Late Game Structures – End Conditions

Written by Matt Pavlovich

We’re excited to return to our ongoing series titled “Game Structures”, a series of topics we’ve organized to help understand the foundations of successful game design. Our initial run of articles covered “Early Game Structures”; aspects of games such as initial starting positions and resources, the value of turn order and key decision points that help to generate replay value and keep players coming back time and time again.

Our follow-up titled “Mid Game Structures” focused on prolonging player engagement the core of the game experience. We covered topics like player ecology to help drive player motivations and player interaction and player strategy which allow player choices to pivot and create new and interesting decisions during games.

During the next few months we’ll be tackling “End Game Structures” which help to drive player satisfaction and help to bring players back to the table again and again. As part of the series we’ll be looking at end game conditions, scoring methodologies and otherwise satisfying elements of game design we’ve encountered in games. To help guide us, we’re going to look at two key questions, how do players factor into the end of the game, and how do the ending conditions of the game factor into the game’s design? Let’s get started.

Question #1: How do player’s factor into the end of the game?

Longtime readers of Games Precipice might have noticed my enthusiasm for mapping design problems to a couple of orthogonal dimensions, and I think the same sort of analysis is useful here. The first axis might be the more obvious: to what extent can the players control when the game ends?

Players have No Control

No Control: Players take the backseat while the game grabs the wheel and drives to the final destination. The “Uber” of end game conditions.

Games at this end of the axis end when they end, and the players can’t do anything about it. The most common way to implement this condition is to have the game last a fixed number of rounds or turns, then the game ends, and a winner is crowned. Terra Mystica lasts six rounds, period. Small World lasts between 8 and 10 rounds, depending on the number of players. Of course, this doesn’t mean that players have no control over the flow or pacing of the game: each player may accomplish a different amount of stuff each turn, and each turn may last a different amount of time. But each player gets only a certain allotment of turns to accomplish their strategy.

“No control” may sound like a bad thing given how much we’ve focused on player control as a per se good in game design, but there are a lot of advantages to having a mechanically enforced end point over one that the players can control. First, it’s completely objective and happens the same way every time, so it will not come as a surprise to any players (assuming you remember to slide the round marker down the track). This means there are no “mah jong” moments where one of your opponents suddenly fulfills a victory condition that you thought was at least three turns away, abruptly ending the game for everyone and creating a lot of satisfaction for one player but confusion for the rest.

A similar advantage of a fixed end-game condition is that it allows players to evaluate their strategies relative to how close the game is to ending. Keyflower, which we recently profiled as part of our Design Analysis series, does a phenomenal and thematically appropriate job of this. A game of Keyflower lasts four seasons starting in spring and ending in winter. Here, the thematic implication is that you “sow” in spring and summer and “reap” in autumn and winter. That is, you plant the seeds of your engine in the early phases of the game, and if you’re not scoring many points, that’s okay; conversely, you shouldn’t be wasting your time creating infrastructure you’ll never use in the later parts of the game, and this is instead when you should be doing most of your scoring. The effect is that Keyflower’s theme and design work together to enforce a fixed end-game condition that informs player strategies.

Finally, some of the practical benefits to fixed end-games might help to make games more approachable for new players. If your game contains six game years, and each game year takes around 15-20 minutes to play, you can confidently tell someone who has never played your game before that it will last 1.5 to 2 hours. And explaining to a new player that “the game ends once everyone has taken six turns” is much easier to grasp than “the game ends once someone has constructed their thirteenth pylon.” What is a pylon? How do you construct it? And what does it mean to have thirteen of them? 

Given all of their strengths, why might a designer not want to implement a fixed end-game condition? The most obvious answer is that they can be tricky to balance. One turn too many, and a game can quickly overstay its welcome; one turn too few, and players can come away from it feeling like they haven’t had a complete experience. Careful playtesting is critical for this sort of design. Another potential problem is that absent a positional balance mechanic, runaway leaders can occur fairly easily. We’ve all played games of Yahtzee where the question on everyone’s mind going into the final throw of the dice isn’t “who’s going to win” but “what are the odds on my large straight hitting because that’s literally the only way I can possibly catch Alex” and the answer is usually “not good enough.”

