Caveman 2 Cosmos

[Toffer90]

This change was to unify building/unit/tech costs.

That is imo only important for human players, unifying it for the AI isn't all that important as the AI will never complain about how the ratios feel in game. It won't complain about having built all buildings all the time because buildings cost too little compared to techs and it won't complain about the logic regarding how much a unit cost compared to this or that building.

I feel it is better, on the higher difficulties, to give the AI a bigger boost on building production, and an even bigger boost on unit productions, than it gets towards research times. Cheaper buildings indirectly boost research times anyhow, and super cheap units makes the AI able to waste more units (As it always does), and to be a force to be reckoned with at war even if the human player is slightly ahead in military techs. Cheaper units also indirectly boost research times as the AI will be able to expand faster because it can train, at a faster pace, all the units it feels it needs before it's ready to expand. The other factors like all the economical benefits the AI gets from the handicap in use also indirectly boost research times as the AI will be able to expand its nation faster than otherwise possible.

The hold up seems more to be that they are not economically strong enough to spread because they aren't as well established on easier gamespeeds when they hit tribalism and don't get there until later. That's about the only reason they would disband a settler which they've been doing, training them then disbanding them.

Then we should focus more on making the AI more economically smart in general.
Noble difficulty is defined as the difficulty where AI and human player is both completely without handicaps and therefore on equal footing in every way. The alternatives would be to make the game much more economically easier for both human and AI players regardless of difficulty in use; or to break with the "equal footing on Noble" tradition and give the AI economical benefits over the human player when Noble difficulty is used by the human player.

 

[Thunderbrd]

That is imo only important for human players, unifying it for the AI isn't all that important as the AI will never complain about how the ratios feel in game.

True you won't get complaints from the AI but you will from players on lower game difficulties who feel the AI should be performing better while you're on a harder setting seeing them perform quite well.

It won't complain about having built all buildings all the time because buildings cost too little compared to techs and it won't complain about the logic regarding how much a unit cost compared to this or that building.

Complain or not it can still get fouled up in the timing of how it goes about things (like spreading cities and building attack forces)

I feel it is better, on the higher difficulties, to give the AI a bigger boost on building production, and an even bigger boost on unit productions, than it gets towards research times.

If you need to give them a better boost on the harder levels than the easier levels need the same ratio.

Cheaper buildings indirectly boost research times anyhow, and super cheap units makes the AI able to waste more units (As it always does), and to be a force to be reckoned with at war even if the human player is slightly ahead in military techs. Cheaper units also indirectly boost research times as the AI will be able to expand faster because it can train, at a faster pace, all the units it feels it needs before expanding its ready to expand. The other factors like all the economical benefits the AI gets from the handicap in use also indirectly boost research times as the AI will be able to expand its nation faster than otherwise possible.

All true, so how is it an argument that it shouldn't get this successful established ratio on easier game settings? I mean, I guess I can see how it creates a stronger ramp. Do they match the player on this ratio on Noble level? That would be a valid argument to then allow them to get an easier ratio as the game difficulty gets tougher for the player.

Then we should focus more on making the AI more economically smart in general.

They aren't economically stupid though, at least not to the degree that an AI can be. They aren't failing to address their needs appropriately, they just do it clumsily because a human player is more capable of deeper insight of a ton of variables. There are limits to how intelligent an AI can get without it taking a day to process a turn. Plus, I'm done with AI work for a while. And with Alberts2 going back on awol status, this isn't an approach we'll be seeing for a while unless someone else figures out how to code AI. I'm sick of getting stuck on projects for months that should've taken a day or two. My appetite for cans of worms is gone for now.

Noble difficulty is defined as the difficulty where AI and human player is both completely without handicaps and therefore on equal footing in every way. The alternatives would be to make the game much more economically easier for both human and AI players regardless of difficulty in use; or to break with the "equal footing on Noble" tradition and give the AI economical benefits over the human player when Noble difficulty is used by the human player.

Yeah, I get this. Players should thus play harder levels to account for the AI not being as capable.

 

[Toffer90]

True you won't get complaints from the AI but you will from players on lower game difficulties who feel the AI should be performing better while you're on a harder setting seeing them perform quite well.

Isn't that the whole point of difficulties? That the AI should perform better on higher difficulties.

Complain or not it can still get fouled up in the timing of how it goes about things (like spreading cities and building attack forces)

I'm not so sure about that being a big issue really... I would need to hear an example illustrating how the AI would do things in an irrational order if unit are too hammer expensive compared to buildings, that the AI cannot cope with the current unit to building hammer ratios that the huiman player currently experience in a game.

