A Better AI.

[Welshie]

Diadem - you don't sell off the extra resources then?

Admittedly, when that's not an option, you're right about the towns, but I find it isn't usually the case.

" If you improve calander resources 9/10 even when you have a very big empire you're just a bad player."

I wouldn't throw comments like that around. I could react badly. I've beaten Emp on big maps/plenty of AIs with dom and space race without building towns on resources. I like to think I am ok. I'll maintain a respectful tone to you and you should do likewise.

[Welshie]

Blake - sounds good.

[Backpack]

Originally posted by Blake
Okay would anyone complain about THIS behaivour when "leave old improvements" is on:

SNIP

(actually I'll give it a threshold... so any resource less valuable than a health/happy wont be connected).

I vote for the system proposed above. I would recommend that instead of "It won't be connected", the system instead go with "won't remove an existing improvement.".

As described originally, I think it would then never hook up non-happy/health resources even in the early game.

The DOWN side of this system, (If its the same system the AI gets) is that the AI will never know to override, and plantation that second dye it cottaged or farmed over when it can sell it to another player for 20gpt or similar income.

So us players will never be able to buy a Plantation or later resource fmor an AI.

Or am I making a mistake?

Thanks for all your work!

Kevin

[Blake]

First it's only for "leave old improvements" which the AI will never enable.
Second, when I say "Connected" I mean that the resource is improved with the objective of making it available to the trade network. Simple improving logic is different, if for instance a late game stone resource is pillaged, the autoworkers will still eventually stick a quarry on it, but only because the quarry is slightly better than a mine - in other words, it doesn't preclude the possibility of connecting the resource, just that the connection improvement will be treated like any other improvement.

[kiky]

Originally posted by Senethro
Also, is there a way to make a shortcut so that Civ4 starts in a particular mod insted of having to load the game twice?

Open CivilizationIV.ini in warlord or vanilla folder.
and edit the following line:

; Specify a Mod folder (Mods\Mesopotamia), '0' for none
Mod = Mods\yourmod

where 'yourmod' is the mod name.


or, you can just double-click the savefile.

[CerberusIV]

Originally posted by Blake
Okay would anyone complain about THIS behaivour when "leave old improvements" is on:
1) Autoworkers will never connect obsolete resources.
2) Autoworkers will only connect resources if there is no existing connected source of that resource.
3) Autoworkers will not perform non-reversable actions, meaning they will not pave over a hamlet (or better) or chop a forest on a bonus (ie uranium), under any circumstances.

1) No problem, although if you are ahead of the AI on tech you can still sell them resources that are obsolete for you.

2) Sounds OK, although if you have several of a resource appear they may not connect the one you want.

3) OK with this one.

Basically this would mean I'd have to watch my workers more closely. Maybe switch them from fully automated in the later game to just building transport network so they railroad everything. That's no big deal.

[player1]

Why it just doesn't do what it is supposed to do?
And that's leaving ALL improvements intact, and let player choose if he feels like to connect his resources manually.

[Blake]

City ruins are classified as improvements. Guess i have to make it preserve them, that's what it's supposed to do.

Sarcasm aside, the reason why, was Industrialization and Aluminum (for example). I build a windmill on some hill - the decision I made, was to windmill a boring hill. Then along comes Industrialization and now the hill is an Aluminum hill - I DID NOT make a decision to windmill an Aluminum hill - if an autoworker comes along and mines it, it is not actually reversing one of my decisions, it's making a new decision based on new information.

Altough I've never seen it in game - it would in theory be possible that a critical irrigation chain has to pass through a unique resource - with absolutely no alternative route. As such the code could result in the autoworkers constantly breaking that irrigation chain.

It's for reasons of irrigation chaining and other hopelessly contrived reasons for not wanting even a single instance of a non-obsolete resource that I'll revert it to city ruins only, even though I think the vast majority of players willing to automate their workers would be happy if the autoworkers ensure a supply of resources as long as they don't pave over towns or stuff.

[Blake]

It has to be said that my AI's are getting a mite unruly.



Carthage has been Razed by the Zulu Empire

Carthage has been Razed by the Zulu Empire

I don't think an AI has EVER succeeded in razing my capital before and yet in the first trial of my new "AI Strategies" build one of the little blighters manages it, chewed through a very solid garrison of 14 mostly anti-melee units and 60% cultural defenses in a single turn. In the very early AD's, using an AI strategy I call "AI Dagger". What'd I do to deserve that?
(That's it Shaka! In future games I'm giving you the Musical Palaces strategy)

Here's another little AI stack going rampagies:


That's in 500AD. It's a stack of 35 units.