Before I move on to other end-game paradigms, I wanted to briefly mention fixed conditions that are not tied to turns or rounds. Some games use timers as a way to make absolute time, rather than in-game time, end conditions: Taboo uses a one-minute timer to end a team’s turn, while Boggle turns last three minutes. Games that use fixed time limits for the entire game, rather than simply individual turns, are rare, but Time ‘n’ Space is a fascinating example of a game that lasts exactly 30 minutes, regardless of what the players actually do. It’s a brilliant way to enforce a hard limit on how much of your evening the game can monopolize, but it’s probably only suited for games with over-the-top chaos and plenty of modes of interaction.

The Ticking Clock

Ticking Clock: The game will inevitably reach the final countdown. Players just need to put themselves in the best position by the time the clock strikes twelve.

The ticking clock is a special case of a player-uncontrolled end condition that’s worth discussing because of its prevalence in cooperative and team games. The basic idea of a ticking clock is that the game ends after a certain structural condition is fulfilled, very often a deck running out of cards. Mechanically, this is an end condition that’s identical to the ones described above, but there are particular thematic or other design reasons why you might want to use one.

Ticking clocks are almost always implemented as “Plan B” (or Plan C, or so on, depending on how complex the design is) in cooperative games to push the game toward a resolution in case the players don’t accomplish their objectives. For that reason, these end conditions tend to be coupled to loss conditions (the building burns down in Flash Point, the ancient demon awakens or the house collapses or a myriad of other horrific things happen in Betrayal at House on the Hill), but they can be neutral game-ending conditions (there are no more fireworks to launch in Hanabi) or even potentially coupled to win conditions (you’ve evaded the zombies for help to arrive in Hypothetical Survival Horror Game).

A very common structure in cooperative games is to have a player-controlled end condition coupled to a win condition and a (perhaps less directly) player-controlled end condition coupled to a loss condition, with the uncontrolled ticking clock lurking in the background. The clock may serve the mechanical purpose of preventing the players from “turtling” until an ideal game state comes along, and the thematic purpose of ratcheting up the tension and making the players feel like the stakes are real.

Players have Indirect Control

Indirect Control: Players may captain the ship, raise the sails or even rock the boat, but ultimately the wind and the current will have the greatest impact in where the game is headed.

Seasons ends after the winter of the third year. The end of Scrabble is triggered when the bag has no more letter tiles. One way in which Concordia can end is the personality deck running out of cards. And Ascension ends once a pool of points, earned throughout the game, is depleted. These conditions are centrally imposed and not tied to the players’ game states or any objectives they might accomplish, so shouldn’t they be examples of “no control” end-game conditions?

The distinction here is that games like this, which I’ll call “indirect control” end conditions, give the players some measure of control over when the end game arrives. Importantly, no player can unilaterally end these games by creating an end-game trigger; they can simply affect the pace of the game by accelerating or decelerating the inevitable. And each player can in turn do their part to hasten or postpone the end of the game as they please. Exactly how much influence the players might have on the end-game conditions depends on the game.

In Seasons, a dial showing the four seasons indicates the phase of the game; once the dial passes winter for the third time, the game ends. The twist is that the movement of the dial is not the same from round to round; instead, it can advance one to three spaces depending on the results of a draft of randomized resources. Further, players can occasionally take actions to manipulate the dial independently of the resource draft. But in general, every turn represents some amount of progress toward the end of the game, and that amount exists within a relatively narrow range. Concordia can actually end in two different ways: one is tied to a player accomplishing a settlement-based task and therefore more directly under the players’ control, and the other is an indirectly controlled condition where the players can take more cards to end the game faster or fewer to drag it out.

Scrabble famously doesn’t end until one player has no more tiles left, so it’s a rare example of a game shifting from an indirect condition to a directly controlled condition at the very end of the game. Finally, points are awarded in Ascension in two forms–the shared pool and by a value represented on each card–so some players might be aggressively pursuing a strategy that hastens the end of the game, while others might be ignoring the end-game condition entirely.

This approach adds an interesting strategic wrinkle to the strict no-control condition but remains relatively accessible and easy to understand. Explaining to players that the game ends when the deck is depleted is an end-game condition that makes intuitive sense whether each player draws one card per turn or chooses to draw between zero and three cards per turn. Although games that use this condition are more complex because they ask the players to make more decisions, that complexity only comes into play toward the end of the game, when hopefully everyone already has a good idea of what’s going on mechanically in the rest of the game. Critically, these games don’t require their players to build an entire strategy around when they want the game to end, as no player can decide alone to end the game, and there’s sufficient player-driven chaos that it would be impossible to guess the end point more than a few turns out.

