Expansive abuse: Oracle Code of Laws fast GM

[Dirk1302]

This is new for me, are you sure?

 

[vranasm]

This is new for me, are you sure?

it's something that is running around here for like 6 months I would bet. First one to mention it was Grashopa I think...

someone who has good knowledge of sdk could look into it imo ;-)

(if we still talk about AI cheating on knowing when you start wonder and eventually not follow with the wonder)

 

[Deity player]

If this is true, would it be possible to beeline Oracle, stop building it 1T before completion and then wait until I can get something like Education from it?

 

[NihilZero]

I'm no code diver so I'm just following the rumour-machine on this one, but I have found it seems to work sometimes (said he in a highly convincing manner ).

If this is true, would it be possible to beeline Oracle, stop building it 1T before completion and then wait until I can get something like Education from it?

Not quite, they seem to eventually start building everything eventually, but they don't priotise a wonder much if they think their chances of getting it are bad (or so the theory goes).

 

[Dirk1302]

@Deity player, yes those are the sort of questions that need to be asked of course if Ai really looks into your wonder building like this, for instance:

What are the criteria it uses, just your current speed of building?

Does it look every turn and does it stop it's own building if you're very fast one turn, beginning again if you're slow?

Many abuses possible, if it is true that the AI cheats this way i think it would be a design error as programming it correctly seems to be difficult and rather unnecessary.

Maybe i'll take a look at the SDK. Problem with the SDK is that looking into it is time consuming if you don't know where to look. If you do as with the War mechanics it's relatively easy to follow.

I'm actually not convinced about this cheating, it could be but i've seen more wild stories running loose around here in the past.

 

[yatta77]

I'm no code diver so I'm just following the rumour-machine on this one, but I have found it seems to work sometimes (said he in a highly convincing manner ).

Uhm... If I want the 'mids (which is almost every game to be sincere ) I start building them early, sometimes I don't even discover bronze working before they are finished (I mean no chapping), and I believe, in years of playing now, I never missed them (usually I finish them in turn 60-85, emperor or immortal, earlier with stone/industrious, later if not).
I don't know if this could be a kind of confirm that this is true? Maybe?
Just a silly note from my experience.

By the way, look who is back!

 

[NihilZero]

By the way, look who is back!

I pop in ocasionally when I am in a wi-fi zone. I am still languishing in the dark ages with no internet at home and I'm not playing too much civ either as I'm in the Costa del Sol, the sun is blazing down and the beaches are jam-packed with topless women. Games have dropped a little in my priorities.

On topic, it's a shame we have nothing more than anecdotal evidence here, but I have heard it from a few people so in the absence of hard facts it's better than nothing, maybe...

 

[yatta77]

I pop in ocasionally when I am in a wi-fi zone. I am still languishing in the dark ages with no internet at home and I'm not playing too much civ either as I'm in the Costa del Sol, the sun is blazing down and the beaches are jam-packed with topless women. Games have dropped a little in my priorities.

I you prefer just that to Civ IV...

Spoiler :

well, I won't blame you

 

[Yamps]

@wonder cheating

Kossin talked about it somewhere in the last Plastic Ducks SGOTM thread iirc. If you start a wonder and the AI doesn't have the needed tech, the AI (may) delay the tech...If they already have the needed tech, it wont work. More info on this would sure be helpful.

EDIT:

BornInCantaloup provided more details:

1) Known : Some people read the game's code and spread such information. Modders can even alter the code.
2) The AIs register the turn a wonder is started only - they don't follow the build's progression. And they act accordingly. "Accordingly" you should understand loosely AIs happen to get some failure gold. I guess that they can target a wonder before getting the needed tech. If the player starts building a wonder that they are targetting they will start building it anyways. If they didn't target it, they will delay it.
(EDIT for clarity :
Present case : If no AI is targetting the Oracle at T90, then they won't go for Priesthood to build it. However, if they get to Priesthood because they are targetting Monarchy and have the opportunity to start on the Oracle, they may do so.)

Mostly guesses from my part in 2)...
Kossin is very knowledgeable on the subject.


ps : 0 hammers is the same as not started : hammers are invested eot. The minimum is 1 (city tile). Therefore, placing the wonder far away in the prod queue does not count as starting the wonder.

 

[FlyingSwan]

@vicawoo.Im liking the idea of culturally flipping Barb citys.Peter of Russia seems like an ideal choice tor your stratagy-PHI/EXP.

