Firaxis: Corruption Breakdown

[Yeti]

Originally posted by Yeti
A city's rank is the number of cities closer to THE Palace than that city is to its nearest capital.
Right?
And that causes both the RCP and remote Palace exploits.

Originally posted by SirPleb
No, it causes only the Palace rank bug.

RCP is from a different cause altogether. Essentially, in a programming sense, it is because the code does not directly use an ordered list at all. Instead for each city it counts the number of closer cities.

Right! That's the "closer" portion of my one sentence Sorry - I was tring to keep the description as short as possible, but I guess that complicated things.

So to clarify:

PTW's Rank System: A city's rank is the number of cities closer (allowing equidistant cities to share ranks) to THE Palace than the city being ranked is to its nearest capital.

PTW's Rank Pros: A pretty strong FP due to lots of repeated ranks (ranks might go 1, 1, 2, 3, 3, 4, 4, 5, 6, 6, 7, 8, 9, 9, 9 - just for example - depends on locations and city spacing and density around the FP vs density around the palace, etc...). A well placed FP works better than a poorly placed FP (which is good because it rewards a thinking / experienced player more).

PTW's Rank Cons: RCP bug, remote palace bug, FP in same city as palace almost totally useless, FP very close to capital minimally useful (hurts AI).

PTW's Rank System FP Placements and Effects:
- FP in capital: No effect on ranks - FP seems useless
- Palace far away with cities around FP - Ranks all low due to bug - FP seems overpowered
- Palace and FP set "optimally" to create two productive cores: Considerable effect on ranks / corruption, with roughly a doubling of the number of productive cities due to duplicate rank generation - FP seems pretty powerful, but this is the familiar scenario we've come to feel is "balanced".

Desired solution: No RCP exploit, no remote palace exploit, FP strength similar to PTW's.

Suggested Rank System: One list. Unique ranks. Rank is based on proximity to nearest capital. Final rank used for corruption calculation is divided by number of capitals (so with FP, effective rank is divided by 2).

Suggested Ranking Pros: FP is still quite powerful. Both exploits are fixed. The FP's strength will be about the same as that of a PTW "optimally" placed FP. There is no longer a way to position the FP to overpower it, or to majorly underpower it (even an FP in or near the capital will be quite beneficial, which helps the AI). FP placement does still matter though, so a skilled player can maximize its effect (reducing corruption in your "best" cities will help your empire more than reducing corruption in some tundra or desert cities, plus of course using the FP to reduce distance based corruption in a set of cities). The addition of more capital flagged buildings will have diminishing returns (I think this is a benefit at least - I would think you'd want decreasing benefits, not something that would trend towards total elimination of corruption).

Suggested Ranking Cons: In many situations the FP will now be more powerful, although, for players at least, usually only by a little bit. This could affect game difficulty since the AI is likely to get a larger increase in FP effectiveness than players (although not an overwhelming change since rank corruption is only half the total corruption picture).

What do you guys think of this?

 

[Commander Bello]

Originally posted by TedJackson

[...]
So, using my ranking method above, for 4 first ring equidistant cities the rank of each city would be 2. Similarly if there were 5 equidistant cities then each would have rank 3 and six equidistant cities would each have rank 4.
[...]

Sorry, but wouldn't cities(5) and cities(6) have the same rank as of 3? Since rounding to the next integer for 5 cities: 5/2=2.5=>3
and for 6 cities: 6/2=3 =>3

Just a thought.. or do I misunderstand something?

 

[Yeti]

Originally posted by Commander Bello
Sorry, but wouldn't cities(5) and cities(6) have the same rank as of 3? Since rounding to the next integer for 5 cities: 5/2=2.5=>3
and for 6 cities: 6/2=3 =>3

I believe he's saying you'd average their ranks. So for 6 cities, all closet to the palace and all at the same distance:

(2 + 3 + 4 + 5 + 6 + 7) / 6 = 4.5, rounds to 4.

 

[Qitai]

Originally posted by Yeti


PTW's Rank Pros: A pretty strong FP due to lots of repeated ranks (ranks might go 1, 1, 2, 3, 3, 4, 4, 5, 6, 6, 7, 8, 9, 9, 9 - just for example - depends on locations and city spacing and density around the FP vs density around the palace, etc...). A well placed FP works better than a poorly placed FP (which is good because it rewards a thinking / experienced player more).

Not true. It goes 1,1,3,4,4,6,6,8,9,9,11,12,13,13,13 rather.

 

[Yeti]

Originally posted by Qitai
Not true. It goes 1,1,3,4,4,6,6,8,9,9,11,12,13,13,13 rather.

