Why AI must evolve after release

To keep the game fresh

So you've bought the game and you really like it but then time passes and it starts to get old. It's not that the game is boring or bad but that you have figured out how to "beat it".  You crank up the difficulty level but then it's ridiculously hard to the point you know you have no chance. 

If the game has multiplayer, you play people on-line but as you get further away from the original release, the harder it gets to find people and in particular the harder it gets to find normal people -- people who play the game as it was designed to be played as opposed to exploiting every minor game mechanic flaw to their advantage which tends to suck the fun out. 

So ends the typical strategy game.  It's usually a question of when, not if, the game reaches the end of its playable lifetime. 

When we made Galactic Civilizations II, we budgeted some money for after-release work.  We did go a little overboard with that with 1.1 and 1.2 in that the guys just never came down out of "crunch mode" and continued to put in crazy amounts of hours to deliver immense numbers of changes.  Now things have returned more to a normal pace and so we look forward to 1.3.

Version 1.3 of GalCiv II will largely involve computer AI work by me.  It's not just about making it tougher as much as it is about making it play differently and more intelligently. It will also likely involve creating additional difficulty levels and fixing the $@#$@ thing that causes some AI players to be set to stupid while others are at hard in order to average things out (i.e. someone setting the overall difficulty level to "Painful" only to find that 3 of the AIs are set to being really hard and the other 5 are set to being brain dead -- your best solution for now is to set the players you want to be intelligent or better if you're looking for a challenge and control the players rather than having the overall difficulty system try to guess what kind of gmae you want).

I have a lot of ideas on how to make the game play better, particularly with regard to the CPU option for the AI being able to use better algorithms.  The expansion pack, Dark Avatar, will have a LOT LOT more of this kind of thing.  But my job is to make sure that the AI continues to provide an interestig experience for players during the summer months.   

55,911 views 25 replies
Reply #1 Top
Keep up the good work; and may your coffe be with you!
Reply #2 Top
Brad,

You've got it exactly right. While I'm hugely impressed with the scope of the game, I've pretty much stopped playing it as I haven't been able to find a setting that makes it challenging without it being abusive (eg. the annoucement that an AI opponent has built a Mfg Capital before I even make my first move).

As my professional/personal interest these days centers on the development of AI for just these kinds of TBS games, I've been quite interested in some of your journal discussions over the last few months, especially concerning scripting vs lack of such. Wish there was a way to contribute to development of the AI, like others can do with modding...
Reply #3 Top
"(eg. the annoucement that an AI opponent has built a Mfg Capital before I even make my first move)."

I like that because to me it signals that different AIs are following different strategies in different games. Also because it seems like a big waste of money to me.

I've spent hundreds of hours on games like Civ2 and GCI getting better and better and exploring different settings, without any AI updates at all. I can't imagine how long they would hold my interest if the AI changed from game to game!

You know, I put this game away for a while till I got a new video card. Maybe it's a bad idea to wait -- I'm depriving myself of the experience of being familiar with 1.2 and then having to adjust to the expansion.
Reply #4 Top
Keep up the good work!

''Additional difficulty levels'' = yes! Do I read your mind that you're also planning to fill the hole of AI bonuses between Genius and Incredible? Anyhow, thanks for the heads up.
Reply #5 Top
Nice. I was wondering where you guys disappeared to. Is there a way to make/change the minor ai? a nerfed version of the main AI's.. so the minors could colonize as well, and we could make mods with our own personal custom minors that actually expand and try to do something consequential.
Reply #6 Top
It's funny, I'm running firefox 1.5, and when you cursed and ended with a @ symbol, it interprets it to be an e-mail address.

I am personally very happy you went over budget on the after market updates. SInce my play computer isn't hooked up right now (summer vacation from school) it's going to be like playing a brand new game when I get back.

I have a question though, is there a seperate team working on Dark Avator right now? And if so, are they just updateing thier AI code as you adjust Dread Lords, or has work not really begun yet?

Keep up the good work, I look forward to investing in more of your products!
Reply #7 Top
Nice to hear about the ai changes!

Is there any possibility you can look into the minor's AI as well? most importantly, thier seemingly "tied ship range" with a corresponding Major race? Re-enabling (through scenerio option, currently does nothing) thier option to colonize would also be awesome imo, more opponants to deal with, especially of the wild card variety, are always a good thing.

Gennerally i run gigantic galaxy games, with common stars/planets - haveing that galaxy filled with more then the standard 9 races who are active would create a much more vibrant, living galaxy.
Reply #8 Top
It will also likely involve creating additional difficulty levels


Hooray! The number 1 must have improvement for me...please please please change that from "likely" to "definitely".

