Water instead of Food

[Ahriman]

I usually play on Emperor, but I agree with cephalo's analysis.

The forest chop mechanic isn't really necessary, and feels out of place in the Dune world.

I also don't think a barren wasteland is necessarily out of place in this mod.

However! Removing the tubers does require some kind of hammer production terrain improvement that can be moved early in the tech tree. Mines in hills alone won't cut it, since half your hills will have windtraps instead.

 

[davidlallen]

How about making "surface" mine buildable *anywhere*, but the more advanced ones deep and core mine can only be built on hills. Is that enough of a tradeoff for removing chop?

Looking at a few autoplayed games, the AI seems to heavily favor solar farms with both of the civics that gives +1 hammer on it, resulting in a bunch of solar farms generating 4 hammers each. It's hard to beat that. I haven't worried about that yet, maybe it is too strong.

 

[Deliverator]

Well, here's a little test patch over 1.2.2. Not everything I wanted to implement is here yet, but this is what I have done:

+ All terrain produces no water (food) unless it has hills in which case it produces +1 water unimproved. I am just testing this hills change - I'm not sure whether we need it or not yet. An alternate idea to give a little extra push at the beginning to make the city square produce 3 rather than 2 water.

+ Bonus placement seemed to be pretty chaotic with no particular logic to it, so I've started from scratch effectively. I've removed all the random factors for now, and just made all bonus placement amounts be related to the number of squares on the map. Once we get the proportions right, we can introduce a bit of randomness for interest.

Placement order goes like this as of now:
1. Aquifer bonus. This bonus is a replacement for the Water resource and will be where you build Well improvements. At the moment this produces +4 water unimproved, but I'd like to get this down potentially to +2. even zero so it is more realistic. I'm thinking the Well improvement can add +4 to the square perhaps, Deep Well +8. I hoping davidallen can do the Improvements as I don't have experience of adding these. The Aquifer bonus is really common. I would like both Wells and Windtraps to be sources of Fresh Water. I've kept the bright blue water for now as it is helpful while tuning to have it super-obvious.

Spoiler :

2. Dew Plants. These are Sword Grass, Burro Weed, Spiked Paintbush, Sand Verbena, Barrel Cactus, Creosote. I have used the grouping stuff in BonusInfos to mean there is a high likelihood these will occur in clumps of two or three. I have produced a nice icon for each of these because I was getting confused. The 3D graphics are unchanged. At the moment, you can build Drip Farms on these to get the water output. Should we have a separate Dew Collector improvement, or just rename Drip Farm to Dew Collectors or what?

Spoiler :

3. Then everything else, we need to review placement for hammer and commerce producers too once we have balanced this.

+ The existing Windtrap mechanic is unchanged. I like it and think it works well. I'd just like to have the Wells as a second source of Fresh Water.
+ Added the new Windtrap model from Lord Tirian.

Spoiler :

+ Improved the Spice eye icon.

@david: Would you be able to add the Well improvements and do some of that lovely autoplay testing? I fully expect to have to rework the BonusInfos numbers several times to get the right distribution of resources.

This is still highly experimental, but I think if we balance the numbers it could be pretty cool. I've checked and the AI seems to grow pretty well - some had a second city down by turn 50 (which doesn't seem that quick admittedly - I think the worm fear theory might be responsible). I'm sure people can pick faults, but let's at least give the idea a fair go. Everything is open to revision and refinement.

Spoiler :

 

[davidlallen]

Oh boy, too many things to do at once. Today I only have a half day due to RL. I was planning to rework the storm as a map effect, which is a pretty huge rewrite. Then I have to figure out how to treat wormsign as a map effect, out of which a real attacking worm will pop, without scaring the settlers.

What is it about adding improvements which seems confusing? The related files are terrain/improvementinfos and units/buildinfos. If you look in the vanilla files so the names are less confusing, you can see the farm improvement and the build farm action. Each improvement lists the bonuses it can be built upon. Each build order lists the time required.

If I can help you figure it out, please let me know, but I don't think I can help on that part of the project today.

EDIT: I have given several times the instructions for turning off worms and storms, it is a one or two character change in the python. You may wish to do this locally as a short term aid to testing the AI expansion rate.