All true, so how is it an argument that it shouldn't get this successful established ratio on easier game settings?

Because I assume Noble difficulty means that the AI and human player are on equal footing, this would mean that regardless of what "Building to unit hammer cost ratio" is set up for the AI and human player on Noble difficulty the human player will always expand long before the AI (see quote below for a good argument for this hypothesis)...

They aren't economically stupid though, at least not to the degree that an AI can be. They aren't failing to address their needs appropriately, they just do it clumsily because a human player is more capable of deeper insight of a ton of variables.

There are limits to how intelligent an AI can get without it taking a day to process a turn. Plus, I'm done with AI work for a while. And with Alberts2 going back on awol status, this isn't an approach we'll be seeing for a while unless someone else figures out how to code AI. I'm sick of getting stuck on projects for months that should've taken a day or two. My appetite for cans of worms is gone for now.

I don't feel that the "game is too easy on Noble difficulty" complaint is something that needs to be addressed right away, it has been an issue in C2C one way or another for a long time.
I did propose to remove the handicap levels below Noble, and add one new one above Noble, some months ago but it was not a popular suggestion.

 

[Taxman66]

From an impartial point of view, it doesnt really matter if the vanilla concept of Noble being equal footing in regards to AI bonuses is kept in C2C.

I suggest a group of settings where Noble has the AI act as competitively in C2C as it does in Vanilla. Then have a few settings above and below that. Perhaps a little blurb stating this concept to be added to the notes as well.

What you really want to avoid are settings, espicially multiple settings, where the AI doesn't offer any challenge against even the most casual and careless player; And lets be honest, this mod is not going to be that type of player's cup of tea anyway.

 

[Thunderbrd]

Isn't that the whole point of difficulties? That the AI should perform better on higher difficulties

True, but I would think they should still be setup at all levels to be able to spread as soon as they get to tribalism like a decent player would do.

I'm not so sure about that being a big issue really... I would need to hear an example illustrating how the AI would do things in an irrational order if unit are too hammer expensive compared to buildings, that the AI cannot cope with the current unit to building hammer ratios that the huiman player currently experience in a game.

The AI may over-invest into some things like production buildings and research buildings before it feels ready to spread and then when it does create a settler it hasn't had enough time to build gold buildings so it cannot support having a new city yet and it lasts that way long enough to then disband the settler.

The human, understanding how to walk finer wires would stop constructing even production and research creating buildings to pause to build the settler and the accompanying military to go with it AND would then understand that it needs to build some gold producing buildings and THEN come back around to production and research when it can afford to, knowing that the spreading of the next city may mean an immediate impediment to research obtainment and may slow down construction and unit training a bit for now but in the end the next city is a stronger investment.

The AI cannot make judgement calls like this. It only knows that if it can construct a production giving building, it's generally more important than a food giving building which is generally more important than a research giving building (unless falling behind in research), which is generally more important than a gold producing building (unless in a gold crisis). One of the first things it evaluates is if it has enough settlers and escorts for them. And it won't train either if it's in gold crisis, but if it goes into gold crisis after those units are trained, which they very well can right now due to rather strong unit upkeep costs, then it won't be stupid enough to send the settler out.

The human would go ahead and send it out and when he experiences a round that would put him in strike at 0% research, he'd put the city on a gold producing process until his treasury is high enough to allow some margin to go back to constructing what it takes to get out of the gold crisis.

Crime can also complicate this because once in crime it can be much more expensive to get out of it, and that, too, can get the AI stuck in the process of having a Settler that never completes getting ready to deploy.

Because I assume Noble difficulty means that the AI and human player are on equal footing, this would mean that regardless of what "Building to unit hammer cost ratio" is set up for the AI and human player on Noble difficulty the human player will always expand long before the AI (see quote below for a good argument for this hypothesis)...

Yeah... you do make a good point here. I'd prefer the same. The ratio is working pretty good as a player challenge right now - feels about right. But it IS a bit more challenging to the AI than the player due to the increased complexity it demands. This is a good discussion. I'm not sure where I actually fall on this. I'm tempted to agree that the answer to someone complaining that the AI isn't spreading until a while after Tribalism would be... play a harder level - that's part of what makes the game easier and no they don't PLAY as well as a human player but at noble they have the same conditions AS a player and as a result it becomes fairly obvious that they just aren't as skilled as even a casual player would be so even that casual player may wish to play a harder level with some handicaps in place.

