Deathmatch Island: The Deep Dive

If you could win unlimited wealth and freedom and all you had to do was murder some strangers, would you? What if you grew close to those strangers through days of adversity and teamwork? Would you fight against the injustice of this sadistic game, or stab your new friends in the back? What if their back was covered in mind-altering red slugs, half the competitors were paid actors, and you can’t remember who you are? What year is it anyway?

Sounds like you’re ready for a ticket to Deathmatch Island. A recruiter will be in touch with you shortly. Be sure to hydrate before the competitor intake process and please don’t wear any metal jewelry, it may react unpredictably with the Field.

Deathmatch Island

Available on DriveThru in PDF

Hardcover and Limited Editions available from Evil Hat

(The cards and maps are great, and the freestanding Island and Cast pamphlets are handy. Probably worth the extra $15!)

The System

Deathmatch Island, by Tim Denee, is the first release based on AGON’s Paragon system by John Harper and Sean Nittner. The essence of the Paragon system is that you mostly just talk through situations as you arrive at them – in DMI’s case, various challenges, locations and other participants – and when you hit the central conflict of the scene, you resolve i t with a single roll versus a target number generated by the GM. The pool of dice you roll is determined by how you approach the situation, and the target is also rolled by the GM based on how their side is approaching the situation.

It’s quite old-school, this sort of scene-level resolution system. Some of the most innovative early games of the Forge era used this method: Dust Devils, Sufficiently Advanced, Primetime Adventures and so on. And, of course, AGON’s first edition. I had only played AGON 1E and it was many moons ago. But structurally DMI felt similar to that experience.

The most notable effect of scene-level resolution is a lack of snowballing like you see in many PbtA and FitD games. Rather than moves triggering more moves and a blow-by-blow escalation in the action, the scene is set, stakes are laid out, and then it’s over.

You can do some interesting play with scene resolution that’s harder to pull off with either straight task resolution or snowballing action. For example, you can be more intentional and thoughtful about framing scenes and what they “mean.” Not that you can’t do these things under other play paradigms! But scene resolution lends itself well to this. It also means if you don’t leverage its narrative strengths, scene resolution games can feel mechanically dull. No comeback opportunities, no yes-but/and or no-but/and nuance. Setting the stakes of the scene can be a bit fraught at the table as well, given the winner-takes-all element.

This, only covered in blood, dodging bullets and throwing knives, and when you land you might need to break someone’s neck.

When you enter a scene and set out the stakes in Deathmatch Island, players then build a pool of dice that aligns with how they approach the situation. Sometimes you’re told what to roll – a white-water challenge will require you roll Challenge Beast, the American Ninja Warrior stat of the game. Sometimes you can choose your approach when dealing with NPC teams. Or rather, your Leader chooses the approach, the Leader being the character who won the last roll.

Deciding whether to take on a challenge or support someone else means announcing your name and describing their arrival as if being viewed on a screen: a chyron or sidebar, along with how they announce themselves into the conflict. Who’s making this announcement? Who’s seeing these screens? Doesn’t matter! It’s all part of the fourth-wall-breaking aspect of the game that makes Deathmatch Island stand out.

Settling the fiction of any scene is kind of weird because of its scene-resolution origins. For example, you only talk up to the point where you describe how you’re going to approach the problem, but you don’t provide any details about how that goes. Then, once the rolls are over, you narrate from worst-to-best. The losing players then narrate (called a “confessional,” which works differently than the more traditional confessional we see in Jared Sorensen’s monster-hunting reality show game InSpectres) their failure in an open-ended way so others later in the sequence can build on them. Folding your chosen capabilities, equipment, advantages and roll into your confessional is a skill that needs practice.

Pro tip: use the limited-edition Beans as your Leader marker. Leaders get the beans!

The best roll that also beats the target number (rolled by Production in a similar but not identical way to the competitors) is the “best” and becomes the Leader of the group. Rank hath its privileges: Leader gets the most Followers, Leader decides how the next roll will be resolved, decides the next point on the map they’ll visit, and so on. But other players can spend an economy called Trust to make the Leader change their mind about these things.

