Of all the facets of Deadzone, probably the most challenging in design terms is the solo rules. Actually I should be a bit more specific. There are going to be two versions of solo Deadzone, and what I’m referring to here is the “full” version. The solo zombie thing is lots of fun, but also lots easier to design.
The “full” solo rules are what you need when you want to play a classic 2-player game and only have one player to hand. They are a set of rules that attempt to replicate the decisions your normal opponent would make. Except we all know that you can’t reduce the skill, cunning and silliness of real humans to a simple set of rules in a book. No, we needs cards too.
The solo rules will use a set of cards (we’re calling them an AI deck) as a quick and simple mechanism to generate tactical choices by your imaginary opponent. A card deck allows me to include a lot of sophisticated numerical shenanigans without a lot of rules in your face. Flipping a card for a decision is also a very simple and familiar kind of mechanic that won’t intrude too much into the rest of the game. What the cards have on them is the nifty part.
The rules need to differentiate between a number of mutable things. Firstly, different models need to behave differently. No point in having snipers running into melee. Secondly, different factions need to behave differently. Real players wouldn’t play Plague and Enforcers the same way so neither should the AI. Thirdly, the AI should respond to differing threats with appropriate responses.
Now I’ve seen all of these elements done badly, sometimes all in one game though the usual problem is ignoring one or more of them. I don’t think I’ve ever seen them all done well in one place.
One of the biggest challenges is predictability. The fewer of these elements you include, the more predictable the AI becomes, and the more the game becomes a case of playing the AI rather than the actual battle. Even with all these elements in place it would be good to have a random seed somewhere in the equation to make the opponent a bit less predictable. What we’re after is an enemy that behaves in character, but not always in exactly the same way. That also allows us to replay the game as many times as we like without it repeating itself.
Video games are an interesting place to look for AI ideas as they frequently use AI systems for single player games. Most of them are surprisingly unsophisticated. A lot of the time you’ve got so much else happening (images, sounds, movement) on screen that you don’t notice the lack of smarts in the AI. On the tabletop it’s another story: you can’t hide behind the shiny special effects. If I want the solo game to be good then it needs clever AI.
The AI deck and rules are, in effect, a decision engine. They are there to replace the human opponent with as close a facsimile as possible. That’s a complex process and could easily be a faff to work out in game, so my aim is to pre-calculate as much of this thinking as possible. When it comes to an in-game decision all you should do is turn over a card or two.
To go back to the challenges, the first and second can be dealt with by categorising the models. We already have the idea of leaders, specialists and troopers for building the force. The solo rules just need this broken down one step further into preference for shooting, fighting or support. A shooting specialist can therefore be made to behave differently from a shooting trooper, and that level of subtlety is a good thing. Faction can also be taken into account in a similar way.
The ability to assess threat comes down to a set of rules and priorities which I’m working on now. Streamlining these will be key. However, streamlining them too far makes them fairly pointless and I’ve seen it taken way too far way too often. Often it’s a rigid rule as well, which again makes it predictable, and predictable is both dull and easily defeated. Easily defeated tends to mean “not played again”, and it’s pointless working on rules that are unlikely to be used. Dealing with perceived threats need not be baroque, but it does need to be variable.
Random seeds can be included in the breakdown of responses by each of the model types. The nice thing about a deck of cards is that you can build this sort of complex variation into it whilst having no impact at all on the slickness of the rules in play – one of the reasons I like cards.
Of course, once the AI has made the decision you need a set of rules about how this is implemented, though here you have an ally: yourself. I think it’s safe to assume that you’re going to win a solo game, so the reason of playing is a pure attempt to have fun rather than polish your ego with resounding victories. For this reason we can actually enlist the player as an aid to help us with some of the options if not the decisions themselves.
Playing With Myself
The best part of developing solo rules is that they are a doddle to test. Any time you need to try something out you’ve got a ready group, just waiting to go. Compare that with trying to road test 6 player DreadBall and you’ll see what I mean. How easy is it to arrange a 6 player (not 5 or 7) game of anything. By that measure, solo rules should be the most refined ones going.
Naturally I will still get other folk to play them as well, and already have a number of volunteers lined up. It’s funny how many solo gamers creep out of the woodwork as soon as you mention there are rules to be had. As a group it appears that solo gamers have long been starved of attention and it’s obvious why when you look at the problem of designing the rules. Getting solo rules right is not easy to do and often avoided because of that. I’m just banking on me being more stubborn than it is difficult😉