Dying Below One Hundred Tiles: The Elimination Threshold and the Tile Cascade

Direct answer. You die in v0.34.24 the moment your territory drops below one hundred tiles, not at zero, and the check runs on the attack result and again on your own tick. When that happens, handleDeadDefender runs a bounded flood-fill over your surviving tiles: every tile that touches the player who finished you goes to that player, and every other surviving tile goes to a non-friend neighbor — the enemies you were bordering. So a third enemy who never attacked you can still inherit land from your death, and the finisher takes only half of your gold (and none at all if you never launched a single attack).

The Decision This Guide Actually Solves

This guide answers the question that keeps coming up in community threads: “I had troops and land, then one bad attack happened and suddenly everyone was standing on my tiles — how?” The player who writes that sentence is almost always someone who treated a small remaining blob as a safe position to regroup from, when in fact the moment their territory crossed under the floor, the game already decided the kill was over. The decisions that change once you understand the exact rule are concrete and immediate. First, a player who is hovering near the floor should stop treating their remaining tiles as a base and start treating the next attack as the last one they will get to answer, because the buffer they are standing on is thinner than it looks. Second, a player who is about to finish someone should spend their movement and their expansion on being adjacent to the victim before the final blow lands, because the cascade hands the connected leftover land to whoever is bordering it. Third, a player who is bordering a dying enemy should hold back on other fronts long enough to make sure they are the non-friend neighbor the flood-fill reaches, since they can claim tiles from a fight they did not start. None of these decisions is about combat ratio or troop cap, which the attack ratio guide and the sit at max troops guide already cover; they are about the moment of elimination itself and the redistribution that follows. The whole point of this page is to give you the exact threshold and the exact ordering of the cascade so that “why did I just lose my whole position” becomes a predictable, preventable event instead of a mystery that plays out in the feed while you are still reading your own attack numbers. It is also the companion to making your own elimination costly: that guide is about how you arrange a death so it hurts the finisher, and this one is about the threshold and the cascade that define what a death actually does once it happens. If you have ever been confused about the difference between being attacked and being eliminated, this page is the one that separates the two events cleanly and tells you which one you can actually influence and which one is fixed by the engine the instant your count crosses the line. That separation is the reason the guide is worth reading past the first section: most of the advice that circulates in beginner threads is really about the first event, how you survive an attack, and it quietly assumes that surviving the attack means surviving the game, when the engine has already scheduled your elimination the moment your territory stopped being above the floor and no amount of troop cap or attack ratio is going to undo that scheduling once it has happened.

Version Boundary: What Is True in v0.34.24

The facts on this page are verified against the v0.34.24 tagged source, specifically the elimination and cascade code in the attack and player execution paths and the gold-amount function in the configuration module. That boundary matters because the community material that motivates this guide — the beginner videos and the early-survival threads — was written against older builds, and at least one of the mechanics this guide depends on is the kind of thing that a future balance pass could quietly change. The specific version facts to hold onto are the numeric floor, the ordering of the cascade, and the gold-share rule. The floor is one hundred tiles, checked in two places: once when an attack lands against you, and again during your own periodic tick, which is what means a player can cross the floor from a defense loss and not just from being the direct target of an attack. The ordering is that the finisher is always first, then the dead player’s enemies, with a hard cap on how many flood-fill passes the engine will run so that a pathological map can never hang the game. The gold rule is that a human finisher receives half of a human victim’s gold, a full gold pile from a bot or nation, and nothing if the human victim had never sent an attack, which the code reports as a distinct “conquered without gold” event. If a future release changes the floor to a different number, or changes the gold share, or removes the enemy-neighbor branch of the cascade, then the scenarios on this page need to be re-checked against the new tag before you rely on them. You can confirm the boundary yourself by opening the tagged configuration and execution files and searching for the one-hundred comparison and the flood-fill, which takes a minute and turns this page from an authority you must trust into a fact you have checked. Treat everything here as “true as of the v0.34.24 tag” and treat the community threads as evidence that the confusion is real, not as evidence of any particular number. When you read the population growth guide or the winning in overtime guide, keep in mind that those pages describe the growth and the clock, while the floor on this page is the line that decides when all of it stops mattering because the player who holds it is already gone.