A push throughout the game is to acquire Followers (aka Glory in AGON). As Followers increase, so does the size of the die attached to your Name. The first time your Name die increases, you get a nickname from the other players. As your character gains Followers, a little ticker-tape is printed out from a small watch they wear.

So that’s the system from 20,000 feet up. Let’s talk about the premise.

The Premise

Deathmatch Island is inspired by a mix of Survivor, Squid Game, Lost, Hunger Games, Battle Royale and any number of WTF psychological thrillers: Severance, The Prisoner, Saw, you name it. There are plenty of media touchstones, and they come together in an unsettling froth.

Totally not Squid Game, please don’t sue!

The characters are paper-thin to start, and completely randomly generated: your name, your career before arriving on Island One, everything. You have no agency, and you wake up on a boat heading into God-knows-what. In our game, I told our players this:

  • I’ve got a 3 shot that’s a cross between Lost, Severance and Squid Game. The less you know about it going in the better.
  • You’ll play a randomized amnesiac character and you’ll learn about them through play.
  • Character death is very much on the table but it’s not a life-is-cheap OSR thing. You really should play to win.
  • There will be a winner, or not, at the end.

This might not be the right approach for every table but it worked very well for us.

If you intend to be a DMI player and want to enjoy discovering stuff, maybe skip this next part.

Also not Lost! But sweaty and green when you arrive. Also confusing. Polar bears possible.

There are three Islands in the “season” of a murderous game show the characters find themselves in. The first two Islands are quite different: the first a Hawaiian paradise, the next an arctic wasteland. You travel to both of them in a small fishing boat. The unnatural weather is the least weird bit of the Islands.

Across each island is a sprawling web of locations the characters move between. It’s functionally a point crawl, where you’re limited to going only to preprogrammed situations and don’t care at all about how you get there. You’re expected to go to three or four of about a dozen possible locations before beginning “Phase 2,” but if you secure a vehicle you can get to more locations before the klaxon sounds.

Location types are challenges, points of interest, or “redacted.” They’re marked as such on the map, so the players know what they’re headed into.

Bikinis provide terrible ballistic protection, cannot recommend.

Challenges are Survivor-style structured contrivances against other teams: races across rope bridges, scrambles through obstacle courses and so on. It’s expected the characters will use everything at their disposal to win: cut the rope, injure or murder the opposing team, whatever.

Redacted locations are “forbidden” by Production, the shadowy entity running Deathmatch Island. They are of course not really forbidden because they’re on the map! Every character has a stat called Redacted that they roll when dealing with anything outside the established point-crawl element of the game. Redacted locations are where players can explore the deep weirdness of the setting and begin developing theories about what’s really going on on the Islands.

That leaves us with Points of Interest, which I think is the meat of the game. A Point of Interest is potentially useful, but ultimately it’s an open-ended opportunity to deal with other NPC competitors. When the GM sets up a season of DMI, the main choice they make is which “crew” set they’ll play; the Island maps are always the same. There are four “crews” of NPCs, each built around a different set of themes to explore. We went with “On The Dynamics Of Applied Authority,” which features a well-organized alliance under the leadership of a police officer, and a smaller, more fractious group that opposes them, with the PC team caught between them.

Regardless of the Point of Interest, what’s really happening in those locations is the evolution of the NPC groups. The GM keeps track of the various groups moving around the Island and the impact of outcomes rolled throughout the game. There are some guidelines about how things might proceed but honestly it’s pretty loose. I came into my first crew encounter feeling like they were tightly scripted. It turns out you can go off-script if you’re prepared to come up with clear stakes for the scenes.

Depending on the flavor of your Island’s weirdness, it might not even matter if you’ve killed an important NPC. Perhaps you only think you killed them. Or they were a clone. Or a copy from a parallel dimension. Or an actor, a convincing robot, a hologram, etc etc.


