Terrain city bonus

MrBaggins · February 7, 2003, 19:55

I'm not sure if this subject has been done to death already (I did do some searches and found nothing)

I've been working on a mod, and examining a number of fundemental game issues along the way, one of these being terrain, and its values... specificially with regard to defining programmatically how and where to settle... translating general AI 'goals' into choices, using weighting.

As part of the democracy game, i've made a somewhat elaborate spreadsheet, which calculates growth, production, science and gold, turn by turn, based on a 'grid' of terrain/river/good that you input; You define the prospective site, using the terrain name and so on, and the spreadsheet does the rest.

I was aware that part of the bonus, is a 'city' bonus, which is modifiable by terrain, but in all cases, at the moment, uses the same values, +10 food, +20 production & +20 commerce.

I was really curious to see whether concept which the Civ2 and Civ3 people have had to eliminate ICS (Infinite City Sleaze)... to negate the food/shields/trade which the city square provides could be done in CtP2... so I entered negative figures for the City modifier in each terrain, corresponding to the positive figures which the terrain normally gives. The technique works, insofaras the city square gives (without a river) a 0/0/0 value.

Since I have the simulator, I started to play with the figures to see what effect that this has on growth and so on.

If you negate the city square, the growth becomes constant... that is... if it starts at 1500... it stay at that level, rather than being reduced when the city grows, as it normally would.

The behavior happens since the city square's value is only ever added once, and the rest of the value is gained from the ring surrounding it. The city square's proportional value decreases as the divisor (the population) increases.

The same behavior occurs with production and commerce.

Something that you could theoretically try would be to keep the +10 food in the city modifier in each terrain, but have a negative modifier for production and commerce that matched the positive value that the terrain normally had. You would need to proportionally increase the production and commerce values throughout, since cities would be producing low and/or negative amounts.

This would, at first sight, solve the ICS issue...

City's Production and Commerce would grow proportionately with size...

E.G. if a size 1 city produces 20 production and 25 Commerce, then the same city at size 2 city will produce 40 Production and 50 Commerce, and 60/75 at size 3.... and so on. So... one size 2 city would be just as productive as 2 size 1's...

Another alternative would be to reduce the proportion that the city square is valued so that growth had less steep decline and/or production and commerce had a more proportional growth.

There are a lot of options here...

---

Just a few random thoughts that may be of interest...

MrBaggins

Maquiladora · February 8, 2003, 11:27

Can you attach the spreadsheet?

Gilgamensch · February 8, 2003, 11:32

MrBaggins,

the ICS had been cured with the maximum values per government, although this is possible to be changed by the user, it isn't so likely for beginners to change. This would mean that they have severe limits after i.e. the 10. city.

But still interresting to know. Might be a nice choice to increase the difficulty level (not sure if it could be done this way?), instead of giving the AI, other bonuses????????

Maquiladora · February 8, 2003, 11:42

Id rather the ICS problem was solved another way different to max cities per government, i always thought that abit weird. Even extreme crime and corruption (under some governments) might be better than max cities per government. Just got to figure out how the AI would handle all these great new ideas.

Gilgamensch · February 8, 2003, 11:48

I actually prefer the way, they have done, as this gives a feeling as more in reality, although default wise they are a bit low.........

MrBaggins · February 8, 2003, 12:15

Two points:

1) ICS still exists. Empire Size limits slow the process, but since you get science rewards for the technique, you are not as subject to the limits as other players, you can continue expanding through governments (and so the caps) quicker.

To very simply see... just imagine 2 empires... one with 3 size 1 cities, one with 1 size 3. The empire with 3 cities will have significantly more production, and commerce than the empire with 1 city.

If the center square held no value (for production/commerce especially) then the two empires would have identical production.

2) Yes I can attach the spreadsheet, but I'm hesitant to do so: It isn't finished to a degree that I'm happy with, isn't wonderfully intuative and so on... The production formula hasn't been finished...

But... here it is.

The 'top' grid, X1-X3 and Y1-Y3 is where you enter the names of the terrains (as found on page2, first column), Rivers are 'Yes' or 'No' answers and the Good columns use 0 for no good, 1-4 for that specific good number.

There is no bounds checking.

The 'results' appear in the grid below, is aggregated in the City/1st Ring section, and then turn by turn values, including culmulative growth is presented below that.

This spreadsheet is intended for approximate productive valuation simulation. It is unfinished, and is not for reuse, or contra-indicated use.

Unrar (probably with winrar). An Excel 5.0 format spreadsheet. Will work with any Excel from 95-XP.

MrBaggins

Gilgamensch · February 8, 2003, 12:29

Quote:

Originally posted by MrBaggins
Two points:

1) ICS still exists. Empire Size limits slow the process, but since you get science rewards for the technique, you are not as subject to the limits as other players, you can continue expanding through governments (and so the caps) quicker.

To very simply see... just imagine 2 empires... one with 3 size 1 cities, one with 1 size 3. The empire with 3 cities will have significantly more production, and commerce than the empire with 1 city.

If the center square held no value (for production/commerce especially) then the two empires would have identical production.

MrBaggins

OK, downloaded but haven't checked yet..........

For your first point:

Yes, it is true, but only to a certain degree. As you can't settle anymore city next to city, as you have the limit of government, it isn't really possible to cheat with it anymore. Because in your calculation you are missing the extra production from the extra workers, plus you could allocate speciallists......... so 3*1 isn't 1*3 really.

And if you would continue your example, for tyranny, it would be a maximum of 10*1 or 1*10, before unhapiness, but adding another city would create a lot of unhapiness... so you would be better of with 1*11.

MrBaggins · February 8, 2003, 12:40

I have to verifiy, but as far as I recall, specialists add... not in an unmodified way, but to the ring that they 'would' have been working... and are hence subject to division, and thus are less proportionately valuable than the city square, which is unmodified.

3 laborers in one city or 1 laborer in 3 cities add the up to the same total bonus. They are actually remarkably balanced.

A city might double in production through growth at 8-9.

It can multiply in production by 8-9 through creating that many size 1 towns....

The player is significantly better off ICS'ing until he hits the cap, pausing until his signficant science advantage gets him to the next government, and then continuing.

MrBaggins

Gilgamensch · February 8, 2003, 12:57

I think in the vanilla version you might be right, but other's have been modified (AFAIR, as far as I remember)......

For the workers:

But with putting in small cities workes, you limit their growth (not talking about farmers).

And Yes, the human is always better at using weaknesses in games, as the AI can't discover those

The radius depends on the MOD, default it is (IRC) 6 and the mods 8.

But don't forget how long it takes to create those settlers.........

MrBaggins · February 8, 2003, 13:03

Growth is a special case...

Growth is capped at 2500, and massively degrades as you grow

Many small cities will always grow quicker than a large city, farmers or not; their growth bonus is unmodified at pop 1, so their effect is far greater than the same number of farmers in a large city (where the growth is divided by pop)

MrBaggins

MrBaggins · February 8, 2003, 13:07

Workers appear insignificant at smaller sizes, because regardless of their better-than-average value in a specific trait; food, production, gold or science... due to the unmodified city square being the most signficant proportion of the cities production... masks the effect... not to mention that the city model is designed to operate best when balanced.

Gilgamensch · February 8, 2003, 13:22

Actually after rethinking all the facts you brought:

Yes, you are right. ICS is still there, only more hidden..... and not longer as strong. It could be nice, to leave it for the first city and just remove it from the other cities. That might be another possibility. (If possible).

It would give you still a faster growing center. So a kind of growth-center (like in good old Populos).

Gilgamensch · February 8, 2003, 13:25

Quote:

the city model is designed to operate best when balanced.

Not sure what you mean???????

MrBaggins · February 8, 2003, 13:33

Quote:

Originally posted by Gilgamensch
Actually after rethinking all the facts you brought:

Yes, you are right. ICS is still there, only more hidden..... and not longer as strong. It could be nice, to leave it for the first city and just remove it from the other cities. That might be another possibility. (If possible).

It would give you still a faster growing center. So a kind of growth-center (like in good old Populos).

Easy: give your the capitol building growth, science and gold bonuses, as well as the small happiness bonus.

Good idea too...

