SMAX - The Will to Power - mod

[dino]

[...] Real life works differently. But this is irrelevant to game design. Let's avoid such arguments.
[...] That is the whole point of any rule in the game - to counter some other tactics.

For me it's too abstract and gamey and such simple tricks not only working, but being necessary, ruin the fun, by overruling the economy buildup advantage, that is the real meat of a 4x game for me.
But even if you don't care about versimilitude much and are fine with these kind of board game like, not simulationist rules, you have to at least admit that AI can't into these kind of tactics at all.
And improving AI in this regard would be a huge task even with full source code, so on top of me liking more grounded in reality rules, I think it's necessary to provide a challenge.

With vanilla rules, I can indefinitely hold off an AI faction of equal power to mine, by dedicating only 10 - 20% of my economy to it, while AI will burn all it's production on the same war.
I want rules that would put me under enough stress, to loose turns advantage against other uninvolved in war AI factions if I get attacked.

Sorry, I don't follow what ruined Civ series?

1 UPT = 1 Unit Per Tile, from Civ 5 and onwards, introduced because of all casuals complaing about stacks of doom. Civ 4 is the last one I've played and I refuse to even try the new ones on that account alone ( there are plenty other reasons, based on online discussions ). I want to play a strategy game, if I wanted an abstract board game, I'd stick to chess. In a strategy game economy and logistics should win the war and tactics should be able to only trip the balance.

Casual player can't stand when he ends up in unwinable position, while giving him tools to save the day even when confronted with superior power, provides him a great satisfaction. This the design goal behind 1UPT, or stack wipes, which allow wins against many times stronger AIs. What is the point of playing with rules like that, if you are a competent player ? I instead want to be destroyed, if I fail at "build" part of 4X.

[...] With collateral damage I don't see a problem.

In a strategy game of this scale, a confrontation between two stacks of units of equal power, resulting in a total destruction of one side with almost no loses on the other, being solely determined by which side happened to be the attacker is not acceptable to me and against AI, it'll almost always the player be an attacker and that alone kills most of the challenge in war part of these 4x games imo.

Of course I may not like it in practice, when I actualy try it

[Alpha Centauri Bear]

Real life certainly has an influence on game design. Civilization series is probably the best example of them all. It does simulate real life!
The thing is when real life concepts and idea and vision are reflected in such games but mechanics is different. Game simulates real life with some under the hood mechanics. That is the reality of game design and it is impossible to simulate 100% of every smallest realistic detail. That is what I meant. I don't call for wiping out each and every resemblance of life connection from the game. On the contrary. I love such elements a lot.

Here is an example.
Unit is given 50% bonus do defend against artillery in open. Game helps explains this bonus as "ability to maneuver under the fire". This explanation is a nonsense. Designers did not introduce a "ability to maneuver under the fire" concept first and then decided how much bonus should be. The bonus was introduced first. Most likely to equalize defense against artillery in all kind of terrain: open, rough, base, bunker. I guess they thought it is easier to modify just one case for bombardment than three = more code reuse or something. Then they tried to "explain" their decision in more or less lifelike terms. That is the normal way to build and advertise a game and add a story to it.
Now some people take this "maneuvering under the fire" for real as a game concept and start actually discussing it and develop it and propose mod changes based on this random wording. THAT is what I warn everybody against. Don't go this path.

[Alpha Centauri Bear]

With vanilla rules, I can indefinitely hold off an AI faction of equal power to mine, by dedicating only 10 - 20% of my economy to it, while AI will burn all it's production on the same war.
I want rules that would put me under enough stress, to loose turns advantage against other uninvolved in war AI factions if I get attacked.

True. Removing collateral damage helps AI attack human. However, it also helps human attack AI and in much greater proportions since human will exploit it better. You are solving one problem but creates bigger one at the same time. I am up for the challenge with my both hands but this one in particular is a questionable fix. Did you think about reducing it to 10-20% maybe? That would almost not hurt AI since it does not stack too much of them together but will still prevent human from humongous stacks of doom.

Sorry, I don't follow what ruined Civ series?

1 UPT = 1 Unit Per Tile, from Civ 5 and onwards, introduced because of all casuals complaing about stacks of doom. Civ 4 is the last one I've played and I refuse to even try the new ones on that account alone ( there are plenty other reasons, based on online discussions ). I want to play a strategy game, if I wanted an abstract board game, I'd stick to chess. In a strategy game economy and logistics should win the war and tactics should be able to only trip the balance.

Casual player can't stand when he ends up in unwinable position, while giving him tools to save the day even when confronted with superior power, provides him a great satisfaction. This the design goal behind 1UPT, or stack wipes, which allow wins against many times stronger AIs. What is the point of playing with rules like that, if you are a competent player ? I instead want to be destroyed, if I fail at "build" part of 4X.