From an impartial point of view, it doesnt really matter if the vanilla concept of Noble being equal footing in regards to AI bonuses is kept in C2C.

I suggest a group of settings where Noble has the AI act as competitively in C2C as it does in Vanilla. Then have a few settings above and below that. Perhaps a little blurb stating this concept to be added to the notes as well.

What you really want to avoid are settings, espicially multiple settings, where the AI doesn't offer any challenge against even the most casual and careless player; And lets be honest, this mod is not going to be that type of player's cup of tea anyway.

Thank you for sharing your opinion on the matter. It is also a valid side of looking at things.

 

[cakedaemon]

Question, but is the latest SVN compatible with saves from SVN 10089? I've just updated my save, but when I try to recalculate the modifiers the game just completely locked up on me, then unceremoniously crashed some eight or so minutes later. .

EDIT: That's odd. Had a second try at it, and it recalculated fine. Nevermind then, must've been some one off glitch!

 

[Toffer90]

From an impartial point of view, it doesnt really matter if the vanilla concept of Noble being equal footing in regards to AI bonuses is kept in C2C.

I suggest a group of settings where Noble has the AI act as competitively in C2C as it does in Vanilla. Then have a few settings above and below that. Perhaps a little blurb stating this concept to be added to the notes as well.

What you really want to avoid are settings, espicially multiple settings, where the AI doesn't offer any challenge against even the most casual and careless player; And lets be honest, this mod is not going to be that type of player's cup of tea anyway.

I've considered the same myself, but I always end up thinking that there is a very good debugging value in having a difficulty where the human player can evaluate AI performance when there are no handicaps either way.

My suggestion to remove "chieftain & settler", and to move warlord between Noble and Prince was quite unpopular.
Maybe it would be more popular to suggest that the AI should play settler difficulty and that the "settler" handicap should be the point where Human and AI are on equal footing?

True, but I would think they should still be setup at all levels to be able to spread as soon as they get to tribalism like a decent player would do.

Assuming a lack of gold is the main culprit for the AI not to expand on the easiest difficulties, then we would need to ensure that there is plenty of gold for all players in the early game (particularly when hitting tribalism).

This would most efficiently be achieved if there was python code that awarded "100 gold multiplied with gamespeed" to players the moment they invent Tribalism. (It would be 1000 gold on Eternity and 50 on blitz).
The code could also evaluate difficulty so that the amount of gold awarded will be less the higher the game difficulty is at (defined by human player difficulty only), then no players will get any gold when human player is playing on Nightmare difficulty.
Or should it perhaps only consider the difficulty of the specific player so that AI get their "100 gold multiplied with gamespeed" regardless of what difficulty the human player use, and the human player won't get squat on Nightmare difficulty?​

I want to avoid doing general economical boosts across the board to all players in XML as then we will quickly exasperate the "too much gold" issue.

The AI may over-invest into some things like production buildings and research buildings before it feels ready to spread and then when it does create a settler it hasn't had enough time to build gold buildings so it cannot support having a new city yet and it lasts that way long enough to then disband the settler.

The human, understanding how to walk finer wires would stop constructing even production and research creating buildings to pause to build the settler and the accompanying military to go with it AND would then understand that it needs to build some gold producing buildings and THEN come back around to production and research when it can afford to, knowing that the spreading of the next city may mean an immediate impediment to research obtainment and may slow down construction and unit training a bit for now but in the end the next city is a stronger investment.

The AI cannot make judgement calls like this. It only knows that if it can construct a production giving building, it's generally more important than a food giving building which is generally more important than a research giving building (unless falling behind in research), which is generally more important than a gold producing building (unless in a gold crisis). One of the first things it evaluates is if it has enough settlers and escorts for them. And it won't train either if it's in gold crisis, but if it goes into gold crisis after those units are trained, which they very well can right now due to rather strong unit upkeep costs, then it won't be stupid enough to send the settler out.

The human would go ahead and send it out and when he experiences a round that would put him in strike at 0% research, he'd put the city on a gold producing process until his treasury is high enough to allow some margin to go back to constructing what it takes to get out of the gold crisis.

Crime can also complicate this because once in crime it can be much more expensive to get out of it, and that, too, can get the AI stuck in the process of having a Settler that never completes getting ready to deploy.