Deathmatch Island plays like a pointedly PvP version of AGON, assuming the players decide early on they want to Play To Win. And there can only be one winner, therefore they’ll have to face off against each other on the third Island. But mostly DMI is a close reskin of AGON, where you gather Followers instead of Glory, spend Fatigue instead of Pathos, and so on. The most divergent bit of the game, and my favorite GM-facing part, is called Theorizing.

There are four categories of potential explanation for all the inexplicable shit that goes down in Deathmatch Island. The underlying theories are that it’s a political experiment, pure entertainment, an experiment, or just Weird (aliens, time or dimension travel, virtual reality, etc.). The GM starts with one “theory point” in each of those categories, which they can spend in play to introduce Interventions. These are physical or situational clues that reinforce each theory. For example, if the Islands are a political project, characters might find a dossier on a prominent political figure, or a bundle of fake passports.

In our game, I generally forgot about these until near the end of Island One. But when they started showing up, oh boy did things start to go off the rails in the narrative! Because the Interventions are all just provocative and not explanatory, they’re ripe for the players to try and connect the dots.

Between Islands, the characters theorize about what they’ve seen and offer up their own explanations of what they think DMI really “is.” The GM takes notes of these theories, turning them into additional points in one of the four categories. It’s a feedback loop: the GM tends to pick stuff that interests them, which the players observe and then reiterate by giving the GM more points in those categories. I went for “weird” because I am a Lost fanboy from way way back, and sure enough they all agreed it was a weird island! Different flavors ranging from “it’s aliens” to “its VR” to “it’s alien VR,” but still.

There may in fact be the occasional hatch you’re not supposed to look in.

Because there is no canonical answer, and the Interventions are strictly provocative, the players will never get to the real answer behind Deathmatch Island. So if you like stuff like Lost, this will work out just fine. But if your players want real answers, they’ll find the experience frustrating. Or, perhaps, they’ll feel compelled to return to the Island and take another swing at it.

Pee Vee Pee

The main tonal difference between AGON and DMI is that the latter is framed as explicitly player-versus-player: there can be only one winner, and odds are good it’s going to be PC competitors facing off in the end. On the way to that final confrontation, on each island there is a Phase 2, where teams move on to straight murdering one another until they’ve killed enough folks to satisfy Production. On Island One, it’s half of everyone who showed up (which might be 100 or more, you’re never told.) On Island Two, it’s all but 10 who will move on to Island 3.

Phase 2 is broken into three steps. The first is that each team Scouts, providing an explanation of what they’re trying to achieve and winning an advantage die in a future conflict. What you actually do when Scouting doesn’t matter, narratively, other than to justify rolling your best capability (“skill”). If you want to roll your Snake Mode, you’ll sneak around a bunch. If you want to chat or negotiate with the other team, you’ll roll Social Game. If you want to hack the game itself – an under-explained option but it does exist! – you’ll roll Redacted.

After Scouting and deciding who got the bonus die, the players decide whether they’re going to Defend against a pair of threats, or Seize the battlefield for the Battle Royale. There are always two threats that will happen if nobody defends against them: one will result in physical injury, the other in psychological injury or betrayal. These scenes are set just like any others in the game, by explaining the situation and setting out the stakes. The other option, Seize, means you can set the stakes of who will die in the Battle Royale, as well as choose which capability can be used in addition to Deathmatch, which is always at play when you’re a-murdering.

While the PC competitors are expected to be working as a team, the one thing they can’t do during Phase 2 is Support, ie give another player a capability die in return for Trust — everyone has to roll on their own.

Trust is a central economy of the game, with PCs tracking the amount of Trust they have with each other. On the first two Islands, they can spend Trust to make another character Bolster them (and give them a die), take their damage (usually Fatigue as a result of losing a roll, possibly a lasting injury later on), or even change the current Leader’s mind about what their approach to a conflict will be. It’s a powerful tool! But it becomes less relevant the deeper into the game you go.

