Unlike GalCiv II, the GalCiv III AI and overall system is extremely moddable. I'm glad they did that but I'm having a hard time making the AI play the game as well as I'd like.
Whether people like it or not, skill at a game is the result of knowing effective, specific, strategies. You can't procedurally generate a strategy. Some of it needs to be hard-coded and that's where we're struggling because GalCiv III is all about data mining and procedural design. It is much more like a Chess AI than what I've done in the past.
For instance, in Starcraft, I'm a mid-level diamond league player. I have very specific build orders, very specific map control strategies that rely on timing and a very tight control of resources. In GalCiv III, presently, I can't even make a ship focus on a particular type of defense. Heck, getting the AI to not decide to build endless constructors is a lot harder than you'd think.
Anyway, you're not here to read me complain about our own game. So let's roll up our sleeves, we've got years to do this. What I can say is that the AI system in GalCiv III is on a much better foundation than GalCiv II. There's much more we can do. But it's going to take time and frankly, a lot of community help.
What I need form you guys are saved games with an explanation of a strategic error you see the AI making. I don't need help in coding what the AI should do because I can assure you, all those functions are there. It's simply calculating that doing something else is a better idea.
GalCiv II way:
If (transportisnearmylanet)then (do specific bad things to that opponent)
GalCiv III way:
Go through and weight zillions of possible things you might do, sort those weights and execute various empire-wide build/movement decisions.
Much of the work on GalCiv III will be making decisions more locally.
I'm sure you dont need to see a save game for this, as the AI seems to always do it. Whenever AT WAR, the AI should amass ships in a theater in max logistical fleets, using single ships as "spares" to supplement losses or damaged ships. The spares should be 1-2 turns away in range, so the main fleet and reserve fleet/ships can merge in a single turn.
As well when at war, the AI should NEVER EVER have multiple ships in 1 hex, that are not combined in a fleet. Even in a single turn where ship's paths may cross. The exception can occur if behind the front, at a great distance. Sure, the human or other AI could "sneak up", but any good player or AI will have sensors setup to watch possible invasion paths.
Quite often, the AI seems to have the "power" to destroy a human player quite easily, or atleast put up a helluva fight. The major problem I see is its infantile use of that military power.
My one comment in this regard is that perhaps the weight on "opportunistic weak target killing" is too high. I know the AI can amass a fleet and attack a fleet. I also know it can amass a fleet to assault a heavily defended planet or starbase. But I'm pretty sure large numbers of its ships get tied up in trying to attack "weak" targets.
Basically they see a constructor or single ship flying through my space and dispatch a single ship to score the easy kill. But they do not take into account the power of my fleet a short distance away that will obviously (to a human) go help. So they do this for a few "choice" targets and end up losing a ton of firepower. If I'm anywhere near correct with my assessment then I'd jack down the priority on opportunism. It is probably best for the AI to attempt to overpower your strongest fleets and planets first, thereby weakening you and making subsequent actions easier. For a human this doesn't make sense but for a buffed up AI it might be the best approach.
The AI doesn't take advantage of diplomacy and that is a huge weakness in it's strength. For instance in any given game.
I the human player can trade all other races (up to 99) 1 technology for x technology the other players have, therefore that one technology gives me potentially 99 different technologies. Now I know a lot of them have researched duplicate technologies and that is only in theory, in actuality it's usually 5-10 different technologies. Because of this (even if the human is trading tech for tech and not using some other exploits to trade for techs) the human player can jump out to a huge knowledge lead even if falling behind in other areas.
With this I suggest doing more of the following, now some of this is already in play but, needs to be more.
If (faction type) focuses on technology then when research technology attempt trade with all races for different technology.
To prevent all factions or AI to focus this way obviously if they aren't focusing on technology this falls further down on their list.
Also if (faction type) sticks to themselves and doesn't expand (sorry forgot the word for it) make that type much more difficult to trade things for as they want more for it or need to align more with other factions ideology types to have them more inclined to trade.
