Sooo... Happiness.

[Recursive]

Anyway, Txurce has a point. The extra worker not only makes the game harder, as it should be, but also distorts the early choices for the human player.

I'm not going to make a case for it, as I don't expect to play in such hard levels, but maybe gameplay can be improved by removing the extra worker.

Well, if anyone wants to discuss it, for reference, the AI unit handicap information is:

• Settler-Prince: 1 Warrior

• King: 1 Warrior, 1 Pathfinder

• Emperor: 1 Warrior, 1 Pathfinder, 1 Worker
  
  
• Immortal-Deity: 2 Warriors, 1 Pathfinder, 1 Worker

(The AI always starts with 1 Settler and Pathfinder in addition to this).

Contrast vanilla Deity AI: 2 Settlers, 2 Workers, 3 Warriors, 1 Scout. VP has reduced the bonuses substantially.

This probably isn't the best place to discuss the handicap bonuses, though.

 

[Stalker0]

Anyway, this seems to me like this is a clear case of "practice and get skilled or play at a lower difficulty", or to put it simply, "get good" + ignorance about game mechanics despite them being explained in detail several times to you.

I do think Bite's comments have gotten out of hand, but there is a legitimate debate buried in the mire.

The argument is that, because the happiness system is "comparative", there is a risk that an AI with strong handicaps will have such a comparative advantage that they will provide a strong unhappiness effect on a human player. This has nothing to do with processes, but simply the ability of the AI to build, culture, and tech faster than the human. Further, this is an area that would not occur in AI vs AI battles...which is where the majority of G's testing takes place. This unhappiness effect could then in theory create the happiness spirals that some human players have noted, but would not be noticeable in a lot of testing.

Now this argument has not been proven, but its a legitimate concern.

But the key thing is....G is attempting to address it regardless. His new tech adjustment will mean that if a player falls behind in tech (which often will happen once you start happy spiraling), then their happiness will actually get better. That is key, because it will provide the player a lift to get out of the hole, and will hurt the AIs that tech too far ahead. So it should in theory soften the blow, and weaken the Anti-rubber band that was present in the previous version of the system. We will of course have to wait for the next version to see if this new system works as we hope it will.

 

[ryanmusante]

This probably isn't the best place to discuss the handicap bonuses, though.

There is a thread in the General Balance section. @Stalker0 would probably know the specifics.

 

[Txurce]

Are all happiness benefits capped -- meaning that going past a certain point is useless? Because if so, it would lessen the worry that AI happiness spirals upward when human happiness is given a buff.

 

[Recursive]

Are all happiness benefits capped -- meaning that going past a certain point is useless? Because if so, it would lessen the worry that AI happiness spirals upward when human happiness is given a buff.

Capped at 10 Happiness, minus the addition of Golden Age points to the meter. Going past it isn't entirely useless, as it provides extra insurance in the event cities grow unhappy (which happens!).

I do think Bite's comments have gotten out of hand, but there is a legitimate debate buried in the mire.

I'm not saying there isn't some reasonable points to be made about changing the difficulty handicaps, there's certainly no such thing as perfect balance and there's ways it could be improved.

But G has responded to legitimate concerns about this before - e.g. halving the culture AIs get from the difficulty bonus because it was excessive, and the change in technology happiness is something that others have brought up as well; it's logical, and I'm looking forward to it. The concern regarding AI processes, which he continues to insist on, has however been debunked repeatedly.

I do agree that balancing happiness via the tech median in this way is a good idea, provided the change isn't so excessive it functions like a rubberband mechanic. Guess we'll have to wait and see.

 

[glider1]

I disagree and think the curent version happiness systems are fine because it was good game design.

Why?

Because if you played on too high a difficulty you got whipped quickly.
If you played on too low a difficulty you won quickly.
If you played on the appropriate difficulty, you have a wonderful challenging game that persists right through to the endgame.
If you played on the appropriate difficulty, the happiness system had appropriate flexibility in it to make the game interesting.

It is good to end games sooner rather than later if one side or the other is guaranteed to loose.

Remember the crappy design of vanilla civ? You are guaranteed a win but must play it out for hundreds of turns?
In VP, if the player is winning they win even more quickly because the AI's happiness is crushed as it is for humans.

It is the correct way to design a game. In chess, one bad move and you are out! So it should be in civ (including bad choice of difficulty level).

Too much rubber banding is bad game design and the more there is, the more it shows a fundamental concept problem.

 

[ryanmusante]

In chess, one bad move and you are out! So it should be in civ (including bad choice of difficulty level).

Too much rubber banding is bad game design and the more there is, the more it shows a fundamental concept problem.

Chess is raw strategy with precision and punishing mistakes for a player. Civ is supposed to be a balance of gameplay and strategy with an overall feel that just about anyone can sit down and enjoy. The hexagonal map and battle possibly more chess like. The remaining parts are more like checkers. Stack gold, stack culture, stack faith.

 

[BiteInTheMark]

The effects of the difficulty handicaps aren't secret and have been the case for some time now. As for them "not working as intended"...they work exactly as intended to produce a greater level of challenge, and from what I've seen those who are skilled enough to play Deity regularly are not constantly complaining about the game being too difficult, especially not because of AI processes - as I expected, you completely ignored the detailed explanation from myself and others why processes don't change the median. Anyway, this seems to me like this is a clear case of "practice and get skilled or play at a lower difficulty", or to put it simply, "get good" + ignorance about game mechanics despite them being explained in detail several times to you.

It's not meant to be "balanced". It's meant to be challenging and give the AI an edge because you have a real brain.

I didnt say its too difficult. Iam simply a bit worried, some effects of the handicap (production handicap, yields by events, yields from founding cities....) can get out of the hand and create fluctations in game difficulty. Do you remember the phase of the mod, where birth of great people and digging artifacts create such giant amounts of yields for the AI, which lead to extreme leads of the AI and produce a huge advantage for tall AI civs? Sure, the handicap have to create a challange, but I think the production handicaps where mainly made to compensate losses in wars, but can maybe create a difference in advantages to AI. Like AI starting with worker in flood plains starting immidatly creating farms VS. AI starting with worker in complete forest area with quarry luxuries.

only a small fraction of Production is converted into one type of yield - remember, you'd need 40%+ of cities to run Arts simultaneously to affect Boredom, and that still wouldn't change the other yields.

I dunno why you think this, but this is definitly not true. Can you explain why you think this?
In real, all you need is one city changing its yields. The one with the median value. Or any city below the median, which rises its values over the median. One city is enough to influence it. How much the medial will change if one city make a step over the median is a other question, but one city is enough. (Sorry, I could not leave it that way.)

 

[Gazebo]

I didnt say its too difficult. Iam simply a bit worried, some effects of the handicap (production handicap, yields by events, yields from founding cities....) can get out of the hand and create fluctations in game difficulty. Do you remember the phase of the mod, where birth of great people and digging artifacts create such giant amounts of yields for the AI, which lead to extreme leads of the AI and produce a huge advantage for tall AI civs? Sure, the handicap have to create a challange, but I think the production handicaps where mainly made to compensate losses in wars, but can maybe create a difference in advantages to AI. Like AI starting with worker in flood plains starting immidatly creating farms VS. AI starting with worker in complete forest area with quarry luxuries.

I dunno why you think this, but this is definitly not true. Can you explain why you think this?
In real, all you need is one city changing its yields. The one with the median value. Or any city below the median, which rises its values over the median. One city is enough to influence it. How much the medial will change if one city make a step over the median is a other question, but one city is enough. (Sorry, I could not leave it that way.)

I don't think you understand how the happiness system works - individual cities do not control the median, the yields from all cities are pooled discretely and then selected. And - guess what? - if the current median rises above the median, there's an extremely high chance that another yield closer to the existing median will take over the median slot. Furthermore, the median cannot rise more than .05% of it's value any given turn. So even if all of your special-case stuff were true, it'd take 50+ turns to see the impact, and it would not rise any faster than normal yield increases.

Sorry, but you're just plain, straight, 100% settling in on unstable ground with your argument.

G

 

[CppMaster]

I don't think you understand how the happiness system works - individual cities do not control the median, the yields from all cities are pooled discretely and then selected. And - guess what? - if the current median rises above the median, there's an extremely high chance that another yield closer to the existing median will take over the median slot. Furthermore, the median cannot rise more than .05% of it's value any given turn. So even if all of your special-case stuff were true, it'd take 50+ turns to see the impact, and it would not rise any faster than normal yield increases.

Sorry, but you're just plain, straight, 100% settling in on unstable ground with your argument.

G

Just out of curiosity, have you seen if cities in regards to yields form a Gaussian function? I mean that there are more average cities and less over-performing and less under-performing? In that case it would be even harder to fluctuate median.

 

[BiteInTheMark]

I don't think you understand how the happiness system works - individual cities do not control the median, the yields from all cities are pooled discretely and then selected. And - guess what? - if the current median rises above the median, there's an extremely high chance that another yield closer to the existing median will take over the median slot. Furthermore, the median cannot rise more than .05% of it's value any given turn. So even if all of your special-case stuff were true, it'd take 50+ turns to see the impact, and it would not rise any faster than normal yield increases.

Why cant it rise more than 0.05% in any turn?

Just out of curiosity, have you seen if cities in regards to yields form a Gaussian function? I mean that there are more average cities and less over-performing and less under-performing? In that case it would be even harder to fluctuate median.

I can check that in my current game, if you want, but it should be looking like a gussian function.

 

[wobuffet]

Why cant it rise more than 0.05% in any turn?

Because that's what the code says. Change in baseline yields is capped.

Gaussian function.

Gaussian (or normal) distribution, y'all. And I'd be surprised if it were such a distribution, frankly.

 

[BiteInTheMark]

Because that's what the code says. Change in baseline yields is capped.

Lets say you start with the value of 4 from the beginning of the game, and always increase the value by its maximum from 0.05%.... after 300 turns you would end with the maximum value of:
4 * 1.0005^300 = 4,64716
I dont think thats how it works. Or is there anything else calculation in?

Edit. BTW:
7 of 9 cities england, 4 of 7 cities morocco, 2 of 5 cities sweden, total of 13 from 33 AI cities (CS not counted)

 

[CppMaster]

Gaussian (or normal) distribution, y'all. And I'd be surprised if it were such a distribution, frankly.

Why? What distribution would you expect?

Lets say you start with the value of 4 from the beginning of the game, and always increase the value by its maximum from 0.05%.... after 300 turns you would end with the maximum value of:
4 * 1.0005^300 = 4,64716
I dont think thats how it works. Or is there anything else calculation in?

Maybe he meant 5% instead of 0.05%. Pretty common mistake.

Edit. BTW:

What is that?

 

[BiteInTheMark]

Why? What distribution would you expect?

Maybe he meant 5% instead of 0.05%. Pretty common mistake.

What is that?

1. Iam collecting data and will be maybe able to show you them at 12 o'clock.
2. Gazebo? Mistake? I dont think so, he knows his numbers. But another thought has come to me. When is the first value set? Founding the first city? In this case you have only one population and get 7 gold = 1 pop is producing 7 gold, if your city grows, you immidiatly fall dow to 3.5 gold for every pop. Is the decrease also limited to the given value? (0.05% / 5%)
3. my game as netherlands, just the turn iam taking the data for every city in the world. Its funny to see exactly this in the turn.

 

[BiteInTheMark]

Ok, here are my data from the game (small, earth, emperor, Turn 249, industrial age)

https://www.file-upload.net/download-13186367/test.txt.html
(The city with the X is a puppet, the shown values are the reduced values)

If someone is interested in playing with this data. Would be interesting to see the distribution of the values and the median. I can do the data set again and changing all working process to control how much the median will change.

 

[ryanmusante]

Ok, here are my data from the game (small, earth, emperor, Turn 249, industrial age)
If someone is interested in playing with this data. Would be interesting to see the distribution of the values and the median. I can do the data set again and changing all working process to control how much the median will change.

If you could screenshot up InfoAddict like is done by ilteroi, that could be useful rather than just unsubstantiated conjecture. Also, stick to no cheats and standard speed/size for balanced results.
https://github.com/LoneGazebo/Community-Patch-DLL/issues/4469

 

[Gazebo]

Why? What distribution would you expect?

Maybe he meant 5% instead of 0.05%. Pretty common mistake.

What is that?

Yields are managed in the hundreds, so .05 is correct.

Also @ilteroi has done numerous proper graph studies of the median and they all show very steady median development.

G

 

[CppMaster]

Yields are managed in the hundreds, so .05 is correct.

Also @ilteroi has done numerous proper graph studies of the median and they all show very steady median development.

G

So @BiteInTheMark is right, that after 300 turns, a value can be at most 4.465 (1.005 ^ 300) times higher?

 

[BiteInTheMark]

So @BiteInTheMark is right, that after 300 turns, a value can be at most 4.465 (1.005 ^ 300) times higher?

You forgot a zero.
If we look at the total maximum grow, its like this: 1 * 1.0005^300 = 1,1618 (The 4 was randomly chosen)
If the median can only grow with a maximum of 0.05% per turn, after 300 turns, the median cant rise more than 16,18%, starting with the beginning of the game. Which sounds a bit weird and unrealistic.

Btw. I checked the cities again with the help of firetuner and auto turn (thanks Gazebo), and again.... at turn 268, 12 cities from 33 are using processes.