A Better AI.

[Elandal]

I don't really see the need for two levels - why not using the best available AI for every level and giving them penalties on the very low levels - I think it is more fun to play against an intelligent AI , which is hampered by penalties than to play against a dumb AI which is blessed with boni.

The optimal situation would be one where no bonus nor penalty was applied to either human or computer player at any difficulty level, just the intelligence of the computer AI rising with the difficulty levels. At least that's how I see it.

Maybe there is a point in playing against a smart mean deity-level AI (that can play Deity without any bonuses) that is crippled by heavy penalties. But in most cases I'd rather opt for no bonus nor penalty to either human or computer, with the AI intelligence level matching the difficulty level.

 

[peacenik]

This stuff is outside the scope of the project... but hey, it's mine so I can do what I want with it? Right? Right? And people have been complaining about start point generation for ages. Since my changes did mess with the start points and since it wasn't great before anyway I decided I could make things right.

Your project, your rules, and we all benefit---a little hard to argue with that. There have been some great ideas in this thread that would not be 'save game' compatible but otherwise doable. Perhaps, say after release 1.0, could you look at some ideas not 'save game compatible' that would be worth pursuing???? I think most would sacrifice save game compatibility in the interest of a better computer opponent. Whatever you decide on this topic, thanks for the massive improvements so far.

 

[Veritass]

First of all, major kudos to Blake and the rest of the AI team for all of their excellent work.

Correct me if I am wrong, but I have been assuming that Blake's code will eventually become part of the normal game. Of course, I think he should be compensated for it, but I for one would like to see it incorporated. It is closer and closer to what the AI should have been from the start.

If/when it gets incorporated, Firaxis will have to re-balance the bonuses to be compatible with it. The tech bonuses and the upgrade bonuses coupled with the new AI make the power curve too steep as you go up difficulty levels.

I have always thought that some number of the strategies should be enabled only on higher levels. For example, the "dagger" strategy should not happen on the two easiest levels. If a new player starting out gets "daggered", they won't be prepared for it; they will lose the game and declare it "not fun".

For the most part, I don't like disabling the new code, as it is essentially hobbling the great work that Blake has done. However, the AI has the advantage that it can micro-manage every city every turn, much more efficiently than a human can. Maybe a small number of these can be made similar to the evolution a human player does: at the lower levels, the human does not know what a GP farm is, so maybe the AI does not identify cities as GP farms until a higher level. Maybe other areas can be identified as well.

 

[Veritass]

My humble suggestion and offer to the AI team. I have always thought that it would be to Firaxis' advantage, or the Better AI team's, to have more automated testing of the code across a wide range of variables. By this, I mean the ability to generate a starting game, then run that game all the way through, the start the same game again with a small parameter change, and run it all the way through again and see how that affected the game.

Of course, this would preclude having a human player affecting the game. The changes would be things like: (1) what if you left five of the civs the same but changed out one civ? (2) what if you had slightly less food resources in the capital? (3) what if the Better AI team changed one parameter? (4) how would the various combinations of civs come out in all combinations of duels? (5) what if praetorians were nerfed to strength five? (6) how does the same strategy work on a different map type?

To this end, I would volunteer my computer, and I'm sure other people would as well. I envision a system where the Better AI team generates games that they want to see run. I sign up and download one such game and start it, and it runs to completion (based on time or victory). Then I send the game save from the victory back to the AI team for analysis. I'm sure I could run a couple of games overnight, and another couple of games through the day, and if enough people volunteered, you could have a hundred different examples running every day.

I know the AI team is small. Maybe they cannot analyze each game in detail, and would concentrate on those games where something unusual jumped out at them. But if all they had to do was load up the save and run the playback, maybe it would give them some useful information.

 

[V. Soma]

you mean auto AI games running?

 

[Arlborn]

My humble suggestion and offer to the AI team. I have always thought that it would be to Firaxis' advantage, or the Better AI team's, to have more automated testing of the code across a wide range of variables. By this, I mean the ability to generate a starting game, then run that game all the way through, the start the same game again with a small parameter change, and run it all the way through again and see how that affected the game.

Of course, this would preclude having a human player affecting the game. The changes would be things like: (1) what if you left five of the civs the same but changed out one civ? (2) what if you had slightly less food resources in the capital? (3) what if the Better AI team changed one parameter? (4) how would the various combinations of civs come out in all combinations of duels? (5) what if praetorians were nerfed to strength five? (6) how does the same strategy work on a different map type?

To this end, I would volunteer my computer, and I'm sure other people would as well. I envision a system where the Better AI team generates games that they want to see run. I sign up and download one such game and start it, and it runs to completion (based on time or victory). Then I send the game save from the victory back to the AI team for analysis. I'm sure I could run a couple of games overnight, and another couple of games through the day, and if enough people volunteered, you could have a hundred different examples running every day.

