A Rank Corruption Discovery and Exploit to negate rank corruption

Discussion in 'Civ3 Strategy Articles' started by Qitai, Aug 31, 2003.

1. Qitai.

Joined:
Jan 22, 2002
Messages:
1,177
Location:
SG.MY.TW.US
I think I just discovered the reason for all the corruption discrepancies and phenomenon that troubled me for a long time. Things like why did I often observed corruption reduction in my old core when I just made a palace jump and also the recent lower corruption I see in my distance 7 ring around the FP in GOTM22 which later increases for no apparent reason.

And it looks like this can be huge exploit, which can reduce corruption tremendously. With this exploit, it is possible to make most, if not all, of my cities rank 1! This effectively negates almost all corruption due to rank.

Before you read this, you should get yourself familiar with Alexs Do you understand corruption? At the minimum, you should understand the difference between corruption due to rank and corruption due to distance.

The discovery is actually very simple. The city rank as we all know really has ONLY one set, not two sets! All rank calculation in Civ3 depends on the number of cities with distance less than Fd from the palace only, regardless of whether the city is nearer to the FP or the Palace. Note that the gist of the above statement is that rank depends on the number of cities with distance less than Fd, and not the number of cities nearer to the palace for the city concern, nor the number of cities nearer to the FP.

Mathematically, the formula should be
(For the full corruption formula, please refer to Alexs thread  Do you understand corruption?)

Distance of a city (Fd) = min(distance from Palace, distance from FP)
Rank of a city (Fn) = 1 + Number of cities with distance FROM PALACE less than Fd (i.e. ignore the source of Fd, just count the number of cities with distance less than Fd from the palace, even if the city is nearer to the FP)

If the above does not sound like something new, then you might not have understand it correctly. Here is an example to illustrate what the above really means.

Let say you have cities with these distances from the palace and FP respectively

Palace  4, 4, 4, 5, 6, 7, 7, 7, 11, 11, 11, 11, 12 etc
FP  3, 3, 5, 5, 8, 8, 8, 8, 8, 9, 9, 10, 10 etc

Let say, I am trying to determine the corruption of the distance 8 cities from the FP. Previously, I had always thought that the rank of the distance 8 cities would be 5. But my discovery says it is NOT.

The rank of the distance 8 city would actually be 9! This is because there are eight cities with distance less than 8 from the capital. They are 4, 4, 4, 5, 6, 7, 7, 7. The number of cities nearer to the FP does not matter.

Similarly, the rank of the distance 10 cites from the FP is not 12. It will be considered as rank 9 again since the same cites - 4, 4, 4, 5, 6, 7, 7, 7  would be closer to the palace.

I am not sure if this example is good enough. If this is not clear enough, I would suggest trying the exploit below and you should be able to understand what the above means.

The Exploit
To exploit the above, basically, all you need to do is to put your palace as far away as possible without any of your own cities near it. This way, all your cities near the FP will be considered as Rank 1! Try this on a recent game you finish by abandoning all the cities near the palace and see how your corruptions near the FP reduces like magic(You may need to remove and put back the population to see the changes). So if the nearest city to the capital is distance 20. Then all cities less than 20 distance from the FP is considered to have a rank of 1. Imaging that!! Even those cities that are considerable distance from the FP now become very useful. I tested it on a recently finish game and I see my previously fully corrupted cities in the outer ring of my FP turning into low corruption power cities.

I have attached an example of this. You can check the city New Argos which is at a distance of 12 from the FP and with a rank in excess of 20 if I calculate it the old way. You can check with Alex's calculator that the corruption level is of a city with distance 12 and rank 1. Note that this sav is mod but none of the mod affects corruption, just to facilitate testing.

RCP impact  The above basically means you just need to do RCP on the palace. Cities around the FP gains the benefit automatically, provided the city distance is not more than the rings at the palace. RCP around the FP does not help at all.

Additional Comment 1  Even a simpler approach of ICS near FP and sparse city placement at palace will help greatly in this battle against corruption since the sparse placement near the palace will help the city corruption near FP.