The One Hundred Tile Threshold, Not Zero

The single most important fact on this page is that the elimination floor is one hundred tiles and not zero. In the attack execution path, the code that decides whether the defender is dead only fires when the target is a player and the tile count the target owns is strictly below one hundred. In the player tick path, the same one-hundred comparison is re-evaluated on the player’s own schedule, which is why a player who is not currently being attacked can still be found dead: the tick notices that the tile count crossed the floor and triggers the same cleanup. The practical meaning of “below one hundred, not zero” is that there is a real and dangerous band of territory — from one hundred down to the point where the last tile is gone — in which the player is already eliminated on the books even though they still visibly own a small connected region on the map. A player with ninety-nine tiles is, in the engine’s terms, the same as a player with zero tiles; the small blob they still see is not a position, it is the thing about to be redistributed. The reason the floor exists at all is that the game needs a deterministic trigger for the cascade, and a threshold is cheaper and fairer than waiting for literal zero, because a player at twenty tiles is not going to meaningfully defend themselves anyway and the surrounding players are already moving on their land. The number itself is a balance choice, and it is exactly the kind of “why did I die with land still on the map” surprise that new players report, because the UI never tells a player that the floor is one hundred rather than zero. Once you know the number, the decision becomes concrete: you are not “nearly out” at eighty tiles, you are already out, and the only question left is who gets to spend your remaining tiles first. Having the check in two places, the attack result and the periodic tick, is what makes the floor robust rather than exploitable, because a player cannot dodge it by arranging for their own count to drop outside of a single attack window. This is also why the floor interacts with defense the way the land combat guide describes: the same defense loss that costs you a border on a normal day is the loss that crosses you under the floor when you are already thin, so the danger is not the loss itself but the level of territory the loss lands you at.

How the Cascade Actually Assigns Your Leftover Land

The cascade is the part of elimination that most players never think about until it happens to them, and it is where “every other player cannibalizes me” comes from. The engine runs a breadth-first sweep over the dead player’s surviving tiles with a hard cap on how many passes it will execute, so the sweep is bounded and deterministic. For each surviving tile it visits, the engine looks at the neighbors. If any of those neighbors is owned by the player who delivered the finishing blow — the finisher — then that tile is assigned to the finisher. If none of the neighbors is owned by the finisher, the engine instead assigns the tile to a non-friend neighbor of the dead player, that is, one of the dead player’s own enemies who happens to border it. Two consequences follow directly from that ordering. The first is that the finisher does not inherit the dead player’s whole territory; the finisher only collects the tiles that are actually connected to the finisher’s own border, because the sweep only reaches a tile through a neighbor it already owns. The second is that a third player who was bordering the victim but never participated in the kill can still inherit tiles, because any surviving tile that the finisher does not reach and that touches that third player’s land is handed to them as a non-friend neighbor. This is the mechanic behind the recurring complaint that a player was “cannibalized by everyone” rather than “killed by one person”: the kill is a single event, but the redistribution that follows is a multi-player event, and the players who get the most leftover land are not necessarily the one who landed the final attack, but the ones who were positioned to be the neighbors the sweep reaches. The ordering is worth committing to memory because it inverts the intuition most players carry in: the most land does not go to the person with the most troops, but to the person whose border is longest against the dead player’s leftover region, which is a purely positional property and has nothing to do with how the kill was actually won. The hard cap on the number of passes is not a detail; it is what keeps the whole mechanic safe, because without it a map where the dead player’s leftover tiles form a long thin ring could keep the sweep going indefinitely, and the cap guarantees the redistribution finishes in a fixed number of steps no matter how the board is shaped.