I know the AI team is small. Maybe they cannot analyze each game in detail, and would concentrate on those games where something unusual jumped out at them. But if all they had to do was load up the save and run the playback, maybe it would give them some useful information.

You mean this: http://forums.civfanatics.com/showthread.php?t=184128


It is abandoned for the time being.

Blake: The changes you made in the starting location are already in the last version disponible? They seems really good.


Difficulty level: What about make 2 of each difficulty level?? like, Settler without Better AI and Settler with Better AI? Sorry but I like every single feature of the Better AI mod and would be just Really stupid to play without some of them...I prefer getting bonuses or the AI getting penalties than play with a less smart AI..

 

[Veritass]

you mean auto AI games running?

Yes, I envision something like the SETI data analysis, where people can volunteer to have their spare CPU time used to analyze specific quantities of data. In this case, one unit of data is a game: the data is downloaded (as a game save), the data is processed (the game is run), and the processed data is uploaded (the final game save).

 

[Veritass]

You mean this: http://forums.civfanatics.com/showthread.php?t=184128


It is abandoned for the time being.

No, I think this was a more ambitious attempt at automating the tweaking of the AI.

I am just suggesting that if the Better AI team thought it would be helpful to run a number of simulations -- of their choosing -- that some of us could volunteer to run through them. If they don't think this would be helpful, or they already have enough compute power to handle this, or they think it is more trouble than it is worth, that is their decision.

 

[kettyo]

I've noticed that coastal starts are much more rare than before.
I had to apply for new starting loc for 10 times or so to get a coastal start for my sea-obsessed civ

 

[kettyo]

Veritass,

You say you're from "Southern California".

There is really an identity like this?

I like 'Eagles' music very much and they refer to it as their home quite often in songs.

Just liked to know if it's not offend you, Thank You.

 

[DrJambo]

Iustus, Blake,

There's something funny going on with AI stack attacking in the most recent build. I've attached a save for you to see for yourself. When you load up this game, focus on my beleaguered city of Madras which has a large Russian stack based outside attacking it each turn. Basically the problem I'm encountering is the AI will only attack with the siege weapons in the stack (i.e. trebuchets or cats). None of the other units (knights, crossbowmen, etc) ever attack the city. Essentially the outcome is the AI takes Madras but it could take it turns before it eventually does, simply by using all of the stack's units.

Can this be addressed?

Also, I notice that the AI now actively uses their Generals (i.e. attaching them to a unit). Is this something you've done as part of BetterAI? I don't remember the AI ever doing this before... Does the AI always do this now?

Cheers

I know my save's been d/l once, so can I hope this is being looked into for the next build? The AI not stack-attacking is bad.

Cheers

 

[Roland Johansen]

Against my own judgment I have decided to rebalance the Starting point code so it no longer matches the 2.08 code but should give more reasonable starts, with the "seafood madness" starts being much rarer.

I did this by making land valued more for start points (less likely to get a capital on a silly little spit of land) and a bunch of other changes... to help keep the starts out of the tundra and stuff.

I also adjusted the start point normalizing code to make it fairer and saner.

I think I've nailed the "civs start really close together" thing too.

This stuff is outside the scope of the project... but hey, it's mine so I can do what I want with it? Right? Right? And people have been complaining about start point generation for ages. Since my changes did mess with the start points and since it wasn't great before anyway I decided I could make things right.

I think that's a good change. Most historically great civilizations started near a river (delta) and with an abundance of food in the neighbourhood. These are also the best starting positions in the game. A seafood starting position is usually less favourable as you need to build many workboats to get the same good terrain as you can with 1 worker. Also, you can't chop forests with a workboat and usually don't have many forests in a seafood start.

You of course decide what is within the scope of the project. Now the mod can be used within other mods that change the gameplay, but if you would at some point decide to go further than that and add your own gameplay modifications, then that's your choise. Allthough other modders would probably like a pure AI-modification version of your mod to be available next to this hypothetical more extensive mod.

In the last game that I've been playing with the 02-12-2006 version, I started at a seafood position. Did this version already have these starting position modifications? I've posted some savegames of that game here (in post 681).

 

[karlis]

1. Thank you Blake and others for great work!

2. Now an issue I with AI. Playing with the latest version of your mod (version of 2006-12-02). Isabella decided to attack me. Everything was well except that she landed one of the stacks on a single spot of land behind a rock. Obviously she did not realise that her riflemen will not be able to attack my city ever. I guess AI should check if the troops will reach any city after landing.

I attach (hopefully successfuly - I am new to this forum) a screenshot and savefile where Isabella did the landing behind a rock.