Additional Comment 2  The above also resolved the question about how is the corruption calculated when its rank is nearer to Palace and distance is nearer to FP or the opposite.

Additional Comment 3  This exploit would make commercial trait useless since this effectively negates almost all corruption due to rank.

I guess GOTM should ban the above exploit this since this will be overpowering. I dont think there is any reasonable way you can ban the method in additional comment 1 though since it starts to gets into the gray area.

Tests are made on Vanilla Civ3 1.29. I suspect it should be the same for PTW. Credit must be made to Alex for having discover and fine-tune the corruption formula which serves as the basis for me to investigate this. Also, Daviddesjs RCP is what really re-kindle my interest to make this investigation.

(All information contain herein are the result of extensive testing made by the author. Although, diligent effort has been made to ensure that the results are true, there is always a possibility that insufficient test has been made. Everyone is welcome to either confirm or show results which contradicts the information shown here)

2. OffaBretwalda

Joined:
Feb 3, 2002
Messages:
1,255
Location:
Mercia
Very clever Qitai.

Did you unwittingly benefit from this in GOTM22 with your Palace jump? Maybe without that boost you would have taken until 420ad to conquer the world (only 900 years faster than me).

How can GOTM ban this; surely it should be addressed by a patch instead.

3. The Last ConformistIrresistibly Attractive

Joined:
Aug 25, 2001
Messages:
27,779
Location:
Yeah, I definitely think this ought to be patched away. Seems to make no sense at all.

4. Qitai.

Joined:
Jan 22, 2002
Messages:
1,177
Location:
SG.MY.TW.US
For a very short while, yes. If I had known this at that time, I might get 400AD victory

5. SirPlebShaken, not stirred.

Joined:
Jan 1, 2002
Messages:
1,415
Location:
Qitai, this is a huge discovery.

I am glad that you pursued the discrepancy which you noticed and were able to deduce the cause - nice debugging! Thank you!

I have loaded a saved game in PTW 1.27, abandoned all cities anywhere near my Palace, and can confirm that the effect you've discovered still exists in PTW 1.27.

I hope that Firaxis issues a patch which fixes this bug for all versions. (I think there can be no doubt it is a bug )

I agree that this should be a banned exploit. I don't think it can be entirely detected. In fact, now that I've seen this effect I can't imagine spending the time to do RCP around my FP. (I already wasn't wild about working out RCP though I'd started doing it to some degree.) I think my personal compromise will be to continue using RCP to some degree around my Palace, not use RCP at all around my FP, and aside from that forget that this loophole exists while playing.

I think in the past I've benefitted from this bug in a couple of games without knowing it, when I Palace jumped. The new Palace region was initially less developed than the old core and some towns would have had less corruption for a while. Probably wasn't a big deal since I of course aggressively grew the new Palace region. (Hindsight now shows that to have been counter-productive )

I think that relatively minor uses of this exploit will be indistinguishable from regular play without this knowledge and can't be detected. But larger abuses will stand out, e.g.:
1) A largely OCP style build around the Palace but an ICS build around FP.
2) Palace in a relatively isolated position.

Joined:
Jul 5, 2003
Messages:
107
Location:
Orange County, CA
Now, I'm really bad at math, so the corruption model in Civ3 makes no sense what-so-ever to me. However, this sounds like a really powerful exploit that is not so tough to pull off in the mid to late stages of the game (as stated in "Additional Comment 1).

Good work on the Civ3 science.

:grad:I'm going to do an ethnography on the online Civ3 community this semester for my graduate-level independant study class. If you have anything you'd like to contribute, let me know in an email: ridcully47@hotmail.com

Thanks

Oh, and... [nods subtly in the direction of his signature]

7. SirPlebShaken, not stirred.

Joined:
Jan 1, 2002
Messages:
1,415
Location:
It is huge. In some cases it could pulled off big time fairly early in the game. E.g. an archipelago map might make it possible to do an early Palace jump to a remote island and then build a very large number of corruption free cities in the original home area.

