Onimusha: Way of the Sword and Capcom's 2026 'Length as Value' Test

In the notebook I keep for the week — an old habit from the years I spent picking apart every passage of play on video — one line made me stop longer than it should have. The main campaign of Onimusha: Way of the Sword is described as running 30 to 40 hours on the Action difficulty. Players chasing full completion need more than 50 hours. Sitting inside that span are 36 boss encounters. Three data points on the same page, with no methodology note, no cross-reference, no distribution by chapter.

I have no problem with an action hack-and-slash game having plenty of bosses. My problem is with the way those data points are placed side by side as if they automatically add up to value. Eleven years of watching sport and esports taught me something uncomfortable: people tend to measure whatever is easiest to count and then call it quality. Minutes played. Shots taken. Hours played. All of them are handsome numbers on a spreadsheet, and all of them can be bent by one clever sampling choice.

Meta in esports is not invented by anyone — it reveals itself when someone bothers to do the maths. In this dossier, what reveals itself is not a competitive meta. It is the way a publisher prices a product in hours.

Context: one single-player action title, three games in one year

Onimusha: Way of the Sword marks the return of a Japanese action franchise sitting inside Capcom's catalogue. The setting is Edo-period Kyoto, and a side-content layer called Curious Cases is designed to lead players into the mythology and legends surrounding the Eastern capital of that era. The protagonist is Musashi, and his progression runs along two axes: his sword and his clothing, both upgraded using iron and leather.

This is a single-player action game. There are no teams, no tournaments, no competitive patch cycle, no meta, no win rates, no pick-ban. Labelling this dossier as esports is a classification error, and I will return to that at the end, because it has real professional consequences.

What I am analysing sits elsewhere. The source is purchase-guidance content: its readers are people who have just started playing, or who are weighing up spending money and want to know how long the game runs. That genre has a structural trait — it prioritises search traffic over methodological rigour. Every figure in it is one author's estimate, not aggregated player data.

The dataset I extracted has six items: a main story of roughly 30 hours; 30 to 40 hours on the Action difficulty; more than 50 hours for players unfamiliar with action games; 10 to 15 hours of optional content; 36 boss encounters; and a platinum trophy requiring at least two playthroughs. Capcom places this game alongside two other 2026 releases, Resident Evil Requiem and Pragmata, and positions it with one short line: its campaign is longer than those two combined.

Structure: one boss every fifty minutes

The simplest division is also the most telling. Thirty hours, thirty-six bosses — roughly one encounter every fifty minutes.

That density is itself a statement about structure. The typical cadence of the hack-and-slash genre tends to introduce bosses in tiers: an opening fight, a few mid-chapter fights, one closer per chapter, and a final encounter. A density of one every fifty minutes pushes the game towards a boss-rush structure rather than an exploration adventure.

There are two ways to read that figure, and both are reasonable at this stage.

The first reading is a design reading. If the combat system carries enough variables — weapons, block timing, recovery resources, enemy states — then high density turns every boss into a new question. The best system does not produce superstars; it produces perfect roles. A good boss fight is not defined by its existence, but by whether the system leaves enough space for the player to rethink how they play.

The second reading is a length reading. If thirty-six bosses spread across thirty hours, then the adventure, exploration and storytelling between them must be thin. For a game set in Edo-period Kyoto, with an entire side layer devoted to mythology, funnelling pacing budget into boss arenas is a trade-off, and buyers should know what they are buying.

The source never defines "boss". A boss in this dossier could be an entity with its own health bar, an elite enemy with its own move set, or a fully staged set-piece. Those three categories deliver three very different experiences from the same figure of 36. That is the largest information gap in the whole dossier, and it matters more than the total.

Length is a property of the player, not of the disc

The gap between the two player groups here runs from 30 to 40 hours on one side to more than 50 on the other. A spread of nearly 67 percent.

This is data about skill, not data about content volume. The same disc, the same storyline, the same thirty-six bosses, and two players walk away with experiences twenty hours apart. The headline "more than 50 hours" therefore tends to hide the most important thing: most of that extra time comes from dying, relearning move sets and adjusting your approach — not from new content.

Looking back at the work I once did with a statistician during the period when leagues had to play in empty stadiums, I see a rather odd parallel. We compared 76 matches played without crowds in bio-secure venues against 76 matches involving the same teams in the previous season with crowds. Home possession rose from 51.2 percent to 54.1 percent, but expected goals per shot fell from 0.11 to 0.08. Empty stadiums give you data, but they take away the thing data cannot measure: noise.

The lesson transfers intact here. Difficulty is the single variable being changed, and the entire length system shifts with it. When one variable explains most of the variance in a dataset, presenting that data while ignoring the variable leads readers to the wrong conclusion.

Players with action-game habits will get a dense, continuous experience. Players who have never touched the genre should prepare for a game around half again as long as what they read on the box. Both groups are right. Only the headline is wrong, because it merges them into one.

Some of the optional content is not optional at all

This is the finding I consider the most practically valuable in the entire dossier, and it is buried fairly deep.

Upgrading Musashi's sword and clothing requires two materials: iron and leather. Those two materials come from side quests and from a challenge called Ace Archer. The source describes them with a fairly strong adjective: essential.