View attachment 143761

View attachment 143762

 

[JMaltman]

1. Thank you Blake and others for great work!

2. Now an issue I with AI. Playing with the latest version of your mod (version of 2006-12-02). Isabella decided to attack me. Everything was well except that she landed one of the stacks on a single spot of land behind a rock. Obviously she did not realise that her riflemen will not be able to attack my city ever. I guess AI should check if the troops will reach any city after landing.

I attach (hopefully successfuly - I am new to this forum) a screenshot and savefile where Isabella did the landing behind a rock.

View attachment 143761

View attachment 143762

She might be attaching a lot of value to preventing you from mining that uranium?

 

[Newcomer24]

Regarding starting points.
I do not like 2.08 starting points.. it's boring when 9/10 times you start on seafood position.
This said.. you should not make this type of starting point overly rare.. it's nice to start on the sea from time to time.. I would suggest to make it something like 1/8 or 1/5..

Roland Johansen.. I do not agree that a seafood position is less favourable.. yes workboats do not chop forest and you need more than one workboat... but first of all.. your city can grow while building a workboat (unlike while building a worker)... also.. seafood resources bring you a lot of gold.

Another thing about starting points.. please do not make them too balanced.. starting on extremely good sposts or poor ones adds variation .. if you always start in a balanced spot you will soon get bored.

Mhm.. considering this is outside the main goal of the mod... maybe it could be made optional?

 

[Quagga]

I do not agree that a seafood position is less favourable.. yes workboats do not chop forest and you need more than one workboat... but first of all.. your city can grow while building a workboat (unlike while building a worker)... also.. seafood resources bring you a lot of gold.

The seafood madness starting points are great for growth, whipping and gold. On the negative side, they create a vulnerability that requires one to build some form of coast guard. It is no good when your size 20 city loses two crab and a fish to pillaging. That said, I too think they are interesting, but I do not want to start in that position every time.

 

[Tiberias]

The AI not stack-attacking is bad.

I'm seeing this as well, and it's not good. The AI won't attack without siege, and when it does attack, it will frequently use siege only; it's saved me from certain defeat, but I don't think it's supposed to be that way .

You say you're from "Southern California".

There is really an identity like this?

This is off-topic, but while California is one state, it's often referred to by region, as in Southern California (San Diego, Los Angeles), Bay Area (San Francisco, San Jose), Central California (Sacramento), and Northern California (no major metropolitan centers). People usually refer to the metro area they're from, though--e.g. L.A., Riverside, San Jose, etc. But let me give you one tip--anyone who says they're from the "the O.C." definitely isn't.

 

[Veritass]

Veritass,
You say you're from "Southern California".
There is really an identity like this?

I like 'Eagles' music very much and they refer to it as their home quite often in songs.

Just liked to know if it's not offend you, Thank You.

No offense taken. I have also commented on this in the thread on Californians. Californians tend to break themselves into Northern and Southern, and occasionally Central as well. I am from Orange County ("The O.C."), but there is an Orange County in Florida as well. Californians, especially Southern Californians, are stereotypically hedonistic, narcissistic, and shallow. As a small example, my wife is from Illinois (Midwest United States). We joke about how she becomes "more Californian" year after year. Here is a list of things that we do regularly as Californians, that she would never do as an Illinoisan.

• We drive cars made in Asia and not the U.S.
• She puts blond highlights in her hair.
• We eat sushi.
• She has had laser skin treatments for her skin.
• We jog and play racquetball.
• We get regular chiropractic treatment.
• We belong to a New Age church (Religious Science).

At least we're still conservatives, but there aren't a lot of those in California. Any more questions should be directed to the California thread.

Edit: Tiberius, I don't identify myself as from "The O.C." That is for TV people.

 

[Pfeffersack]

My vote is on less seafood-starts.And if such a start is created, then please not one with 6+ seafood (I believe the personal record in my games were 8 or 9 I spotted for an AI city on a small peninsula)...two or three is enough by far.

 

[Zophos]

To a respond to a post quite a few posts back about razing:

Russians "razed" Moscow before Napolean tried to get there. But Moscow still existed as a city even after this "razing" - its not like they completely wiped Moscow off the map.

As far as fires burning down a city (like Chicago in the 1800s I think it was) - generally these resulted in the cities being modernized as the city structures that were burnt down were replaced with brand new modern buildings.

For the 2 cases here, what happened was not like the Civ version of razing where the city and its inhabitants just disappear. Being able to just disappear a city makes building an empire completely different in Civ as compared to rl - it makes conquering other civilizations MUCH easier than it would be otherwise.

However, cities HAVE been razed. I don't mean in WWII - Rome did it to Carthage, for instance, at the end of the Punic Wars. Burned it down, killed the inhabitants or sold them into slavery, and sowed salt in the earth. Go try and find Carthage now.

And now back to your regularly scheduled thread about AI mods. [/deadhorse]