A Better AI.

[kniteowl]

LOL I find it funny that your quoting me from the other Website's forum, BTW great AI improvements I Find myself getting beat to the Oracle by the AI at 950BC on Prince Level...

I'm not going for the Oracle unless I'm beelining to Priesthood from, now on. Those Auto-workers really do chop rush wonders in your new mod...

Improvements... well I can think of a few military ones but I believe your delaying AI military improvements in the distance future.

Economic Wise - I was wondering if you could get the AI to beeline in the tech tree. Like the the English bee-lining to early Rifling for their UU.

I'd personally like to see AI's beeline to certain techs that have advantage to them eg- UU, Wonder rushing, Free Great People.

So you'd expect the English to be the 1st to get Rifling, The Spanish 1st to Guilds, Russians 1st to Military Tradition Etc. I believe they already more likely to go to war in time period where thay have a UU.

Question 1 - What Techs do the AI normally choose if they win the Oracle or Liberlism Slingshots? The Most Expensive Techs? or the most Useful ones?

Question 2 - Do the AIs know how to effectively use National Wonders?

Eg- Build Wall Street in Shrine City, OXford in Science City, Ironworks Production City & West Point and Heoric Epic In Military Production City.

Personally I Havn't checked this, In my Next Game I'll Beeline to Communsim and Build Spys to Find out.

 

[Wlauzon]

One thing I have seen time after time is the AI's making tons of frigates etc, which quite often just sit in some city doing nothing. A couple of days ago just before I captured a city it showed about 2 melee units and 20+ frigates, caravels, etc etc.

Maybe the AI cannot tell the difference between units that can defend and those that can't?

 

[kniteowl]

Another Question on City Location.

As Far as I know the AI already knows how to settles Cities in proper locations with your modifications but what I want to know is.

How do AI's Rank or Measure their City Locations In regard to Specific Resources, Food surplus and Distance and likely hood of their Rivals Settling before they do?

Will The AI more Likely to settle their 1st City on a Military Resource, Eg- Copper & or Horses *after Research BW and AH* in order to Defend against Barbs IF they don't have one of those Resources in thier capital's Fat cross already, Includes 2nd culture Expansion or have an early Adequate UU (eg - Quechua, Skirmisher.) AND/OR possibly have the Protective Trait (I'm not sure how effective the protective trait is against the Barbs but the extra 1st strike isn't anything to under estimate, I've Beaten Axes on open ground with Drill1 and 2 Archers quite alot)

Will the AI more likely to Settle near High comerce and happy resources EARLIER to benefit from them. This includes the folowing - Gold, Elephants, Gems (if the gems arn't covered in jungle) Sliver and Furs (If their not on Tundra or Ice Tiles and near food surplus resouces/tiles so you can work them) High Commerce for Research Advantage, Happy Resources for Bigger cities advantage.

Will The AI know how to Settle in city locations in the Direction of their Closest rival in order to settle the Best locations before their Rivals. This means Settling their Furthest Cities 1st and closest Cities Last if it benefits them and not at the Cost of High Maintenace that will Cripple their economy Early and must also Take Creative rivals into consideration.

Do AIs, Generally Creative AI's know how to cut off a pieces of land/continent using cultural Pressure, closing their broaders in the short term in order to settle the cut off land in the near future preventing their Rivals from Settling. Therefore dominating more of the continent/land.

I Don't know if Creative AI's or AIs in general do this on purpose or by accident (eg- cutting off land) even if they do cut off land on purpose they still keep open-broaders and let their Rivals cross their Broaders and Settle on un-occupied land, where that Land should Actually be Theirs. Logistically it'll almost be impossible to defend that City anywayz if one of your cities ended up on the other side of your rival broaders.

A human Would Close Broaders and Settle their Cut off territory then open broaders later once they've settled that area.

The only AI I know who has actually done this effectively is Toku, But he's An Isolationist anywayz LOL

The normal human would take these elements into consideration and settle their cities, Not sure if it'd be too difficult for you to write a code on this if it hasn't already been done properly.