I would like to look at what economical factors the AI considers before feeling ready to train a settler one day, it sounds like there should be a better evaluation of national economy in that consideration than what it currently have.
There might have to be a stricter economical evaluation for training the settler, and escorts, than what there is for actually sending the settler out to found the city, an economical buffer zone so to speak, to allow for a worsening of the economy between finishing the settler and actually sending it out.
Not something I want to look at right away, the python solution I presented above could be a good solution for now.

This is a good discussion.

Agreed, I felt it was necessary when I saw the solution raxo presented, it was not a bad suggestion in itself but I felt it was a bit inaccurate and therefore unwarranted.
It is usually better to discuss before acting than to act prematurely, regardless of what we agree on in the end.

 

[Taxman66]

My point is there appears to be too much of a hang up over the 'names' of the skill levels.

If there is a level where there are no handicaps (either way), then so be it. Said level doesn't have to be called 'Noble' and it doesn't have to have the same number of levels above and below it. If zero handicaps is a poor challenge, but not completely useless then put it at or near the bottom (with a position appropriate name). If it is completely hopeless even against careless rookies then does it really need to be there? If you still want it for debuging or an All/Only AI test then put it at the bottom and call it something like 'Embarrassingly Easy'.

 

[Thunderbrd]

I want to avoid doing general economical boosts across the board to all players in XML as then we will quickly exasperate the "too much gold" issue.

One shot boosts aren't going to help because it evaluates its gold status by it's income vs budget rather than by how much it has. Players tend to enjoy a more gold stressful scenario than is healthy for an AI. That's a catch 22 that makes it truly tough to make a decent AI function as well as it should with the same level of strain given to the player.

Maybe it would be more popular to suggest that the AI should play settler difficulty and that the "settler" handicap should be the point where Human and AI are on equal footing?

I would probably be alright with that.

There might have to be a stricter economical evaluation for training the settler, and escorts, than what there is for actually sending the settler out to found the city, an economical buffer zone so to speak, to allow for a worsening of the economy between finishing the settler and actually sending it out.

I've often thought there should be more than one layer of recognition of income stress rather than the one that currently exists. FYI I think it's something like 40% gold or less on the slider without being in the positive is the definition of 'economic stress' - something like that. There IS a global that can adjust this definition somewhere.

The same point defines when it should stop training most units too. It also is used as a control that past a point in the priority hierarchy it'll consider going on Gold production if I recall but there've been some changes to that.

 

[Toffer90]

One shot boosts aren't going to help because it evaluates its gold status by it's income vs budget rather than by how much it has. Players tend to enjoy a more gold stressful scenario than is healthy for an AI. That's a catch 22 that makes it truly tough to make a decent AI function as well as it should with the same level of strain given to the player.

I suspected as much.... hmm...

I would probably be alright with that.

If no one object to it then I'll implement the change in 10-20 days.

I've often thought there should be more than one layer of recognition of income stress rather than the one that currently exists. FYI I think it's something like 40% gold or less on the slider without being in the positive is the definition of 'economic stress' - something like that. There IS a global that can adjust this definition somewhere.

The same point defines when it should stop training most units too. It also is used as a control that past a point in the priority hierarchy it'll consider going on Gold production if I recall but there've been some changes to that.

If that one global is the only layer used in economical AI considerations then it explains the rigidity of the AI code in that area. It is quite possible to complicate the consideration a bit without it increasing turn times noticeably, but that is imo not something we should stress about right now; low priority. If you feel like working on some related code one day, give me a holler, and I'll assist to the best of my abilities.

 

[Thunderbrd]

I suspected as much.... hmm...
If no one object to it then I'll implement the change in 10-20 days.
If that one global is the only layer used in economical AI considerations then it explains the rigidity of the AI code in that area. It is quite possible to complicate the consideration a bit without it increasing turn times noticeably, but that is imo not something we should stress about right now; low priority. If you feel like working on some related code one day, give me a holler, and I'll assist to the best of my abilities.

Agreed. I will let you know.

 

[tmv]

Maybe it would be more popular to suggest that the AI should play settler difficulty and that the "settler" handicap should be the point where Human and AI are on equal footing?

Much better IMO.

 

[Dancing Hoskuld]

Maybe it would be more popular to suggest that the AI should play settler difficulty and that the "settler" handicap should be the point where Human and AI are on equal footing?

And confuse veteran Civ (any version) players even more. Noble has been the standard "level playing field" over all versions of Civ.

