Once upon a time, a game called Caylus (2005) was released, and it was quickly lauded as the best game ever. It soon climbed the ranking charts on BGG, and there was much hysteria about whether it would surpass Puerto Rico (2002) as the new #1. (Spoiler: it did not.) And so when I sat down to play it on December 14, 2005, I had high expectations.

They were not met.

OK, I’ll admit, I didn’t understand at the time that the hype was over the worker-placement mechanism, which created a whole new industry of eurogames, including ones that I quite enjoy such as Agricola (2007), Le Havre (2008), and A Feast for Odin (2016) — which seems to suggest that I particularly enjoy the somewhat smaller Uwe Rosenberg worker-placement game industry. But I also love Viticulture (2015) and much of Richard Breese’s Key series, depending on how you define “worker placement”. I mean how many games can say they were the basis of a mechanic tag on BGG? (Dozens, I suppose, with pointless arguments over whether a game was “first” or not, when it’s really a question of whether it was “inspirational”*.) And how many games can say they created stupid arguments over terminology started by people too lazy to understand the words they’re using?

But here’s what I did understand: Caylus was kinda boring.

Blasphemy? Maybe. It just dragged on and on though. Yeah, I’m sometimes a short-timer on games, and was more so in my early days in the hobby. But Caylus has a particularly bad problem in this regard: it can literally run twice as long as you expect!

The troublemakers are the provost and the bailiff and how they interact. Depending on how the players use the provost, the bailiff will either move one space a turn or two … and his movement controls when the game ends. So if you get stuck in a game with players who are constantly backtracking the provost, you find that Caylus outstays its welcome. I don’t know if that was the case with my first game of Caylus in December 2015, or my second game in February 2016, but I know I’d soured on the game by that point because I didn’t play for another year, and in all time I only played the hottest game of (late) 2005 four times.

I’d say “It’s not you, it’s me”, but it’s definitely you: Caylus had a crippling problem with game length.

Breaking the Game (Length)

Any game with a widely variable game length is problematic, because you just can’t assess how long it might take. But it’s worse than just an unknown table time: games with widely variable game lengths often end up being not-fun. Quite simply, a game that’s intended for a specific length is going to be the most fun at that length. If it runs too long, it’s going to be boring, and if it runs too short, it can be unfulfilling. And I think it’s even worse when players are making the choices about game length; because they have a choice, the game isn’t going to be a random length (which will settle toward the average) but instead it’s likely to be an extreme length (either super short or super long).

The Ascension (2010) deckbuilder is another game where I see this same problem: where player choice can dramatically change the game length. There, the game ends when all the honor crystals are given out: that’s a pretty standard end-game mechanism, where exhaustion of victory-point token is one mechanism for game-end. Race for the Galaxy (2007) does it, and so do many others.

The problem in Asension is that honor crystals are mostly distributed for one very specific sort of play: killing monsters. (Lifebound heroes also occasionally distribute honor when played, but they account for a smaller percentage of the honor total.) Unfortunately, this means that players can rush the game-end — and I tend to see this a lot in the computer version of the game, where the AI players mostly focus on monster killing because the monster-killing Heavy Infantry card has a better cost-to-victory-point ratio than most other cards. Even if there aren’t monsters on the board, Heavy Infantry can kill cultists, and the honor crystals continue to run out of the honor pool at a very rapid speed. In the end, other players are left with a half-constructed deck whose engine never got going, extinguishing a lot of fun in the game. Meanwhile, I’ve seen the exact opposite problem in in-person games: no one is willing to buy monster-killing cards and so the center row clogs with monsters that no one can defeat. And that makes the game drag on and on.

When a game tends to always run too short in its computer incarnation and always too long in its physical incarnation, that’s a prime sign of a problematic game length. Because it shows that the game length is largely controlled by player choice (and unfortunately, the different player choices in the physical and computer game versions tend to push it toward the least desirable lengths for each of those mediums).

(And I should say, unlike Caylus, I loved and still love Ascension, but I think that’s because its extremes of game length are smaller, probably accounting for a few rounds of difference at most, not an entire doubling of the game length.)

