first but continuously, reminded that you are never truly safe, in order for the moments when you find a bit of safety to really shine.

Having played the game now for a week or so, what I was truly surprised to find thinking back on my wins and my losses, was that I remembered far more vividly the runs where I made it out with some spectacular gear than I did the runs where I lost that exact same gear. In fact, my most memorable losses were never the ones where I lost the most stuff but rather the ones where I played the best games, tactically positioning and thinking on my feet and being just barely outplayed by my rival runners. Maybe I had great gear in those runs too, but if so, I don't remember it. What I remember is the thrill of losing a battle well fought.

There is a particular class of people who react to the mention of Bungie as a company by getting a wistful look in their eyes and saying something like "oh... Bungie..." as though they had vanished like the Mary Celeste after releasing Halo: Reach. ↩︎
A brief recap for the uninitiated. Marathon's visual design is not just striking, but probably the best a AAA game has looked in a decade. It's not the graphical fidelity, although I have nothing to complain about on that score, but the establishment of an overarching design language that feels futuristic while being in conversation with real-world visual design^[3]. It's a game that can be appreciated as visual art beyond the level of "wow pretty landscape", which is nearly unheard of in western AAA.

It is also extremely similar to the work of the digital artist Antireal, who as it turned out was followed on X by the Marathon design lead along with other members of the team. When some pixel-for-pixel Antireal art was found in the game's asset base, the scandal broke full-force, and a junior designer ended up fired. Fortunately, the story has as happy an ending as it can: Antireal was hired on as a design consultant, she's expressed satisfaction with the arrangement, and the game's visual design has only improved since then. ↩︎
The Designers Republic is its own story, a firm so colossally influential on the look we now think of as "Y2K" that it became a visual cliché. In Christa's words, the ability to build on that design in a fresh and interesting way "speaks to how well it is pulled off in Marathon". ↩︎
Even then I entirely missed Goldeneye 007, to this day the definitive video game incarnation of the storied imperialist running dog. ↩︎
For those unfamiliar with this relatively new genre, it functions similarly to a battle royale in that multiple players or teams are placed on a shared map where they can shoot one another^[7]. However, the goal isn't specifically to kill other players; it's to accomplish missions and collect resources from the map, often by doing challenges not involving other players such as killing powerful NPCs. You bring resources you've accumulated, including weapons and armor, into future runs to accomplish harder goals, creating a risk/reward dynamic both in the challenges you attempt and in how tempting a target you make yourself for other players to hunt. ↩︎
This is, I admit, not entirely true. I still haven't played the most recent two chapters of Deltarune, and too many people have told me how excellent Of the Devil is for me to ignore it just because I find the art style unappealing. But sometimes one's fingers itch to play a game with a sense of motion^[9]. ↩︎
Different extraction shooters vary in how much they encourage player-on-player combat, although a universal constant is that killing a player allows you to loot the corpse and take all their best items. Marathon is reportedly pretty far towards encouraging bloodthirsty play, as is Escape from Tarkov^[8]. By contrast, ARC Raiders is known for being relatively low-conflict. ↩︎
The progenitor of the "extraction shooter", and another foundational shooter that I've never played nor had any real interest in. ↩︎ ↩︎
Despite its motive fundamentals, though, Marathon is decidedly menus rather than parkour. This should come as no surprise given that Bungie's other active game^[10] is explicitly called out as menus in the original comic. ↩︎
I was surprised to hear that some Destiny 2 players are grumpy^[11] that Marathon exists instead of Bungie spending more time on their pet game, and are making their views vocally known on Twitch chats and social media. Specifically, I was surprised that there were still enough Destiny 2 players in existence to make this amount of noise, given that the peak monthly player count in 2026 was just a third of what it was in 2025. Personally, I think it probably makes sense for a company^[12] to work on a game that people actually seem interested in playing. ↩︎
apoplectic ↩︎
subsidiary of Sony Interactive Entertainment LLC ↩︎
Surely this is a distinction that needs no further explanation. ↩︎
Framing this in terms of "levels" is actually an oversimplification. In practice, it begins as more of a question of which role a player is taking on. Are you going in with the intent to complete quests? To accumulate resources? To hunt other players? Where is the fun for you? Even this is too simple, though, because the roles can shift on a dime based on what happens in the field. A run that was intended to be a zero-stakes^[15] player hunt can immediately become a high-stakes meticulous exfiltration if you find loot you care enough about. Any run can become a PvP run if there are players in between you and your objective. ↩︎
An important part of Marathon's ecosystem, which as I understand it doesn't exist at all in Escape from Tarkov^[8:1]. The player has access to (at least) one faction's^[16] "free kit" every day, which is a preset package of a gun with no upgrades, a handful of low-quality healing items, and some ammunition. This is an awful kit, and what's more it forbids the player from bringing any additional equipment with them, but it provides a place to start from zero resources and it even gives players with a vault comfortably full of equipment the emotional comfort of risking nothing.