This means the layer described as optional, running 10 to 15 hours, is effectively split into two tiers. One tier is genuinely optional: the Curious Cases carrying mythological content, treasure hunts, chance encounters, and the Lion Dogs — the creatures the source calls the most endearing in the game. The other tier functions as a power gate: side quests and the Ace Archer challenge, where upgrade materials sit.

The direct consequence: players who run straight down the main story risk walking into later boss fights under-geared. This is not a hypothesis awaiting community data. It follows from the material structure as described, and it is testable within the first few hours of play.

With one boss every fifty minutes, being under-geared at the midpoint stops being a small matter. It creates a loop: players with weaker skills need more materials, players who need more materials must do side content, and their completion time inflates. The 67 percent gap between the two player groups I analysed above does not come from hand skill alone. It also comes from reward structure.

The same reading explains why the Curious Cases are rated highly by the source while sitting outside the optimal path. They carry narrative value and completion value, not power value. An action game that places two kinds of reward in two different content tiers is telling players something clear about what it wants them to experience first.

Two playthroughs and the doubling problem

The platinum trophy requires finishing the game at least twice. That doubles the time commitment for the completionist crowd.

There is a detail easily missed in that structure. The second playthrough is almost certainly a speed-optimised run rather than a run of new content. The story is known, the boss layout is known, the route is known. Don't ask how good the player is; ask how the system protects them — and here, the system protects players with two things: upgrades carried over from the previous run, and accumulated knowledge of enemy move sets. If both conditions hold, the second run is far shorter than the first. If they do not hold, completionists will have to replay a thirty-hour game from scratch, and that is a variable the source does not confirm.

I flag this as a design risk of medium probability and low impact. The platinum-hunting group is small but loud, and their frustration scales with how much they feel time was taken without anything new given back.

Three games, one year, and an unverifiable comparison

Capcom's positioning line for this game is that it is longer than Resident Evil Requiem and Pragmata combined. The source provides no length figures at all for those two titles.

A comparison without a denominator is not a measurement. It is a way of reserving shelf space. That does not make it false, but it does make it unverifiable today. The only way to verify it is to wait for aggregated completion-time data across all three titles after release and place them on the same scale.

The catalogue structure itself is the more analysable part. Three single-player action games from the same publisher launching in the same year is a bet on the market's wallet capacity. On one hand, it shows the publisher still backing the buy-once model rather than the live-service model. On the other, it puts three products into internal competition. When three games compete for one entertainment budget across twelve months, the line about being longer than the other two stops sounding like pure boasting. It becomes a shield.

I hold this observation at medium confidence. It is an inference from structure, not from sales data.

The contrarian angle: where I could be wrong

I argue that most of the risk sits in pacing and in the way the data is presented. But there are at least three points that could collapse my argument.

The first is the boss-definition problem I raised. If the thirty-six encounters in the dossier include both major fights and shorter elite enemies with their own move sets, then one boss every fifty minutes stops being a concern. A layout with quick rhythms, many short peaks and a few long ones is a very common structure in Japanese action games, and it works well. In that case, my worry about repetition is aimed at the wrong target.

The second lies in the quality of the data itself. Every time figure in the dossier is one author's estimate, with no methodology, and there is an unresolved internal contradiction: one place says the main story is about 30 hours, another says 30 to 40. If aggregated community data after release returns a figure above 40 hours with genuine variety, my entire pacing concern loses its footing, and the thing I have to revise is not the conclusion about the game but the conclusion about the source.

The third concerns my own reading of the reward structure. I assume upgrade materials come mainly from optional content. If the game offers a plentiful supply of iron and leather along the main path, my concern about being under-geared at the midpoint disappears, and the 10-to-15-hour layer returns to being genuinely optional. In that case, the finding I rate most practically valuable in this article becomes the weakest one.

I leave all three open. They are the places an honest analysis must point out before someone else points them out.

A classification error with professional consequences

There is a technical matter I want to record, because it relates directly to my job.

This dossier was tagged as esports at the upstream classification layer. There is no competitive element anywhere in the data: no teams, no tournaments, no patch cycle, no win rates, no pick-ban, no prize pools. This is a single-player action game, and every data point in it belongs to single-player content.

The consequence is not small. If dossiers like this flow into competitive analytics datasets, they contaminate every comparison of patch cadence, roster strength and tournament cycles. I have filed a technical note recommending this group be reclassified as single-player and separated from the esports data pipeline.

On the content side, treating a single-player action game as a sports event reveals a habit in the media industry: anything young people play gets called esports, and anything with numbers gets called analysis. Both labels are convenient. Both blur the line between two forms of content consumption with entirely different economics.

What to watch over the next sixty days

I will make three verifiable calls, and I accept being checked against them when the data lands.

Community-aggregated main-story completion time on the Action difficulty will land between 32 and 38 hours, inside the 30-to-40 band the source gave but below its upper bound.

The loudest complaint in the first two weeks after release will centre on one of two themes: a sense of repetition across some boss fights, or upgrade materials locked behind optional content.

And the claim about being longer than the other two games combined will go unchecked by numbers for the first month, because it was designed to survive exactly the period before the numbers appear.

The length of an action game says little about its quality, just as minutes played say little about a footballer. What says more is the structure inside that span: how much of it is a new question, how much is repetition neatly arranged, and how much is the player generating it themselves by dying until they understand. Thirty-six boss fights is a big promise. The real question is how many of those thirty-six times force the player to think differently.