House of Nettles: #game design

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. ↩︎

2025-12-03T22:13:12Z

Bruno Dias wrote:

Against 'Metroidbrania': a Landscape of Knowledge Games

Knowledge games tell you things, even if they ask you to make significant leaps of logic with the information they present – as in Animal Well, where some of the critical knowledge has to be arrived at by analogy, by seeing things in the environment and relating them to the player’s affordances.

A few other typical features of knowledge games:

Players are asked to build an internal model of a narrative or system, rather than just internalizing discrete bits of information. For example, The Case of the Golden Idol asks players to reconstruct sequences of events.
Knowledge is useful more than once and/or far away from the site where it's gained. In Animal Well, learning the "secret" affordances is useful throughout the game, for example; the final level in Case of the Golden Idol asks the player to understand the full story, not just the events of that single level.
Knowledge is a central resource – in a 'pure' knowledge game, the only resource. So, for example, an immersive sim having a post-it note telling you that the password is 451 does not have the knowledge game nature.
…

I enjoyed Bruno's post about the broad category he describes as "knowledge games" a lot, even if he allows the concepts of "genre" and "mechanic" to remain muddier than I would prefer^[1]. But my biggest takeaway was learning about the term "metroidbrania", which is so ridiculous I find it kind of fascinating. It suggests that "metroidvania" is becoming a term so divorced from any intrinsic semantics that it becomes a purely syntactic signifier.

Some of the games listed as belonging to this purported genre are almost luridly disjoint from anything that is typically implied by the (already broad to the point of near-uselessness) base term "metroidvania". I defy anyone to tell me what Her Story has in common with either Metroid or Castlevania^[2] beyond the fact that it is a video game. Some of the games do involve movement through a virtual space, but that's not the same as the distinctive many-keys-that-fit-many-locks pattern that the term implies.

But the fact that the term is silly isn't as interesting as the way in which its silly. It suggests that injecting a word into "metroidvania" functions as an affix converting it into a term for a genre of video games. It works much the same as adding "-ly" to an adjective to make it an adverb, or to use an example that's much more recent, adding "-gate" to a word to make it indicate a scandal.

As such, I propose that we standardize on this. Let us no longer argue about "roguelike" or "roguelite"; these games are now "looptroidvanias". The dual meaning of puzzle games will haunt us no longer now that we can say with full clarity "metroidbrania" or "tetroidvania". JRPGs are now "statroidganias", platformers are "metroidvaniups", and racing games are "fastroidvanias". Finally, to avoid confusion, we'll rename the games that are now called "metroidvanias" to "castletroids".

I also think the omission of Christine Love's Digital/Analog/Hate Plus trilogy from the list of database thrillers is borderline criminal. While it's true that they aren't as mechanical as many of the games that came after them, they're clearly important early examples of the same narrative loop. ↩︎
You could, not entirely wrongly, argue that the term "metroidvania" has already passed the point where even [the canon] is closely related to the first entries in either of the namesake series. But Obra Dinn is still no closer to Hollow Knight (to pick an inarguably canonical game) than it is to Metroid. ↩︎

Hades II Has a Legibility Problem

2025-10-04T11:17:43Z

This post contains mechanical spoilers and boss names for Hades II.

I wasn't just hoping to like Hades II, I was fully expecting to. I generally think there are two possible failure states to a sequel to a game like Hades where much of the fun is encounter and build design that can be expanded on almost indefinitely: either it turns out to be more of the same, which may be a bit uninspiring but can't be too bad if I liked the original (think Dark Souls III); or it takes the game in a wildly different direction that doesn't quite work, which may not be as fun as such but is almost always interesting (think Tears of the Kingdom). And at first it looked like Hades II was going to be more of the same, and I was content with that. But the more I played, the more it seemed to have an issue at the heart of its combat that just wasn't there in the original.