The AI Should take these elements into consideration, Which City to Settle 1st, Which City to settle 2nd etc. The most important City Locations should be settled 1st. I don't know how you'd rank these elements in city location order. I'll leave that up to you

 

[Smidlee]

Your assumptions are broken as well as the game then. Its not supposed to be the nature of the game, at least its not how they sell the game. If the game is like that on purpose then they should say that and not give us the "illusion" that any path we choose for win, we are able to win with some work. Even against the human players in MP. CIV is not sold as an agressive and military game only, so if you need to be agressive to win the highest levels, something is wrong.

If you gain nothing from war (being aggressive) then there wouldn't be much left to the civ game. I can find no way around this since civ has always been about having powerful cities;thus you can build them or conquer them. Civ4 did make war more costly than in any other civ game I've ever played.

Since the main advantage human players has over AI is war then it make sense that you are force to max out this advantage (being a lot more aggressive) in higher levels.

 

[Wodan]

Since the main advantage human players has over AI is war then it make sense that you are force to max out this advantage (being a lot more aggressive) in higher levels.

Sorry... I disagree.

The main advantage the human players have over the AI is intelligence.

Taking advantage of the AI weakness at war is an exploit. It may make sense with the current programming, but that doesn't mean we can't hope they'll fix it. Which will bring warmonger strategies down on part with other strategies, and bring balance to the Force.

Wodan

 

[Smidlee]

Sorry... I disagree.

The main advantage the human players have over the AI is intelligence.

That's why it's called artifical intelligence since computers don't have real intelligence... It's war which requires the most intelligence so this is where humans have the advantage. It easyer to script the AI building order (Space race is the easiest victory) than it's war strategy.

 

[Wodan]

It's war which requires the most intelligence...

May I ask why you think so?

It easyer to script the AI building order (Space race is the easiest victory) than it's war strategy.

Perhaps. The answer to which is easier, however, is irrelevant.

In fact, if it is indeed more difficult to program, then that underscores the urgency with which we should petition Firaxis for it. It's human nature to slack off and "take the easy route". We can demand and insist that Firaxis hold themselves to a high standard; that's our perogative as customers. And, I think their own pride in their craftsmanship as well as desire to increase future sales will help.

The only even partial relevance that difficulty has is that Firaxis will have limited programming resources devoted to an already sold game. However, they are the ones who will compile the list of potential enhancements, prioritize it, and select which ones they will implement, which ones are not a good idea, and which ones are a good idea but will have to wait for the next expansion or Civ5. Regardless, from our perspective, voicing our opinion here is unchanged, so that they make note of it for that list.

Wodan

 

[Arlborn]

Since the main advantage human players has over AI is war then it make sense that you are force to max out this advantage (being a lot more aggressive) in higher levels.

But that is exactely what I'm complaining about...Read again my post perhaps you get what I mean then.

 

[Blake]

Improvements... well I can think of a few military ones but I believe your delaying AI military improvements in the distance future.

Actually I'm working on it now and making good progress but it's pretty much all changing so what the AI does now doesn't matter much...

Economic Wise - I was wondering if you could get the AI to beeline in the tech tree. Like the the English bee-lining to early Rifling for their UU.

I'd personally like to see AI's beeline to certain techs that have advantage to them eg- UU, Wonder rushing, Free Great People.

They already beeline the freebies/wonders techs, with a random factor.

I thought about putting extra-extra emphasis on UU/UB techs but it might make them too predictable.... ah whatever I'll do it.

Heh I'm looking closely at the code for evaluting buildings... as far as I can tell it's near nonsense... no wonder the AI doesn't care about Drama or Compass, the buildings aren't valued at all... Humm... fixed anyway, all buildings are now valued, UB's are valued extra.

Another thing I thought of was making the AI put more weight on techs which would be good trade fodder (probably with a random factor involved) to help defeat the human tactic of researching good trade fodder...
Too evil?

Question 1 - What Techs do the AI normally choose if they win the Oracle or Liberlism Slingshots? The Most Expensive Techs? or the most Useful ones?