Another benefit of indirect-control end conditions are that they can increase replayability. Even though each game might end once the hypothetical gold mine is depleted, the fact that that mine is depleted at a different rate every game and under different conditions makes the experience of playing feel less static.

The “indirect control” approach seems to work best in card games or other games with a fixed set of components, like the Scrabble bag, where the end-game condition is tied to the deck running out of the bag depleting. Additionally, there needs to be some cost or reward associated with making the game go faster or slower: perhaps you can draw as many cards as you want to speed up the end of the game, but each additional card past the first costs some resources. Otherwise, an entrepreneurial player could simply fulfill the mechanical condition as soon as was convenient and end the game with a thud of an anticlimax.

It’s a system that in a lot of cases represents a nice balance between the static determinism of no-control games and the too-many-moving-parts possibility for direct-control games (which I’ll cover in the next section), but it’s not a perfect solution in every case. Most importantly, the connection between the theme and the mechanics of the end game can be tough to justify. What’s so special about going through the deck three times; why not two or four? If we’re delaying the end of the game by drawing fewer cards or choosing to move the turn tracker fewer spaces, what does that correspond to thematically? A game like Seasons at least attempts to explain some of these mechanical elements thematically, but it mostly boils down to “time travel magic,” which will not be a universally applicable explanation.

Finally, on a related note, it’s much tougher to apply these end conditions to games that aren’t based on the idea of cycling through a deck or counter or a track a certain number of times. In principle, it might be possible to design a territory control game that ends when a total of fifteen of the map’s twenty-one territories have been conquered. Worker placement games, I think, would have an even tougher task trying to pull this off.

Players have Direct Control

Direct Control: A player who grabs the reigns can steer the group to his or her preferred final destination.

On the other side of the spectrum lie games that end when a player fulfills a particular condition, as I’ve alluded to a few times already. Race for the Galaxy ends when a player has a total of 12 developments and planets in his tableau. Chess ends when one player checkmates the other, when one player resigns, when an outright victory becomes impossible, or if the players agree to a draw. Fluxx is an intriguing example of being so directly player-controlled that the win condition itself is undefined at the beginning of the game, and the players change the win condition as frequently as they actively try to win. Team games like The Resistance often end when one team performs its objective (here, a first-to-three in either succeeding or failing missions) and cooperative games like The Grizzled often ask the players to accomplish its objective and thus end (and win) the game before the ticking clock ends it.

As you might expect, direct control can be great for player control as it gives the players a lot of agency in terms of how long the game lasts and on what terms it ends. From a player’s perspective, it can be a lot more fulfilling to end the game because you accomplished something than because the designer decreed that it was time for the game to end.

Does the game stop when the end condition is triggered? Sometimes it ends immediately, sometimes the current round finishes (or, in games with multiple phases per round, the current phase but not the current round finishes), and sometimes everyone just gets one more turn, like in Ticket to Ride. This is nice because it gives players a chance to tie up loose ends and perhaps have a small value play, to throw a Hail Mary if they know they’re going to lose anyway, or just to have fair warning for the end of a game instead of never looking at their pieces again. It’s more emotionally satisfying when a TV network announces a “final season” of a cancelled show instead of unceremoniously failing to renew it (or, worse, yanking it from the schedule mid-season).

Sometimes, final plays can even become fun collaborative puzzles that all of the players can contribute to solving even in competitive games: your strategy is already played out, and it’s not like you can score any more points or the other player can derail your strategy, so why not help the other players to get as many as they can?

As I mentioned earlier, triggering the end of the game is not necessarily the same as winning it. We’ll be exploring this idea in detail in the next article in this series as we look at the various ways in which games might be won or lost. For now, I’ll briefly mention that some direct-control games have their end conditions coupled to win conditions, and some are merely coupled to “let’s do some math” conditions.