Ah, got it now. Wasn't familiar with the term.
100% agree. I always say that mere combat fiddling is nothing compared to sheer number/quality of units. Still you want some combat balance to provide you a cushion of protection from slightly stronger opponent. They should reach certain level of economical superiority before they start breaking through your defense. After that there is no stopping. So holding them tactically, politically or otherwise is the way to protect and develop economical advantage. Both global (economics, logistics) and local (tactics, maneuvering) are important.

[...] With collateral damage I don't see a problem.

In a strategy game of this scale, a confrontation between two stacks of units of equal power, resulting in a total destruction of one side with almost no loses on the other, being solely determined by which side happened to be an attacker is not acceptable to me  and against AI, it'll almost always the player be an attacker and that alone kills most of the challenge in war part of these 4x games imo.

Of course I may not like it in practice, when I actualy try it

You got your tool now. Go and try it. At the same time feel free to propose a fix that'll make it more challenging to player. Like do not apply collateral damage to AI's units but to human's only.

Have you played Thinker mod a little? Or this one since it is based on Thinker?
In Thinker AI is so economically savvy that I am usually a weakest faction for long long long time. That is already pretty challenging. I believe no collateral damage modifications can counter such economical power superiority.

[dino]

Removing collateral damage helps AI attack human. However, it also helps human attack AI and in much greater proportions since human will exploit it better. You are solving one problem but creates bigger one at the same time. I am up for the challenge with my both hands but this one in particular is a questionable fix. Did you think about reducing it to 10-20% maybe? That would almost not hurt AI since it does not stack too much of them together but will still prevent human from humongous stacks of doom.

As I explained, there is little that prevents me from safely using homongenous stacks of doom already and my stacks almost never suffer collateral from AI attacks, while I wipe enemy stacks quiet often. Removing them completely only removes this huge advantage from me, in return I gain an advantage of being able to use slower, but cheaper infantry units with +25% bonus on base assault, instead of only speeder chasis and that is only if I can't produce elite infantry anyway. Without collateral, a disorganised stream of AI units should have more chance of building up at my gates, into a stack that can overcome my defense.

I've played some early games with thinker, it's economy buildup is huge improvement over vanilla, but I still can destroy wastly larger AI forces.

Still you want some combat balance to provide you a cushion of protection from slightly stronger opponent.

This is why I fully agree with increasing armor values, I was only worried that maybe you are increasing them a bit too much in mid-late game.
As for collateral I hope it won't change AI vs AI dynamics much, it should make it easier for AI to conquer a base, but also equally easier to take it back.
Lets rest this discussion for a week, or two, since like you've said, I have my tool and there is nothing more to add, it's time to try it.

[Alpha Centauri Bear]

Yep. Yep. I see your point about collateral damage now.
AI sucked absolutely in vanilla. In Thinker it sucks significantly less against occasional troops. You cannot just sneak a pure attacker to city walls anymore. That adds some challenge. However, it still sucks against fully prepared assault.

[Alpha Centauri Bear]

Removed AQUATIC mineral bonus. Now it seems to get into the desired shape.

[Alpha Centauri Bear]

Version 24.
Automated release procedure from project.

[Alpha Centauri Bear]

Need to rethink cost of non combat units. Unit cost formula primarily targets combat units and it does it well. It is not possible to use a single simple formula to cover for non combat modules cost and effect, though. So they may need to be tweaked specially. Here are my thoughts on what these units should cost.

infantry colony
Still think it should be somewhere 6, maybe 5 but not lower. 4 is too cheap. I played with 6 for long time now and it seems to be on target. Bases do not get depleted of population anymore and overall expansion now is more controlled and not explosive like. Still one can dedicate whole resources for colony building and expand exponentially but this impacts economical development. Players's choice.

speeder colony
9 is fine. It lowers down to 7 with Fusion reactor and then further down to 5 with Quantum. With slower expansion there is still plenty of colonies to build after Fusion reactor discovery.

foil colony
Generally developed sea bases seem to be slightly better than land ones. However, initially they grow very slowly due to lack of minerals. Even with Recycling Tanks included they are stuck on 2 minerals production until sea formers start building mining platforms. Their mineral production stays pathetic until they reach significant size. Surely coastal bases have no problem building hem at this cost but it is an impediment for AQUATIC factions. I think it would be fairer to price it same as infantry one.
scheduled change

infantry former
4 seems to be a cheap enough. This is a lowest cost that doesn't pose problem for +25% abilities. Besides, it is comparatively better than some combat 2-1-1 unit.

speeder former
6 is perfect. It lowers down to 4 with Fusion. Later on clean super rover former will be somewhere 5-7 depending on abilities cost.