Ahh yes. You're quite right Qitai Repeats are followed by skipped values since all of those "tied" cities are closer than the next one.

I think all of my points still work though (but good to get the details right!).

What did you think of my final suggestion for a solution? You're a battle tested Civ veteran with a thorough understanding of Corruption... do you think this change would strengthen the AI by too much? Or perhaps just be too different (or too strong) for players?

 

[Grille]

I could live with it.
But I disagree about the effects listed as pros and cons. (They are partly made in comparison to the old FP vanilla/PTW mechanics?)
It's probably always much better to not have FP and P close to each other. You'd throw away the distance benefit when FP next to P.

Also, I'd think FP will be less (not more) powerful in general (compared to PTW).

 

[akots]

Originally posted by Qitai
...I do not think Firaxis is looking for a major overhual here, but just plucking loopholes in the current model. Perhaps those can be ideas for Civ4.

The problem is that sharing a rank between several cities is not possible as it was in PTW. For example, if in PTW a player had 10 cities, last one can be rank 9 or even less if carefully placed. In the current model, city number 10 will have rank number 10. Then, as a result of this, number of ranks has increased which makes corruption extremely high. On the other hand, rank corruption equation did not change. So, it looks like fixing loopholes might not be possible at all without changing the equation. And some not so "major overhaul". Just someone has to repeat high school algebra and do some thinking and comparison.

 

[SirPleb]

Originally posted by TedJackson
but rather that those cities should have the same rank (possibly arrived at as the average of the sum of the number of cities at that particular distance, rounded up) not simply give them different ranks based on the founding date.

I agree that this would be nicer but it would actually be a fair bit of work to implement. The simple approach of averaging the ranks would lead to a new problem related to OCN:

Suppose that OCN is 11, and that there are 9 cities equidistant from a capital, beginning at rank 9. I.e. those 9 cities go from rank 9 to rank 17. If we average their rank and treat them all as being at rank 13, the player ends up paying a penalty. 13 is over our OCN, so the rank corruption is much higher than cities under OCN. But if we assign ranks in an arbitrary sequence, two of the cities will be ranks 9 and 10, under OCN, and will get reasonably good rank corruption.

Similarly, a clever player might use this as an exploit. If OCN is again 11, and the player has 9 equidistant cities beginning at rank 6, their "average rank" will be 10. That will draw all 9 of these cities under OCN, where four of them actually should have been over OCN. This new exploit would surely be found and abused

To properly come up with the same corruption for all cities at the same distance, I think it is necessary to average the resulting corruption percentage, not the rank. That can be done but is getting into somewhat complicated code. I'll have another look to see if I can find simple code which would do that, but I'm not hopeful.

 

[SirPleb]

Originally posted by Yeti
Suggested Rank System: One list. Unique ranks. Rank is based on proximity to nearest capital. Final rank used for corruption calculation is divided by number of capitals (so with FP, effective rank is divided by 2)
...
What do you guys think of this?

It seems to me that this is effectively the same as the suggestion I made earlier about creating distinct lists, i.e.:
"Finally, a comment for those who would like the FP weakened for the human or strengthened for the AI:
The code Tavis showed us creates an interesting opportunity. If after the line:
'if (pLoopCity == this) continue;'
we insert the new line:
'if (pLoopCity->pClosestCapital != pClosestCapital) continue;'
this would cause the Palace and FP rank lists to become distinct! I.e. non-overlapping. This would not reduce the effect of a far away FP. But it would greatly increase the strength of a nearby FP."

It seems to me that the only difference in results between a single averaged list and the distinct lists would be rounding differences, and small (but immaterial) reordering differences when one capital's region is more densely settled than another. Does that seem true to you also? If not, how does a single averaged list produce different results? If that is true then I prefer the distinct lists because they seem easier to describe in English.

In any case my feeling was that this kind of change could greatly increase the AIs mid and late game strength and should be deferred until later if it is to be tried at all.

 

[SirPleb]

Originally posted by akots
The problem is that sharing a rank between several cities is not possible as it was in PTW. For example, if in PTW a player had 10 cities, last one can be rank 9 or even less if carefully placed. In the current model, city number 10 will have rank number 10. Then, as a result of this, number of ranks has increased which makes corruption extremely high. On the other hand, rank corruption equation did not change. So, it looks like fixing loopholes might not be possible at all without changing the equation.

Why not? I think that the two small code changes I proposed will accomplish exactly that, fixing the loopholes and behaving like PTW - have you reviewed my fix and concluded that it won't?