The game has a legibility problem, and not just in terms of the density and chaos of on-screen interactions (although that certainly doesn't help). This problem is structural: it pervades the game's design sensibilities. Whether it's a consequence of the game's wildly popular early access period or a result of the studio's anxiety about their first-ever sequel to their only massive hit, it pushes the player into a mode of play that's less fun and more frustrating. Legibility isn't just about being able to keep track of what's going on, it has far-reaching consequences for what types of play are safe in practice and thus which builds are viable. In a game whose primary mechanical driver is the joy of assembling new and interesting builds, that's critical.

Some of this springs from the new boss design philosophy in Hades II. In the original, bosses were generally built around melee attacks that did damage in some hitbox around the boss and ranged attacks that flew outwards from the boss. There were occasional situations—most notably when fighting Theseus after vanquishing Asterius—where large AOEs^[1] appeared that the player had to avoid, but these were very much the exception. The result is fights that are focused on the player establishing a rhythm where they dive in, hit the boss, position themselves to evade the boss's next attack, and repeat.

With the exception of some minibosses and Hecate (the first true boss), Hades II doesn't work this way. Bosses in the sequel are much more focused on two types of attacks that rarely appeared in the original: AOEs that are telegraphed by regions of the arena turning red usually independently of the boss's positions, and "waves" of damage that the player is expected to avoid using the invulnerability frames. When melee attacks do appear they tend to come out very quickly and often have their own AOEs, while ranged attacks come in one of two flavors: they're either very fast (often coming out of something that looks like a melee attack), or they're heavily telegraphed while also tracking the player's position relative to the boss.

I think each of these attack archetypes have serious issues, and I'll address several of them each as part of case studies of particular bosses below. First, though, I want to discuss the net effect of all of them in concert. Hades II is still about positioning in a certain sense, but in a much more global sense: the player is constrained by the areas the bosses allow them to exist in and the time the arena is clear. Quick melee attacks that hit all around the boss and tight boss tracking means the player's position relative to where the boss is facing is much less relevant in comparison to their distance from the boss. Arena-wide AOEs force the player far away, and damage waves require the player's dodge to be off cooldown.

When it's rarely safe to attack from nearby, ranged attacks end up structurally advantaged. The best place to be for nearly every boss is the middle distance: close enough to hit them with a longer-ranged attack but far enough away to avoid the radius of their quick melee hits and have time to dodge through any wave attacks they send your way. Only the strongest melee builds are viable, those that do enough damage to make the exchange rate on taking hits from the boss profitable; it's just not realistic to plan to do damage up close and avoid damage reliably. Of course, there are a sequence of inputs the player could do to survive, but the legibility isn't there to prompt the player room to make those inputs without tons of memorization and close attention to what the boss is doing^[2]. To understand this in more detail, let's look at some case studies.

Case Study: Cerberus

I like this fight on the whole, but that's because the mistakes it makes are gentle enough that I can trick myself into thinking it's a classic Hades-style fight about carefully keeping behind the enemy as he moves around and giving distance during critical sections. But that strategy only works because the things that punish it are mostly relatively low-damage and mostly by the time I finish the mourning fields my damage output is high enough that enduring those hits is viable. It does not forgive the sins of the design.

There are complaints to be made the half-dozen distinct AOE animations this fight employs or the fact that Cerberus's lunge attack shoots a projectile for no discernable reason, but by far the biggest issue is the way the attacks where he slams his paws into the ground work. There are two variants of this attack: in one, he lifts one paw high in the air and slams it down, creating a shockwave that goes all around him in a big circle. It's odd that this attack is specifically telegraphed by one side of the boss hits well past the other side, but it's a heavy-looking attack with a long windup by Hades II standards so it's not beyond the pale. In the other attack, he lifts both paws much lower and more briefly before bringing them down. This creates two smaller shockwaves as you'd expect, but those shockwaves still reach fully behind Cerberus and so will effectively hit the player if they're standing close to him in any orientation.

Because this attack comes out so quickly—faster than the recovery time of some of the slower points in weapon combos—it means that if you're regularly attacking the boss from directly behind you're just going to take damage from it sometimes. The solution? Tank the damage, avoid combos entirely, or rely exclusively on ranged attacks for your damage output.

