More funnel-shaped spider-sandtraps in player-created maps

[star mouse]

Gramphos' example should also take city overlap into account. Overlapping tiles shared with another city should count 1/2, or zero.

[gremalkin]

A good illustration Gramphos..

I would also suggest that mineral production be taken into account however.. Your example would make the best starting place all grassland..

With expanding borders due to culture more relevance should be given to tiles closer to the city..

and resources would have to be taken into account..

I don't see why this should be such a problem for the AI.. Placing cities near to oppenents however gets complicated...

[Gramphos]

Originally posted by star mouse
Gramphos' example should also take city overlap into account. Overlapping tiles shared with another city should count 1/2, or zero.

Read my answer to tniem:
[COLOR={qcolor}]I forgot the overlap punishment. In the beginning I thought of half the points for all overlapping squares, but never tested it, and forgot about it while posting. If that would move down an existing "own" AI city below the lower limit that position should not be used.
I will add that Ralf's system don't take this in account either.[/COLOR]

[tniem]

Originally posted by gremalkin
I would also suggest that mineral production be taken into account however.. Your example would make the best starting place all grassland..

Which with irrigation would cause a city to grow really really fast. But remember when Gramphos/Firaxis has their equation fully balanced other things are going to be taken into account. Including strategic location, resources, and other elements. So other placements might be more prominnent.

With expanding borders due to culture more relevance should be given to tiles closer to the city..

I brought that up a couple posts before this one and Gramphos has already responded that should be included.

and resources would have to be taken into account..

Resources would only be taken into account with the above suggestion. The resources worth should be higher the closer to the city it is. It would go out to four or five but only with a low percentage as it will take a lot to expand that far. So resources should be taken into account divided by the distance or something like that.

I don't see why this should be such a problem for the AI.. Placing cities near to oppenents however gets complicated...

And that lies the problem. So the AI can pick a good spot for a city in terms of growth but can it figure out what place is going to be the easies to defend? How about the placement that strategically spring boards it for that invasion to the south? Or prepares a city to be used to colonize the globe?

The problem is city placement requires some strategy and planning, more than what is just the best spot. Because many times you will sacrifice a few points of growth in one city to make sure that you have a spot without over lap for a couple other cities. Can the AI handle that type of complexity? It hasn't before and I really don't think it will now.

[Gramphos]

Originally posted by gremalkin
I would also suggest that mineral production be taken into account however.. Your example would make the best starting place all grassland..

Once more: It is yet to be balanced. I take some production in account. All the civ to bonus resources of Civ 2 adds extra points to a square. I knowq that the "best" city is optimized for growing, but I think that the values should be modified for what the AI nead while founding the city. If it need a high production it should decrease grassland by one point, and increase coal.

With expanding borders due to culture more relevance should be given to tiles closer to the city..

See my reply to tniem. I said there that the Civ3 resources had to be taken account in the range from the city. It is very hard to balance this system for Civ 2, and it is mush harder to make a system for Civ 3, as we don't know how that game works yet.
As I'm not doing a Civ game (I have no time, but this thread realy wants me to), I see no meaning in balancing this. The entire example was ment to convince Ralf that his sytem doesn't work, as there is a lot more to take in account.

and resources would have to be taken into account..

Yes I know that. I took account for Civ2 resources, but as I don't know enough about Civ 3 to make an example about that I can't take Civ 3 resource-system in account.

I don't see why this should be such a problem for the AI.. Placing cities near to oppenents however gets complicated...

Yest I know, that was how I started with this example. I tried to convince Ralf that finding a good spot is the smallest problem of placing an AI city on the map. I've been thinking on that with the enemy cities a while, and one of the most important things is to boost production, and stay out of sight. Therefor a penalty for positions next to enemy roads, and the production boost spoken on before should be used (together with other things). Also the city on river bonus should be increased with one more.

Ralf has still not answered on this example, that I made for him . I never tried to make a perfect balanced system, and I'll not do that unless I make my own civgame one day.

Final note: I wrote a better reply to this before, but just as I should submit it the computer crashed, and I wrote this from my memory.
I hope I didn't forget any good points from the original reply. (I liked the language of that more)

[Ralf]

Originally posted by Gramphos
Ralf has still not answered on this example, that I made for him . I never tried to make a perfect balanced system, and I'll not do that unless I make my own civgame one day.

Sorry Gramphos. Better late than never. This is only a short reply however. As I said previously, Im all positive to improved expansion-algorithms by taking more into account (although there are limits). By the looks of the colored mini-maps, they seem to have achieved some good progress, compared with Civ-2/SMAC. But I think we are talking about two things here: Computer-generated maps vs human-edited standalone- & scenario-maps. The "sandtrap-model" mostly concerns the latter alternative.