Intensive algorithms has done a reasonable job of getting the AI closer to me on Genius, but it's still not that much of a challenge. Meanwhile, Incredible continues to absolutely kick my arse.
Reply #9 Top
Insert plea:

Please fix the load/save game so that the game does not add extra movement points and halt production and research.
Reply #10 Top
Level between Masohistic and Crippling is badly needed.

Too big difference.
Reply #11 Top
I agree p22.
Reply #12 Top
I think GC3 should have some kind of self-learning AI system, preferably with a central exchange server where the game could send it's newly learned succesful strategies to, and from which it would download strategies discovered by other copies of the game.
Reply #13 Top
Insert plea:

Please fix the load/save game so that the game does not add extra movement points and halt production and research.


Please fix this.

There have been many posts about this and it is a hassle for those that like to play difficult games in larger maps as it essentially corrupts the reloaded game by adding the extra movement phase.

When we have to leave the game and reload it over the course of a long game it is irritating that there are several non production spikes.

I know that the AI suffers also, maybe even more so, but shouldn't a saved game reload not corrupt the pace of events?

I find that this also leaves the AI at a disadvantage many times when it is trying to defend itself, fixing it would aid the AI in most of my games.

This is the only "bug" that I really see outstanding.

thanks for the consideration.
Reply #14 Top
I second the load/save issue.
The most annoying bug in GalCiv2 (proabably because it always appears when loading saves).
Reply #15 Top
The darnest thing with 1.2 is currently the AI not following Rally Points bug. But I got some reassurance from a SD staff member that this is currently under investigation.

Bring on the better AI with 1.3 Iam sure we can live with it until Dark Avatar
Reply #16 Top
AHH, now that explains alot of what I have been thinking lately. I bet that would fix the races that build their planets poorly also.

Thanks again to you all. Wish I had a more leisurely schedule but this week is dedicated to "Cedars Point" as our 8 year old calls this park that we are at for our summer adventure.

W/R
Suralle Straykat
Kat Lord @ Large
Reply #17 Top
I encourage you to also consider helping with some of the more mundane parts of the game. When I say mundane, I mean things you need to do where there is no real thought or strategy involved, just extra work. Not in any particular order:

1 - Make the explore setting work better/more efficiently
2 - Take some of the manual work out of updating starbases
3 - Simple way to colonize, like right click and say colonize and the next colony ship built goes there, an ordered queue might be needed with this.
4 - Have the initial settings for things like Taxation, Spending, Spending Distribution be remembered or have a place to change the initial settings
5 - Not part of the mundane issue but allow building older buildings, not just the latest researched version.
Reply #18 Top
More AI tweaks and the CPU option - I like what I read here.
Reply #19 Top
How much are Dual Core Processor's being used currently by GalCiv2? The market of people that have them is growing and I would like to see an option we can check that will take advantage of it.
Reply #20 Top
I'd like to see more balanced options for "alignment defining" events. It seems like the "evil" otions are the only ones that come with benefits, and the "good" options almost always come with penalties. All ethical options should come with balanced pros and cons.

For example: Higher production if you press the indiginent life into slavery but a penalty to loyalty. Lower population limit if you try to coexist with the natives, but a bonus to morale.

The way it now, it seems the game system is trying to make us all into its evil minions.
Reply #21 Top
I think the alignment options are fine. Choosing good shouldn't have immeadiate benefits. Otherwise an evil civ would pick them...

The only problem is that the down stream consequences of evil are not signficant :

1) Fundamentalists are a joke, i can generally retake the planets in a week or two.
2) Limits to trade, trade is a joke. I routinely leave the UP. The only good thing about trade is improving relations.

Make trade better, and give the fundamentalists some ships. (or give them the ships in orbit)
Reply #22 Top
Love everything I'm reading about 1.3. Hope it is as good as the talk.
Reply #23 Top
Choosing good could provide a moral bonus. This helps you keep the moral at 100% for maximum population growth and which in the long run would make good civs economic powerhouses (short term pain for long term gain).

v1.3 will be addressing the load/save bug which many have complained about. Please also address the bug that occurs when you get a new planetary improvement tech (via research or tech trading) on the same turn an instance of previous version of the tech is completed on one or more of your planets. For example, you get the tech for building a "Factory" on the same turn your "Basic Factory" completes. You will see a "Factory" instead of a "Basic Factory" without any production.
Reply #24 Top
bump.
Reply #25 Top
Or you could just make it multiplayer. I really appreciate the effort you put into your game, and I think the marketing and software decisions you've made thus far have been insightful to say the least. All but the decision to not include multiplayer that is.

The challenge of a realistic AI is only part of the equation. What about the simple fact that I like gaming with my friends instead of recanting tales of my game to them over coffee? I like the personal feel of having adventures with my friends. Cmon man, you know you want to make it multi. If you do a turn based fantasy strategy and dont make it multi. .. I will die. Waiting so long for Master of Magic remake. So long.. I know you know where I'm coming from. Do the right thing.