Win-coupled end conditions tend to exist in the classical abstracts (when I get all of my stones to my store in Mancala, the game ends, and I win) and in some card games (if you get rid of all of your cards in Rummy or Uno, you win; the first person to reach an agreed-upon number of points over many hands of Cribbage wins) though they’re not uncommon in Euro-style games either, with Settlers of Catan (when someone reaches 10 points, the game immediately ends with that player’s victory) being a prominent example. Among Euro games with player-controlled end conditions, uncoupled end conditions tend to be the norm: (our favorite) Power Grid ends when one player has a certain number of cities in his power network, but the winner is the player who actually powers the most in the last round of the game. That structure serves as a check on a player trying to win through a “rush” strategy and ensures that the winner is the player who actually has the best-constructed network and not simply the most overextended one.

Such player-controlled, win-uncoupled end conditions can provide a tense and exciting endgame, and they can be strategically more interesting than either “it’s the end of the seventh year, so the game is done” or “I end the game, so I win.” In particular, there are fascinating strategic implications with deciding to end the game: sometimes a player can meet the condition to end the game, but it’s ambiguous as to whether he should because he might have a better chance of winning if he lets the game play out another round or two.

In certain rare cases, a directly-controlled end condition is actually coupled to a loss condition: if I’m the clumsy (or simply unlucky) player to grab the wrong brick in Jenga, the tower topples, the game ends, and I lose. Having no way to actually win the game, aside from being the player not to lose the game, is an odd choice that violates our maxim of seeking net-positive satisfaction: the person who loses is deeply dissatisfied, left questioning that last move and wondering if the strategy or simply the execution was off, while the other players are not actually that satisfied, collectively shrug, and are just glad they’re not holding the brick.

One crucial idea here, regardless of whether the end condition is coupled to a win condition or not, is that one player can end the game when he meets the end-game condition. Depending on the extent of player interaction, the other players might have a lot, a little, or nothing at all to say about it. In our earlier article on player interaction, I defined one axis of interaction, similarly to this axis of control, as spanning games with no interaction, implicit interaction, and direct interaction. The closer to the “no interaction” side of the continuum, the more a game might be thought of as “multiplayer solitaire,” and the associated end-game condition is a big reason why. Most of the player interaction in San Juan comes in trying to anticipate which roles your opponents are likely to select and being flexible enough to play off their strategies. But ultimately, a player’s decision to build 12 buildings and therefore end the game is an individual one that the rest of the players cannot do much about.

Therefore, I think that the lack of influence on an end condition that is ostensibly under the players’ direct control is one of the biggest frustrations associated with so-called multiplayer solitaire games. This brings up one of the most important drawbacks to direct-control end conditions: if you’re going to tell the players that they, not the rules, get to decide when the game ends, then a game ending when someone looks up and announces “game over!” is going to feel terribly anticlimactic. On the other hand, games that both 1) give the players a lot of control over when they end and 2) make that decision feel like all of the players are collectively invested in it are necessarily games full of interaction. And as we discussed in our article on player interaction, a direct interaction paradigm is neither an optimal design for all types of games nor something that every player is expecting or comfortable with.

Last man standing

One special case of these interaction-intensive, direct-control end conditions involves “king of the hill” or “last man standing” scenarios where the object of the game is to eliminate all of the other players, and whoever is left is the winner. This case is most common in Risk and other similar war-themed games, where one empire is trying to achieve dominion over the entire world, but you also see it in games like Bang!, where the Renegade can only win if he’s the last character alive.

The most obvious reason why a designer might implement this end condition is because it fits a game’s theme uniquely well: if your game is about conquering the world, then the most intuitive way for it to end is to have someone conquer the world (and, naturally, the winner is the player who conquered the world). We’ve discussed extensively why we’re not the biggest fans of player elimination as a mechanic, and of last-man standing end conditions in general: the Reader’s Digest version is that being eliminated 30 minutes into a 3-hour game (as can, and does, happen in Risk) is one of the biggest ways to drive dissatisfaction, not to mention downtime.

But last-man-standing is making a more thoughtful comeback in the form of micro-games like Love Letter and Coup. In Coup, you’re as likely to be out of the game by your second turn as you are to win, but that’s okay because each game lasts about ten minutes. Love Letter presents an innovative nested approach to player-controlled end conditions: each “hand” of Love Letter is win by a last-man standing condition, but the “game” of Love Letter is won by the first player to win five hands.

