I was asked a very insightful question a few days ago by fellow solo gamer Stephen Miller that me think about a difference between two approaches to making Automas (an artificial opponents made using a specific method) that I’ve used. It occurred to me that making a conscious choice between these two approaches can be very useful in guiding the design choices when making an Automa.
The context for Stephen’s question was that we were discussing two related solo variants for the game Village. They’re nice solo variants, but I mentioned that there were a couple of things that I had considered changing: “When I read the two solo variants I was also thinking about going away from the die rolls which can lead to some very swingy behavior, and I'd probably try to go the way I've done in Between Two Cities and Scythe, where there's a wee bit of intelligence in the actions of the opponent.”
Stephen noted that I didn’t mention Viticulture Automa and asked “Is that specifically for Village, or is that a general leaning in that direction? As in - If you were to have done Viticulture's Automa today would you have it as purely on a card flip, add to the current season, play, or made it more responsive to the current game state (presumably 'to how long the game's gone on for' in the case of Viticulture)?”
A photo of Village. Image credit: Josh E..
Does it play the game?
My short answer to Stephens’s question is that if given the chance I wouldn’t remake Viticulture with a more intelligent Automa.
But then, why would I like to make the Village opponents more intelligent? The answer to that question will come to us from the answer of another question, which you can ask of any artificial opponent for a multiplayer game:
Does it play the game?
What do I mean by this? Well, I’m asking whether the opponent use the mechanics of the game (likely in a simplified manner) to achieve in-game effects or whether it’s just there as an obstacle for the player making it harder for him to score points.
In Viticulture the Automa does not play the game. It doesn’t recruit workers, it doesn’t make and sell wine, and it doesn’t build structures. It’s simply there to block you from taking actions in an unpredictable manner and to provide a target score to compete against.
In Scythe, on the other hand, the Automa does play the game to some extent. It gains military power, moves its units around the board, uses them to attack the player, and it scores points based on the territory it conquers, and stars it places. All this is just like a human player (though it follows simplified rules, so as not to burden the player running it).
Since Viticulture Automa isn’t actually playing the game, and thus doesn’t actually compete with the player using the game’s mechanics, we could allow it to be dumb as a door. Scythe Automa is another kettle of fish. It does play the game and we wanted to make playing against it feel like playing against a human and thus it needed a bit of intelligence.
Had we not given it some intelligence it would sometimes have spent half the game trying to drive its units into the edge of the board or attack repeatedly when it had no attack power.
Intelligence vs. raw power
An alternative to intelligence is raw power, which can be a nice solution if you’re OK with the Automa not feeling like a human opponent. Dumb, but strong can definitely provide an interesting challenge, and of course any Automa will be dumber than a human player and will need to be stronger, so what I’m discussing is more of a matter of degree than quality.
Tiny Epic Galaxies is an example of a game where the artificial opponent is dumb, but powerful. It’ll happily spend a long series of actions futilely trying to colonize with the space ships it hasn’t launched yet, and then after all those futile colonization action it’ll launch them, instead of the other way around, which would actually have accomplished something.
The advantage of the dumb but powerful approach is that it’s easy for the human player to manage, the downside is that the difficulty level can be highly erratic, which is the case in Tiny Epic Galaxies, and since I care a lot for making games tense, I prefer to add some intelligence.
The two Village opponents
With that explained we’re almost ready to answer Stephen’s question, but let me first spend a few paragraphs on the Village solo modes, since that’ll help provide context for my answer.
The first one is very simple. It’s there to provide an obstacle and a timer mechanism: It might initially seem like it doesn’t play the game, but that’s not quite true. It does play the game a little bit. It gains family members, places them on the map, and uses a simplified version of the game’s time mechanism to age and kill off family members.
All of this is done make it contribute to the game end game timer mechanic and to block the player from having free reign over the village chronicle scoring mechanism.
The village chronicle scoring/timer mechanism. Image credit: Ender Wiggins.
The designer of this variant (Dale Buonocore) seems to have used the same design principles for the opponent as I do, that is he has identified which parts of another player’s presence that impacts you and then he has the opponent mimic those.
