Tweet
Faction Settings and the AI
I recently was looking at some old games I'd played, and I decided to check out which factions had been most successful at fighting other factions (when both factions were controlled by the AI), and also to
see which social engineering choices the AI liked. I play a peaceful, builder-style game so the AI's had lots of opportunity to kill each other and develop their own societies.
In 7 games (5 SMAC, 2 SMACX with mixed factions), the Usurpers, Believers, and U.N. (UN???) were the factions which most often killed other AI factions. The AI-run Morgans were *always* killed by other AI's except for one game in which they were very isolated geographically, and the University was usually killed by other AI's.
Looking at the interests of the killee and killer factions (the 1, 0, 1, 1, 0 string which denotes aggressiveness and interest in power, wealth, etc.), I noticed that the killer factions tended to be aggressive and interested in power and growth, while the killee factions were the opposite. The Morgans in particular - the faction which did the worst - are peaceful and not interested in power or expansion. Interest in tech and wealth did not seem to matter.
That seems obvious when you put it like that, but it's useful when you're designing your own factions. What's the point of a faction which gets killed by other AI factions in the opening of every game?
So you may want to avoid the peaceful-no power-no growth combination, and it's probably worth tweaking the Morgans to give them an interest in growth, for example.
The AI's Social Engineering choices
It's worth noting that these are the SE choices of AI factions which survived the other factions, and most of them were recorded in the end-game. So these choices may be the AI's reaction to my playing style, rather than universal selections on the part of the AI. I suspect not, because these selections would be better against a rush-style player than against a builder-style player, and the AI employed them against a builder-style player, but who knows?
Given a choice, the AI almost always chose Planned economics and Power values. It also occasionally (20%) chose Free Market economics.
It had a reasonable variety of choices of politics. Democracy and Fundamentalism were both well-represented, and Miriam and Yang chose Frontier from time to time. It never chose Police State if it had a choice. Then again, considering how much it likes Planned, Police State would be a very bad choice (-4 efficiency).
Now Planned/Power is a pretty bad choices for the end-game. Few human players will choose Power unless they have the Cloning Vats, which is rare for the AI, and while Planned is great for occasional use at
carefully selected times, especially in small factions, it is really bad as an ongoing policy for a faction of 50 or 60 bases that is in a research race. So this suggests some tweaks if, like me, you like
playing solo builder games.
The obvious suggestion for Power is to reduce the Industry penalty from -2 to -1.
Planned has to be changed a lot before it makes sense to use it the way the AI uses it - all the time throughout the end game. First, the efficiency penalty has to go. It is a very bad penalty for the AI, because it doesn't build bases that are individually very strong, so it needs lots of bases to compete in the research race. But inefficiency prevents the outer bases from contributing. Furthermore, I will claim that "inefficiency" is not a very "realistic" disadvantage for "planned" economies anyway. In the game, "inefficient" economies still work well if they're small, but they work badly for large economies, whereas historically planned economies worked as well (as badly) for large economies as small ones. On the other hand, planned economies produced massive pollution, and they were notoriously bad at producing services. So I suggest to replace the -2 efficiency on planned with a -2 planet (all that growth and industry in a hurry hurts) and a -2 economy. This sounds bad, but it's a much smaller penalty for a big faction than the -2 efficiency.
That leaves Planned as +2 growth, +1 industry, -2 planet and -2 economy. That's too good (it'd be way too painless to permanantly pop boom with democracy/planned/creche), so I'd take it down to +1 growth, +1 industry, -2 planet, and -2 economy.
Comments? I *think* those SE settings are reasonably balanced, but I haven't tested them and could easily have missed something. A multiplayer game using them should be viable, but they're mainly intended to make the endgames of solo games tougher.
I'd be interested to hear other people's observations of AI behaviour, particularly the SMACX factions, or the results of someone playing a game using these settings. (I will, when I have time, which probably won't be soon
.
see which social engineering choices the AI liked. I play a peaceful, builder-style game so the AI's had lots of opportunity to kill each other and develop their own societies.
In 7 games (5 SMAC, 2 SMACX with mixed factions), the Usurpers, Believers, and U.N. (UN???) were the factions which most often killed other AI factions. The AI-run Morgans were *always* killed by other AI's except for one game in which they were very isolated geographically, and the University was usually killed by other AI's.
Looking at the interests of the killee and killer factions (the 1, 0, 1, 1, 0 string which denotes aggressiveness and interest in power, wealth, etc.), I noticed that the killer factions tended to be aggressive and interested in power and growth, while the killee factions were the opposite. The Morgans in particular - the faction which did the worst - are peaceful and not interested in power or expansion. Interest in tech and wealth did not seem to matter.
That seems obvious when you put it like that, but it's useful when you're designing your own factions. What's the point of a faction which gets killed by other AI factions in the opening of every game?
So you may want to avoid the peaceful-no power-no growth combination, and it's probably worth tweaking the Morgans to give them an interest in growth, for example.
The AI's Social Engineering choices
It's worth noting that these are the SE choices of AI factions which survived the other factions, and most of them were recorded in the end-game. So these choices may be the AI's reaction to my playing style, rather than universal selections on the part of the AI. I suspect not, because these selections would be better against a rush-style player than against a builder-style player, and the AI employed them against a builder-style player, but who knows?
Given a choice, the AI almost always chose Planned economics and Power values. It also occasionally (20%) chose Free Market economics.
It had a reasonable variety of choices of politics. Democracy and Fundamentalism were both well-represented, and Miriam and Yang chose Frontier from time to time. It never chose Police State if it had a choice. Then again, considering how much it likes Planned, Police State would be a very bad choice (-4 efficiency).
Now Planned/Power is a pretty bad choices for the end-game. Few human players will choose Power unless they have the Cloning Vats, which is rare for the AI, and while Planned is great for occasional use at
carefully selected times, especially in small factions, it is really bad as an ongoing policy for a faction of 50 or 60 bases that is in a research race. So this suggests some tweaks if, like me, you like
playing solo builder games.
The obvious suggestion for Power is to reduce the Industry penalty from -2 to -1.
Planned has to be changed a lot before it makes sense to use it the way the AI uses it - all the time throughout the end game. First, the efficiency penalty has to go. It is a very bad penalty for the AI, because it doesn't build bases that are individually very strong, so it needs lots of bases to compete in the research race. But inefficiency prevents the outer bases from contributing. Furthermore, I will claim that "inefficiency" is not a very "realistic" disadvantage for "planned" economies anyway. In the game, "inefficient" economies still work well if they're small, but they work badly for large economies, whereas historically planned economies worked as well (as badly) for large economies as small ones. On the other hand, planned economies produced massive pollution, and they were notoriously bad at producing services. So I suggest to replace the -2 efficiency on planned with a -2 planet (all that growth and industry in a hurry hurts) and a -2 economy. This sounds bad, but it's a much smaller penalty for a big faction than the -2 efficiency.
That leaves Planned as +2 growth, +1 industry, -2 planet and -2 economy. That's too good (it'd be way too painless to permanantly pop boom with democracy/planned/creche), so I'd take it down to +1 growth, +1 industry, -2 planet, and -2 economy.
Comments? I *think* those SE settings are reasonably balanced, but I haven't tested them and could easily have missed something. A multiplayer game using them should be viable, but they're mainly intended to make the endgames of solo games tougher.
I'd be interested to hear other people's observations of AI behaviour, particularly the SMACX factions, or the results of someone playing a game using these settings. (I will, when I have time, which probably won't be soon
.
Comment