I've had a draft of this sitting around for a while. Then this post by @potatojin showed up and made this a lot easier to finish because now I can just refer to that for half of what I wanted to say: I Like Short Songs.
To summarise his post:
- Short games are awesome.
- A short game is one that you can get good at, because you can play it over and over again. Replayability is excellent.
- Designing a small game, you can hold the entire thing in your head and really make sure it works, rather than just building up a structure and hoping.
I've heard a few times recently games the length of movies described as "short". Let me be clear here: 3 hours is not short. A short game is one you can play multiple times in one sitting. If you play a game and it takes 3 hours, you're not about to play it again today. It's much harder to get good at a game if you can't play it more than once in a row; a tight feedback loop really helps with learning, being able to try out different approaches, see how they work, rethink them, and try again while the previous attempt is still fresh in your mind.
Let's use the classic Sid Meier aesthetic for games; "a series of interesting decisions" - for different aesthetic goals the following is going to be largely meaningless. What makes a decision interesting? I see two aspects: ambiguity and consequence. If there's no ambiguity, if it's obvious what the correct option is, there's no real choice at all. If there's no consequence, if a decision has no real impact on the game, it's similarly a fake choice.
It's implicit in the word "decision" that there's some level of ambiguity and consequence. These aren't binary values though; a decision can be more or less ambiguous and more or less consequential. So we can ask how much ambiguity, how much consequence a decision should preferably have. This is going to have different answers for different people, but for me: I like decisions ambiguous enough to not be obvious, but not so ambiguous that I'm just choosing blindly; there need to be clear reasons to pick one option over another. I like decisions to be quite strongly consequential, to feel like they have a significant impact on the game; I don't enjoy micromanaging for incremental gains.
This last is a reason I like games to be short. In a smaller game, there's room for more consequential decisions. Shorter games have a higher density of decisions; either you cram the same number of choices into a smaller amount of time, or you have fewer choices but more consequential.
Consider the board game 7 Wonders: it lasts for 18 turns, and on each turn you make one main decision. So on average, each of these decisions contributes 5.6% of your total influence on the game. There are 3-7 players, so each turn you're determining about 1% of the outcome of the game. That might seem like a small number, but it's a big chunk compared to a game like Civilisation, where you make thousands of little decisions across hundreds of turns and several hours of play, and most of them don't matter in the slightest. (It's odd that for all that "interesting decisions" are famously Meier's idea of what makes a game good, I find his own games pretty terrible in that regard, being over-long and riddled with tedious micromanagement.)
There's a possible counterargument here that I'd like to acknowledge. I'm measuring the consequence of a choice by what proportion of the game it affects, how much it affects the outcome. But in a longer game, there's scope for pivotal decisions with a larger absolute effect, that still have an impact hours later because the game is still going. However, there can't be many such decisions; I'll take big decisions in a small game over a few giant decisions mixed in with hours of micromanagement.
I'm going to make an analogy with chaos theory, and I'm not going to be very precise - vague and inaccurate analogies to chaos are traditional, even among chaos theorists themselves. Roughly speaking, the study of chaos deals with systems that are neither easy nor completely impossible to predict. Systems balanced between order and disorder, that produce interesting complex behaviour according to deterministic rules. As you evolve a chaotic system, you get feedback effects - positive feedback amplifying perturbations, or negative feedback damping them. There's a delicate balance there, it's easy for them to fall into order or disorder.
Similarly for games, they're best when balanced between predictability and randomness. Positive feedback helps whoever is winning to maintain their lead, and negative feedback helps their opponents to catch up. Both of these are valuable effects: If there's no way to catch up, then the outcome of the game is determined near the start and there's no consequence to subsequent decisions - order. But if it's not an advantage at all, if it's too feasible to make a comeback, then the outcome ends up being essentially random - disorder.
This balance pushes games towards being short. It's easy enough to make a game behave chaotically for a little while, but it's very difficult to maintain it for an extended period - to have whoever's in the lead benefit from it, but still have their opponents having a chance of dethroning them; to have the game continually progress without falling into order or disorder.
But even if you break the balance in either direction, it's not so harmful if the game is short. When positive feedback kicks in and the winning player can't possibly lose, that's dull for them and horribly frustrating for any opponents, but it doesn't matter if the game's over pretty soon after that - it's when it drags on for another hour that there's a problem. Similarly, if there's no way to hold onto a lead and the game constantly swings back and forth, that's frustrating if it goes on for a long time and nobody can make progress, but it's okay in a short game because you only need to be ahead for a short time to win.