Fixing the Game Length

Don’t take this rant to mean that variable game length is problematic: it’s not. Variable game length can add both tension and strategy to a game. If you don’t know when the game is going to end, that’s tension. And if you can slightly move the needle, to make the game last a round less or a round more, that’s strategy. The problem is a game like Caylus that can last twice as long or a game like Ascension that can have enough variability to move it outside of its zone of best-play.

So how do you have variable game length without fully telescoping game length?

Better design.

That’s often done by replacing a player choice with a randomization, because randomization will eventually move towards the mean (if the randomization occurs sufficiently!).

Ra: The Dice Game (2009) offers a fine example. Players throw a hand of five dice, and get to reroll them a couple of times. At the end of each player’s turn, any die showing a sun is added to the counter that advances the game toward the end of the round (and eventually, the end of the game).

But, that could have created a lot of variance, as the timer could have advanced 0-5 slots on each turn, so game designer par excellent Reiner Knizia also put in a limiter: if more than two suns turn up, then the timer isn’t advanced at all, meaning that in reality it advances 0-2 suns each turn. This keeps the game length bounded and makes sure it doesn’t spiral into any less likely randomizations.

Even with total player choice, game length can be tightly constrained, as is the case with the excellent Poker-derivative, Havoc: The Hundred Years War (2005). In a normal game, there are up to two rounds of play before each of the first eight wars, but players can make the game go short by starting wars early. How do you balance this? Self-interest. A player only calls a war early if he thinks he can win, but the longer it’s been since the last war … the more likely that someone feels like they are that especially lucky player.

However it’s more than just self-interest. As opposed to Caylus, the correlation between game length and winning in Havoc is much clearer. In Caylus, the movement of the provost and the bailiff is not just really abstracted, but it’s far removed from winning. A player can make the game go long or short without feeling like they’re making a commitment. But in Havoc, a player is making a real commitment to win a war when they shorten the game, so they make very sure they know what they’re doing**.

Overall, the way to fix a broken game-length design is just to apply game design. It needs to be carefully considered, like the rest of a game’s play; problems occur when a designer thinks the issues related to game length aren’t important. The tight limitations of Ra and the strong incentives of Havoc demonstrate two ways to manage a game’s length, but they’re just two possibilities among an infinity of choices.

The Problem with Player Numbers

Before I close out on the problems of game length, I want to complain a bit about a closely related topic, player numbers (which I ranted about a decade ago: some things never change).

The problem is that too many games don’t account for player numbers when determining their game length.

If a game manages to stay the same length for most player numbers, that’s great. Consider Pandemic (2008): the size of the player deck is set, and so the game is bounded by the same length if you have two players or four.

If a game goes a little bit longer with more players, I think that actually might be ideal: individual players don’t lose too much of the experience, but still you don’t have to massively change the game length.

But the games that chunk up an extra 30 or 45 minutes, those are problematic, especially when they don’t admit it on the box. Heck, box labeling might fix most of those problems, as “30 minutes / player” is actually more informative than “1-2 hours”, but it’s still a subpar situation where you’re trying to get a game to the table at a gathering, and you have to calculate whether it’s playable in part based on how many people are interested. Give me a game that only runs 3-4 players any day over one that runs 3-6, but which adds on 20-30 minutes for every player you add.

And then there’s the occasional game that gets exponentially longer the more players you add. Bleh!

(Also the worst: games that just list one play length, irrespective of the number of players. So you sit down for your 60-minute game, and two and a half hours later you’re cursing out the four friends unlucky enough to be playing with you, as well as scaring the cats.)

Just as designers need to think about game length in their design, they also need to think about the effects of player counts on that game length, rather than just throwing their games into the wild and hoping for the best (because then those are the games that then get thrown in the trash).

That is my game length TED talk. Thank you for listening to me rant.

* The pedants will argue “Caylus wasn’t the first worker-placement game.” Probably the case, though I violently disagree with the idea that Keydom (1999) / Aladdin’s Dragons (2000) was anything but an auction game. But, the “first” is rarely an important marker in any case: the “popularizer” is what matters.

** Could a bad player still ruin the game by shortening every round of play? Yes. It’s a vulnerability in the game, but one that will only be exploited by someone purposefully breaking the game, and they’ll likely be disinvited from the next game.

– The image of Caylus is courtesy 3EBC on BGG. This version of it has been cropped to highlight the problematic pieces.

Liked it? Take a second to support Shannon Appelcline on Patreon!