Asynchronous endings

Finally, a few games don’t end at the same time for all of the players, even in the absence of a player elimination mechanic. I only know of this occurring in race games, where every player is trying to reach the end of a path, but it’s possible to envision a bizarro-San Juan where every player gets to construct 12 buildings, and the game ends not when the first player hits 12, but instead when the last player does.

Two actual examples are Tokaido, which allows each player to personally achieve their end condition (finishing the pilgrimage from Kyoto to Edo) at their own pace, and Around the World in 80 Days, which has the fascinating idea of in-game days being separate from mechanical turns, and the winner is not the person who circumnavigates the world the fastest within the conceit of the game’s mechanics (i.e., in the fewest turns) but rather the most efficiently within the thematic universe of the game (i.e., in the fewest days). Both games do have incentives to finish the race quickly in the form of bonuses at certain spaces on the road (and Around the World in 80 Days is slightly meaner, furthermore penalizing players who don’t get back to London quickly enough) so that players can’t meander and play a “back game” to wait for optimal moves to show up.

Clank! follows a similar itinerary as Around the World in 80 Days but one in which players have the goal of trying to collect as much loot in the dungeon as possible. A player seeking a rush strategy may decide to forgo larger treasures to “start the clock” on their opponents. The first player to leave the dungeon and return to the surface with at least one treasure is finished with their normal turns and on each future turn advances a pawn on the Countdown Track in the game; a mechanic that adds pressure to players still in the dungeon and increases risk involved in continuing to plunder. After five turns the Countdown Track results in the demise of any players who have yet to surface. There is a strong correlation between how deep into the dungeon a player ventures and their potential score, so the overall effect is a fun challenge at judging when someone has gotten too greedy and quickly rush to the surface with the hope they can’t make it back in time.

Question #2: How is the end condition connected to the game’s design?

Now that we’ve (more exhaustively than I anticipated!) discussed how much control players have over how a game ends, let’s move on and talk about the second axis, how closely is the end condition tied to a game’s design? We’ll approach this similarly to how we talked about a game’s theme in relation to its design and in particular how “persuasive” the end condition is given the game’s theme and mechanics.

Incidental end conditions

Incidental end conditions: Some races have arbitrary lengths and finish lines. Little known fact; the marathon was actually assigned a length of 26.2 miles to encourage consumer sales of the confusingly titled “26.2” bumper stickers.

Some games end at an arbitrary point because they have to end at some point and not last forever. If we were a blog that argued definitions between what makes a “game” versus an “activity”, some might say one difference between the two exists here in that an activity ends at the choice of its participants while a game has pre-defined end points. The key here is that the end condition is essentially independent of either the game’s mechanics or its theme. Playing strictly by the rules, the game it ends at the point that the designer chose, but the game could end at some other point and still feel like the same game and accommodate a wide variety of house rules.

For example, Apples to Apples ends after a player collects a certain number of green cards. This number is ostensibly encoded in the rules (and decreases with increasing player count) but there’s no thematic or mechanical reason that, for example, a 5-player game should take 7 green cards to win instead of 10 or 3. Similarly, Dixit ends when a player scores 30 points, though this could just as easily be 15 or 40 or “everyone is satisfied with the amount of Dixit they’ve played” and it would not change the mechanical functionality of the game or the experience of playing it. (Dixit can also end if the card deck runs out, an example of what I’ll refer to later as a “mechanically persuasive end condition.”)

Typically, games with incidental end conditions are lighter party-style or trivia games, though some strategically deeper games also use them. Lost Cities can be played as only one game, but the “recommended” way to play is to total scores from three games. And the aforementioned Love Letter canonically ends when someone wins 5 hands but it could just as easily be some other number without compromising the experience of playing it.

The primary purpose of an incidental end condition is to give a game a sense of identity where one might not exist otherwise. An end goal and a final objective, even a largely arbitrary one, can turn what would otherwise be an activity into a structured game, and it can keep players working toward something instead of the game just ending when everyone loses interest. Of course, incidental end conditions don’t work well for heavily thematic games, which should end when it makes contextual sense for them to end, or mechanically complex games, which often have more finely tuned balance considerations that could be upset by a longer or shorter experience.

Designer’s Choice (Balance-Justified) end conditions

Designer’s Choice: If the game doesn’t have a natural end point, it’s your judgement call as the designer.