Case Study: Polyphemus

I think this may be outright the worst fight in the game. It's not tremendously difficult, but the difficulty that is there is all in mechanics that encourage the player to memorize the fight and play conservatively (which is to say, not very fun). In that way, it's kind of a microcosm of the game's combat design as a whole!

This fight has, by my count, five distinct animations indicating waves of damage flowing outwards from the boss. The bread and butter one comes when his jump attack lands, where a bright red circle expands out from the landing spot—but the circle is only visible after a second or so, because it's intentionally obscured by the cloud of dust as he lands, laying a trap for players who aren't used to the fight yet or anyone focusing on the adds he summons and trusting their peripheral vision to alert them of threats. There are also smaller versions of this circle that appear overlapping one another as he walks. These are distinct from the small circle he creates when he punches the ground, though!

There's also a very different animation that happens when he throws a rock at the ground. This creates multiple thicker red projectiles which fly off in all directions in several layers, one behind another, so that the player can't easily dodge through them like they can the slower circle. Why a rock thrown at the ground would diegetically create such a dramatically different effect than a cyclops landing, let alone produce two distinct waves, are clearly not questions the designers are interested in, but these questions are important for a player to be able to easily pick up on what's happening.

The fifth and final animation is a series of circles that expand outward from him in a semicircle after stomps the ground and explode upward after a few hundred milliseconds. This attack may be his most vexing. It's diegetically incoherent again (if punching the ground creates a small symmetrical effect why does stomping it produce a massive asymmetrical effect?^[3]), it covers a full 180 degrees around the boss so moving to the side doesn't work even if the player has internalized that it's a directional attack (so why make it directional in the first place?), and the circular pattern with a width that's a significant proportion of the player's dodge distance makes dashing over it (the only real possibility for evasion) finicky. It's only made worse by the fact that the coloration before the explosion is dim, making it simply hard to distinguish visually on top of everything else, particularly in the Rivals version of the fight.

Any study of Polyphemus would be incomplete without mentioning that his jump prior to landing and producing the first wave I described above also, completely inexplicably, does damage if you touch him at all while he's in the air. The game otherwise never has contact damage for enemies—only their attacks deal damage. The result is that any attempt to stand anywhere near him during this attack is going to be punished, particularly because he can turn nearly full-circle in less than a second and jump directly behind him.

Again, this doesn't mean the fight is all that hard. Once you know how to dodge the attacks, most of them (other than the stomp) are pretty easy to avoid. It just requires playing it extremely safe and largely avoiding any attack that doesn't have good range. And if your build after Ephyra is focused on melee attacks, especially repeated flurries? You're in for a long, dreary boss.

Case Study: Melinoë

Before I wrap up, I want to talk a bit about the way Melinoë's kit itself contributes to the game's legibility issues. Legibility is about the player being able to form an accurate mental model of what will happen in the game, and part of what happens is always going to be their own actions. It's even, by this definition, about the build management aspect—whether players accurately predict how different boons or items will interact. And there are numerous places in the game where there are just arbitrary limitations-by-fiat to keep players' builds from becoming too strong. I think this is almost always a mistake in build-oriented games like this. When much of the appeal is specifically trying to find combinations of abilities that are crushingly powerful, running into a place where that power is curtailed in a way that feels artificial feels like getting your shiny new toy taken away. If a player manages to make a build that can push massive Scorch buildup if they can get in successive hits (already not a trivial feat!), seeing it cap out at 999 is a real letdown.

Some of her weapons suffer from the same kind of "no fun allowed" moments. In particular, the Black Coat and the Aspect of Artemis sword and sickle both have the ability to block incoming damage while charging their omega attack. But some sources of damage—again, just by fiat—can't be blocked. Not because they're ambient like fire or coming from the wrong direction, just because these particular attacks are flagged as unblockable. You even see a little "UNBLOCKABLE" notification pop up in the game! There is of course no way to tell which attacks these will be until you try and fail to block them, so it's just another thing a player has to memorize if they want to use those tools effectively.