foil former
Same considerations as for foil colony. Sea former is more effective in terraforming than its land counterpart. However, limited sea base mineral production makes it difficult to build. So, I guess, it makes sense to price it same as infantry one to reduce initial sea base struggle.
scheduled change

infantry supply
Currently 8 to balance its resource crawling use. Delivering 4 minerals from rocky mine pays off for its price in 20 turns. It is still lucrative but not that insanely when it cost only 3.
I am thinking to raise to 12 maybe to reduce its profitability even more.
Should its cost drop by reactor? I think that is acceptable. It'll match ever-growing base production proportionally.
scheduled change to 12

rover supply and other more expensive modifications
Probably won't be used for mineral supply but for SP building only. In this case it doesn't matter how much it cost.

foil supply
Speed is marginally relevant for reaching to crawling tile and also marginally relevant for SP aiding. With than regard I don't think we need to make it more costly on basis of speed. From another hand sea squares are less attractive for harvesting. Rocky mine provides 4 minerals, 7 with bonus. The best things to harvest on sea are only nutrients (3, 5 with bonus) and energy (3, 5 with bonus). All together there is no point in pricing it more expensive than infantry.
scheduled change

foil transport
Currently 6 but I am thinking whether it is too high cost especially for aquatic factions. From the other side it is a perfect vessel for early pods exploration that can pick up unity rovers and artifacts. Gun foil with cost of 1 can be used as a perfect first line explorer.

cruise transport
That naturally should cost more than foil version due to its speed.

infantry probe team
Is it even needed? I thought it may be cheaper option for defensive probe team that sits in the base all the time. However, it becomes too cumbersome to handle different types of them. Probably should remove predefined design. Human can still design and build it if desired.
scheduled change: remove predesigned unit

speeder probe team
Cost 6 seems to be on target taking further reduction in cost by higher reactors.

foil probe team
It cost same as speeder now. No change needed.

[Alpha Centauri Bear]

Rethinking cost of abilities in light of reactor based cost reduction.

Now with reduced cost some abilities becomes quite cheap. Need to rethink their price based on their corresponding reactor era.

Fungicide Tanks = 1.
Super Former = 3.
Price should reduce with higher reactors.

Hypnotic Trance = 32 (2 flat).
Empath Song = 64 (4 flat).
Shouldn't be affected by reactor same as native units. Price should be tied to native unit cost directly.

Clean Reactor = 32 (2 flat). Has nothing to do with anything but support. 2 flat rows pay off in 20 turns.

Non-Lethal Methods = 64 (4 flat).
Essentially provides second police unit without maintenance. Even better than clean reactor as extends police control beyond POILICE rating.

Algorithmic Enhancement = 4.
Essential enhancement that is going to be reduced by higher reactors anyway.

[Alpha Centauri Bear]

Rethinking technology cost.

There are ~90 technologies and 400 turns to retirements on highest difficulty. Should be no more than 4 turn per acquiring technology. Researching it is just one of the acquiring methods together with pod popping, trading, and stealing. Assuming one gets about half of them in non research way average research rate throughout the game should be somewhere 8 turns on hardest and somewhere 4 on easiest.

Starting conditions are same regardless of difficulty and map size. So should be initial technology cost. Minimal labs at the beginning with two bases is 2 if player is exceptionally unlucky. Otherwise, with rivers/energies/monoliths/SE it could be up to 4. With 3 labs as average staring research power a 12-24 cost tech can be researched in 4-8 turns.

Higher difficulty allows faster development for AI and assumes the same for the player. So difficulty should affect how fast technology cost grows. Somewhere 7% on easiest and 12% on hardest.

Map size determines when expansion stops. It should not affect the rate of exponential price increase but should define when this exponential rate turns into linear and further speed of this linear growth.

Proposed changes.
Replace <map size square root> to <map size> in original formula.
Set first tech cost to 18.
Set tech cost growth rate to 7%-12% based on difficulty.
Limit tech cost to <first tech cost> * <cost growth rate> ^ <number of researched techs>.

Last rule defines exponential growth. It will outgrow default quasi linear formula at some point and won't limit it anymore after that.

[Alpha Centauri Bear]

Tried new tech cost formula. It does meet all criteria.

• Cost of each tech depends on start game parameters only and never changes throughout the game. That 100% eliminates tech trade hold off exploit.
• Late game cost growth is proportional to map area as planned.
• Early game game cost growth between levels is relatively smooth: 20, 70, 220 for first three levels (this is 7 * 3 = 21 technologies, a quarter of all tree).

There is a big BUT, though. This is drastic change in game concept. It is not documented anywhere, new players are unaware of it, UI doesn't highlight it anywhere, blind research engine doesn't take it into attention when choosing tech for research. There are all sort of missing links everywhere in the game. And even if all those links are fixed there is still a conceptual disconnect. Players coming from vanilla need to be educated and trained on it. A huge work that I don't want to do.

[Alpha Centauri Bear]

Here is new formula visualized for those curious. Chart is by virtual tech sequence number assuming whole level got researched before going to next level.
Vanilla chart is calculated using discovered tech cost formula on wiki for highest difficulty (5) and normal map. No correction for time or for being behind. Of course, it is a pretty loose approximation. I've tested it and it seems to match more or less. Starting values are actually somewhere at 20-28, ending are 4200 or so.

My idea was to simulate accelerated growth in the beginning and then more linear after whole map is populated. Accelerated and linear formulas are used and then minimum is picked. Of course, the timing when map is populated depends on map area. You'll see the intersection point moves left and right when you change slope of linear part. That being said this is soooo inexact science that it shouldn't even match the actual development. It'll adjust itself: development or cost will catch up with each other.

I've tried to model accelerated growth with exponent but it seems to grow very sharp at the intersection point. May produce noticeable rate change. The quadratic one is somewhat smoother.

[Alpha Centauri Bear]

Version with revised tech cost is released.

[Alpha Centauri Bear]

Some more pondering about tech cost as it seems to be not a simple fix.

First of all, big variations in tech cost do not seriously break the game due to accelerating labs income. That is how game designers got away with vanilla formula in first place.
Example.
Let's suppose labs income doubles every 50 years. Also let's suppose a player discovers all techs in 200 turns. Let's double all tech costs now. Does it also double research time? Nope. It'll take somewhere 250-300 turns to get them all - you can do the math yourself. Moreover, smaller discrepancies won't be even noticed on the other game variations background (map size, difficulty, game play).

The exact formula is not important. We just want to bend it here and there to remove stagnation pockets and speed-runs. If we want to get 85 techs by turn 300 we need to acquire one tech per about 4 turns. A lot of this acquisitions come from trades. So very very roughly one should able to research a tech per 6-10 turns throughout the game.

Here is my analysis on research pace problem. Let me know if you experienced something different.

• The first few technologies are on the money. They could be researched relatively quick in 4-10 turns.
• After about 5-10 techs their cost keep growing linearly but faction development is still in initial stage. Research rate climbs to 30-50 turns per tech. That is the first problem: early research stagnation.
• In the mid game faction development catches up with tech cost growth and research rate goes back to normal range of 4-10 turns again.
• Development accelerates toward the end game but the tech cost seems to be progressing slower. As a result late technologies got discovered in 2-3 turns easily. Some very advanced factions can discover even few techs per turn. That is the second problem: end game speed-run.

The conclusion is that we need some modified function that accelerates slower at the beginning, about same in the middle, and then faster at the end.

There are also two other concerns: world size and difficulty.

World size does not affect factions growth rate until available space is completely populated and expansion stops. From then on all factions combined progress is more of less linear due to population growth in already existing bases. The speed of such linear progress is roughly proportional to world size, obviously. Of course, this is not exact but rough approximation should do.

Difficulty is the tricky beast. A human player usually does not fall too much behind mainstream research. Trading, buying, and stealing helps to catch up quickly. For the same reason it is difficult to get too far ahead. Lack of trading, buying and stealing targets as well as total hate from other factions makes it difficult to advance further. That leads us to an interesting conclusion that human player research speed is defined mostly by that of other AI players! On a higher difficulties you generally get techs faster! I may speculate here as I never played on easy levels. Please correct me if this contradicts your experience.
Anyway, I think it makes sense to alter human player tech cost with difficulty but we should not alter it for AI players as they set up a research mainstream and everybody should research so many techs in allotted time anyway regardless of difficulty. So the world average research progress should not be affected by difficulty.

Now to my exercises. I tried exponential, quadratic, and cubic approximations. I've adjusted them so that their end game growth rate is the same and about twice as much as vanilla end game one for standard map librarian difficulty (solving problem #2). At the same time I adjusted them so that techs between 10-20 are about 3-5 times cheaper than vanilla (solving problem #1). The end game linear growth is proportional to world size. That defines how long accelerated growth continues before it reach desired end game growth rate. The bigger the world the longer accelerated growth continues.

All three approaches works more or less same for tiny and standard worlds. Mostly because growth stage ends somewhere in the middle. They start to diverge for huge maps as emulated growth stage may continue beyond the end of the game. Exponential is too fast, quadratic is too slow, cubic seems fine.

This being said, all adjustment parameters can be tweaked to match playing experience better. This is a matter of play testing. I am playing with cubic implementation now and I've got 25-30 techs by turn 100 which is absolutely perfect development rate.

[Alpha Centauri Bear]

Bug fix in combat calculation. Please update.