Is the AI actually cheating??

[theNiceOne]

Originally posted by wrgj
I must admit that I do not have the courage to wade through the seemingly thousands of posts, so I do not know that there are a lot claiming "cheating". I only asked the question.

...and you deserve credit for that. There are way too many "Another blatant AI cheat" posts that turn out to be unfounded accusations here and on CFC, so a few of us have become a bit sensitive. But you deserve credit for have a question in your thread name.

It seems now that we can answer "NO" to your cheat question. Partially by the hut explanation, and partly by Nor Me' excellent and simple test which shows that there is no such cheat.

Case closed.

[Catt]

Originally posted by bigfree1
It'll be interesting to see if wrgj can come up with the neccessary proof (saved game files) to show catt that he(I'm assuming he) was not pulling anyones leg around here.

While his boasts are awfully high, I tend to lean toward wrgj's account, if only because his defense of his posts' have been steady and lack attacks against those that are questioning him.

I am not by any stretch of imagination demanding proof from wrgj about his game -- I am wearing my disbelief quite clearly on my sleeve and trying to give him a bit a of a hard time, though

And I agree -- on wrgj's attitude to my pushing, prodding, and (good-natured) belligerence.

Originally posted by Nor Me
I was finally able to spare the 15 mins of civ playing time it takes to test this. I waited for an AI to get an advance, reloaded the autosave 1 turn before they got it, gave them the advance and watched as they took several turns to research a cheaper advance.

Thanks for testing and sharing, Nor Me.

Catt

[wrgj]

For those who are interested in my far fetched claims at having a great game in Deity, I will document, make an interesting little commentary and explanation, and I will come back and post it. I do not think that this is relevant to this tread, so I am thinking of putting it on a new thread.
[Note to Catt: I was only due to finish Sistine in 470BC with help of many workers, if this helps your calcs. Finished Pyramids in 1650BC, Great Lib in 1125 BC.]
-----

I went back last night (depriving myself of vital sleep - but now I have become intrigued) and looked for old saved games.

The Lincon affair was much earlier on in the game and in the very first go at this map when I wasn't trying for Great Library. I therefore deleted the file when tidying up files.

I hold that it happened as I related, as I investigates at the time, and related my findings to my Civ buddy (who was with me at the time).

In my investigations last night, I went back to the same Lincon (with no contact) and found the saved game where he completed Code of Laws, and repeated the same steps, and he did not then get Philosophy in the proceeding 2 turns. It is therefore not a basic mechanism of the game, but an unusual occurence, and could be explained by huts, considering the earlier stage of the game.

Case closed on this.

But...
The reason I came and posted was that there was a rush of techs discovered around 500BC.
I looked at this (and lost more sleep):

Gunpowder recently discovered in 650BC and held by: Rus, Ger, Zul, Per, Egy, Azt.

By 570BC, some of these civs discover:
Astro, Chem, Printing, Music.
(Surprising - 1 appeared per turn!)

No one has Astonomy in 630BC. (Presumed found in 610BC)

Chemistry discovered in 590BC. Held by both Russia and Greece (and others)

Banking found in 550BC. Held by Russia and Greece.

Metalurgy found in 510BC by Russia.

=====
Hence,

* must have been Greece to get Banking in 550BC;

* did not have astro in 630BC, breaking rule of 4 turn discoveries if banking got in usual way.

* Greece surrounded by 5 other civs, and little access to uncovered map area. (I have most of known world map.)


Conclusion
The above events, combined with the earlier occurence with the Amreicans, led me to believe that the AI could POSSIBLY switch research. The number of techs produced seemed unrealistic, and last night I proved it to be impossible, without:
(1) nearly all the civs researching different things at 650BC,
(2) Greece uncovers a hut and gets banking in 550BC.

As this seemed unlikely, I then posed the question here. It appears that it was indeed something that is not unreasonable to assume, even though it is unlikely to occur very often. Together with the evidence (also completed by 'Nor Me') that AI players cannot ALWAYS switch tech research, and that their play is usually optimised, I conclude that it is not possible, and that we are free of a game bug!

[wrgj]

PS: When I tested, it was in the same game conditions, in particular, in deity. So just in case Nor Me tested on a lower level, this is confirmed. (Although it would be freaky to see different game algorithms on different playing levels.)

[BRC]

Greece uncovers a hut and gets banking in 550BC.

Isn't it impossible to get a non-ancient age tech from a hut?? I thought that I read this somewhere.

[Hurricane]

Are you playing on version 1.29? In the earlier patches, the AI was able to trade techs between each other at any time (even during the human player's turn), which would speed up tech progress even more.

[Nor Me]

Originally posted by wrgj
* did not have astro in 630BC, breaking rule of 4 turn discoveries if banking got in usual way.

Banking requires only Education. So this is possible.