8. AggieDeity

Joined:
Jan 11, 2002
Messages:
6,278
Location:
Amsterdam, Netherlands
This explains a lot a of situations in which I found it weird that my old core stayed strong after a palace move. Great discovery!

9. anarresanarchist revolutionary

Joined:
Apr 22, 2002
Messages:
6,069
Location:
www.civ3duelzone.com

This messes everything up! I am sooooo angry! Argh!!!!!!!!!!

:saiyan:

10. lz14Prince

Joined:
Jan 26, 2003
Messages:
576
Location:
Auckland
This is a huge bug. Sounds more like a beast to me. Great discovery !

11. SirPlebShaken, not stirred.

Joined:
Jan 1, 2002
Messages:
1,415
Location:
Imagine what it can do to MP team play. Two players who are teamed up each build FP. Each also builds a useless but high food town and lets it just grow. When they've met each other and their FPs are done they give each other the useless towns. Then each does a Palace jump. Yikes.

12. carlosMMDeity

Joined:
May 14, 2003
Messages:
8,565
I second anarres.

13. SkyfishEmperor

Joined:
Sep 24, 2002
Messages:
1,212
Location:
-
Well actually the reason anarres is so mad is that he is right now building RCP cores and FP second cores in 1vs1 PBEM games. This discovery does mess up his games quite a lot...

In a case of SP games its been shown and proven the AIs are just very easy to beat...however when playing against another experienced deity level player, every little detail counts and RCP was widely and rapidly adopted as every little edge needs to be used in such games.
The whole strategy behind those games will need to be reviewed now in the light of this piece of news...
To know more visit the www.civ3duelzone.com

14. Grillepanel insect

Joined:
Dec 19, 2002
Messages:
2,709
Location:
Kiel, Germany
Great discovery, Qitai!

Hehe - this mechanic explains a lot. In two recent games (Greeks on large, Germans on small map), I used my second city (which got a handmade FP ASAP) as first RCP center.
Later, I GL-rushed a Palace elsewhere to have a second RCP core. But there was a huge difference in both games: Playing the Germans, I had actually two almost complete rings around the desired future capital built before moving the Palace- I had to claim the area before the ai could do so and the RNG denied me my GL for a while( ). There was no very big boost of income as I finally moved the Palace, the new capital was comparably close to my first core.

Now playing the Greeks, I noticed an incredible boost of income when moving the Pallace. I rushed it in a captured Aztec city (w/o any "direct income boost wonder" like Pyramids, Smith's etc) on the other side of the world. No other cities of mine were nearby at that time. In relation to Aggie's description, I think I accidentally "used" that rank exploit.

While a do think that this feels exploitive, I would also call it a nasty bug. I don't play competition games, but I could very well imagine situations where a far-away Palace move would be a considered strategic option (even if just used as an "insurance" against CF for a captured city w/ a bunch of really nice wonders, for example). This is really a screw-up. When would a player be allowed to move the Palace? Even forbid any Palace-move for competition games (or "political correct" SP games)?

I agree, Firaxis should fix it.

15. RibannahFighter Druid

Joined:
Aug 6, 2001
Messages:
2,449
Location:
Castle Gobs
I have never understood why the city's distance's rank should have anything to do with corruption. It comes across as a forced attempt to counter the advantages of ICS.

But the real exploit in all of this is not so much a wide first ring around the capital, which has to be weighed against the time that is invested to make it so, but - as before - the palace jump, which appears to be even more advantageous than we already thought.

16. Qitai.

Joined:
Jan 22, 2002
Messages:
1,177
Location:
SG.MY.TW.US
I think the concept of rank corruption is fine. It does balance against ICS quite well. Alternatively, the test of time scenario of making settler extremely expensive is another good approach to make ICS less attractive. ICS is indeed too powerful as applied in earlier versions of Civ.

And yes, palace jump is an exploit in my view, no doubt about that. But forbidden palace is too unbalancing especially against the AI I think. I highly suspect AI don't even know the value of using a GL to move the palace/FP to create a second core. In fact, it may well be more balancing to drop this whole concept of a forbidden palace.