Maybe the most frustrating legibility issue is actually part of Melinoë's universal loadout. The cast^[4]'s primary purpose, the only thing it does by default when not omega-charged, is binding enemies in place. But whether it actually binds them and to what extent is completely arbitrary, varying from enemy to enemy and in some cases even from move to move (the miniboss Phantom will move much slower in a binding circle during its slash attacks, but will ignore it entirely otherwise). If a player wants to plan its deployment effectively the only solution is once again to memorize which enemies get stopped, which get slowed, and which ignore it entirely and recall this in the heat of battle. And even with that done, there's no getting around the sheer misery of the first time you lay down a binding circle and an enemy walks right through it and whacks you in the face.

What's Legibility For?

Despite finding these issues tremendously frustrating when first making my way through the game, still running into them with some regularity as I learn the Rivals versions of the boss fights, and having numerous other resounding complaints about the game^[5], I'm still playing Hades II and by and large enjoying it. Legibility is, in a sense, a beginner's resource: it's there to help players understand what's going on before they have a lot of iterations, not so much to help them succeed as to allow them to feel like their failures can be understood and fixed, and so to encourage them to brush themselves off and try again. But I tried again regardless, and now that I have a lot of iterations under my belt, most of the issues I was running into early are consigned to memory.

So am I making much ado about nothing? If legibility is a problem that solves itself as the player continues playing, and Hades II's build management is compelling enough to get me—someone the friends who have listened to me rant would probably call "a Hades II hater"—to push past these issues, is this a design issue at all?

It is. What's going on here isn't just a problem that's limited to the game's first thirty hours; that's just where the symptoms are the most visible. The game's legibility issues don't exist in isolation. They weren't created because of a sequence of poor design decisions that coincidentally happened to point in the same direction. On the contrary, they serve an underlying design goal, albeit one that evinces a profound anxiety in the designers.

That goal is to slow you, the player, down. We can see this that this is a design concern very directly by looking at the Eris mechanic. Not the boss fight^[6], but a mechanic most players likely never saw: if you start making too much progress on your path to the House of Hades too early, Eris will show up and curse you, making enemies do more and more damage every room and making a "clean save" run if not impossible then nearly that. No fun allowed!

I suspect the ham-fisted nature of this approach is the reason it only appears in runs that are truly dramatically early (I don't know the number, but I'm pretty sure it stops appearing within the first ten). But the goal is still there: the designers want to make sure the player engages in all their precious dialog and meta-progression. They want Hades II to feel like a bigger game than the original, and they apparently don't have confidence in the depth of their build options or challenge runs to bring that about organically. This despite the fact that Heat was such a brilliant challenge design it was adopted by all sorts of run-based games in the five years since Hades!

In a sense, dialing down legibility solves this issue perfectly. It creates barriers to progression that will throw players back to the crossroads a few times, but because these barriers are largely solvable by sheer memorization, players can progress once they've faced a boss a few times. I'm sure the graph of average player progression per run looks beautiful on someone's office wall.

Unfortunately, these designs come with real costs. Fights become feast or famine: when so much of them is raw memorization, they fall over to players who have memorized them. And when many of the illegible attacks specifically (and surprisingly) punish attacking up close, they severely limit entire the game's build diversity, which was the core strength of Hades and remains so in Hades II. The player is forced into two choices: play carefully from range or produce massive damage. There is no other option. That in turn feeds back into the encounter design and produces bosses that are contorted more and more towards making chipping in from range as interesting as possible for the builds where that's viable, and leaves builds where it's not in the lurch.

I can't help but wonder if the game's year-and-a-half-long early access period contributed both to the manifest insecurity of the design and to the specific shape the solutions took. When the vast majority of playtesters are the most devoted players of the original game, it makes sense that the designers would worry about the game being too easy for Hades players to blast through. When you have the same cohort of playtesters for so long, it makes sense that legibility issues would fade into the background. It would take a strong design vision to push past that kind of feedback, and a strong design vision is not among this game's virtues. It's sad for me to see, but... I guess I'll just do one more run.