 

[SirPleb]

Originally posted by Yeti
PTW's Rank System: A city's rank is the number of cities closer (allowing equidistant cities to share ranks) to THE Palace than the city being ranked is to its nearest capital.[/B]

PTW's Rank Pros: A pretty strong FP due to lots of repeated ranks (ranks might go 1, 1, 2, 3, 3, 4, 4, 5, 6, 6, 7, 8, 9, 9, 9 - just for example - depends on locations and city spacing and density around the FP vs density around the palace, etc...). A well placed FP works better than a poorly placed FP (which is good because it rewards a thinking / experienced player more).

Originally posted by Qitai
Not true. It goes 1,1,3,4,4,6,6,8,9,9,11,12,13,13,13 rather.

Originally posted by Yeti
Ahh yes. You're quite right Qitai Repeats are followed by skipped values since all of those "tied" cities are closer than the next one.

Nonono, I think you were right in the first place Yeti

The thing is that the "tied" cities are NOT necessarily closer than the next one. One of them definitely is since it was measured from the Palace. But the other of the tied cities may or may not be closer to the Palace. All we know for sure is that it is equally close to the FP.

 

[Yeti]

Originally posted by Yeti
Suggested Rank System: One list. Unique ranks. Rank is based on proximity to nearest capital. Final rank used for corruption calculation is divided by number of capitals (so with FP, effective rank is divided by 2)
...
What do you guys think of this?

Originally posted by SirPleb
It seems to me that this is effectively the same as the suggestion I made earlier about creating distinct lists, i.e.:
"Finally, a comment for those who would like the FP weakened for the human or strengthened for the AI:
The code Tavis showed us creates an interesting opportunity. If after the line:
'if (pLoopCity == this) continue;'
we insert the new line:
'if (pLoopCity->pClosestCapital != pClosestCapital) continue;'
this would cause the Palace and FP rank lists to become distinct! I.e. non-overlapping. This would not reduce the effect of a far away FP. But it would greatly increase the strength of a nearby FP."

It seems to me that the only difference in results between a single averaged list and the distinct lists would be rounding differences, and small (but immaterial) reordering differences when one capital's region is more densely settled than another. Does that seem true to you also? If not, how does a single averaged list produce different results? If that is true then I prefer the distinct lists because they seem easier to describe in English.

In any case my feeling was that this kind of change could greatly increase the AIs mid and late game strength and should be deferred until later if it is to be tried at all.

Ugh. That made me think Don't do that!

OK, I thought for a bit, and unless I'm missing something, I believe there actually are differences. Your suggestion reduces corruption in those cities that end up closer to the FP by giving them a new separate ranking list. Mine is more indiscriminate in reducing corruption

For the "ideal" situation - an FP in the middle of a group of cities and your palace where it always was at the center of another group of cities - you end up with two nice cores, each with their own rankings now, and therefore each with a ring of productive cities. In this situation I agree that it would work, disregarding rounding issues, the same as my suggestion.

However, at the extremes I believe they would work differently. For the case of an FP in the same city as the Palace, your system would do nothing for ranks. Mine would cut ranks in half (basically it would make your palace twice as strong and double the size of your initial productive core).

At the other extreme, an FP off on the very edge of your empire, or on a separate island / continent with only a few cities, your system would provide new rankings only for those cities closer to the FP than to the Palace. This would result in a HUGE decrease in corruption for those few cities, which would suddenly drop down to ranks 1, 2, 3, etc... But it would only help a small portion of your cities. You could end up with the FP benefitting 3 cities, and all the others having the exact same ranks they always did around the Palace. The system I'm suggesting would again cut ranks in half. The lowest corruptions would of course be right around the FP and the Palace, so FP placement does matter, but regardless of where they are, the other cities in the empire would still benefit.

Now... since they appear to be different, the next big question is, which is the preferred behavior

I think your solution does a better job of coming close to how things work in PTW. My solution would make it easier to get a benefit from the FP, no matter where you put it, which would make the AI stronger. It would also help players some, but I think the AI would benefit more. I'm sure every once in a while a player would be in an odd situation where there are a few nice highly productive (shield wise, not corruption wise) cities on an island or other continent that they'd like to slap an FP in, and then they'd prefer my solution because they could make those cities almost corruption free without gimping the rest of their empire. But most of the time there is more likely to be an issue with a tougher AI.

 

[SirPleb]

Originally posted by Yeti
Ugh. That made me think Don't do that!

Well, your post got even, now my head is hurting

You are right, they are different, I hadn't considered all the consequences. I think I'd prefer the distinct lists approach because of the difference when one center has very few cities around it and the other has many - with the averaged approach the result in that situation is a bit like the communism effect. Also, there'd be a non-intuitive result when new cities are added to the less populated region - building them would increase corruption in the larger region. I personally don't like effects like that where reducing corruption in one part of the world has a side effect in a different part of the world.