The Gold and Asset Transfer, and When You Get Nothing

Elimination is not only about land; it is also about the gold and the units a dead player was carrying, and the rules for that are separate from the tile cascade and equally surprising. The gold side is handled by the conquer function, which computes a gold amount and adds it to the finisher. The amount depends on the victim’s type. A bot or a nation hands over its entire gold pile, because there is no human player behind it to keep a share. A human victim hands over only half of its gold, because the game treats a human’s gold as partially protected even in defeat. And there is a third case that catches many players off guard: if the human victim had sent zero attacks across the whole game — meaning they never actually played, or they were AFK in the spawn area — then the gold transfer is skipped entirely and the finisher receives nothing, which the code records as a distinct “conquered without gold” event. The unit side is smaller but real: if the dead player was a human and the finisher is on the same team, the finisher captures the dead player’s leftover warships and transports, so a friendly finisher does not lose the naval units the victim was carrying. What the finisher does not get is the dead player’s structures, which are handled separately and are not simply transferred to the finisher by the gold function. The decision implication is that finishing a human is worth roughly half of their visible gold plus any same-team naval units, not their entire economy, and that finishing a player who never attacked is worth almost nothing in gold terms, which changes the value of “who do I finish first” when you are choosing between a fresh, unengaged human and one who has been trading and expanding all game. The practical read is that the gold of a kill is a small, often-missed bonus on top of the land, and that a finisher who is chasing the gold rather than the land is optimizing the wrong variable, because the land is what compounds and the gold is a one-time, halved, and sometimes absent pickup. The table below summarizes the three gold outcomes so you can price a kill at a glance instead of reconstructing the rule under pressure, and it is the piece you want in your head the moment you are deciding between two targets that look the same on the map but carry very different gold and unit values.

Scenario One: The Player at the Floor Decides Whether to Commit

Imagine a player who is mid-game and has just spent a large share of their troops defending a border, and they are now sitting on a modest but clearly visible region of land. The trap this scenario exposes is that the player reads their own tile count and assumes they are safe to keep playing, when the count is already at or near the one-hundred floor. The decision the player has to make is whether to commit their remaining troops to another attack or to bank them and hold. The wrong move is the one that treats the visible land as a base: the player sends the troops on a counter-attack, the counter-attack loses more tiles than expected because the defense loss on their own side is higher than the attacker’s, and the tile count drops under the floor. At that point the player is eliminated, and the cascade starts redistributing the region they thought was theirs. The correct move, and the one this guide makes possible, is to run the numbers before committing: if your tile count is anywhere near the floor, you should assume that one more defense loss of a normal size will put you under, and you should either hold your troops to preserve the buffer, or deliberately trade down to a position where you are not bordering a strong finisher. The scenario is realistic precisely because the floor is invisible in the UI; the player does not see a red “you will die at one hundred” marker, they see a normal amount of land and make a normal-looking decision that the threshold quietly turns fatal. The concrete change in play is that a player near the floor stops making new commitments and starts deciding which neighbors they are willing to let finish them, if they cannot hold. To make it concrete with numbers, if you are at one hundred and twenty tiles and your recent defense losses have been around twenty to thirty tiles each, then one more unopposed defense is enough to put you under the floor, so you are effectively one bad attack away from being dead even though the map still shows you a normal region. You can read this scenario as the defensive side of threat assessment: the threat is not only “how strong is the attacker” but “how thin am I right now”, and a player who is thin is in danger even from an attacker who is only moderately strong.

Scenario Two: The Finisher Chooses Position Over the Blow

The second scenario is the mirror image, played by the player who is about to deliver the finishing attack. A player has an enemy at a low tile count and is deciding how to spend their own tiles and their movement over the next few minutes. The naive play is to focus purely on landing the killing blow, because landing the blow is what everyone thinks “the kill” is. But the cascade shows that landing the blow and collecting the leftover land are two different skills. The finisher only inherits the tiles that are connected to the finisher’s own border, so if the finisher’s current border does not touch the victim’s leftover region, the finisher will land the kill and then watch a third neighbor sweep up the tiles the finisher could not reach. The better play is to spend two or three minutes before the final attack extending your own border so that it wraps around as much of the victim’s region as possible, because every tile you put adjacent to the victim’s land is a tile the cascade will hand to you instead of to their other enemies. The tradeoff is real: while you are repositioning, the victim may be fighting someone else, or may be reinforced, or the map may have moved, so you are spending safety to buy land. But when the victim is genuinely near the floor and the kill is close, the position play usually wins, because the difference between finishing with a wide border and finishing with a narrow one is the difference between inheriting the region and inheriting a strip of it. The decision rule is concrete: before you commit to the final attack, check how much of the victim’s leftover region actually touches your border, and if it is a minority, delay the blow and extend first. The timing window matters too, because the position play only works if the victim is close enough to the floor that the repositioning will not take them out of the danger zone; if the victim is still far above the floor, the safer move is to keep applying pressure elsewhere rather than telegraphing your intent by wrapping the region. This is the same positional reasoning that the conquest gold guide uses when it asks whether finishing a target is worth the spend, except here the value you are optimizing is land inherited through the cascade rather than the direct gold of the kill, so the two guides should be read together when you are deciding whether a low-tile enemy is worth committing to at all.