Perhaps paradoxically, free kits go a long way to acclimating players who don't come in with the taste for blood to player combat. A free kit may be decidedly disadvantaged in an out-and-out firefight, but raw firepower is way less important than positioning and tactics. Free kits make it viable to hunt players even when you expect to lose the first four encounters. As I'll get to later in the post, loss has its virtues. ↩︎
Successfully exfiltrating with a free kit also gives the player a cosmetic recolor of their current shell^[17], leading to days when a player has access to one of the more aesthetically satisfying free kits and so wants to do runs with it over and over. This sideways benefit cleverly undermines the potential for free kit use to feel like a failure; even once the player has as many skins as they care to get, they're acclimated to free kits as a low-pressure alternative to bringing in real gear. ↩︎
class ↩︎
Certainly not if I keep writing more footnotes than I write body text. Sometimes ideas are a straight line, and sometimes they're a tree. Today is a tree kind of day. ↩︎
The tension of the draw step^[20] as driven by the land system is in fact the heart of why that game engine is as successful as it is and has as much depth as it does. Compare to the Vs. System where tournament results were famously always a list of the best players in order of skill and no one ever wanted to start playing only to lose constantly; compare also to the World of Warcraft TCG where the variance was so concentrated in the opening die roll to determine who got the first turn that the winner of that roll was wildly favored to win the game. Magic sits at a comfortable point where the very best players have something like a 65% win rate against other pros; real, respectable, but not dominant. ↩︎
It's also worth noting that the amount that a given card draw determines the course of the game is never quite so dramatic as it seems in the moment. Part of the genius of Magic is its lenticularity^[21]: all the decisions you've made in the game up to that point contribute to how much weight the next card has to your game, and an expert player has room to line those decisions up in a way that is completely invisible to players who see each new card as a wholly disconnected event. ↩︎
When I was young, I was a huge reader of Mark Rosewater's blog. These days I'm a lot cooler on him as a writer and even as a designer, although I think a lot of what I find off-putting in his latter role is due to the tension between his dual roles as design educator and product cheerleader for a company^[22] subject to inevitable capitalist rot. I deeply wish he found more space to grapple with what was wrong with his own explanations in the past rather than presenting direct contradictions as simple fact. But I do think this article in particular contains fundamental insight into a critical aspect of design, not just for games but for anything whose goal is to provide a path for people to move from beginner to expert. ↩︎
subsidiary of Hasbro, Inc. ↩︎
The point of a game is also engaging with people. In fact, the human connection is probably more important for me than any individual aspect of my own personal play. Even single-player games (even novels and films) I conceptualize as interpersonal acts—I'm connecting with the creator, with everyone else who's experienced the same thing through the perspective of their own prior experiences, with the novel angle on the human experience that every piece of art brings. Because our manifestation of sentience is itself catalyzed by human contact, all conscious articulated thought is intrinsically interpersonal.

I don't consider this incompatible with the idea that the point of the game is the emotions, though. Those emotions are an aspect of the conscious interpretation of the game as a space, and they're the deepest expression of the game's effect on us that can be shared between different players. The emotional outcome of playing a game is the foundation of any possibility for discussing that game with others. ↩︎
Personally I find the impulse to try to make genre names perfectly accurate pretty Quixotic. Terminology as used in practice is always imprecise, and while I find thoughtful discussions of the specific ways in which games fit into our squishy ideas of genre fascinating, I think of the specific names more as opaque identifiers of those shared concepts than anything I'd expect to be accurately descriptive in itself. ↩︎
Boiling games down to basic relationships between coarse-grained mechanics is a pretty bad way to understand them as entire works of art, but I find it very useful as a way of placing them within the larger landscape of design and especially genre. By virtue of themselves being abstractions across many games, genres are often^[26] expressed in these heavily mechanistic terms, and thus to make themselves legible to players most games involve at least some thought of how they'll be expressed in the "elevator pitch" style of conjunctive description. ↩︎
Even the "survival horror" genre, whose name is so focused on its emotional content and which at first blush suggests a strong aesthetic grounding, ends up being defined primarily in mechanical terms: they're games about carefully managing resources too limited to deal with all the game's threats. The fact that this lends itself well to a mood of tension and fear almost feels like a happy accident. ↩︎
There's an interesting question lying latent here: how much is our understanding of "the point" of a genre driven by our expectations of other genres that surround it? Extraction shooters come from a particular moment in game design as a medium, in the shadow of battle royale and roguelike as ascendant genres. It makes sense that we name and frame it in contrast to those, when the key distinction is taking stuff between runs. But you can also imagine a world where something more like the traditional dungeon-delving RPG is the genre du jour and "leaving with more than you came with" is taken for granted as the point of a unit of play. How would we describe an extraction shooter in that world? ↩︎
You can think of this as motivating the player to beat their own high score. The game is, in a sense, divided up into little stretches of challenge between checkpoints; if you make it a little further each time you try to move from one checkpoint to another, you'll get your souls back, demarcating a success (even if a qualified one). In this way, the genre superimposes the arcade-style play pattern of "repeat this challenge until it goes from insurmountable to possible to complete" onto a each section of an action-RPG. ↩︎
I've done this almost three times: once for Elden Ring, once for Dark Souls, and not quite to the end of Sekiro^[30]. The substantial majority of these runs weren't even very hard—these games are made to be beaten. The difficulty sliders that people howl for do exist, they're just part of the play of the game itself. If anything, it's easier to beat them at level 1 but with everything other than that set up to be the easiest it possibly can than to beat a casual playthrough you're not trying particularly hard to optimize for power. ↩︎
Sekiro is an odd game because it doesn't have a single definition of "level" that clearly matches other games. Instead, it has three separate levels for attack power, health/defense, and unlockable skills. I only leveled the latter, which means that my damage output was such that I'd need to spend 15 minutes fighting near-perfectly to clear the final boss, something I have not yet found the stamina to train myself to do. ↩︎

House of Nettles: #marathon

Marathon and the Thrill of Losing

The Sweat

Extraction Shooters and Roguelikes

Soulslikes and Dungeon Delvers

Losing Rules