Caveman 2 Cosmos (ideas/discussions thread)

[pi4t]

That's true, I drew it as an ideal case to simplify and to emphasize on how blaze's graph was wrong rather than draw it how it works in practice as that would further confuse how it works.

But the thing is, your graph doesn't show an ideal case. It shows a case that can't possibly happen, and the reason it can't happen is precisely because of the issue I'm talking about. Whereas Blaze drew a graph which was essentially the right idea. The only thing he got wrong is that he drew the lines at a diagonal instead of doing sharp steps. Which is incorrect, but happens to be irrelevant to the issue I'm raising.
.

The difference in cost here isn't unreasonable, and delaying doing a hurry by up to 7 turns, and then delaying the second hurry up to 15 turns (assuming decay period is 8 turns) just to reduce inflation costs either means it was not important to hurry the production of these things, or it means you have a too weak economy to hurry production when it is needed. Playing this system by only hurry production once every X turns and only the turn before decay happens is possible but it is 100% not the best way to play considering one delays getting stuff to pay less in inflation (which is the whole point of this inflation, to give the player pause when it comes to reckless use of hurry production as it costs money).

I agree that with the specific numbers I used, it's not a problem. I only hurried a small number of things, enough to establish what formula the game is actually using. With this data, we can now calculate what will happen when we hurry larger numbers of things, without having to test it manually by clicking "end turn" hundreds of times.

I also agree that it should cost more to hurry lots of things all at once. My proposal is designed to keep this element as close to the current situation as possible.

I haven't figured that part out yet, I'm mostly correcting some things you said that seemed way off, like inflation being exponential/quadratic and changing each turn when it is actually quite linear and changing only at given turns, then you suggested we should make inflation exponential with this formula suggestion I = kN^2 which very much made me think you had misunderstood something fundamental about how inflation was coded, especially considering your initial sentiment about inflation costing more than you liked.

I mean that the total inflation cost is quadratic. That is to say, the area under the graphs we've been drawing is quadratic. For example, in my test 4, after hurrying three things, I was paying 32gpt for a few turns, then 21gpt for 8 turns, then 10gpt for 8 turns. The total money I lost to inflation in the time it took to drop to 0 is roughly 8*11*(3+2+1). If I had hurried ten items in one turn, I would have been paying something like: 110gpt for a few turns, then 99gpt for 8 turns, then 88gpt for 8 turns, etc, all the way down to 10gpt for 8 turns. The total money I would have lost to inflation would have been 8*11*(10+9+8+7+6+5+4+3+2+1).

If you're familiar with the formula for the triangle numbers, you'll realise that this means that when you hurry N things in quick succession, the total money you lose to inflation will be roughly

8*11* N *(N-1)/2

Which is roughly 44N^2.

especially considering your initial sentiment about inflation costing more than you liked.

I would like to emphasise that it's not exactly that inflation costs more than I'd like. I would be quite happy for inflation to remain the same, or even get higher, in most situations. The problem I have is that, because of a mathematical quirk in the current calculation, there is a specific situation in which inflation just keeps going up and up forever. That situation is when you trying to rush something more than once every 8 turns (on marathon speed) over a long period.

I guess I should fix the issue (of decay only occurring on turns that are dividable by a given integer constant tied to gamespeed), since you're so insisting about it being a problem. Would be easy enough by introducing a vector cache to all players that each time one hurry adds a new element that is X turns away from when the hurry production action was done, and if multiple hurries happen the same turn it won't just keep adding the same turn, it will add the turn now+2X, then now+3X, and so on, as hurry count will only be able to decrement once a turn (i.e. there can't be two identical elements in the vector). So on the start of a players turn the first element in the vector can be compared with current turn number, if equal decrement hurry count and delete first vector element. That way hurry count can be decremented on any arbitrary round rather than on a fixed multiplication table of turn counts.

Meh. That's only a secondary issue. I didn't even know about it when I made my original post, and at this point I think you've talked about it more than I have. It seems to be causing confusion, so I suggest we ignore it for now.