Scenario Three: The Third Enemy Inherits Land From a Kill They Did Not Start

The third scenario is the one that surprises players the most, because it is a player who is not even involved in the kill. A player is bordering two enemies who are about to finish each other, and that player is not spending troops on the fight at all. The naive reading is that the fight between the two neighbors is none of their business, and the player keeps expanding somewhere else. But the cascade is watching the dead player’s surviving tiles, and every surviving tile that the finisher does not reach and that touches the bystander’s land is assigned to the bystander as a non-friend neighbor. That means the bystander can inherit a meaningful chunk of the victim’s region without having sent a single attack, simply by virtue of being the neighbor the flood-fill reaches. The decision the bystander has to make is whether to hold their border adjacent to the dying player, or to pull back. Holding is usually correct when the bystander is otherwise safe, because the land is free and the cascade is the only cost, and the land the bystander inherits is land they did not have to spend troops on. Pulling back is only correct if the bystander is itself near the floor, because the moment the bystander’s own count crosses the floor the same cascade turns on them and they become the victim instead of the heir. So the third scenario resolves into a clean rule: a safe bystander who is bordering a player who is about to die should keep touching that player’s land, because the game will redistribute it to them for free, and a bystander who is itself near the floor should not gamble on being the heir, because being the heir and being the next victim are only separated by their own tile count. The bystander play is the purest expression of the cascade, because it is the case where a player’s gain from a death is entirely positional and requires no combat at all, which is why it is the case most players miss when they are busy watching the actual attack they think is the real event. A strong reading of the board is to treat a dying player who is bordered by three or more enemies as a shared resource that the surrounding players will split, and to position so that you are the one whose border touches the most of it, rather than assuming that the player who lands the blow gets to keep the prize.

Failure Modes and Counter-Play Against the Cascade

The failure modes that follow from the floor and the cascade are specific, and each one has a counter. The first failure is the invisible-floor trap, where a player near one hundred tiles makes a normal-looking commitment and crosses the floor without realizing they are already eliminated on the books. The counter is to treat one hundred as a hard warning line: if you are below roughly one and a half times the floor, you are in the danger band and you should stop making new attacks until you have rebuilt the buffer. The second failure is the finisher’s narrow border, where the player who lands the kill inherits almost none of the leftover land because their border did not wrap the region. The counter is the position play from scenario two: extend your border before the final blow so that the connected leftover tiles are connected to you. The third failure is the bystander’s overcommitment, where a third player tries to claim the cascade land but is itself near the floor and ends up being the next victim. The counter is to check your own tile count before you decide to hold adjacent to a dying enemy; if you are in the danger band, the free land is not free, it is a trap that the cascade will close on you. The fourth failure is the gold misread, where a finisher expects the victim’s full economy and is disappointed to receive only half of a human’s gold, or nothing at all because the human never attacked. The counter is to price the kill correctly before you commit: a human finish is worth about half their gold plus any same-team naval units, and an unengaged human is worth almost nothing in gold, so the value of the kill is mostly in the land, not the gold. The fifth failure is the single-minded finish, where a player commits all of their safety to landing the blow and ignores that the land they actually want is connected to a neighbor they are not touching at all. The counter is to run the position check before the final attack and to accept that sometimes the best finish is the one that gives up the gold in order to keep the connected land. The common thread across all five is that the cascade is a deterministic, multi-player redistribution, and the players who lose to it are the ones who treat it as a mystery instead of running the two checks: is my own count above the floor, and which of my neighbors will the flood-fill reach when this player dies. These failure modes overlap with the ones in the costly elimination guide, but from the opposite side: that guide is about how a dying player arranges the death, and this one is about how the three parties who survive the death should have positioned themselves, so a player who reads both knows the full shape of the event from all three seats.