This will help with the human getting the early advantage with technology early on in the game with tech trading enabled.
Along with the whole diplomacy stage of things, have a counter offer in there when allying with other races that they will agree, however, only if you declare war with other factions that are at war with them. (help fight their wars).
If one faction/player is way ahead in knowledge than make other factions more and more unlikely to trade technologies with that faction unless they are allied.
Yes. I too agree with the above posters. They are correct.
Also, the AI shouldn't build up their worlds until they've colonized as many planets as they can. They start building Xeno Factories and what-not too early.
Finally, the AI don't build ANY military units for close to the first 100 turns. Makes them predictable, and thus exploitable....
I see this as well. When you see the AI putting ships on a tile, it WANTS to put them into a fleet. Why it isn't is something I'm trying to figure out. As you can imagine, it's frustrating.
The only reason to not fleet up is to retain the ability for the faster ships to get there before the slower ships. Alternatively perhaps the AI calculates the logistics incorrectly and doesn't think it can fit the whole group into a single fleet so it does nothing.
I have seen this behavior as well. Currently I have war with the Yor and it has dozens of (single) ships in an area. Another Doze are fleeted up into fleets of 2 and 3 ships. I see it merge the 2+1 ships but then do nothing. I am late game and I am severely over powered in relation to the ai. I am playing on all ai on gifted.
...or make tech trading something that needs to be researched first (by both parties).
That would defeat the purpose of tech trading?
As a practical matter, if we could define how many auto-save slots the game maintained, submitting save games for AI debugging would be easier.
Currently, game only supports 2 slots i think.
Didn't SK suffer from this issue too at some point? No idea if it's related, but I guess it's worth mentioning on the off chance that it's the same root cause.
Not that I agree with it (or disagree) but I think he means that the ability to tech trade would be a something you have to research, similar to a research treaty.
Just a few things that come to my mind.
I assume in the code you have some "evaluate value of" sort of thing.
As others said, the AI should "evaluate: put ships in a fleet" higher, and also "evaluate: built higher movement speed ships" higher.
In this regard, it may be really helpful if a function is put in as in GC2, where one gets a template and then "auto-applies" weapons and technologies onto it, like "focus beam, focus kinetic" etc. as it was in GC2. This will for once help the players to upgrade their own ship designs; but it will also help the AI to built different types of ships that are not pre-scripted.
Also, before starting a war, the AI should perhaps calculate a "fake battle" between their best fleet and the best fleet of the player, and re-run the calculation a few times until it gets a decent outcome. If it can't get it to work, maybe the AI should reconsider declaring war.
Yeah, I love how the AI DOWs me, I create a single fleet in 6 turns, and that one fleet wipes out their entire armada.....silly AI, silly AI.....
I just managed to pull off a victory on Suicidal difficulty on a large map in 98 turns. The result is on the Metaverse and if needed I have a save game to prove that no cheats were involved. Note I have modded Carriers (badly) because they are poorly balanced but at that point in the game nobody had them anyway. I had a custom race but avoided use of the coloniser trait which marigoldran has shown could use a nerf.
The strategy was a combination of fast-tracking alliances and a military rush of any AI that I couldn't get an alliance with. Actually, I declared war on the Altarians during the colony rush, because well ... I hate them more than the Yor!
Based on that I just want to reinforce the points made above. The AI had better technology but I was using fleets with far higher logistics and they didn't manage to get a fleet together with enough ships. If the AI had built larger fleets I may not have pulled this off.
The other surprise was how easy getting a couple of quick Alliances was, and although I did get lucky with race spawns, I feel the AI should be progressively less interested in Alliances if it's going to lead to victory by the player.
Here's a save-game file (GC3 version 1.03) -- I hope this helps.