The one thing missing from Dale’s opponent compared to what my Automa Approach would have included is competing against the opponent. Instead of competing against the opponent, you’re competing against your own high score with the opponent as an obstacle. I’m not saying that my choice is the right one here, I’m just saying that personally I prefer to have a win/lose criteria instead of playing against my own high score.
Dale’s solo opponent has been extended by Schiavonir. I haven’t playing against this version of the opponent, but from the rules it’s clear that it’s designed to actually play the game and replace a human opponent instead of “just” being an obstacle, and it goes further in this regard than I do in my Automa Approach.
Back to Stephen’s question
Dale’s version only plays a small subset of the game and it does this without intelligence. Because of this I initially felt that since it played that part of the game, then the right thing to do would probably be to add intelligence.
But, writing this post has made me think things through, and I’ve realized that since Dale’s opponent only plays the game to such a limited extent, then removing that to make it a non-playing obstacle style opponent might be better than adding intelligence. So, what I’d try to do could be seen as making the Village opponent very similar to the one I made for Viticulture.
That said, without having tried it out, I don’t know whether I could make that work, and it’s important to me to state that I’m not stating that my approach is better than Dale’s, I’m just armchair quarterbacking, while Dale has actually made a well-functioning opponent.
The village chronicle scoring/timer mechanism. Image credit: Ender Wiggins.
For Schiavonir’s version, however, it’s a completely different matter, since that version plays the game to a much larger extent than Dale’s, and thus as discussed above it would likely benefit from some intelligence to avoid a swingy difficulty level and doing stupid stuff like taking the travel action without having anyone to do the travel.
To play or not to play?
I’ve now answered Stephen’s question, but I haven’t really explained why I sometimes prefer to make an Automa that plays the game and sometimes prefer to make one that doesn’t. So why do I think that a non-playing opponent is a good fit for Viticulture, but not for Scythe?
The answer to that question is more or less handed to us if we consider the level of player interaction in each game. In Viticulture you basically have two points of interaction: Action space blocking and play order/bonus selection blocking. The consequences of your actions only apply to your own player mat and has no direct impact on the other players. OK, I’m simplifying things here, since some cards you play in Viticulture can impact other players, but this has little enough impact that I chose to ignore it in Viticulture Automa. It’s also worthwhile to pay attention to the player mats of the other players since that can help you to predict their actions and scoring potential, which should be factored into your decisions.
While Scythe can also be played as a low interaction game if it’s played by two carebears, then it features a lot more options for interaction. You fight for area control, steal resources, race to get first to the central location (“The Factory”), and more importantly the threat of attacks is often a factor in your strategy.
This difference in the level of player interaction is why I think that Viticulture is served well by an “obstacle” Automa, while Scythe needed a “playing the game” Automa. This difference is at the heart of my answer to Stephen, and if you’re considering to make an Automa, then I think that a consideration of the level of player interaction and of whether the Automa should play the game or not, should be a core part of you design process.
For simple games you can of course to go both ways, and I could have implemented an Automa for Viticulture that played the game. Since it’s my philosophy to keep the Automas as simple as possible while still mimicking the core player interactions, I think that going for an Automa that doesn’t play the game was the right choice, since that’s so much simpler and good enough mimic the game’s core interactions.
When I worked on an Automa system for Between Two Cities it was clear to me that the game had a lot of player interaction and thus would be best served by an Automa that plays the game. Therefore we made one that did just that. It was clear however that some players felt it overwhelming to use it before the game’s scoring system had become second nature to them (which luckily it quickly does), so we chose to include a simpler mode that doesn’t really play the game, but mainly generates a score to compete against.
This simpler mode did the job it intended to do and it was fun to play, but it also had a rather different feel than the multiplayer game
There’s another part to my initial comment about making the two Village opponents more intelligent, and that relates to their simple and easy die roll system for action selection, but this post is already quite long, so I'll ignore that for now and hope that I’ll have the time to cover that in a separate post.
A blog about solitaire games and how to design them. I'm your host, Morten, co-designer of solo modes for games such as Scythe, Gaia Project and Viticulture.
13 Nov 2015
- [+] Dice rolls