"Area of effect" attacks that damage the player if they're in a broad area when the attack activates and does nothing if they're outside that area. ↩︎
The Dark Souls archetype of boss design does ask for close attention to what the boss is doing and, at least in the extreme cases, memorization of its windups and timings. I like those games a lot, so what's the big deal here? A run-oriented game puts bosses in a different context than a bonfire-oriented game does. When you're spending 6+ minutes to reach the first boss and 24+ to reach the last one, the pain of losing a run because you just didn't know what was going on (as opposed to not reacting correctly or having poor strategy) is much higher than when it takes less than a minute to retry. Note that the Elden Ring: Nightreign nightlords generally have dramatically more telegraphed attacks than a typical Elden Ring boss to compensate for being in a similar structure. ↩︎
There's definitely some use in this game, as in the original Hades, of the Street Fighter conceit of martial attacks producing energy projectiles. By and large, this works fine: a punch or kick in the air, especially one that's visibly charged with energy, can be easily read as a precursor to a ranged attack. The issue arises here where the animations don't do enough to sell the specifics of the attack: a stomp doesn't feel as directional as a kick to the ground would, and expanding circles don't feel as clearly like an outcome of a shockwave as a literal wave would. ↩︎
I detest this name, by the way. Hades already had something called "cast" which is far more similar to the starting weapon's special, to the extent that it took me probably ten hours to stop accidentally taking cast boons that I thought would apply to my special. What's more, there's a resource named "magick" in the game and it has nothing to do with the cast (or at least no more than any other move). For the life of me I cannot understand why they didn't call this move "bind", since it's described as creating a binding circle. ↩︎
I'll save my bitter disappointment in the writing for my review proper. ↩︎
I don't have enough to say on this to do a whole case study. I think the fight is largely pretty solid, with the notable exception of the telegraphed aim lines. The fact that these rotate to face you much faster than the actual attacks that they're telegraphing is deeply misleading and makes the fight feel very unpleasant until you learn to evade right before she starts shooting. I also think it's the only fight I've seen so far that's substantially improved in the Rivals version. ↩︎

Power, Skill, and Silksong

2025-09-18T11:21:00Z

While this blog post does contain specific mechanical spoilers about Hollow Knight: Silksong, they're only at the very end, clearly marked, and hidden by default. Most of the post is spoiler-free.

Many games, video and otherwise, are structured in a way that presents the player with both challenges tools they can use to address those challenges. In video games specifically, much hay is made online of the challenges themselves: everyone talks about Ornstein and Smough, Absolute Radiance, or Balteus. But to the player in the moment, the shape of the tools they use determines as much or more about their actual experience.

Tools aren't just in-game items. They're anything that aids the player or allows them to engage with the game, from their character's stats to movement mechanics to the very concept of "make your character not be where the attack is". In sports, the players themselves are the most important tools. In chess, the pieces are tools but so are rules castling and stalemates. It's an intentionally broad term to discuss a broad set of game structures.

These tools are not only the ultimate determiner of how difficult the challenge is^[1], a well-designed arsenal gives players the opportunity to sculpt their experience, creating a mode of play that fits both the needs of the challenge at hand and their own personal preference. Often players end up self-sorting into two rough camps: those who choose one set of tools (a "build") and stick with it for every challenge and so experience challenges that may be easy for others as very difficult when the tools they've chosen don't line up well, and those who see each challenge as an opportunity to puzzle out the exact optimal set of tools and so take down the challenge as easily as possible. I myself fall into either camp depending on the specific context^[2].

One of the first thing a player will ask when choosing which tools to use is, "what's the most powerful?" And in some cases the answer to this is straightforward. In Dark Souls, the Straight Sword Hilt is certainly among the weakest weapons you can use by any objective measure. But in many cases the answer is unclear. There is no consensus "best" weapon in Dark Souls nor its successors, because what "power" even means depends on the player's play style, their goals, and to a substantial degree, their skill.

Defining Skill

I want to take a moment here to clarify what I mean by "skill", at least for the purposes of this post. I don't really want to get into the weeds discussing video game difficulty here and now, but I do think that discourse has made it difficult to mention skill as a concept without raising everyone's hackles and bringing in a bunch of extra baggage. So I'll try to be explicit about how I'm using the term here.