17. SirPlebShaken, not stirred.

Joined:
Jan 1, 2002
Messages:
1,415
Location:
I also like the concept of rank corruption, I think it adds nice balancing and gameplay.

The more I've thought about this bug the worse it seems to me. In one regard it is one of the worst we've seen since the game's release. Unlike most previous exploits, this one is a kind of Pandora's box. Hard to close after opening. There's seems to be no simple rule to get rid of it even on an honor system. (Except a rather drastic one, "no FP allowed.") I can tell myself I won't use this exploit but can I really do that in practice? I can certainly refrain from planning to use it. But if I do a Palace jump, can I stop myself from thinking "Is this fair? Will I be gaining production which I shouldn't? Is that influencing my decision here?" And when building new towns can I stop myself from thinking "Will that affect corruption to give me an unfair advantage? Or will I perhaps be shooting myself in the foot unfairly?"

As a comparison, when RCP became well understood I tried for a while to not think about it while playing. But I wasn't able to not think about it. Now that I understand it, it has inevitably become a factor in my play. I think this is a shame because I think RCP is a bit imbalancing and a kind of exploit. I don't think the game's designers intended that any particular geometric placement of cities should have an inherent advantage. To me it seems that the game's underlying feel is that overlap (inter-city spacing), distance from cultural center, local geography, and competition for space with rivals are the factors intended to affect city location choices.

This bug regarding FP centered corruption seems similar to RCP in that regard but worse in its impact on the fun factor. I'm doubtful I can play without it crossing my mind. I'll wonder whether I'm doing it (for gain or loss) accidentally. And as soon as I have that thought, it unavoidably becomes in a sense deliberate choice, there's no way out at that point.

I too think that removing FP entirely would be more balancing for the game, as well as eliminating this problem. The AI doesn't know how to use FP well.

Now that we understand this bug I'm wondering if the bug accounts for a few cases where some AIs seem to do significantly better than others - perhaps when an AI gets "lucky" with a cramped peninsula start and then builds FP in a more open direction away from the peninsula, the resulting gain gives it a noticeable boost. This would be a random factor happening more often in AI play than for a good human player who would usually be trying to maximize both regions.

Removing the FP concept from the game would also in effect make the AI more competent. The result would be to slide the higher difficulty levels a fair bit upscale. For a human to win a large pangaea with many rivals at deity, without an FP, would be considerably more difficult than it is now.

In any case, I very much hope that Firaxis issues a fix for this bug, and issues one for both Civ and PTW. I think it would be nice if they'd fix RCP at the same time, but if not, this one seems the most important thing to fix. I do hope they'll consider one more patch to Civ to address this (vs. just fixing PTW.) If it is not fixed in all versions, comparative play will be difficult. And perhaps (depending on the approach chosen for the fix) impossible.

18. Qitai.

Joined:
Jan 22, 2002
Messages:
1,177
Location:
SG.MY.TW.US
Sir Pleb, you are making me feel guilty for posting this information . I was thinking if this should be revealed when I discovered this since I had the exact same thoughts that once this is known, it will affects everyone's decision, even if it is sub-conciously. And there is no clear line to distingush intentional use of this information versus playing normally. But I decided it is best be publish in the end since I do not want to hold this information to myself. Apologies to all if this information has mess things up for you.

19. alexmanKing

Joined:
Feb 28, 2002
Messages:
778
Location:
Maryland, USA
Whoa! Good job discovering this bug, Qitai! It is for sure a bug, and I will be disappointed if they don't fix it when Conquests comes out.

Depending on how exactly Firaxis has implemented the code, at this point it might be easiest to change the FP to affect distance corruption only (plus the current 10% increase in OCN), with the Palace being the only building affecting rank. This would devalue the FP, thus helping the AI, while still keeping the challenge and planning required set up a secondary core.

Joined:
Jan 1, 2002
Messages:
1,415
Location: