[Freeciv-Dev] Re: Why does ICS work?

["Ross W. Wetmore" <rwetmore@xxxxxxxxxxxx>]

Good analysis. 

But I think the real life point is that exponential growth is a fact of
life. All successful populations grow this way in an environment of
infinite resources. Unfortunately there are always controls sooner or later. 

The real restriction on ICS is the fact that sooner or later you run out of
squares to build on (crudely change this by setting the minimum distance).
And probably before that you run into escalating military costs to protect
your periphery. 

In the real civilization, the spread of corruption in a low level form of
government reduces the effective payback. This is based on both an extent
and a  total city/population count. Eventually, cities lose all their
resource production and even with higher forms of governments, citizens
still become unhappy, i.e. are born unhappy==city in Revolt. Then you need
to build more military units or pacifier improvements at a horrendous cost
per civ to counter this. The real civ also adds a cap at about 20 cities
that sharply increases these effects with each new city added after this
point. 

Also, later technologies like Automobile cause the effects to increase. One
could do things like double the disease rate when you discover Trade (i.e.
partly countered by improvements like Aqueduct) if it was felt that this
sort of control was needed earlier. The disease rate stops cities from
growing at rollover. Another deterrent against unimproved cities is to
increase the calamity factors like famine, earthquake, pirates that are
countered by Granary, City Walls and Barracks, or add some new ways to
target village sprawl.

But the Civs that grow fastest are also the simplest. They sacrifice
defence potential and cooperative efficiencies for sheer mass. It is a
strategy, but not a sure winner, and it has to achieve its success early on.

Note that a trade route or a marketplace in a size 8 city makes it
equivalent to a size 12 city, and a Bank to a size 16, a Stock Exchange to
??. What do you have to add to your 4 size one cities to achieve the same?
Two or four more cities per improvement? And what is the proportional cost
to improve? Note a Marketplace costs 2 settlers, and *doesn't* reduce the
payback at any point. This partly answers Jason's question about how
doubling Settler costs relative to growth improvements changes the balance.

Also, when your settlers are busy building cities, they don't improve the
terrain, and terrain improvement is both another multiplier and defense
enhancement. In the time it takes a Settler to build and grow a new one
(say 30 turns), it can add how many new food, resource and luxuries to your
Civ compared to the 10 or so from citizen doubling? That's partly why
active settlers still eat food, or a farm-o-rama strategy would dominate.

The key to developing the more effective units and technologies is to make
your workers and resource squares more productive.

Note that in an infinite resource scenario, the edge is slighly in favour
of ICS, but if you only have twenty squares, a single size twenty or maybe
30 city will be far more productive than 10 single cities at a far lower
total cost.

It is also the case that "We Love the President Day" growing at one citizen
per turn in every city, TOTALLY wipes out any exponential growth curve from
pure food growth/rollover.

Getting to Republic with a defensible core of size 3 cities and an
appropriate balance of luxury and pacifier improvements is *THE* major
growth strategy. If you can do that before the ICS-types find you and
overrun you, or you can hold them off until you hit critical mass, they are
dead meat.

Cheers,
RossW
=====

At 12:25 PM 01/06/12 -0700, Arien Malec wrote:
>--- Jason Short <jshort@xxxxxxxxxxxxx> wrote:
>> Do you really think doubling the cost [of settlers]
>> will slow down ICS?  I'm not quite claiming it doesn't,
>> but I'd like to see a mathematical explanation to support
>> this claim.
>
>Good point: to fix ICS, we should proceed from an explaination of why
>ICS is a good strategy. Ideally, we'd have a good game-theoritic model,
>and could test fixes against that model, but I can't contribute there
>:-). I'll start with a plain text explaination....
>
>Smallpox works as a strategy because of three factors:
>
>1) A city of size one can work 2 squares, and it only costs 1 worker to
>create a settler. A city of size=2 has 3 workers: if the city creates a
>settler who founds another city, there are now 4 workers
>
>2) Growth through creation of cities is exponential (1->2->4->8...).
>(It's really more of an S-curve, since there are limits to exponential
>growth: total land, presence of enemies)
>
>3) There are upkeep costs (happiness, cost of buildings, support of
>settlers to create infrastructure) associated with vertical growth that
>there aren't with horizontal growth
>
>It is 1+2 that is really deadly.
>
>(1) is very fixable: Michael Kiermaier and I wrote a patch that allows
>settlers to have a pop cost of 2: so a city of size 3 with 4 workers
>becomes a city size 1, with 2 workers + a settler, who founds another
>city of size 1 with 2 workers, and you're back to 4 workers total.
>Status quo. If you really want to put the breaks on sprawl, you could
>make the pop cost of a settler 3, so that you lose workers when you
>create cities. (There are probably Real World arguments why this should
>hold....)
>
>(2) is inherent in the game. You can try to slow it down (increasing
>the min distance between cities, increasing the production cost of
>settlers, etc.), but you can't stop it.
>
>(3) may be fixable. Increasing the support costs (e.g., by increasing
>unhappiness) for small cities levels things between horizontal and
>vertical growth somewhat. But I believe that fixing 1&2 would do more
>to balance the game than fixing 3.
>
>The problem is not so much to stop horizontal growth, but balance
>things to make vertical growth a viable strategy. If we solve problem
>(1), and we can't make problem (2) go away, we need to create
>expontential benefits to vertical growth to balance those for
>horizontal growth.
>
>Currently, 4 cities of size 1 produce as much trade as one city of size
>8 (all other things being equal, and ignoring corruption). Given that
>it is significantly easier to get to 4 cities of size 1 than 1 of size
>8, it would help if the size 8 city had some sort of trade bonus...
>Something also S-curve shaped, peaking between 6-14, and trailing off
>afterwards.
>
>Arien
>
>__________________________________________________
>Do You Yahoo!?
>Get personalized email addresses from Yahoo! Mail - only $35 
>a year!  http://personal.mail.yahoo.com/
>
>