But it does require the 6 AI's to be researching at least 4 different paths similtaneously.
alexman once came up with a formula for which tech the AI would research next.
In a relatively recent game (AU203, wrote up here) , an AI appeared to beeline to Military Tradition. I checked that they had researched it before banking which is odd as all the techs to Banking should have been more desirable than Chemistry. So I came up with the 2 player debug mode scenario I adapted above and used it to test a few things.

1) The AI obeys the 4 turn minimum even on deity.

2) I found no random factor. If 2 techs allowed identical things it would always choose a particular one of them even if I swapped the things they allowed.

3) In fact I found no factors other than those alexman had already found. I checked wonder obsolescence (an early theory; they had the oracle).

So I suspect that the AIs are cooperating. i.e. the only way they cheat is by knowing what the others are researching so they can avoid it. This is obviously not going to show up in a 2 player scenario. None of the techs researched similtaneously in your game give anything that important. If the civs are roughly equal this would lead to them researching as efficiently as you saw.
In my game one of the AIs could research much faster than the others so once it, say, discovered Invention and another AI civ was researching, say, Theology, it would keep on researching the path to Military Tradition while other civs researched the other paths more slowly. This is a sensible move for the leading civ so they went on to win.

[wrgj]

Hurricane: Yep v1.29.

Education -> Banking
Ah, yes! What an oversight! Too involved in the answer to bother with the question!

It was pretty impressive, still, to see Gunpowder discovered in 670BC, then Russia discovering Chemistry and Metalurgy in the following 8 turns!

As to the order of research, I believe that each civ had a deterministic path by which they descend the tech tree. Even if it is effected by what everyone else has, and therefore the cost vs. their preferences for research, on average, they should go the same route: so when going through to either Military Tradition, Physics, or Democracy depending on whether they have a military or other bias, and whether they have democracy as a preferred govt, having 7 leading-edge AI civs effectively guarantees that some civs research each route. (And, on the side, a religious civ building a wonder will go for music.) These preferences, which are present in the mod (prefs for govt, characteristics of civs) are not taken into account by Alexman's calculation post, as it is only valid for the civ HE was using.

[Theseus]

Holy tech arguments, Sidman!

Catt, un-frigging-believable. Remind me never to face off you in your day job. ( )

wrgj, welcome aboard. You know your sh-t... but I think you now see that there are a bunch of people around here who are a) pretty passionate, b) know as much as just about anybody knows about how this damn game works, and c) are pretty in your face followed by pretty cool and welcoming.

Yeah, 'poly takes as much time (if not more) as playing... but I've found I really like this new form of gaming experience.

Good thread, at the end of the day.

[wrgj]

R.E. the dubious game of mine mentioned 17,000 times above - I assumed that I'd be able to use workers to chop forests towards wonder production. Now I realise that these shields aren't appearing in my wonder city as this is obviously being interpreted as rushing as wonder. Can someone confirm if you can/cannot clear forests for wonders?

This invalidates my claims that it appeared that I'd be able to get Copernicus'. I should still get Sistine or Bach. I have an exam on Wednesday, so I'll be late with my post with details on the game. (I was distracted by my strategic failings this weekend.)

[notyoueither]

You cannot rush a wonder in any way. Neither chops of forests nor disbanding units will work. You'll have to earn it. You don't have to earn it 'honestly' against the AI though...

You can prebuild it. You know that Sistine is coming up and you would die for it. Have one of your decent production cities start building a Palace. When you discover Theology, switch the target city to Sistine. That will give you a head start.

[vmxa1]

You can not rush wonders with anything other than a leader. You can prebuilt them. The only way to increase the speed of the build is to increase shiled production on the tiles that are being worked. IOW mine those tiles, work tiles that produce the most shields.

[Mad Bomber]

Originally posted by wrgj
R.E. the dubious game of mine mentioned 17,000 times above - I assumed that I'd be able to use workers to chop forests towards wonder production. Now I realise that these shields aren't appearing in my wonder city as this is obviously being interpreted as rushing as wonder. Can someone confirm if you can/cannot clear forests for wonders?

This invalidates my claims that it appeared that I'd be able to get Copernicus'. I should still get Sistine or Bach. I have an exam on Wednesday, so I'll be late with my post with details on the game. (I was distracted by my strategic failings this weekend.)

Huh...Is this your first game of CIV 3 or have you not noticed this in all of your previously brilliant games? At first I would have given you benefit of the doubt, assuming you were trying a OCC or 5CC in which building wonders at this rate would be difficult but now I can only say that this thread is complete BS.

BRC:

It is impossible to get Banking from a hut.

[Spec]

Is the AI actually cheating??

Of course, like in anyother game.


Spec.

[joncnunn]

Sir Ralphs explation is more likely to be correct.

One of the times I'll check carefully to see if I can make a tech deal is if I'm planning a poping a goody hut that turn.