The Importance of Scalability
One of the most important decisions you will make when designing a game is the number of players it will support. In fact, this will probably be (or at least it should be) decision number two or three for the project.
The impact is massive. It determines the components needed (and therefore production costs), the length of the game (more players typically means longer), its complexity, pacing (how do you keep 3 players from falling asleep as they await their turns?), and who will purchase your game. After all, many people such as myself will rarely consider a game that doesn't play with two. I'm currently designing a game that plays only with two and it terrifies me to think how that will limit the game.
I'm going to make some bold statements. After all, debate isn't fun when somebody meanders forth with a soggy opinion.
Statement #1: Your game should be excellent in every variation specified on the box. If you say two to four on the box, your game should be fun with two, three, or four players.
For many, the "two" on "two to four" is a joke. This is often thrown in hastily by a publisher seeking the broadest possible audience. A friend recently noted his 2-6 game was originally designed and tuned for 3-5, but it was changed by the publisher. 7 Wonders, a wonderful game loved by many (myself included), boldly states 2-7 players on the side of the box. That 2 is optimistic at best.
I learned this lesson the hard way when I initially released Farmageddon on The Game Crafter. I assumed that nobody bought it to play with two players. In reality, many people bought it for two players and they weren't pleased with the result. There were a few problems I had to resolve.
Many of the cards simply didn't work with two players. Crop Rotation used to change the order of play. In two player, it skipped the other player's turn. Not fun. Foul Manure was a two turn blocker. In two player, it was a pointless card.
The pacing was off. Drawing only two cards per turn meant the game took forever. Action cards were too potent. Because you only had one player to target, the game became mutually assured destruction.
I took this feedback to heart and set out in earnest to improve the experience. Over the course of its development, two player Farmageddon is one of the most thoroughly tested aspects of the game. Now, every card works regardless of the number of players. There are no special two player rules on any cards. To improve the pacing, players draw more Crop cards every turn and must play more as Fertilizer. This keeps the game moving at a steady clip. The Action cards vary more in function now and aren't pure offense. The Action deck features offensive, defensive, and support cards. Finally, players draw fewer Action cards, which means if you spend all your bullets hastily, you'll have far less ammo.
Two player Farmageddon is different, but it's good. How can this be applied to other designs? What are the core questions one must ask? I encourage designers to always establish goals at the outset of every project for the play experience. When testing for player number variations, test against these goals. Here are some to consider:
The pacing of the game is fun. It doesn't take too long for the game to end, nor do players grow bored from waiting. In 7 Wonders, players take simultaneous turns. In Bohnanza, players trade regardless of whose turn it is. You're never far from a decision.
The complexity is sufficient. One of the reasons many auction/bidding games require 3 players is because a two player auction isn't very fun. On the other hand, one of the reasons games like Ascension or Magic: The Gathering are less fun with more than 2 players is that there is too much information to track with too many players.
The amount of uncertainty is appropriate. War games are, by their very nature, confrontational. If the theme wasn't war, people would potentially call them a "take that." Because of this, many war games feature only two sides to reduce the amount of uncertainty and back and forth. You know who your enemy is. With King of Tokyo, the amount of information on the board is low and the probability is easily gauged. Therefore, the amount of uncertainty with even 6 players is just fine.
Choices aren't obvious. It's not much fun in a game when your choices become obvious. As I noted above, this was a problem with Farmageddon. If you have only one opponent and a pile of weapons, of course you're going to use it against your one opponent. Design your game such that it isn't just WHO to target, but HOW to target and with WHAT. King of Tokyo does this very well. Often, when two players are left the game is a race to the finish. Will you go for points? Go for the Knock Out? How will you do either?
If you take the time to ask the questions and devote efforts towards testing the answers, you'll find that your 3-5 label isn't a lie, but a bold proclamation for variety and fun. Also, in making your game work for every player number variant, you may find that you'll improve the game overall. Testing for two players revealed some of Farmageddon's worst elements.
Statement #2: Your game should scale with as few tweaks as possible. There are varying degrees of acceptable modifications here on a scale. There are exceptions in all things and degrees of variance with each of these states. Nothing is set in stone, but it's good to create a starting point.
Worst: Fake players (i.e. you must manage an AI)
Less Worse: Add or Replace Rules
Okay: Vary Existing Rules
Better: Modify the Presentation or Game Content
Best: No changes
Making no changes between the number of players is clearly ideal. This means your players read the rules once and are set forever in every situation. This isn't always feasible. Forbidden Island does this. With only two players, you have fewer class abilities (less flexibility), but you draw flood cards less quickly. It's a great trade-off and you don't need to change anything.
Modified content is the next step. If your game has scarce resources, you may need to vary the supply per the number of players. If you have a war game, you may need to alter the number of armies allowed. Bohnanza states which bean types to remove for certain games, Small World provides different boards, and Die Speicherstadt has fewer cards auctioned each round. They are the same games with slightly modified components.
Varied rules can be used sparingly. In Farmageddon, I varied the number of Crop cards drawn at the start of the turn (+1), increased the amount of Fertilizer to be played (+1), and reduced the number of Action cards drawn (-1). You still have the same rules, they are just executed slightly differently. It's not too bad, but even I have to reference it if a great deal of time has passed since I've last played with 2 players.
With new rules, you need good justification for doing so. I'm currently doing this for Empire Reborn. Though it has tested well and is pretty straightforward, it leaves me uneasy. In a three and four player game there are sufficient units to cover the map. However, with only two armies, instead of creating a map for only one play type, I chose to modify the rules. In a two player game, each player controls one main Army, like in any game, as well as one ally army. The ally army is severely curtailed in capabilities and therefore adds little overhead or too much additional management. They essentially help fill the board with warm bodies. I'm currently exploring a different way of creating the boards to more easily allow a board that scales for 2, 3, or 4 players. This would replace the current two player rules, relegating them to variant status.
I explain the two player rules with under a half page of quick bullet points, but I've seen some games with pages of rules for different player numbers. This is dangerous ground and is burdensome for your players.
Finally, fake players, in my opinion, are the worst way to allow a game to work with multiple players. Requiring players to fully manage their own team as well as remembering to maintain a neutral third entity is not much fun. Creating a game with excellent pacing is already difficult enough. This just kicks your pacing squarely in its shins.
More statements can be made, but I believe this is a good stopping point for now. What games scale the best in your opinion? What scaling problems have you encountered? What assertions did I get incorrectly? Comment below!