I want to avoid doing general economical boosts across the board to all players in XML as then we will quickly exasperate the "too much gold" issue.

With pre v38 it was difficult to get gold even playing on Noble. With v38 and v38.5 it is trivial. Thus we already have a "too much gold" problem. As far as I can see it is mainly because of the change to the cost of subdued animals. I have hundreds of the things and am still raking in the money.

 

[Taxman66]

DH, a brief amount of confusion for c2c newbies is, IMO, a much lesser evil than a group of multiple difficulty selections that are not competitive to even the most casual player. Additionally, as I've said before, this mod does not cater to that type of player anyway.

Just as important is that vetern players are going to expect a level of compitentcy from Noble (i.e. mid level) difficulty that just isn't there with a no adjustments game.

You can also mitigate that confusion with a comments in various places (download notes, tool tips, web pages etc...)

 

[Noriad2]

My suggestion to remove "chieftain & settler", and to move warlord between Noble and Prince was quite unpopular.
Maybe it would be more popular to suggest that the AI should play settler difficulty and that the "settler" handicap should be the point where Human and AI are on equal footing?

I still think that having Nightmare as a game option is unpolished. Having it as an option looks like a work-in-progress. There should be one place to set the difficulty, and that is the settler-to-deity pulldown menu. Also, I just saw another youtube Let's play of an average competent player who complained that anything below deity is not worth playing as you just roflstomp the AI.

In my opinion,

There should be one level where the AI is completely passive for new players who just want to check out the game, the rules and the eras without risk of the AI attacking them. This should be "settler", as it is now

Most people learn civ 4 from Beyond the Sword, and only then they check out mods like C2C. So there is no need to keep a large number of learning levels in C2C. To be honest, people who are so bad at civ 4 they can't handle C2C Chieftain after learning the C2C rules on Settler are unlikely to install C2C in the first place.

Thus, I'd drop either chieftain or warlord, but don't change any names. I'd shift one place down all the difficulty parameters of levels higher than the one dropped, and move current Nightmare parameters into the new deity and remove the nightmare option . Then, the AI should play standard on warlord instead of noble.

 

[raxo2222]

I still think that having Nightmare as a game option is unpolished. Having it as an option looks like a work-in-progress. There should be one place to set the difficulty, and that is the settler-to-deity pulldown menu. Also, I just saw another youtube Let's play of an average competent player who complained that anything below deity is not worth playing as you just roflstomp the AI.

If it was changed so it is just like any other handicap, then Nightmare could be accessible by flexible difficulty BUG option too.

Edit: Doing such change will break existing games.
At best Nightmare games would fall back to regular Deity games.

            <Type>HANDICAP_DEITY</Type>
            <ReplacementID>HANDICAP_SO_DEITY</ReplacementID>
            <ReplacementCondition>
                <Has>
                    <GOMType>GOM_OPTION</GOMType>
                    <ID>GAMEOPTION_NIGHTMARE_MODE</ID>
                </Has>
            </ReplacementCondition>

This is how Nightmare handicap is defined.
It is in module by the way.

 

[Thunderbrd]

With pre v38 it was difficult to get gold even playing on Noble. With v38 and v38.5 it is trivial. Thus we already have a "too much gold" problem. As far as I can see it is mainly because of the change to the cost of subdued animals. I have hundreds of the things and am still raking in the money.

Wow... NOT my experience at all. With every city I've had to wait a significant amount of time to recover my economy so I can grow forward and at 11 cities I'm still hovering at 75% research. Interesting variations I'm hearing here.

 

[raxo2222]

Wow... NOT my experience at all. With every city I've had to wait a significant amount of time to recover my economy so I can grow forward and at 11 cities I'm still hovering at 75% research. Interesting variations I'm hearing here.

Maybe he is building all gold giving buildings?
On what game speed you are playing?

On slower game speeds its easier to accumulate gold.

 

[Thunderbrd]

Maybe he is building all gold giving buildings?
On what game speed you are playing?

On slower game speeds its easier to accumulate gold.

Monarch Diff, Epic Speed. The AI and I are pretty close to the same # of cities throughout most of this game so far as well.

 

[raxo2222]

Monarch Diff, Epic Speed. The AI and I are pretty close to the same # of cities throughout most of this game so far as well.

Hmm so two levels above Noble.
Epic has 6000 turns.
On what map size do you play?

@Dancing Hoskuld
On what map size/handicap/speed you are playing?