 

[Ahriman]

How about making "surface" mine buildable *anywhere*, but the more advanced ones deep and core mine can only be built on hills. Is that enough of a tradeoff for removing chop?

Sounds feasible. Though; really, what is it about a hill that makes it better for mining? I know we're so used to that cos its been that way since civ1, but the purpose of that was to give hills *some* kind of resource yield, since flat land is better for farming. But now that food comes from hills thanks to windtraps, you don't need hills to be the things that produce production.

I don't mind civics giving big bonuses, because of course part of the opportunity cost of choosing that civic is that you can't get the benefits of other ones.

I would like both Wells and Windtraps to be sources of Fresh Water

What is the purpose of fresh water access in this design? It seems like it does nothing.

I think +1 water on hills is a good idea, if only because it will help the AI prioritize hills as valuable tiles when choosing city spots.
Maybe reduce down the bonus from windtraps by 1 though.

I'd also argue against the +1 water from the city tile. Since all the water is heavily designed to come from only a handful of tiles, initial city growth will be very very fast, as long as you have a worker around to build the appropriate improvement.

I still worry that this design will leave large amounts of the map useless, and lead to bad city placement choice by the AI.
It also means that that the limits to growth of a city are pretty much hard-coded; the food capacity of a city is determined solely by hills and bonus resources. There is none of the strategic decisionmaking in vanilla of trading off food/commerce by choosing whether to build a farm or cottage on a normal tile. In vanilla, you can choose to make a city a commerce haven by building lots of cottages, or run a specialist economy (or just grow very fast) by building lots of farms. There's no scope for that in this design.
Detracts from fun I think, but taking a lot of control out of player hands. Improvement choices are relatively uninteresting; you build as many windtraps as possible, wells and dew collectors on the bonus resources, and then as many cottages/turbines as possible.

Maybe have some city buildings that give a +food bonus, or a +food% bonus? That way, by building city infrastructure you can still somewhat expand the upper limit to your city size, and make it somewhat endogenous.

 

[Deliverator]

@david: Well, there is some buzz around this mod it seems. Better to be busy than not... I wasn't sure how to specify which technology makes the Well/Deep Well available mostly. I may actually try 3 levels of well. Also not sure how your Windtrap/Fresh Water thing works. I suppose I also thought it might be better for you to integrate seeing as you've done a lot of the improvements work, and are on top of the current growth/economics issues.

@Ahriman:

What is the purpose of fresh water access in this design? It seems like it does nothing.

I was thinking it will be possible to irrigate only those tiles with Fresh Water access. So use Fresh Water to control where improvements can be built. You may be right though - may be it is redundant.

I still worry that this design will leave large amounts of the map useless, and lead to bad city placement choice by the AI.

Part of tuning the Bonuses is making sure that there are few useless areas of the map. The AI we can watch and test. There are various AI weights we can play with.

It also means that that the limits to growth of a city are pretty much hard-coded; the food capacity of a city is determined solely by hills and bonus resources.

Maybe, but you have limits on city growth even with food producing tiles... There has to be some limit.

There is none of the strategic decisionmaking in vanilla of trading off food/commerce by choosing whether to build a farm or cottage on a normal tile. In vanilla, you can choose to make a city a commerce haven by building lots of cottages, or run a specialist economy (or just grow very fast) by building lots of farms. There's no scope for that in this design. Detracts from fun I think, but taking a lot of control out of player hands. Improvement choices are relatively uninteresting; you build as many windtraps as possible, wells and dew collectors on the bonus resources, and then as many cottages/turbines as possible

Having played with this, I really think this is something that will attract people to the mod, but as Kael's article points out you then need the gameplay to keep them. This is not a finished design - It is a starting point which I hope will evolve with playtesting and ideas from people. Fully agree that we want interesting strategic choices, but I think we can work that in. Start simple and then add layers of interest.

Maybe have some city buildings that give a +food bonus, or a +food% bonus? That way, by building city infrastructure you can still somewhat expand the upper limit to your city size, and make it somewhat endogenous.

Certainly water infrastructure buildings are part of the idea.