If this would fix your main issue with inflation, then I seriously misunderstood your first post on the issue, I thought you wanted to really change the math from linear to quadratic inflation impact from hurry count, or that you wanted some more serious rewrite of the system as a whole where inflation decay would not be statically incremental like it currently is and would still be with the cache I mentioned (cache only increase accuracy of the decay increments to fit exactly the turn the hurry production occured on) into something more dynamic with increasing rate of decay the higher inflation get or some such.

As you figured out in the next post, this doesn't fix my main issue

I think I finally understood your first post, sorry for being a tard. I now think you were suggesting that instead of inflation slowly decaying on a fixed turn period, all of it should decay away on the first decay point, i.e. if one hurry production 20 times on turn 50 and the decay time is 5 turns, all inflation from those 20 hurries should be gone after 5 turns, rather than take a 100 turn to go away, and to compensate for it lasting much shorter you suggested that the impact of hurry should be exponential rather than linear. This is basically what you suggested, right? My thoughts on that is simply that I oppose the suggestion as it makes it impossible to feel one has runaway inflation, it gives too much feeling of control to the player, makes it too easy to predict and calculate how much inflation one can generate as it will not have long lasting effect even if one miscalculate.

Quadratic rather than exponential, but yes, that's the idea.

I'm having some technical issues with the forum, so I'm going to post this and then write a followup post after.

 

[pi4t]

Ok, technical issues are sorted. I respectfully oppose your opposition on two grounds:

1) Inflation can still run away with you under my proposal. In fact, it runs away with you in a nastier way than it does now. The reason is this: when you hurry N things in one turn, you take the same economic hit with my proposal as you do now. In both cases, the total money you lose over the next 100 turns is roughly some constant times N^2. But with my proposal, you take that hit all in one go over the next (say) five turns, giving you less time to deal with the situation. From a player's perspective, the difference will be that if you overspend and break your economy, the result will be a catastrophic failure rather than a slow decline.

This has got rather theoretical, so perhaps some numbers would help you visualise this. Say that my expenses are such that one "unit" of inflation is 10gpt. Under the current system, when I click the hurry button for the 20th time this turn, my income will go down by 10, just as it did the last 19 times I clicked the button. It will stay 10 down for the next 100 turns, and I will have to deal with the consequences of that. Under the new system, when I click the hurry button for the 20th time, my income will drop by about 200. It will only stay that way for 5 turns, but if I overspent that could be enough to bankrupt me. Which, of course, leads to disappearing units and other long term fun problems. Notice, incidentally, that in both cases I lost about the same amount of money in total: 1000 gold.

The current system is more likely to lead to players going "Oops, I shouldn't have hurried so much stuff 50 turns ago", while the new system is more likely to lead to "Oh, heck, I should not have clicked that hurry button just now." Arguably, the slow burn is better, all other things being equal. I happen to think it is, myself. But they do both do run away inflation, so the loss is smaller than you seem to anticipate.

2) The proposed system will fix a major flaw in the current model. This flaw means that in C2C, when you have large amounts of money and income, the optimal strategy - no matter how much money you're making - is to hurry one thing every (hurry increment) turns and do no more hurrying except in an emergency, and let everything else sit in the treasury for a rainy day. The reasons for the flaw are subtle and I'll try explaining them again below. But before we get bogged down in those details, just suppose for the sake of argument that this flaw exists. Wouldn't fixing it justify this change?



The flaw:

I tried to explain this before, but apparently we talked past each other. So I'm going to try a different way, using graphs again. Let's suppose that the hurry increment is 5 turns, and 1 unit of inflation is 10 gold per turn.

First, let's think about what happens when I have a relatively weak economy. I only manage to stockpile enough money to hurry something once every 10 turns. In order to avoid inflation, I space things out as evenly as possible.