More often, Euro-style and other strategically complex games end at points that are not truly arbitrary but rather represent the designer’s idea of when a balanced, fully experienced game should end. To use the common metaphor of the engine for player strategies in Euro games, many games end once the engine has been constructed and fired up but long before it runs out of gas and needs new parts. As we alluded to in our earlier discussion of dissatisfaction, one useful trick game design is to have the game end just a turn or two before it feels like it should, so it both avoids overstaying its welcome and makes the players want to come back for more.

There’s nothing in the mechanics of Settlers of Catan that would make it impossible to have the game end when someone reaches 8 or 12 points, nor is 10 points thematically significant to Settlers. And sure, you could very easily house-rule that Settlers ends when someone reaches 8 or 12 points instead of 10. The distinction between these games and ones with incidental end-game conditions is that the end condition is not intimately tied to the design of either category games, and changing the end condition is easily possible to imagine for both, those changes would disrupt the experience of playing games with balance-justified end conditions.

Due to the caps on the number of settlements that any one player can have and the single time that any of them can be upgraded, the marginal difficulty of gaining points in Settlers keeps increasing with higher point totals. In other words, although 10 points is only 25% more than 8 points, it’s much more than 25% more difficult to reach 10 than it is to reach 8. In effect, decreasing the end threshold from 10 to 8 would make it much easier to win Settlers accidentally and provide a dissatisfying, anticlimactic end before everyone really felt like their engines were firing. On the other hand, increasing the end condition from 10 to 12 would turn the game into a brutal slog, where the Longest Road changed hands every round and players started buying up the development card deck in chunks in a desperate bid to unearth the Library.

This flavor of end condition can be applied to either player-controlled or player-uncontrolled mechanics. Other player-controlled, balance-justified end conditions include Dominion ending when three piles are depleted (an example of indirect control in contrast to Settlers’ direct control) and one player running out of money in Last Will. Player-uncontrolled, balance-justified end conditions include many games that end after a certain number of rounds, including Small WorldAgricola and Elysium.

These end conditions are pretty simple to implement, especially in the absence of a more compelling mechanical or thematic reason to end the game. They can be finely tuned to suit a game’s unique design and benefit most from playtesting to arrive at an optimal number of turns or player objective. However, they are not necessarily intuitive for players to keep track of or build toward, and they could be a sign of an underdeveloped theme.

Mechanically persuasive end conditions

Mechanically Persuasive: Every book ever written has a final page. The author of a great book invests in making that last page turn meaningful.

This category includes games that end when there’s just no more game to play, maybe because the pieces have made it to the end of the track, there are no more cards in the deck, or one player has all the marbles. This is a “stronger” end condition than the previous category because there’s a physical reason connected with the end of the game, even if there isn’t necessarily a thematic justification to back it up.

Mechanically persuasive end conditions are particularly prevalent among abstracts and card games, which aren’t particularly concerned with theme. Checkers ends when one player doesn’t have any more checkers, trick-taking card games like spades and bridge end when the deck has been run through and there are no more tricks to be taken, and professional games of poker end when one player has control of the entire pot. Scrabble is over when a player has no more tiles to play or draw.

Some thematic games also rely on mechanical cues to end: Hotel Samoa ends once the players have exhausted their bidding tiles, rather than achieving some goal related to developing their hotels. Hotel Samoa features a market-based bidding structure, where players use tiles to place bids on improvements to their hotel and the rate they’re charging to prospective guests. Once a player has used a particular tile, that player can’t use that tile again for the rest of the game. And, naturally, once the supply of tiles is exhausted, the game ends. It would be possible in principle–though it wouldn’t make much sense–to end the game when every player had two tiles remaining, and unless the players were armed with some cardstock and Sharpies, there wouldn’t be anywhere for the game to keep going after all of the tiles were used.

Similarly, Speculation is over once a company token reaches the end of a track, rather than being tied to a particular market event. In Speculation, tokens representing fictional companies move up and down a track as their valuation changes, and the game’s end condition is tied to those tokens reaching the end of the track. You could decide that the game ends when pieces are a few spaces away from the end, though it’s obvious from looking at the game’s mechanical design when it’s supposed to be over.

Connecting an end condition that was otherwise merely balance-driven to a physical component is a good design choice because it gives the players a constant visual and tactile reinforcement of how much the game has progressed and how much it has left to go. Furthermore, because some of the simplest and most universally known games use these end conditions, they can help with the Clarity axiom of approachability. Even though you’re probably designing a more innovative deck of cards than the standard 52-card deck, your players have played a ton of games that end when the deck runs out, so it’s comfortable and familiar for yours to end under that condition too.

If you’re going to use a balance-justified or a mechanically persuasive end condition for the game that you’re designing, how do you know where to set the cutoff? How many times do you need to run through the deck, what’s the right track length, how many turns does each player get? As good friend of the blog and French game designer Xavier Lardy put it, “Ideally, a game should end ‘just before’ the moment the players would wish it to end, to create a desire [for] replay. A game that lasts too long after the game has been played for some or all the players ([where] we already know the winner) makes your game unpleasant.”

Thematically inseparable end conditions

Thematically Inseparable: The final piece of a puzzle is inevitable. It’s predictable, achievable, satisfying and it wouldn’t make sense to end the experience any other way.

The strongest end conditions are ones that make sense both thematically and mechanically. Risk is a game about world domination, and for all of its faults with positional balance and player elimination, it ends precisely when it’s intuitive for it to end: when one player’s empire stands unopposed. Burgle Bros ends appropriately when either the players have been caught or they successfully complete their heist and escape. And Lewis & Clark has a “sudden death” condition where the first player to complete the expedition and reach the Pacific Ocean ends the game and wins.

There’s a pattern here: each of these examples, and many other games in this category are themed around the real world and/or historical events. A game could also be successful in this space if it relies on an established fictional canon, such as Battlestar Galactica or the Game of Thrones board game, by drawing on the “victory conditions” present in those worlds (like the Cylons destroying the human fleet or one of the houses sitting uncontested on the Iron Throne). And of course, it’s not inconceivable that a thematically inseparable end condition could exist in a completely novel, fictional world, and if any of our readers have good examples of games like this, I’d be happy to hear them.

Thematically inseparable end conditions are great because they help the players feel like they’re doing what the game tells them they’re doing, which is a design paradigm that we love at Games Precipice. They can also be very hard to persuasively incorporate into games, especially ones that are not set in a version of Earth or an existing fictional reality. Finally, if an end condition is driven too much by thematic concerns and not by balance, it could be a really dull march for the last ten turns when everyone knows that the Alliance is eventually going to steal the Death Star plans, it’s just a matter of getting the Bothan Spies on the right space at the right time.

Case studies in the control-design space

I’ll wrap up with a closer look at two games that exemplify a wide range of these conditions and illustrate that games can end in vastly different ways and exist in large areas of the space.


Cooperative games always make fascinating case studies for ending conditions, and nearing its tenth anniversary, Pandemic was one of the originals and still one of the best. As I mentioned above, cooperative games tend to have a player-controlled end condition coupled with a win condition, one or more end conditions with varying degrees of player control coupled with loss conditions, and a ticking-clock loss condition lurking just to keep the players moving and the intensity up. Pandemic exemplifies all of these paradigms almost perfectly.

The players in Pandemic win by developing cures for four different diseases. Developing these cures is directly in the players’ control: it is the goal of the game and the objective that the players should be building toward. (To accomplish their goal, the players accumulate cards specific to each disease and take them to research stations.)

Pandemic is set on a map of the world, with nice attention paid to network nodes and edges corresponding to how people actually move around the world. I’ve already noted the tendency of Earth-based games to have very tightly integrated end conditions that reflect the design well, and I see Pandemic’s win condition as almost, but not all of the way, toward the thematically inseparable end of the second axis. Pandemic is about curing diseases that threaten the world, and, fittingly, it can end once all of these diseases have cures. But in a few ways, this end condition is more of a mechanically persuasive condition than a thematically inseparable one: the four diseases intentionally do not represent real diseases, and the game could have been designed with three or six diseases and remained thematically sound. However, cartoon electron micrographs of the four pathogens are displayed prominently on the game’s board, and the colors associated with each disease are reflected on the board, cards, and various other components, leaving little doubt in the players’ minds that their goal is to cure all of them.