Playing Thalan. Challenging difficulty, huge galaxy, 7 AI stock factions all Gifted. Using mods: 1. AIKickstart75 v1.0 by rspiccaver https://forums.galciv3.com/467746 2. MediumMissileBoatFix by TurielD https://forums.galciv3.com/467065/page/1/ 3. WeaponAugmentFix by TurielD https://forums.galciv3.com/467065/page/1/
Altarians declared war on me.AI Problem: Altarians keep sending a single ship with kinetic weapons, no defenses on suicide flights into Thalan ZOI: Smashing Ideal-L M2-13 (next to Thalan ship Sensor Bug M1-1, east of Petroni Starbase 1)Altarians have done this before so should know what's awaiting this attacker, and it won't be good for them.Note the Thalan fleets lying in wait: Thalan ships have armor, beam weapons and missiles.Zip file includes save-game file and contents of my Mods folder for reference.https://www.dropbox.com/s/cycvov2sa2nkwtt/thalan%20-%20lone%20altarian%20attacker%20save.zip?dl=0
An aside: rspiccaver's mod has made the AIs much better at the colony-rush in this game. This is the first game I've played where the AIs have colonized more planets than me (I've long stopped using the Pragmatic free constructor-to-colonizer conversion).
Just wondering if you stopped because you think it is an exploit or because the mod you mentioned makes it undesirable?
I stopped because I felt it was making my games too easy (I was getting so far ahead using it, I felt bad). I've played only this one game so far with the AIKickStart75 mod. I'm really behind the AI now, so this game has become quite challenging (read: really fun! ).
Actually, I know what's causing this, because it happens to players as well. The AI uses Rally Points for everything, so what's happening is that multiple ships are landing on the Rally Point at the same time, and if there was already a fleet of 2 ships there, the game just shrugs and says "there's a fleet here already, what do?" and leaves all the other ships as singles. The game needs to recognize that it needs to keep adding ships to any fleets that show up on the rally point until they are full, and that should fix the problem. It's probably happening because too many ships are landing on the space at once, and since the game can't add ALL of them to a single fleet, it doesn't add any of them, since it can't decide which ships to exclude.
I don't think that the ships are landing on that rallypoint "at once" - but rather one by one, because the calculations for each ship is done individually as well. What we need is a RallyPoint that creates a fleet when 2 or more ships stand on that point, and everytime a new ship enter this point the game has to check if there's still space enough available in those fleet(s) and if true, add new incoming ships to them. If untrue, create a new fleet and put the ships there. It should further order the fleets internally to optimize speed, range, and also type in either military offensive or transports.
Here is a really good video series on Suicidal difficulty from Macsen where the AI is being utterly dominated. It's well worth watching to get ideas to help the AI. Picked up a few tips for my own games as well!
https://www.youtube.com/playlist?list=PLHMuUHsmNFlQEvNuuDLZ2uBLJ0HNEY2si
Frogboy could you please answer this for Icemania and my information
Originally posted by I'm Spartacus: Ice - you remember Brad Wardell said he had seen one other game with AI as good as GC2 Dark Avatar - but he did not name the game as I assume it was a competitors - read some speculation it was civ 4 - suppose we ask Brad what it was since he's got more expertise than any of us?
This is from the steam forums Brad and makes sense to me as well - comment from you?
At first I thought you were saying "Why don't people just make AI better instead of giving it bonuses?" and wrote 5 paragraphs about it not being that simple. Then I realized you meant thereason more people don't play at high levels is precisely because we're not given better AI to play against. Which is a great point, and I definitely think it should be brought up with Brad Wardell.
YES. I second this.
it delays trading and forces you to choose between researching something that can be used to immediate effect (speed boost) vs. something that might be useful (especially in the case that both parties need to research a tech translator)
In the case that both parties would need it...some factions might not care to trade tech, with fewer factions trading techs there's less benefit for a human exploiting it.
Also factions should be more guarded with techs.
There are many great features available to you once you register, including:
Sign in or Create Account