Most useful mostly with some weighting on cost. I've increased the weighting towards cost in this case.

Question 2 - Do the AIs know how to effectively use National Wonders?

They don't know that they can only build 2 per city.... so no... not really.

One thing I have seen time after time is the AI's making tons of frigates etc, which quite often just sit in some city doing nothing. A couple of days ago just before I captured a city it showed about 2 melee units and 20+ frigates, caravels, etc etc.

Maybe the AI cannot tell the difference between units that can defend and those that can't?

No... the navy stuff is just plain crap.
I'll think about a way to fix it... maybe get the ships to dump their cargo and sail out to sea if the city is threatened....
And get the AI's to train less ships.

 

[WuphonsReach]

Random is good... especially it lessens the chance that an AI will end up at a local maxima. (Where the short-term goals end up not being the optimal solution over the longer term.)

As a for instance, I hardly ever see AI civs that leave forests in their fat crosses for later use as lumbermills or for health benefits. So while the chop-rush is good, I think that some of the civs need to value forests over a chop.

I think tech-trading needs to be toned down in some fashion. One way would be to make the AI simply less likely to trade at all at the higher levels, even to other AIs. At lower levels, everyone trades for parity or for slightly over parity. At higher levels, AIs should be greedier (weighted by friendliness) where they want 2:1 or 3:1 return on their trade.

 

[Harrier]

Wrong post - deleted.

 

[Mergle]

Another thing I thought of was making the AI put more weight on techs which would be good trade fodder (probably with a random factor involved) to help defeat the human tactic of researching good trade fodder...
Too evil?

Isn't the danger that everything then gets weighted and so it all cancels out? Is the best approach to ensure that different AIs aim for different things, so that the human can't choose eg paper knowing that none of the AIs will have researched it. With variation, they'll not duplicate each other and inter-AI trading will be more effective. (I can hear the cries of anguish from those who think there's too much of it already. Can't see it myself - I generally let the AIs do most of my research for me by beelining the unprefered techs).

(PS thanks for the welcome Arlborn. I've lurked for ever but this project has forced me into the open!)

 

[Pudd'nhead]

Insulting (yes I think that is the case) Aelf of wanting a simple static puzzle is going to far, when you should know better when you have read his threads. Please be open to other opions and don't draw out of the line conclusions.

I am sorry if you feel it was insulting, it certainly was not my intent. All I was trying to do was postulate that maybe we each view the impact of these changes differently, simply because we each play the game for different motivational reasons. I don't see that I ever suggested one reason was "better" than another.

I haven't read every thread/post that aelf has ever made. I was just responding to his response to me, and only what he said in that response, which is recreated in its entirety here:

Because I want to play more balanced game? Because if I want to play an all out aggressive game I'd play some RTS? If people who don't like to have to war for the entire game can only win on Prince, that says a lot about what it takes to master this game.

For me, it's not about not wanting challenge. I still have two very difficult levels to move on to. It's about wanting the potential for a more balanced gameplay to stay.

What I said I care about is not overall challenge in still having game setups I cannot win at, but in CONSISTENT challenge from beginning to end of a given game, irrelevant of whether I win or lose. While some sees challenge remaining in "having difficulty levels to move onto," I don't -- because I see having higher difficulty levels as a way to handicap the game experience, and not as the thing itself to be conquered (I don't know how anyone can read aelf's post as not saying that). If you'll allow me an analogy, I play civ casually, like a play golf: each hole, each stroke, is where I find the fun of the game. While I have played in a league & have a handicap, I frankly don't care that much what it is, or if it gets better, as I do just meeting the challenges of each stroke; the motivation for me is mostly in each stroke and/or hole of itself. I'm not motivated to get my golf handicap to any specific number, though I know others do & I don't see anything wrong with that! Likewise in civ, If I happen to imporve in my game play, moving on to the next level is for me just a handicap adjustment to rebalance the experience of a single game, it is not the challenge, the raison d'etre of the game itself. I find most of the appeal of this game are the "mini-challenges" within any individual game -- ideally at all stages of the game -- and not in an accession through dificulty levels, across multiple games, "mastering" it over time.