I'm not talking about skill as in some sort of innate talent at video games or even specifically hand-eye coordination or twitch reflexes. What I mean is something that comes with practice and, especially, shedding the fear of the unknown.

Every player, no matter how good or bad their reaction speed, is worse at the game in the initial phase when they're still getting used to it. Everyone loses some percent of their possible effectiveness from needing to manually think through which actions to take, from not fully understanding the challenge they're currently facing, and from not knowing how to match those actions up with that challenge. I'd go so far as to say that for almost everyone, this lost effectiveness is so large that it makes up the vast majority of the reason they fail the challenges with which a given game presents them.

The other side of this coin, then, is that players can increase their effectiveness simply by paying attention to their experiences and internalizing them through practice. This scale of practice-driven effectiveness is what I mean here by "skill". I certainly don't mean it as any measure of the player's "quality" in a moral sense or whether they've "really" engaged with the game. There's nothing better about spending the time practicing a game so you can beat it blindfolded except inasmuch as that experience is intrinsically rewarding to the player.

Power and Skill in Dialog

What's the more powerful ring in Dark Souls, Tiny Being's Ring which increases the player's max HP by 5% or Red Tearstone Ring which adds 50% attack power when the player is below 20% HP?

It's tempting to say that the answer is Red Tearstone Ring hands down. It's a staple of all the most challenging ways people play the game: Soul Level 1 runs, hitless boss fights, and speed runs. Tiny Being's Ring on the other hand is available as a starting item and is handily outclassed by numerous other rings which the player could use in that slot.

I don't think this answer is wrong at all, but I do think it's incomplete. For an unskilled player, Red Terastone Ring might as well have no ability at all. When the whole game is full of unknown monstrosities with unfamiliar moves, the safest course of action is to never be below 20% HP if you can help it. Any time you find yourself that low, your first priority should be finding a safe spot to heal. Tiny Being's Ring may not have a dramatic effect, but if that little HP bump saves you from one hit that would have killed you otherwise it's at least doing something.

We can think of power as a way of talking about what's most effective at helping a given player overcome challenges with the least amount of effort. From this perspective, how powerful something is depends on the player who's using it. This sort of "power" doesn't make sense as a metric in its own right—for a low-skilled player, Tiny Being's Ring is stronger than Red Tearstone Ring. For my friend who never uses spells in Hollow Knight, the underwhelming Grubsong^[3] charm is stronger than the excellent Shaman Stone^[4].

But this is a little unsatisfying, because games are fundamentally social even when they're single-player and tightly individualizing the concept of "power" cuts off any ability to discuss it with others. What I think most people mean in practice when they talk about it is something more like power pegged to some sort of implicit "average player", although where this player falls specifically in terms of skill and preferences is left unspecified, sometimes to the result of confounding the whole conversation. If this average player is pretty good at Dark Souls, maybe Red Tearstone Ring is better for them because they might be able to survive long enough to get in some hits at 20% HP, but it's still much weaker than one of the better options like Ring of Favor and Protection^[5]. I think if you posited this to a player, they wouldn't argue.

The power of tools affects skill as well. I talked about a tool's power as its effectiveness in helping a given player overcome challenges with the least amount of effort. In turn, exerting effort to overcome challenges is how the player builds skill. So we can see tools as lowering the skill bar for facing a given challenge, and we can see their power as how much they lower it^[6].

Another Perspective on Power

I want to look at this from a slightly different angle, though, inspired by my recent experience playing through Silksong and discussing the merits of various builds with my friends. I want to look at which tools are best the first time through a game. In other words, I want to view a tool's power as the lowest overall skill required to overcome a challenge with that tool across all starting skill levels. Red Tearstone Ring rates poorly on this metric because it only becomes strong in the hands of players whose skill levels are already high, but Tiny Being's Ring rates poorly as well because it doesn't lower the skill requirement much for anyone.

This is particularly interesting in the context of Silksong because it's full of tools that are at their best when the player is already skilled. Hollow Knight's spell mechanic is the same way: you accumulate "soul" by hitting enemies, and you can spend it either to heal or cast spells. While spells can do a lot of damage, using them concretely eats into your ability to survive fights.