Map and Mode Adjustments

The same floor and the same cascade behave differently across maps and modes, and the adjustments are about the geometry of the board rather than about any change to the numbers. On a large, open map with many players spread out, the cascade tends to be a two-player event, because a dying player usually borders only one or two enemies and the flood-fill has a narrow set of non-friend neighbors to reach. On a small, dense map with many players packed together, the cascade is a multi-player event, because a dying player borders several enemies at once and the leftover land gets split across all of them, which is the board where “every other player cannibalizes me” is literally true and where the bystander scenario is most powerful. The number of players matters more than the map size, because the cascade is only as wide as the set of neighbors the dead player has. In team or faction modes, the non-friend branch changes, because a teammate is a friend and is excluded from the cascade, so the leftover land goes to fewer players and the finisher’s share is larger; that is also the mode where the same-team naval capture matters, because a friendly finisher collects the victim’s warships and transports. The mode adjustment that changes the decision most is the number of non-friend neighbors a dying player has, because that number sets how many players will inherit from a single kill, and a player in a dense mode should expect their death to feed several enemies at once, not just the finisher. The practical rule is to read the board the same way the flood-fill reads it: count how many non-friend borders your own region has, because that count is exactly the number of players who can inherit your leftover land the moment you cross the floor, and it is the single most useful number to keep in mind when you are deciding how thin you can afford to be. One more adjustment is the presence of terrain and chokepoints, because a dying player whose leftover region is split by terrain into two disconnected pieces will have those pieces assigned by the flood-fill as they are reached, which means the side of the region that is more exposed to a particular enemy is the side that enemy is most likely to inherit. The midgame survival against low threat guide covers how to stay above the floor in the middle of the game, and this section is the endgame and the mode-level read of the same underlying number, so the two are complementary rather than competing.

The Pre-Death Checklist to Run Every Time You Drop Toward the Floor

The whole guide compresses into a short checklist that a player should run the moment their tile count starts approaching the floor, and it is the same checklist whether you are the one dying, the one finishing, or the one bordering both. The first item is your own count: is it above the one-hundred floor, and if it is below roughly one and a half times the floor, you are in the danger band and you stop making new attacks until the buffer is rebuilt. The second item is your non-friend neighbor count: how many enemies border your region right now, because that is the exact number of players who will inherit your leftover land if you cross the floor, and it is the number that decides how wide your own death will feed. The third item is your finisher’s border: if you are about to finish someone, how much of their leftover region actually touches your border, and if it is a minority, you delay the blow and extend first so that the connected tiles are connected to you. The fourth item is the bystander check: if you are bordering a player who is about to die, are you yourself above the floor, because if you are in the danger band the free land the cascade is about to hand you is the same mechanism that will hand your land to your enemies a turn later. The fifth item is the gold read: if you are the finisher, price the kill as half the human victim’s gold plus same-team naval units, and not their full economy, and price an unengaged human as almost nothing in gold. The sixth item is the version read: are you playing the version this guide was verified against, because if the floor, the ordering, or the gold share has changed since the v0.34.24 tag, then the numbers on this page are the ones to re-check before you rely on them in a serious match. Run these six checks and the elimination stops being a surprise: you know the exact tile count that kills you, you know the exact number of neighbors who will spend your leftover land, and you know the exact position that makes the finisher’s kill worth the most land instead of the least. The cascade is not a mystery and it is not random; it is a deterministic flood-fill with a numeric floor, and once you read the board the way the flood-fill reads it, the “why did everyone just take my land” question becomes the “I see exactly which three players were about to inherit from me, so I will hold a buffer” question, which is a question you can actually act on and plan for well before the floor is ever in question.