General Framework
September 2, 2022
I want to start this series with the general mindset I employ in Avalon. The following paragraphs form the foundation of my gameplay, and are the most strongly-felt pieces of my overall game methodology. As a side-effect, this post is more introspective. As a result, it also involves more justification than I plan to provide in future posts.
I approach Avalon the same as any other multiplayer game that I enjoy. I enjoy games most when I am doing my best to compete and win — whether it is against other players, the computer, a scenario, and so forth. The outcome isn’t as important to me as knowing that I put my best foot forward. I have the most fun when I play with others who think similarly, even if we disagree about everything else. Competing isn’t fun if my heart isn’t in it, and is even less so when others’ aren’t. Accordingly, I play strategies designed to maximize my long-term winrate: individual wins and losses aren’t a big deal.
To every rule, an exception. (And Avalon is no exception, haha.) I do have one idiosyncrasy that I might discuss at length later, which is throwing fails. I throw a fail at every possible opportunity. I suspect that this might be to the detriment of my overall gameplay, but I’m not so convinced yet. It’s simply too fun to be a part of a failing team, and the arguments that ensue, so I throw fail with abandon.
Of all games, I enjoy repeat player games the most. I love developing, then refining strategies against players doing the same. Finding and executing optimal strategies is deeply satisfying. As a repeat player in an adversarial game, it is important to develop a balance of play: being obviously Good or Evil in Avalon is a fantastic strategy for a one-off game as Good, but is disastrous to one’s long-term winrate. If you fail to balance, you’ll find yourself essentially automatically losing 50% of games you play: ~40% as Evil, plus ~10-15% as Merlin. This will come to the very real chagrin of your teammates.
Although your role will become clearer as the game progresses, from the beginning of the game and for as long as possible, you should try to keep your role uncertain for three reasons: 1) as Evil, you would like to remain plausibly Good for as long as possible; 2) as non-Merlin Good, you would like to remain plausibly Merlin for as long as possible; 3) as Merlin, you would to remain plausibly not-Merlin for as long as possible.
I don’t think this will be a controversial or profound statement,
* but I balance my gameplay by playing as Good. I enter every game doing the things that I would do if I were a good player. Most of the time, this doesn’t involve any acting, because most of the time, I’m good-aligned. My behavior deviates as the game progresses and as I gain information: who is aligned how, and who is what role. I like this approach for two reasons: 1) this requires the least acting the least often from me (I don’t esteem myself as a particularly good actor), and 2) I’m good roughly 50% more often than I’m bad.
Good should prefer rejecting, particularly 1) on early missions and 2) while off-team. This is especially true in 7-player, and a little less true in 10-player. If you’re new, a general rule of only approving missions you’re left off if you have a compelling reason to is a good starting point.
Let’s take a 9-player game, for example. When you’re clueless good, an arbitrarily chosen 3-person mission that does not include you has a (5/8)*(4/7)*(3/6) = 17.8% chance of including no Evil. Even if Merlin always picks a clean mission — which not every Merlin will — your odds only improve to 20.3%. Those are simultaneously horrible odds and essentially the best they will ever get for 3-player missions.
The other reason to bias towards rejection is because seeing more missions tends to benefit Good. More missions gives Good more time to discuss and more things to discuss. Of course, it also gives Evil more information to suss out a leaky Merlin with, but you do have to win on the board before you can dodge an assassination. Even if you’re clueless good, or can’t read signals at the table, this gives your teammates more time to connect with each other, for Merlin to spread information, and for Percival to find Merlin. Remember the odds above: to improve on that 20.3% chance, Good needs to figure things out.
You might say, well, I can just be on the team, and it’ll be okay. In the example 9-player game above, even if you’re on the team, if no Good but Merlin has priors, you still have only got 40% odds that it’s a good team. So even if you’re on the team, you may still want to reject: odds are, without priors, it’s not a winner.
So, it’s really important that Good gets information out to the table. Having more time to discuss and seeing more votes on more teams helps Good do that. Even in a field where people reject heavily, you’ll still be able to extract more information than in a world where everyone simply approves the first team they see. Of course, there are additional considerations. For example, who the force is, who the next force would be if this team were to pass, or voting to hide Merlin. These and more all serve as reasons to overcome what to me is a default state of rejecting teams.
Of course, more discussion provides Evil with more data points about who is and who isn’t Merlin. This isn’t fatal. A lot of this can be shored up with tighter play. While some leaks are simply inevitable, once the marginal utility of additional discussion wears thin, Good can plug these leaks up by simply shutting up and passing winning teams.
* At least, I would have said as much until this week, when I ran into someone who intentionally balances by playing as Evil. I question the wisdom in that. EV aside, I think it’s detrimental to others’ gameplay experiences. Other players must first overcome their priors on an Evil-balancing player to include them in missions at all, and then feel particularly bad if that player proves Evil, especially if that player goes for rub-ins. I enjoy winning but I also prefer when friends enjoy themselves. Ideally, losses shouldn’t sting.