Island Three

After Island Two, the characters are sent alone to experience individual vignettes on Island Three. This is where shit gets really strange. In our game, one player entered a giant sphere where he uncovered memories of a sibling he’d forgotten he had! Another entered a Dance Dance Revolution style dance-off against her NPC ally until one collapsed. Yeah, it’s like that.

A surprisingly awful way to die.

Three of our four competitors survived to go on to the Inner Circle and the final conflict. But before they enter, they receive a questionnaire with one question: will you Play to Win, or Break the Game? Depending on the table, the vibes throughout the game, how hard the GM pushed the characters to watch out for themselves, this could be a fraught question! Because if it’s mixed, the players who chose to Play to Win (ie kill everyone) will get a big bonus die against the suckers who thought they could Break The Game.

The endgame is a tricky balancing act. Mechanically it’s pretty thin, once you get past the delicious “will you break the game or go on to win?” question. You roll once to Scout and the winner gets a big advantage die. The winner also decides how the survivors are divvied up between two groups. There is one roll to Skirmish and take out everyone else in your cohort. Then one more roll for the big Finale. As far as mechanisms supporting the things the game wants you to feel, that’s not happening at the dice level.

If you’ve been building toward heavy character investment all game long, then the endgame feels much more intense. And that’s where the balancing act comes in. This is in fact the whole game’s balancing act: life is both incredibly cheap, and incalculably valuable (as Production tells the characters early on: You get only one life, use it wisely!), and you have to keep both those ideas in your mind at once. If you die in DMI, easy enough, you just pick up another of the survivors. Characters are paper-thin, and rely on little hook-setting moments throughout to build personal investment.

Last Thoughts

After playing through a season of Deathmatch Island, our table decided it’s in the liminal space between a dicey RPG and an ultralight story game. You advance your character and there’s some limited agency (which, similar to how other old-time indie games like Lacuna never outright told you, you can expand if only you think to try), but honestly I came away from the experience feeling like the game was closer to something like Ross Cowman’s City of Winter. It is, ultimately, a point-crawl with limited choices where the real juice of play lies in the narrative and player investment.

Honestly it’s just gorgeous in person. Photo from Backerkit.

The physical production of the game deserves to be called out. This is maybe the most beautifully produced gaming product I’ve seen in a decade or more. The book is clean and easy to navigate, but also just filled with tasty, evil little aphorisms from Production (“A diamond is a chunk of coal that survived under pressure”) and the occasional illustration of an anodyne, sinister Production products: a “melee weapon” (box cutter) or “small luxury” (stale cookies). The Limited Edition comes with glossy foldout maps of the Islands, booklets for all the Islands and Crews, a deck of motivation cards (we dealt ours out secretly and revealed only at the end), trust-building and ice-breaking exercises, and so on.

What struck me at the end of our season was how fourth-wall-breaking the experience is. The “Win or Break?” form is filled out by the players. The trust-building checklist is held by the players. The players can come back to DMI for another season fully informed as to what their previous characters ran into in previous runs. Over and over, the game plays hard with the player/character divide. This might lead to the fun kind of bleedy for your table – or it might be the bad kid, given how aggressively PvP the game is.

The text of Deathmatch Island is also a bit coy about some aspects of play that players may find frustrating. I haven’t played a game centered on lightly held secrets in a long time. I didn’t tell my players how scene resolution worked until they had to do it, which was fun! I didn’t tell them what the real story behind DMI is, because there isn’t one. I never told them that their Redacted skill is what you roll when you’re trying to do things outside the rules of the game (or The Game as it were – the line between the RPG and the Island is as thin as the one between the player and the character). Can you tell the players these things without leading to a total disruption of the experience? Should you? It’s an interesting, and to my mind, open question that each table will have to decide for themselves.

There’s a safety checklist the game comes with (as a diegetic artifact from Production of course), and you can play with the option that nobody “really” “dies” in the game. But those are merely content warnings. The core of the game itself, which plays fast and loose with the barriers between players, characters, game rules and the reality show game, is more dangerous and exciting than any safety tool can really address.

Leave a Reply