There's a few kinks to work out (I'm learning a bunch of new ways to partially corrupt save games !) but I should have something safe to release soon... well the AI's might kill you but that should be the worst that would happen .

[Krill]

Look what you're doing to us lower level players Blake! Give us a rest! Or a Worker at the start, at least!

[Diadem]

Originally posted by Blake
Sarcasm aside, the reason why, was Industrialization and Aluminum (for example). I build a windmill on some hill - the decision I made, was to windmill a boring hill. Then along comes Industrialization and now the hill is an Aluminum hill - I DID NOT make a decision to windmill an Aluminum hill - if an autoworker comes along and mines it, it is not actually reversing one of my decisions, it's making a new decision based on new information.

Well, true. But it's still destroying an improvement you might want to keep.

Aluminium is a very good resource, and you'll want to mine it 99/100. But Uranium for example is pretty sucky, and you won't want to mine it often. In fact if uranium pops up under an already existing mine I might even consider windmilling that square

What is important to keep in mind is that you *know* when new resources appear. And when they do, every player is going to check out where they are. It takes perhaps 20 seconds extra to decide which ones to improve and send some workers to them. Everybody who is too lazy to even do that most likely won't have 'automated workers leave old improvements' on anyway.

On the other hand, the potentional damage of wrong automated worker actions is quite big. Especially since you run the risk of not noticing it for a long time. So we have here a situation where overriding the 'leave old improvements' gives you very little gain, and potentionally very big losses. So I see really no reason to do it

[Arrian]

It has to be said that my AI's are getting a mite unruly.

Holy crap.

Soon, I think it may be time to review the AI bonuses.

-Arrian

[rah]

I've already lost three Monarch games since the patch. Moved down a level and still having problems.

Thanks Blake for adding surprises to the game.

The AI now has a good middle game.

Looking forward to new surprises.

[Penjo]

Of course, you're coding up AI counters to each strategy you're teaching it, right?

Right? =)

[Blake]

Soon, I think it may be time to review the AI bonuses.

So when reviewing the difficulty levels should Noble still be the fair difficulty?

Um yeah, two things to add. Firstly, not all AI's will employ the Dagger strategy, Monty and Capac probably have like a 70% chance because they have an early UU, while Gandhi... (has just been fixed to recognize that his UU isn't a combat UU).. while Gandhi as a peacemonger will have a 0% chance. The typical early-UU-less AI might have a 20% chance of so. Aggressive AI increases the chances dramatically such that even Gandhi will have a 10% chance while Monty is pretty much a solid 100%.

Next thing, while the AI's do build big stacks, they aren't really bigger than the stacks I build. Good old Shaka turned up with a stack of 25 odd units, but I did have 14 units in my capital and another dozen marching to the site. I was killing Mansa Musa at the time and Shaka blindsided me. Even at Monarch difficulty, the AI was NOT out-researching nor out-uniting me, I just couldn't take on two AI's at once. As such, at least for me, I still consider Monarch to the be the fair difficulty - and - extremely entertaining. I believe, that it is possible to "win against the odds", it's not like old Deity which is kinda rigged against the human, the AI's at monarch now keep up and compete a lot better, but it's not really rigged against the human... it's just a FAIRER game, where you should expect to lose quite often rather than winning every time, and it's fair because the losing isn't based on the game being totally rigged from the start...

So... I, as a long time Monarch player, still enjoy Monarch difficulty - it's far more possible for me to lose, but I find that enjoyable anyway. It doesn't feel rigged, instead it feels like the other players have a few tricks up their sleeves. Usually I'm middle of the pack, which feels right. Being consistently top of the pack should only happen with a super start.

So I think what you want to ask yourself is this... Do you want a game where you win 100% of the time? Or do you want a fair game?

For now, I don't feel the need to adjust the difficulty levels, as I find the game continues to get more entertaining and engaging.

Of course, you're coding up AI counters to each strategy you're teaching it, right?

Right? =)

Not exactly. Sometimes in order to win the game you have to take risks. It can be too expensive to prepare for a massive stack of swords, instead it can be better to pray that they target someone else, buying time to get crossbows or whatever, since it's a lot cheaper to defeat them with crossbows.
The idea basically is, that AI's will get eliminated by other AI's. Some AI's will survive through friendship and such, and will focus on space or cultural wins. An AI successful in conquering may grow big enough for domination or diplomatic. In any case, I do want for the AI's to sometimes die for the cause of a domination win, since this is inately balanced by the space and culture wins. In games with human players, players get eliminated. The same with AI's. Certainly the warmongers will often exhaust themselves against longbows but I do want to see AI's dying.