I had a phase a few months back of oracling music,the plan with that was to see how quickly I could flip an AI capital.I,d build my 2nd oracle city as close to the AI cap as possible and settle the GA from music.The hope was when I got CoL the relgion would be founded in that city for even more culture.Now Im thinking getting early CoL and then running artists to bulb music would be better.

trying to bulb music on immortal would bebut I managed it on monarch.The idea of trying to flip an AI cap didnt work out that great anyway(it takes ages as I recall) but youve given me some inspiration to try it again to see if I can make it work.

 

[vicawoo]

What was a new idea to me is to let the city first grow to size 2 before running the artist, so you dont delay grow so much while waiting for the border-pop. Smart, never thought about that.
Would you say delaying the oracle to research the wheel would take to long (on Immortal or Emperor?) I found it helpful to keep the workers busy and to get traderoutes and faster movement early. When I tried it on your save, I got the oracle around 1360, had my road network finished by then and got a lot of commerce from the traderoutes. However, the overall expansion in your save was a lot faster then mine.

It would be great to tech the wheel before the Oracle, but that tends to lose you the Oracle a lot.
I find that a AH->priesthood build with stonehenge support will get the Oracle most of the time on immortal (I'd like to say at least 70-75%), at least with chinese starting techs. I'm surprised most people don't get any stonehenge fail gold, stonehenge is going to finish before oracle in most games.

The wheel->pottery gets you more commerce in the long run, but in terms of Oracle timing it's slower.

Could just do a series of no action tests for the Oracle to see when it finishes, with industrious civs and without them.

 

[kossin]

it's something that is running around here for like 6 months I would bet. First one to mention it was Grashopa I think...

someone who has good knowledge of sdk could look into it imo ;-)

(if we still talk about AI cheating on knowing when you start wonder and eventually not follow with the wonder)

The first mention was dingding, it's something that came from the Chinese Civ forum.

 

[Abegweit]

The first mention was dingding, it's something that came from the Chinese Civ forum.

IMO, it's an old wives' tale. Soren was very adamant that the AI would not cheat.

 

[Um the Muse]

IMO, it's an old wives' tale. Soren was very adamant that the AI would not cheat.

Yeah? Then how would you explain being able to get both Buddhism and Hinduism on levels higher than Prince? There's even an article on the War Academy on doing it, I believe. Iirc, you put one turn into one of them, switch to the other. Get it and switch back. I don't remember if that's exactly how it goes, but I know that I tried it and it worked on Monarch. Btw, it's still not worth going for a religion that early, as far as I could tell.

 

[Abegweit]

That's not dingding's claim, you know. It was that you need to put a turn into the wonder, not into the tech.

Sorry dude, but this sounds like another tale.

Of course we won't know until someone actually looks into the code.

 

[kossin]

IMO, it's an old wives' tale. Soren was very adamant that the AI would not cheat.

Ever read how every AI unit sees through the fog of war for as many tiles in every direction that they can move? That's cheating in my book

 

[Mylene]

@AZ mentions in a few videos that the AI can see you building a wonder, certainly more than a fairy tale
I'd trust his experience over guides, it's also mine that they know it from a few tests i ran via the worldbuilder.

I wouldn't want to see how a blind AI acts

 

[Dirk1302]

I'd like to know where he got that info. I have heard people claiming that a coin is rigged when it throws tails 5 times out of 5, could be but not necessarily. I'd certainly trust relevant samples from SDK code over AZ's experiences.

5 years ago most people including some very good players believed that barracks helped to stave off AI war declarations on higher levels. Certainly a more reasonable thought than the AI cheating with wonders in this way. Also here experience was mentioned to prove it as a fact but in the end it proved to be 95% untrue.

 

[Mylene]

Well Dirk, some time ago good players also said that distance and logistics is a minor factor in AI DoW calculations. I remember @AZ showed them it isn't, just via his playing experience.
And everyone who plays Deity a lot should have known it, you know if someone is close to you the chance is always higher for a backstab.
This just means some of the Deity players don't have the level of experience that @AZ has, you can see him try out so many things early in a game, which makes for different experiences than just playing through a Deity map with estabilished tactics

 

[Dirk1302]

I don't think there are many players that think distance is unimportant in war declarations. This is so obvious from experience that further evidence is not really needed. If a coin tosses 1000 tails in a row i would believe it was rigged.

I remember the discussion you're referring to, people including myself believed that distance was important because of the land target rule, if you join > 7 tiles with a neighbor the chance you're being picked increases dramatically. This was found some time ago already in SDK code by DanF. AZ posted a save where he was attacked by a neighbor while not sharing > 7 tiles. Diplomatically that shouldn't have happened which led me to investigate the situation in SDK and indeed distance can override diplo factors without this > 7 tiles rule, you learn something new every day . But i don't think i remember anyone claiming that distance was unimportant then.