So, are spells powerful in Hollow Knight? By this metric, which I'll call "initial power", I think they're good but not amazing. Their damage output is big enough that learning to use them effectively can shorten some fights enough to make up for lost heals, but not so strong that you can rely on only using them and never healing without investing a lot more time into building skill.

By contrast, we can also talk about "final power": the power of a tool once the player has built up their skill and understands the game's challenges. (We can frame this as the tool with the biggest overall reduction in skill required to overcome a challenge with that tool across all skill levels.) Hollow Knight's spells rate very high here, because once you're getting hit infrequently enough they can massively accelerate fights and reduce the overall risk.

The Part Where I Talk About Silksong

I'm not going to spoil anything you can't find in Act 1, but I will talk frankly about certain mechanics.

Reveal spoilers

There are so many Silksong mechanics whose power scales up with the player's skill: skills work just like Hollow Knight's spells, of course, but tools are also easier to stomach incorporating into your routine if you're confident you'll win a fight before having to go farm more of the crafting materials needed to create them. Some crests place restrictions on healing which are mitigated by better positioning skill (as well as of course taking less damage). The upgraded Hunter crest directly scales its damage with how infrequently you get hit. At the far end, Barbed Bracelet can double your damage at the cost of doubling all damage you receive in what amounts to this game's own Red Tearstone Ring.

But this line of thinking was most proximately inspired by thinking about the Reaper crest. If you're not familiar, this crest gives Hornet a moveset with very long-range swings including a pretty straightforward downward aerial attack (which is incredibly important in these games for both platforming and dealing with enemies) in exchange for being quite slow.

I heard someone (tentatively) call Reaper a "noob trap", but I don't think that's right. I think rather that it has a high ratio of initial power to final power. Reaper's long range isn't just comfortable (although it is also that), it's forgiving. It allows players to be messy in their positioning and inaccurate in their guesses about what enemies will do next with much more generosity than other crests. And when you're fighting new enemies or traversing a new platforming challenge, that's extremely valuable. On the other side of the same coin, the cost of pushing less damage is relatively lower early on because it's not always obvious when it's safe to land multiple attacks anyway.

Reaper (along with tools like Warding Bell that similarly forgive sloppiness) are perfectly suited to playing in an exploratory style, learning the rhythms of enemies and fights that you're seeing for the first time. They provide initial power which lets players build the familiarity they need to take full advantage of other tools that are better at higher skill levels. It's not so much a "noob trap" as a "noob aid". More importantly, the concept of "initial power" and "final power" help us understand why that is.

The phenomenon of the Dark Souls Soul Level 1 run is a sort of inverse example of this. The player chooses to specifically limit their toolkit by never leveling up (character level itself being of course a tool) in order to dial in a harder difficulty on a game they've already played, and so experience it from a different perspective. ↩︎
When I played Shin Megami Tensei III: Nocturne, I would look up each fight and meticulously plan out builds hours in advance, carefully breeding the exact right demons to meet every challenge I faced. On my first run through a Soulslike game, I instead tend to stick pretty assiduously to a single build. Interestingly, I'm playing Silksong more like SMT than like Dark Souls: I have a main loadout I mostly use for traversal, but I'll switch that up to some degree for most boss fights. ↩︎
You gain soul (a meter used for healing) every time you get hit. The rate on it is appalling. ↩︎
Increases both the size and damage of all spells. I got a lot of mileage out of this bad boy in my playthrough. ↩︎
Increases max HP, stamina, and equip load by 20%. Incredibly strong, but notionally "balanced" by the fact that it breaks forever if you take it off. Here's by Dark Souls hot tip of the day: don't take off Ring of Favor and Protection. ↩︎
Again, I want to be clear: I don't think this means that players who use all the tools available to them are "worse" in any sense at all. Finding ways to lower the skill bar is the game just as much as building skill to meet that bar. A player choosing where in the interaction of power and skill is most satisfying for them is player customization just as much as choosing a Greatsword because it makes them look like Guts. ↩︎