Comments on the AI

[Ajantaka]

Over the years I have had 2 complaints - no, 3, the Spanish Inquisition has 3 complaints - against the AI.  First, it builds large air forces and does not use them save against colony pods, formers or any target that one aircraft can defeat by itself.  I have asked about this before but been told, I think, that it cannot be corrected, nor can the AI planes be made to bomb improvements.  Is this correct?
       Second, the AI cannot conduct a competent war across oceans, unless the distance is only 1-2 hexes.  This is especially destructive of fun games as it is my experience that the original colony pods are scattered over the world, meaning that there will always be a number of factions seperated by large expanses of water.  Yes, sometimes you get 1 or 2 factions in your immediate neighborhood but this still leaves the problem of AI incompetence over distance of water, expecially in the middle or late game.  The only partial solution is to play with very large land masses and little water, which I don't like as handling navies and such is fun and is 80% or so of the world is land the factions get so large and have so many units that a turn in the later game takes an eternity.  Can this problem be solved?  Has it been solved? 
       Third, some factions are remarkeable do-nothings when it comes to building improvements and most are when it comes to buildings; I'm looking at you, Spartans.  When I get into their territories I find few farms, boreholes ect.  Granted that some factions would, by the nature of their ideologies, build considerably less than others, a very low level of this dooms them to eventual defeat.  Can this be solved?  Has it been solved?
       Fourth, the Spanish Inquisition has 4 problems, and this one is totally trivial.  When I go to all the trouble of liberating a captured faction leader I would like an explosion of fireworks or some other such display to reward me for all my clever action, not just a line saying the geek has been freed.
       Finally, again, this is still, 16 years later, the finest turn based game ever made.  Sorry for the length of the two posts.

Thanks

[Yitzi]

Over the years I have had 2 complaints - no, 3, the Spanish Inquisition has 3 complaints - against the AI.  First, it builds large air forces and does not use them save against colony pods, formers or any target that one aircraft can defeat by itself.  I have asked about this before but been told, I think, that it cannot be corrected, nor can the AI planes be made to bomb improvements.  Is this correct?

With .exe modding (which, in order to change the AI, would require me to first analyze the AI, a big project), it could be changed.  However, air units are fairly expensive for their power (and probably should be even more so), so I'm not sure it's such a good idea for them to be used against hardened targets.

Second, the AI cannot conduct a competent war across oceans, unless the distance is only 1-2 hexes.  This is especially destructive of fun games as it is my experience that the original colony pods are scattered over the world, meaning that there will always be a number of factions seperated by large expanses of water.  Yes, sometimes you get 1 or 2 factions in your immediate neighborhood but this still leaves the problem of AI incompetence over distance of water, expecially in the middle or late game.  The only partial solution is to play with very large land masses and little water, which I don't like as handling navies and such is fun and is 80% or so of the world is land the factions get so large and have so many units that a turn in the later game takes an eternity.  Can this problem be solved?  Has it been solved? 

I don't know if kyrub has done anything about it; if he hasn't, doing so would mean figuring out what it should be doing to wage cross-ocean war.

Third, some factions are remarkeable do-nothings when it comes to building improvements and most are when it comes to buildings; I'm looking at you, Spartans.  When I get into their territories I find few farms, boreholes ect.  Granted that some factions would, by the nature of their ideologies, build considerably less than others, a very low level of this dooms them to eventual defeat.  Can this be solved?  Has it been solved?

It certainly could be with proper analysis; not sure if it has.

Fourth, the Spanish Inquisition has 4 problems, and this one is totally trivial.  When I go to all the trouble of liberating a captured faction leader I would like an explosion of fireworks or some other such display to reward me for all my clever action, not just a line saying the geek has been freed.

You provide the movie, and get enough people to decide it's important, and I can probably figure out how to have it play the movie.

[Ajantaka]

Thanks for the answer to each part of my post and for all the work you have done on patches.

[benschwab]

       Fourth, the Spanish Inquisition has 4 problems, and this one is totally trivial.  When I go to all the trouble of liberating a captured faction leader I would like an explosion of fireworks or some other such display to reward me for all my clever action, not just a line saying the geek has been freed.

In terms of gameplay, it would be nice to have a faction be able to surrender to another faction if it looks like its going to lose:
:  "Hey Lal, Miriam is going to overrun me.  You're running Green right now and have tried to support me.  Do you want all my basses, units, energy, and technology?"
:  "But won't that just make Miriam declare war on me and put me on the same continent as her?"
  "They're really nice bases.  I mean they don't have a lot of facilities but you can't find better parks anywhere on planet."
 :  "I think I can take Miriam, I'm two generations ahead in technology and have the Hunter Seeker Algorithm, and I love me some parks.  Sure."
And if this happens, Deidre is being sheltered by Lal.