MrBaggins · February 8, 2003, 13:39

Quote:

Originally posted by Gilgamensch

Not sure what you mean???????

Well... Terrain becomes more valuable, as it can be improved. Specialists have... useful at first, although not overridingly so, fixed values, that don't improve as the game progresses.

It would be nice if specialists could be improved by advances as well as terrain.

Gilgamensch · February 8, 2003, 13:43

Quote:

Originally posted by MrBaggins

Easy: give your the capitol building growth, science and gold bonuses, as well as the small happiness bonus.

Good idea too...

No, not the capitol, just the first city............ Might be more fun even

MrBaggins · February 8, 2003, 13:46

Well... you couldn't do it easily in your 'first city'... there is no way to single this out... unless you create duplicate terrains for every terrain, keep the city bonus for those terrains, and when you create your first city, terraform it with that terrain.

Alternatively you could go the simple route, and just give the capitol building bonuses... you can only have one... and its usually your first city.

Gilgamensch · February 8, 2003, 13:51

Quote:

Originally posted by MrBaggins

Well... Terrain becomes more valuable, as it can be improved. Specialists have... useful at first, although not overridingly so, fixed values, that don't improve as the game progresses.

It would be nice if specialists could be improved by advances as well as terrain.

That's actually a thing I don't like so much. Like 'entertainers', with the introduction of TV, you can reach far more people......

Same for the rest of the specialists.......... they should increase when you advance, but I guess the AI couldn't handle it

Gilgamensch · February 8, 2003, 13:56

Quote:

Originally posted by MrBaggins
Well... you couldn't do it easily in your 'first city'... there is no way to single this out... unless you create duplicate terrains for every terrain, keep the city bonus for those terrains, and when you create your first city, terraform it with that terrain.

Alternatively you could go the simple route, and just give the capitol building bonuses... you can only have one... and its usually your first city.

Not possible to add this as a bonus to the terrain?

The reason why the capitol wouldn't be a good choice, is that I wouldn't like to see the bonus being moved with it.

I mean take England: biggest city London.
France: Paris
Germany: Berlin
Russia: Moscow

You see what I mean? It is normally their first capitol being olsa the biggest.........

MrBaggins · February 8, 2003, 14:12

The AI just needs a little help is all... especially since we can force it to do certain things...

Like for instance settling: at the moment the AI chooses where to settle based on a scoring system, where terrain has one 'usefulness' value, and is increased by the presence of a river.

It pays no attention to its goals... the AI does not consider distance enough, or goals of needing production or science at all.

The best settling system is one where a range of distances are considered, and each site is graded... yet that grading is modified by the ratios of the goals that it wishes to pursue:

Say an AI starts, at the beginning of the game with a growth desire of 1, a production desire of 1 and a commerce desire of 1.

If it finds out that its got enemies close, its important to grow quickly, to land grab, and to defend itself... so it increases its growth desire, but not its commerce desire and production desire... at first: its more important to have more cities than fewer 'perfect' cities. If its alone on its continent, and its a small continent, then it doesn't need to concentrate on growth so much, but on commerce; scientific desire should increase: it needs to develop a navy to be able to colonize other land masses... and also is relatively free of danger from other civs... and so should prioritize scientific development.

These are just examples, but you get the concept.

Discover 'high level' information. Use that to form weighted decisions. Use those weightings to form low level decisions. Enforce them through SLIC.

Theoretically, the AI could get better than placing cities, than a player who doesn't use a spreadsheet or calculator, to figure the odds...

That is the absolute fundemental flaw of the AI (and from where most of its troubles originate)... it just doesn't have a clue about city placement... the decision system is too simplistic.

MrBaggins

MrBaggins · February 8, 2003, 14:14

Quote:

Originally posted by Gilgamensch

Not possible to add this as a bonus to the terrain?

The reason why the capitol wouldn't be a good choice, is that I wouldn't like to see the bonus being moved with it.

I mean take England: biggest city London.
France: Paris
Germany: Berlin
Russia: Moscow

You see what I mean? It is normally their first capitol being olsa the biggest.........

If you add it to the terrain, it effects all cites. (unless you do a ton of work)

All of the cities you mention, would have the 'capitol' building, with the possible exception of France.