 

[davidlallen]

Among the different big problems with the mod right now, storms/worms scaring settlers is the worst one. Among the people available to work, I think I am the only python expert; so I would like to work on the storms/worms right now.

The windtrap improvement and city windtrap use python to add an invisible feature. You can see this feature in terrain/featureinfos, FEATURE_WINDTRAP. All it does is add fresh water. The actual food/commerce bonus is associated with the improvement, terrain/improvementinfos, IMPROVEMENT_WINDTRAP. I don't know how to change the "radius" of the fresh water, but maybe we don't need to do that.

The key measurement that I make for economy is based on running a huge, noble difficulty, autoplay for 250 turns and then looking at the total commerce of each civ. There is a python function which writes a spreadsheet, but you can just as easily use ctrl-Z and then the economics advisor to see how all the civs did. What I look at is the total commerce displayed on the left. In a vanilla archipelago game, the total commerce of most civs is around 250-350. Today using 1.2.2, settlers are scared off by worms and each spice gives +3 commerce; typical total commerce values are around 100-150. With worms and storms turned off, typical total commerce values are around 200-250.

You can also visually count the number of cities; with worms and storms turned off, the average is around 4-5 cities with an average pop of 10 each. With worms and storms on, the average is around 2-3 cities.

We haven't tracked down the CTD yet, but it usually occurs around turn 300. I have noticed that when I get the total commerce up into the 250-350 range, sometimes the crash moves earlier. I guess this makes sense since whatever tech/unit/etc is causing the crash, can now be built earlier. So if you pump up the economy and start seeing this crash, you cannot get the actual commerce data, but you can assume it is high enough.

EDIT: also you asked how the techs are associated with improvements. This is the PrereqTech field on the build action. I suppose that is a little confusing. So units/buildinfo BUILD_FARM says that the prerequisite tech for farms is agriculture.

 

[Deliverator]

How do you set the game up to autoplay? I've never done it...

 

[davidlallen]

See dune wars\docs\README\AIAutoPlay Readme.txt for one way.

What I do is based on using the built-in python shell. Press the ~ character (you may have to search on international keyboards). This gives you a python prompt. Type:

CyGame().setAIAutoPlay(250)

I usually alt-tab to minimize the game, and alt-tab back to it every once in a while. Depending on how fast your machine is, you may or may not be able to surf the web, etc while it is running.

 

[koma13]

Among the people available to work, I think I am the only python expert

Not true. My python skills vastly surpass my knowledge in sdk and xml.

How do you set the game up to autoplay? I've never done it...

Press Ctrl+Shift+X...

 

[davidlallen]

Not true. My python skills vastly surpass my knowledge in sdk and xml.

Excellent. Have you been able to do much tuning on the early game combat unit selection? Your ideas on that from two weeks ago seem very promising.

 

[Deliverator]

Cheers guys. I'm using the Press Ctrl+Shift+X way. Amazingly it only using 50% of my trusty old machine's CPU.

Well, I could learn Python/SDK if I wanted to, but I code for a living so for me doing art is more recreational... it's more a preference thing I guess.

 

[koma13]

Excellent. Have you been able to do much tuning on the early game combat unit selection? Your ideas on that from two weeks ago seem very promising.

Yes, I know it's long time overdue. Right now I'm a little stucked with the hover spam. The problem is imho that hovers are the only sea units in early game... on a Archipelago map script... with coastal cities only.
I made an experiment where I changed thopters to a 2nd domain sea unit and ai start spamming thopters... I'm eagerly waiting for cephalo's new map script to see if that is leading to an improvement. Else I would have to look into sdk to reduce that or work with AI_Weight but I would prefer a solution without these arrangements.

I also have the problem that the transition between my new early units and dw's midgame units is still very rough. I want to smooth that out a little bit.

 

[Deliverator]

(Shouldn't this discussion be in the unit progression thread )

Edit: Also, with these AI Autoplays does setting the difficulty to Noble handicap the AI at all? Why not set it to Deity and let them at each other?

 

[davidlallen]

with these AI Autoplays does setting the difficulty to Noble handicap the AI at all? Why not set it to Deity and let them at each other?