I've done this before when I was going to lose anyway.  I don't have to be best friends with the faction I'm surrendering to, I just have to hate them less then the faction that would otherwise take over all of my basses.  It's a great way to be passive aggressive and make war far less profitable.

In terms of aesthetics, it would be nice that instead of capturing the enemy faction leader every time you defeat them, they could flee to another faction.  "Santiago has fled to the Hive and started a Spartans in Exile government."  If I defeat the Hive and any sympathetic faction (based on how one faction feels towards another) is too far away to be fled to, I get them both at that time.  "Santiago has been dragged out of her hidyhole and is now a prisoner of the Peacekeepers."

       Finally, again, this is still, 16 years later, the finest turn based game ever made.  Sorry for the length of the two posts.

Thanks

Now that I think about it, you're right.  What have game makes been doing this last decade?  After I got SMAC, I got Civ 3, 4, and 5.  4 did some things I liked better than 3 and some I liked worse.  I dislike 5 altogether.  Maybe epic isn't as profitable... but what artist is merely concerned with profit.  I read an interview of Will Wright where he was asked what advice to give to young computer companies.  I can't remember the exact quote but his answer was: Don't go public.  The investors on Wall Street will steer your company somewhere you don't want to go.

Maybe I've changed.  Maybe it's harder to impress me with toys now than when I was a kid and once impressed with a toy, I'll always be impressed with the toy.  I fondly remember spending what I thought was late nights in high school playing SMAC.  Even sweeter since it was on my father's computer and I only visited him every other weekend.  If I wanted to get a game done in a month's time, I had to spend several hours each night playing.  Whenever I load the game again, I get wonderful nostalgia.  Maybe that's part of it.

[Yitzi]

       Fourth, the Spanish Inquisition has 4 problems, and this one is totally trivial.  When I go to all the trouble of liberating a captured faction leader I would like an explosion of fireworks or some other such display to reward me for all my clever action, not just a line saying the geek has been freed.

In terms of gameplay, it would be nice to have a faction be able to surrender to another faction if it looks like its going to lose:
:  "Hey Lal, Miriam is going to overrun me.  You're running Green right now and have tried to support me.  Do you want all my basses, units, energy, and technology?"
:  "But won't that just make Miriam declare war on me and put me on the same continent as her?"
  "They're really nice bases.  I mean they don't have a lot of facilities but you can't find better parks anywhere on planet."
 :  "I think I can take Miriam, I'm two generations ahead in technology and have the Hunter Seeker Algorithm, and I love me some parks.  Sure."
And if this happens, Deidre is being sheltered by Lal.

I've done this before when I was going to lose anyway.  I don't have to be best friends with the faction I'm surrendering to, I just have to hate them less then the faction that would otherwise take over all of my basses.  It's a great way to be passive aggressive and make war far less profitable.

In terms of aesthetics, it would be nice that instead of capturing the enemy faction leader every time you defeat them, they could flee to another faction.  "Santiago has fled to the Hive and started a Spartans in Exile government."  If I defeat the Hive and any sympathetic faction (based on how one faction feels towards another) is too far away to be fled to, I get them both at that time.  "Santiago has been dragged out of her hidyhole and is now a prisoner of the Peacekeepers."

The non-AI stuff could be done, the AI stuff would require AI analysis first.

Now that I think about it, you're right.  What have game makes been doing this last decade?

Focusing on graphics instead of solid gameplay and (with some notable exceptions) story.

[Dio]

With .exe modding (which, in order to change the AI, would require me to first analyze the AI, a big project), it could be changed.  However, air units are fairly expensive for their power (and probably should be even more so), so I'm not sure it's such a good idea for them to be used against hardened targets.

I agree with the statement that analyzing the AI is a large project. Furthermore, it helps to specifically determine which aspects you want analyzed because some portions are easier than others to analyze.

[Lord Avalon]

Isn't AI always the hardest part (if you want something decent)?

What is the plan of action? Does it attack in a schwerpunkt? Probably simpler than a two-pronged attack. If there's more than one base that's a potential target (along a border), which one does it go for? How do you have it amass forces before attacking, rather than committing units piecemeal? Taking a border base might open up a lot more targets - now what to do? What is the status of the invasion force: does it consolidate and wait for reinforcements or continue to attack?