For me the fun of golf is that that the back nine is just as challenging at the front. That's not true in vanilla civ, not by a long shot, the late game drags merciclessly, mostly due to AI stagnation. All I was saying is that if Blake's changes make that closer to reality, then I'm all for it, and I don't care much if it means my "handicap goes up by 10." And I think it his changes will do that, by definition, if the opponent strength at a given absolute difficulty level is more due to imprioved AI (which impacts the end game) than by artificial bonuses (which at some point are overcome by more efficient development). But I can understand if others who take more pride in what their "handicap" is (which I think there is NOTHING wrong with) might be more upset if the "groundskeeper" has done things to mess up the par for the course. But while I accept that POV, I do think that others that have it should accept that there are people like me who just want all the holes to be fun to play & don't really ultimately care what the par is.

Anyway yes lets move on as suggested, I just had not caught up with the replies til now & thought I'd better clarify since some took personal offense at what I wrote, even though no judgement at all was intended. FWIW, I have no doubt that aelf and voek & others who took offense are probably better payers than me and get much better scores than me -- just like most of the guys in my golf league -- I hope that admission helps some to resolve bad feelings.

 

[Pudd'nhead]

Taking advantage of the AI weakness at war is an exploit. It may make sense with the current programming, but that doesn't mean we can't hope they'll fix it. Which will bring warmonger strategies down on part with other strategies, and bring balance to the Force.

Wodan

That's a very good point. It seems like a lot of the concern about these changes has been "play balance," and that players will have to war all the time. But, if the AI can be tweaked to recognize the human player's force buildups and/or movements, and react by improving defence, and it gets better at exploiting weaknesses in the human's defence -- which it apparently already does, given the accounts here like the attacks on multiple fornts, sea invasion of washingtion, etc -- which will force the human player to have to keep more troops in reserve, then ... isn't this actually making the pure offensive military strategy LESS achivable, not more?

Once again, thanks Blake. I've just finished up WOTM2 and hope I find a little time this weekend to try it out myself!

 

[kettyo]

First of all i would like to ask what is the current (v208) war-declaration routine for the AI?
I think it could be much more improved.

 

[Jet]

Blake, the link in your first post still points to the 11-04 version.

 

[V. Soma]

Blake, the link in your first post still points to the 11-04 version.

try the other link down in the OP.
Still, that first link could be corrected

 

[Arlborn]

Blake, if you didnt do any more declarations about the Cultural victory discussion, even after my new points, and of a few others, were made, I suppose it means that you wont be working in anything like that anytime soon(or never)?
Real pity if its true, I hope that one day at least you could think of it, or even a other nice modder...

Anyway keep the good stuff coming.

 

[Blake]

Blake, if you didnt do any more declarations about the Cultural victory discussion, even after my new points, and of a few others, were made, I suppose it means that you wont be working in anything like that anytime soon(or never)?
Real pity if its true, I hope that one day at least you could think of it, or even a other nice modder...

Anyway keep the good stuff coming.

As I said earlier the Cultural Victory is not really a viable victory type.

It also requires far too much planning. The CIV AI still, fundamentally, does not know what it did last turn, it doesn't know what it's doing this turn and it has no plan for next turn. It wouldn't be too hard to enable an AI cultural victory with broken save game compatibility so that I can tell the AI what it is doing.

There are a few nasty hack methods I can use to get the AI to commit to a certain strategy and I may yet resort to such methods. For example I could have the AI commit to a cultural victory from the very start of the game, actually quite easily, without breaking save game compat. The method would be both horrific and hilarious to programmer types, I might indeed go implement it for precisely this reason.

 

[Blake]

Dont let it be known that ally ou have to do is use the puppy dog eyes and I melt and go implement the requested feature, but I will indeed be programming in an AI cultural victory.

With luck I'll have one of my AI's win a culture victory before I win a culture victory (yep, I've never won one, not outside of a SG anyway).