I have used noble as a basis for comparison. I am sure that the commerce numbers would vary if the difficulty level varied. So by doing all my runs at noble, I can see how things are progressing over time.

 

[Deliverator]

Here were the results from my first autoplay with the water test patch posted above. Noble, Huge map, 250 turns. The screenshot shows the 7 city guy.

Num Cities / Biggest City / Total Commerce / Worked Tiles
5               16                 247                  79
3               8                  110                   29
3              12                 148                   38
7              17                 201                   56
2              15                 112                   15
5              11                 190                   60
4              14                 175                   37
4              13                 186                   40
6              15                 224                   36    * Score Leader
4              18                 174                   45
2              15                 121                   20

(Apologies - didn't know how to do a decent looking table.)

Not sure if these numbers are good, average or poor...

I'm going to do the following:
+ Reduce Aquifer to +2 unimproved.
+ Add the Well and Deep Well improvements.
+ Reduce Wind Trap to +2 water bonus.

And then see how that affects things...

 

[Ahriman]

I'd say poor for turn 250. The variation is also somewhat disturbing, particularly given how far apart all those civs are; it suggests that good/bad start local terrain have a huge impact on outcomes.

The clumped water resources also really don't look great; the food ones are fine, but the adjacent aquifers less so.

The screenshot also suggests another problem; the AI's tendency to build cities close to each other with lots of overlapping fat cross is a big problem when all the food comes from only a few tiles.

The resolution on the screenshots also isn't quite enough; are the cities building windtraps and cottages ok?
Is this with or without tuber forests; it looks like without, which is a problem hammerwise unless you also have something like the proposal for mines buildable anywhere; it means very low hammer output in the early game.

 

[Deliverator]

Well, bear in mind this was run with the sandstorms and worms still in play. Tubers are unchanged also. It may be necessary to get a better balance between food, hammer and commerces bonuses. I'll run some normal 1.2.2 games for comparison.

According to david's post above:

Today using 1.2.2, settlers are scared off by worms and each spice gives +3 commerce; typical total commerce values are around 100-150.

In which case these numbers don't seem too bad. If I can find where david describes how to disable storms and worms then I can run that test also.

On resource clumping, there is an iUnique value which will ensure that there is an empty ring of tiles around each instance of this resource. It is simple enough to set it to 1 for Aquifer - the only issue is that it very quickly reduces the number of available tiles for placement. Each Aquifer placed would take up 9 available squares. There is also a hardcoded game rule that no bonus can be adjacent to a bonus of another type. Is it just the graphics that bother you about the clumped water resources or you just don't like them being clumped?

Edit: For comparison here's the same numbers from a normal 1.2.2 game with worms and storms switched on.

Commerce / Worked Tiles / Num Cities / Biggest City
204          44          6          17
217          63          5          15
247          123         7          13
181          54          4           15
226          76          6           14
211          66          5           17
199          58          4           14
139          23          3           12
233          45          7           10
149          31          3           14
115          28          1           15

 

[Deliverator]

Normal 1.2.2 vs Water Test 1
Average Commerce: 192.82 vs 171.64
Average Worked Tiles: 55.55 vs 41.36
Average Number of Cities: 4.64 vs 4.09

So I definitely need to boost things for the next test. Still at least I'm not too far off. Perhaps using the AIObjective and AIWeights a bit will help.

 

[davidlallen]

These results look very promising. I find it helpful to do one run with the same settings on a vanilla archipelago map and get the same statistics. This gives a "baseline" for any comparison.

If I can find where david describes how to disable storms and worms then I can run that test also.

I should sticky this post with the instructions.

I have gotten the storm as map-effect working. Naturally, on the *first* turn of the *first* game I tested it on, I put a scout on autoexplore and what did he do? Walk right into a storm and die. Having *some* fear of storms and worms is good. I will ignore this for now; it is more important to get settlers sent off. Losing one or two workers and scouts due to storms they could have avoided is a much smaller problem we can worry about in the future.

I will try to get worms as map-effects working, and put up a patch. I have a bunch of other changes in this patch. We will have a temporary merge problem with your changes, but that is solvable one time.