Naval invasion seems even more difficult. Has the enemy territory been scouted for better landing sites? Then a bunch of transports have to be built, loaded up and sent all at once - not to mention having naval escorts might be a good idea. Can it send a land force and then attack somewhere else via sea?


Oh, and then there's the whole AI has contact with you because a transport managed to reach your shores, eventually gets pissed off and declares vendetta, but never actually invades because it's far away across the ocean.

[Nexii]

Although to play devil's advocate, if the AI did war effectively I'd argue the game would be near impossible on Transcendi.  The production bonuses the AI gets are massive and with exponential growth are difficult to overcome. 

I think if you set the priority/aggression flags of an AI faction in their text file, you can make them play more 'effectively'.  From what I've observed, the Hive is probably the strongest of the default factions winning the highest amount of AI FFA games.  You could try setting the other 6 to have the same flags as the Hive though since their bonuses differ it might cause some oddities.

If anything I think the AI needs a bit of help buildering.  I rarely see an AI tech/transcend that well once the tech curve speeds up.  Or it makes odd decisions like to never replace Headquarters or build Creches which are very powerful.  Advanced terraforming use is another.  Helping the AI there might take less effort than something seemingly more complex like proper use of air/naval units, unit prioritization based on relative cost, diplomacy, etc.

[Yitzi]

Although to play devil's advocate, if the AI did war effectively I'd argue the game would be near impossible on Transcendi.

That's a good thing.  That way, the top-tier players could still get a SP challenge...and if you don't want a near-impossible enemy, you can always play on a lower difficulty.

[Buster's Uncle]

Seconded.

[Eadee]

Well, primary I'm a pen&paper rpg-player. And from those games comes the following:
 I HATE it when NPCs use different rules than PCs. Its allright if you give your NPCs better stats or better equipment, but I want the same rules for everyone else it just feels wrong.

However I understand that programming a GOOD  AI is hell lot of work and can take forever, therefore programmers tend to give AIs bonuses rather then making them smarter. But I'd actually prefer a game were AIs and players follow the same rules and the diff-setting simply changes the kind of moves of the AIs tend to do. And everyone who manages to create such well-developed AIs has my deepest respect.

[Nexii]

I don't think any 4X game has ever had a truely fearsome AI.  I suspect that intuitively it's much more difficult to program since there are so many aspects (explore -> horizontal growth, build -> vertical growth, discover -> techs/SP, conquer -> military/diplomacy) to consider.

And yes a more difficult Transcend would be a good thing for sure.  But at current bonuses any game where you reside near another AI would not be survivable past early game let alone winnable.  The exponential growth in the early game seems the most difficult to overcome.  That may be because with fewer early game options there are less 'mistakes' the AI can make

[Green1]

Semi old thread, but some do. AoW3 can be pretty brutal at tactical level. the strategic game has improved over patches, though. But, I don't think the holy grail of AI - replication of the MP experience without the hassle of MP that scales to your abilities and is still sandboxy unlike MP - has ever been achieved.

But AI is one of those slippery slopes. At CFC, they have folks playing on Deity just to play on Deity. I am with Eadie, though. Not that I am not "hardcore" enough or some mumbo jumbo the leet kiddies regurgitate. It is a matter of fun. I like competent and if I get beat, I want to know it was on me, not an overpowered mess.

But you also run into the Cyanide Blood Bowl problem. An AI that is good for one faction may suck balls for another. Yang is historically a decent AI but Morgan has been lackluster. Also, a good AI has to be modular. Consider Ete's Annihilators. With better AI, it would crush things. Not build 50 tons of worthless 3res garrisons. The AI components would have to be separate and the game would have to be able to adjust decisions based on strength and circumstances.

Oh, and in case someone reading doesn't know, the AI for AC does not go by fog of war or hidden units. It knows where everything is. Imagine the coding for that.

[Yitzi]

And yes a more difficult Transcend would be a good thing for sure.  But at current bonuses any game where you reside near another AI would not be survivable past early game let alone winnable.  The exponential growth in the early game seems the most difficult to overcome.  That may be because with fewer early game options there are less 'mistakes' the AI can make

Although there's no reason that the AI's bonuses can't be focused somewhat later in the game...

Or that the high AI aggressiveness on high difficulty levels needs to stay how it is...

[Nexii]

True, the innate AI bonuses could start low and scale up with TECH or something like that. 

I found that the pacifist faction flag also tends to make the AI play worse than erratic/aggressive flag...but have to test this out more.  Aggressiveness is sort of a double-edged sword too.  It can be good or bad to play aggressive depending on the situation.