The map/scenario-builder should be given an optional AI-civ expansion direct control. Direct human control, that is. One cannot "hardcode" the AI to found new scenario-cities accurately according to historical facts - each city differently placed depending on scenario and map-scale. How could they possibly do that? Some human help-guiding is obviously needed.

[Gramphos]

Thanks for taking you time to reply.

I don't say that your system is totally bad (as long as you can turn it off), but I think that it wont solve the problem mush more then standard algorithms (hopefully editable priority) can do, as you still have to tack account for other cities and enemy units.

I would also like to add that all cities in the real world aren't located in perfect placed, so why should them be that in civ?

And I agree that a Scenario could have used for defining how the AI's should expand during the game, no doubt in that. Your idea seem useful for that, but still these other then terrain factors are to be take count of.

Finally I want to apologize for almost taking over you thread. That was not my meaning.

[Steve Clark]

Originally posted by Ralf
The great thing with the Civ-3 game, is that you can have it all. Both random main-games, and very detailed; higly prepared; scripted; pre-edited scenario-games.

Nicely said, perhaps the "big picture" answer that seems to be lacking here. I would venture to say that at least 90% of the dozens of scenarios I've played take it one step further...ALL cities are predefined, predetermined, predictable, etc., etc. For historical-based scenarios, they would have to be (e.g., you need NY and Charleston in the right places for the Amer. Rev. War and likewise, Thebes and Kadesh for Ancient Age scenarios).

So for scenarios that perhaps don't need predetermined locations for AI cities, you still need some tight constrant else it wouldn't be a scenario but a random main game. For example, in many scenarios, you win by taking over X number of objectives (AI cities). Whether the cities have been predefined or not, they HAVE to be there in a controlled environment. I would take it one step further in that they would have to be in specific locations in order to maximize strategic and tactical play-balance.

[star mouse]

Here's my revision of the penalties for overlap. The 12 distant tiles count 1/2 if they overlap, and the 9 central tiles count 0 if they overlap

- not in city radius
- count 1/2 if they overlap
- these tiles are not counted if they overlap. Perhaps we could also make these tiles count -2 regardless of terrain.







If you have played Master of Magic, you may recognise this scheme. In that game, overlapping tiles counted 1/2, and you had to place all cities at least 4 tiles apart.

[Gramphos]

Now I've taken another screen (quite boring one (also a random generated startup position)).

---

8 plains * 4 = 32
13 ocean * 2 = 26
1 whales * 2 = 2
1 fish * 2 = 2

---

32+26+2+2=62

---

Next to ocean +1

---

63 points for that one. I think it is a good position, If you irrigate, but not unless. Therefor a system also must check how mush food the city area can produce in:
1. No improvements, just 1 pop.
2. No improvements, maximum reachable pops (of that food) in Despotism
3. Full improvement, maximum reachable pops (of that food), Democracy

I repeat, I'm not balancing this, just trying to give other aspects of what is important for a city area.

[Ralf]

Originally posted by Steve Clark
So for scenarios that perhaps don't need predetermined locations for AI cities, you still need some tight constrant else it wouldn't be a scenario but a random main game. For example, in many scenarios, you win by taking over X number of objectives (AI cities). Whether the cities have been predefined or not, they HAVE to be there in a controlled environment. I would take it one step further in that they would have to be in specific locations in order to maximize strategic and tactical play-balance.

Good point. I have been so busy argumenting about efficiently exploited AI empires and historically accurate scenario empire-expansions, that above play-strength/balance argument temporarily was overlooked. But I agree of course. If the scenario-builder have access to all the excisting (and possible) city-locations (and also all the recource-locations), he can obviously script-control the scenario in much more effective ways.

Above can of course also be achieved by setting up all cities once and for all - then remove the settler-unit all together. All contenders are then forced to play with what they already got. But this only works well in scenarios with very short timelines, like for example WW-1 or WW-2.
In long timeline scenarios however, where every Civ are expected to expand and establish new cities, then the ruleless freedom of founding these cities makes it much more difficult for the scenario-builder to script effective enough counter-strategies. One can truly say; "the more ruleless, the more clueless", from the scenario-builders point of view.

I have previously emphazised that the scenario "sandtrap-expansion" model only concerns AI-civs, but personally I would love to optionally hand over the decision of possible HP city-locations in the hands of the scenario-builder as well - as long as I can choose available expansion-directions. If that can improve & strengthen gameplay an AI-civ resistance in advanced script-guided scenarios (and I know it will), then so be it.