Gilgamensch · February 8, 2003, 14:37

Quote:

Originally posted by MrBaggins

These are just examples, but you get the concept.

Yes I do. And it would make sense

Quote:

Discover 'high level' information. Use that to form weighted decisions. Use those weightings to form low level decisions. Enforce them through SLIC.

Theoretically, the AI could get better than placing cities, than a player who doesn't use a spreadsheet or calculator, to figure the odds...

That is the absolute fundemental flaw of the AI (and from where most of its troubles originate)... it just doesn't have a clue about city placement... the decision system is too simplistic.

MrBaggins

But that would leave another problem:

How often would you re-evaluate those factors? Very turn, every time unknown terrain would have been discovered? How do you want to 'create' the long term goal? Do you want to 'attach' it to the leaders personality?

And know, I haven't read the SLIC-docu's yet, but I often 'tried' to create things myself (basic/C++/pascal/....), so I know a little bit about the basic's

Gilgamensch · February 8, 2003, 14:39

DL, problems with Apolyton ??????????

Gilgamensch · February 8, 2003, 14:44

Quote:

Originally posted by MrBaggins

If you add it to the terrain, it effects all cites. (unless you do a ton of work)

All of the cities you mention, would have the 'capitol' building, with the possible exception of France.

Not really

Germany the capitol was moved back only 'recently' And the old capitol (past 2nd WWW) was Bonn, and that wasn't really growing quickly............

Why for France exception?

MrBaggins · February 8, 2003, 14:49

France 'capital' = Versailles

MrBaggins · February 8, 2003, 14:56

Quote:

Originally posted by Gilgamensch

*snip*

But that would leave another problem:

How often would you re-evaluate those factors? Very turn, every time unknown terrain would have been discovered? How do you want to 'create' the long term goal? Do you want to 'attach' it to the leaders personality?

*snip*

Strategic considerations should be reevaluated in a strategic time period: say every 50 turns. Certain conditions, such as the declaration of war, should also cause reevals.

The goal is based on 'optimal development' given the situation. We might add on personality later...

You should allow the AI to see what it could not ordinarily see (cheat basically)... so it can be adaptive and predictive, and not reflexive.

MrBaggins

Gilgamensch · February 8, 2003, 14:56

OK, beaten me, but it belongs to (translated) Isle of Paris. Which most people 'consider' nowadays as Paris.........

not really true, but...........

Gilgamensch · February 8, 2003, 15:01

Quote:

Originally posted by MrBaggins

Strategic considerations should be reevaluated in a strategic time period: say every 50 turns. Certain conditions, such as the declaration of war, should also cause reevals.

The goal is based on 'optimal development' given the situation. We might add on personality later...

You should allow the AI to see what it could not ordinarily see (cheat basically)... so it can be adaptive and predictive, and not reflexive.

MrBaggins

That sounds feasible. Is it actually possible to check the percentage of newly discovered land? (start of turn/end of turn)
Reason why I mention: Imagine AI receives a map-exchange/offer and at that point it would need to re-evaluate....or it would do major exploration.....

The point of giving the AI this advantage, might be to big. It should be the same as for the human. Espacially, think of small maps, where it would really matter.........

MrBaggins · February 8, 2003, 15:07

All it does is act as a balance to the connective reasoning capability of the human.

Ultimately, we may be able to reduce some of the inherent AI production bonus... as we make it artificially smart(er?).

I really like the idea of a smart AI. A lot. and I don't mind if it has to 'cheat' to appear smart. Its better than cheating just to get more units/gold/pw.

MrBaggins

Gilgamensch · February 8, 2003, 15:17

Quote:

Originally posted by MrBaggins
All it does is act as a balance to the connective reasoning capability of the human.

Ultimately, we may be able to reduce some of the inherent AI production bonus... as we make it artificially smart(er?).

I really like the idea of a smart AI. A lot. and I don't mind if it has to 'cheat' to appear smart. Its better than cheating just to get more units/gold/pw.

MrBaggins

Getting it smarter is nice, but too much cheating could be an overkill. OK, it would be less cheating for the evaluation of the map, then actual boni for the production. But would be nice to have it without (don't think it is possible, without major changes)