Tweet
I'm also of the opinion that the AI could have been made better given a wet weekend's work.
One thing which would massively help it would be the instruction to
1) Improve tiles which are being worked, and
2) Don't improve tiles which aren't being worked.
I can't see a particular problem with this, just have a city controller AI flag which squares it would like improved, and how.
These days I play Alpha Centauri mostly, and I have seen the AI in that build a mine on flat (i.e. the wrong type of) terrain which wasn't in the radius of it's base.
And the battlefield AI wouldn't be so hard either. In Smac the AI attacks with up to 20 (occasionally more) troops at a time, but it stacks them all on the same square and then sends them. They die from counter-attacks. Have them spread out a little and they would be dangerous.
Or an algorithm to count the number of squares on an island - just a simple floodfill, like you get in MS paint - and increase the priority of naval techs if it is small. Or to reduce settler production if there are no more city sites (Area/(21*city-number) = 1, if you want to be lazy).
Or to steal tech when an opponent is ahead, rather than at random or not at all.
Or to work out whether the base will grow next turn, and if so to work out if doing so will make the city discontent, and finally if so to create an elvis.
Or not to build marketplaces until they will pay for themselves. Likewise banks, stock exchanges, and on a more abstract level libraries and universities.
Or not to attack with warriors when your opponent is using tanks (yes, I've actually had this, a warrior and a catapult from a trireme while I way conducting a modern-age war. And yes, on Diety.)
I'm one of those players who trounces the AI at Deity, and I wonder how many people who play on lower levels know the extent to which is cheats. Open up the editor and count the columns of minerals & food. If you had nearly halved production costs, nearly double growth, king-level happiness, no maintainance, wouldn't you be rather incredible at the game? Watch a game with the map revealed. I think you will be astounded by how bad the AI actually is.
Yes, it has to cheat in order to put up a really good show. But ask players like Ming what sort of effect optimizing a city each turn, so that it grows a turn sooner, produces a turn sooner, or *doesn't go into revolt* makes. A computer is perfect for this, it doesn't take any great strategies.
I wish the AI were open, then I would put my code where my mouth is. Sadly, from the 10-minute look I had at Clash, they don't need AI help.
Anyway, enough ranting.
One thing which would massively help it would be the instruction to
1) Improve tiles which are being worked, and
2) Don't improve tiles which aren't being worked.
I can't see a particular problem with this, just have a city controller AI flag which squares it would like improved, and how.
These days I play Alpha Centauri mostly, and I have seen the AI in that build a mine on flat (i.e. the wrong type of) terrain which wasn't in the radius of it's base.
And the battlefield AI wouldn't be so hard either. In Smac the AI attacks with up to 20 (occasionally more) troops at a time, but it stacks them all on the same square and then sends them. They die from counter-attacks. Have them spread out a little and they would be dangerous.
Or an algorithm to count the number of squares on an island - just a simple floodfill, like you get in MS paint - and increase the priority of naval techs if it is small. Or to reduce settler production if there are no more city sites (Area/(21*city-number) = 1, if you want to be lazy).
Or to steal tech when an opponent is ahead, rather than at random or not at all.
Or to work out whether the base will grow next turn, and if so to work out if doing so will make the city discontent, and finally if so to create an elvis.
Or not to build marketplaces until they will pay for themselves. Likewise banks, stock exchanges, and on a more abstract level libraries and universities.
Or not to attack with warriors when your opponent is using tanks (yes, I've actually had this, a warrior and a catapult from a trireme while I way conducting a modern-age war. And yes, on Diety.)
I'm one of those players who trounces the AI at Deity, and I wonder how many people who play on lower levels know the extent to which is cheats. Open up the editor and count the columns of minerals & food. If you had nearly halved production costs, nearly double growth, king-level happiness, no maintainance, wouldn't you be rather incredible at the game? Watch a game with the map revealed. I think you will be astounded by how bad the AI actually is.
Yes, it has to cheat in order to put up a really good show. But ask players like Ming what sort of effect optimizing a city each turn, so that it grows a turn sooner, produces a turn sooner, or *doesn't go into revolt* makes. A computer is perfect for this, it doesn't take any great strategies.
I wish the AI were open, then I would put my code where my mouth is. Sadly, from the 10-minute look I had at Clash, they don't need AI help.
Anyway, enough ranting.
Microprose released patch 1.2 for ToT today, after it was determined that approximately 0.001 % of gamers will go into violent, possibly fatal convulsions, when hearing the sound of an elephant attacking, immediately after they hear the game sound of cannon fire. All ToT players are earnestly urged to update their files. The patch ensures that some sort of goblin will always be interpolated between the cannon and elephant sounds. "While we were at it, we improved the AI combat logic, made it easy for scenario designers to animate their own units, included instructions for better controlling the random generation of maps 1, 2 and 3, and added a map editor," said a spokesperson for MP. "We tested out the map editor on maps produced by our rock solid civ2-to-ToT scenario converter, also included in the patch." Foreign language versions of the patch are expected in about two to three months; foreign language players are urged to turn off sound effects interim...
Comment