But in either case the AI would gain considerable strength. So if someone wants either of these changes it should probably move off to a separate discussion thread.

 

[Yeti]

SirPleb:

Excellent point about it being bad when doing something in one place affects corruption somewhere else. Those situations will always exist to some degree, but minimizing them is good.

I give your suggestion points over mine for:

- More similar to PTW.
- Easier code (therefore safer and more appropriate for a patch).
- Fewer side effects (less change in AI power, fewer situations where a change in one place affects corruption in another place).
- Benefits the smart player more (where you put the FP remains important).

I would give my suggestion points over yours for:

- Makes it easier for the AI to use the FP well.
- Helps total empire corruption more in some situations.

Two of those are fairly mutually exclusive, yet both show up as plusses - It's almost impossible to find a solution that rewards the more skilled player over a less skilled player, without making it hard for the AI. But again we come back to the first point in your favor - NOT making things easier for the AI keeps things more consistent with PTW. As people have pointed out (and I agree) we don't want to be creating balance issues or forcing the adjustment of the difficulty levels or frustrating players who find they can no longer play at the level they're used to.

I think we should try to wrap this up soon so that Firaxis can move on

As of right now, I think SirPleb's suggestion is the best. Do others favor a different solution or have a new one to suggest?

 

[HellboundHeathcliff]

Ultimately it will be Firaxis decision. They put this out for the community to work on and a few of you have done an excellent job.

Since this should be the last adjustment to the corruption model, it should require the least changes that will give the intended effect. It should be *SAFE* to implement and closely model that of C3/PTW. I therefore vote for SirPleb's implementation.

You could put up a poll with a full explanation in easy to understand terms for the uninitiated OR just wait for feedback from Firaxis.

I have been following this discussion but staying quiet since i profess no up to date programming skills or a full understanding of the mechanics involved. Again... excellent job

 

[SewerStarFish]

1) Probably the best thread in the history of CFC

2) Tavis: You DA MAN

3) I really like the idea that FP is a lesser Palace, and wish Courthouses and Police Stations were more effective. The whole idea of an expansion is to improve your playing experience: having a better late game AI (e.g. an AI balanced throughout the ages) outweighs me learning a new strategy of FP use. (NOT FLAMING, but Conquest Scenarios were worth the price alone, the old FP lovers could play/ if a lesser FP is unacceptable)

4) At the least this is a great Civ4 Palace/FP wishlist thread.

 

[SirPleb]

Originally posted by SewerStarFish
The whole idea of an expansion is to improve your playing experience: having a better late game AI (e.g. an AI balanced throughout the ages) outweighs me learning a new strategy of FP use.

I'm thinking that I haven't made my personal position on this very clear. I agree! I'd like the same thing for my personal use. But against that I'm more worried that increasing the mid/late game AI strength by an untested amount could have a negative impact on the majority of players.

Something I think could be interesting to try (after next patch) is a test of a "distinct list" model, e.g. with that one extra line of code. Tavis, would I be right to guess that Firaxis has a "self-play" test mode for the program? I.e. where AIs just play each other and info gets recorded for subsequent analysis? No need to answer that if you don't want to of course. But, if you do have that, it might be interesting to play a few AIs with the first model against a few with the stronger "close FP" model, to see how much stronger the modified ones play

Or maybe it really isn't a CivIII issue and should wait for the ground-up rethink in CivIV

 

[Justus II]

I will also vote for the distinct lists, for a couple reasons. First, I think intuitively most people (at least most people who wondered how it worked in the first place) would have assumed that that is exactly what it did, creating a new ring as a "second palace". Second, and more importantly, is reducing the "law of unintended consequences" where building the FP increases corruption around the capital, or in the example Yeti sites, when it is built in a remote location with few cities, that it would actually reduce corruption around the original palace. Neither situation would seem to be obvious to newer/ "casual" players, and I think should be avoided.

As for the desire to improve the FP for the AI, I would love to see it, but not at the expense of it's functionality for player, except perhaps as an option later.

 

[Qitai]

Originally posted by Yeti
What did you think of my final suggestion for a solution? You're a battle tested Civ veteran with a thorough understanding of Corruption... do you think this change would strengthen the AI by too much? Or perhaps just be too different (or too strong) for players?

It would not be intuitive I think. The halfing of the rank has no parallels to what I would expect of a FP in real life. So, I don't like the idea solely from that angle.

 

[Yeti]

Originally posted by Qitai
It would not be intuitive I think. The halfing of the rank has no parallels to what I would expect of a FP in real life. So, I don't like the idea solely from that angle.

Aye I've actually changed my mind as well. I'm not even supporting my own suggestion any more!