The most dramatic way for Pandemic to end is when eight disease outbreaks occur. Essentially, an outbreak happens when a particular city should have its infection burden increased past its capacity, and the disease spreads to neighboring cities. Clearly, this end condition is coupled with a loss condition. This is not something that the players can control directly, but they can take steps to mitigate the threat, most often by treating a disease if it becomes too concentrated in a particular city. In terms of design integration, this end condition is strongly mechanically persuasive (with an Outbreaks track at the side of the board, and a token sliding farther down toward the scary red skull and crossbones each time an outbreak occurs) if not necessarily thematically inseparable (would eight disease outbreaks really cause the public to panic if six or seven didn’t?).

Finally, Pandemic has two sets of components that both represent ticking-clock end conditions: the four sets of colored cubes used to track the disease events and the stack of city cards used to travel and develop cures. If either runs out, the game ends and the players lose; according to the official rules, these lose conditions represent the disease spreading too much and the team running out of time. Both are indirectly controlled by the players, but there are many more ways to affect the number of cubes remaining (again, mostly by treating diseases) than the number of cards left (which are generally drawn at a fixed rate at the end of each turn but may be modified under a few special circumstances). We certainly appreciate designer Matt Leacock’s efforts in making these end conditions more thematic, but they are mostly mechanical end conditions as well.

Twilight Struggle

We posted our design analysis of Twilight Struggle almost two years ago. I haven’t gotten the chance to play it since then, but I still can’t help thinking about it in nearly everything I write about game design. In my original notes for this post, I seemed to recall that “Twilight Struggle can end in six thousand different ways”; to my surprise, the game’s rules list only four.

Three of these are player-controlled “sudden death” conditions; the player who triggers the end condition automatically wins two of them. A player can end and win Twilight Struggle by reaching the end of the tug-of-war style Victory Point track or having a sufficient amount of influence in Europe when a European scoring event is triggered. The Victory Point track is perhaps the single least thematic thing about Twilight Struggle and is a necessary abstraction to wrap up the space race, cultural influence, political developments, and every other aspect of the Cold War that is not reflected on a situation room-esque map of the world. It’s a balance-justified end condition that’s not intimately tied to the game’s mechanics or theme. Having enough influence in Europe to “control” it is much more thematic, and absent any official proclamations from Twilight Struggle’s designers to the contrary, I think the way the Cold War ended in real life.

It certainly didn’t end in the third way that Twilight Struggle can end: triggering DEFCON 1 and a nuclear war. This is a rare example of a Jenga-style directly player-controlled end condition coupled with a loss condition–whoever starts the nuclear war fittingly ends and loses the game–and one of the sneakier ways to win Twilight Struggle is to force your opponent into reaching DEFCON 1. This end condition occupies an interesting sort of middle ground between indirect and direct control (you would never choose to trigger the nuclear war, but occasionally your opponent can make a play to make you do it) and is the most tragic yet thematically appropriate way the game might end.

Finally, a player-uncontrolled safety net exists to catch games of Twilight Struggle that don’t end in any of the above three ways. At the end of round 10, corresponding to in-game 1989, all of the regions are immediately scored on their levels of influence, and the player with the most victory points wins. This is a mechanically persuasive end condition (there are no more cards to play) with a dash of thematic justification (there is no more history to play).


Here, we’ve looked at a variety of end conditions across two different continuums: how directly the players can affect when the game ends, and how much sense the game’s end makes in the context of its theme and mechanics. We’ve seen how certain types of end conditions naturally make the most sense for particular designs and illustrated a couple of case studies with many different end conditions that span wide swaths of the design space we created.

Although some discussion of win conditions has inevitably spilled over into this post, especially because so many sudden-death type end conditions are tied to a player immediately winning, Alex will follow up next with a more detailed discussion of winning conditions, and he’ll finish out this series by taking a look at various scoring systems.

Further Exploration:

Blog – Most Dangerous Game Design – Game Over – Game End Conditions

Blog – Make Them Play – 5 Ways of Ending Your Board Game


4 comments on “Late Game Structures – End Conditions

  1. Pingback: The Spoiler-Free Legacy Mailbag (with Hot Takes) – Games Precipice

  2. Pingback: Structures de fin de partie - Conditions de fin de partie | Xavier LardyXavier Lardy

  3. Alan

    Just found the site and I’m working my way through all the content. Great stuff so far!

  4. Gabriel

    Waiting for the next article! Great work you are doing there!

Contribute to the conversation, leave us your thoughts and feedback