Here's a graph of what the inflation does. The X axis is the turn, the Y axis is the money lost to inflation on that turn. To calculate the total money lost to inflation, we'd need to find the area under the graph.

Spoiler Graph :

As you can see, inflation stays low. It goes up to 10 when I hurry something, but it's already dropped down again by the time I can afford to hurry again. This is working fine. Overall, I pay 5 gold per turn inflation on average.

Now suppose I improve my economy a bit more, and now my treasury is filling faster and I can press the hurry button every 8 turns. What happens now?

Spoiler Graph :

The "multiple of 5" quirk means that the graph is rather more weirdly shaped this time. But inflation still stays low, so this isn't a problem. If you're curious, I pay 3.75 gold per turn. Slightly less than before, but eh, it's 1 gold. Who cares? This situation is still fine, this is not the flaw I talked about above.

Now I improve my economy a bit further, and can now afford to click the hurry button every 5 turns. What happens now?

Spoiler Graph :

Now my inflation is at 10 all the time. This is still fair - I'm hurrying things twice as much, so I'd expect my inflation to double. Obviously, I'm paying 10 gold per turn.

Now what happens if I improve my economy one step further, and can now afford to click the hurry button every 3 turns?

Spoiler Graph :

Oh dear. This time, the inflation isn't double what it was before. It's not triple, or even ten times what it was before. It isn't stablising at all, it just goes up and up forever. This is the flaw I'm talking about.

It doesn't matter how you do it, if you hurry things on average more than once every five turns, the money you lose per turn to inflation will go up and up until you change your strategy. So it doesn't matter how much gold you're making, it is mathematically impossible to hurry things faster than once every 5 turns over a long period. If you try, you'll find yourself losing more and more income to inflation, until you don't have enough money to hurry things more than once every 5 turns any more whether you want to or not.

Rather than doing this, if you're making a long term plan, you're better off just never hurrying things more than once every 5 turns in the first place. If you do that, you're hurrying things at the maximum possible long-term rate, and you're also making a bunch of money every turn that can be spent in diplomacy, or on unit upgrades, or whatever.

Hopefully that makes sense?

 

[Toffer90]

But the thing is, your graph doesn't show an ideal case. It shows a case that can't possibly happen, and the reason it can't happen is precisely because of the issue I'm talking about. Whereas Blaze drew a graph which was essentially the right idea. The only thing he got wrong is that he drew the lines at a diagonal instead of doing sharp steps. Which is incorrect, but happens to be irrelevant to the issue I'm raising.

It is ideal as it ignores implementation specific noise and inaccuracies, it was only meant to correct the sloped graphs of Blaze without introducing factors like inaccuracies in the implementation that would only serve to confuse about the nature of the mechanic; I guess I misinterpreted the audience and caused confusion rather than preventing it by presenting a simplified picture.

Test 4: Hurrying three things in quick succession
click the hurry button. Inflation 10.
click the hurry button again. Inflation 21
click the hurry button a third time. Inflation 32
decay. Inflation 32
decay again: Inflation 21
decay a third time Inflation 10
decay a fourth time Inflation 0

Just want to present another test case to give some nuance.
Noble difficulty, handicap is only option that mess with inflation impact:
At the start of the test, my inflation was 0. I was losing 52 132 gold per turn total expenses and gaining 52 365 gold per turn with a treasury of 14 593 653 gold.

Test 5: Hurrying three things in quick succession
click the hurry button. Inflation 218
click the hurry button again. Inflation 437
click the hurry button a third time. Inflation 662
click the hurry button a fourth time. Inflation 881

At the end of the test I was losing 53 033 gpt total expenses and gaining 51 464 gpt with a treasury of 14 566 170
I.e. I my expenses grew by 901gpt.... which is strange and indicates a bug (likely due to some cache not updating correctly) as treasury upkeep got reduced and inflation can't explain the increase in expense...
(Edit: after a recalculate modifier the inflation expense from 4 hurries were 797 gpt rather than 881 gpt, so there's some miscalcalculation or cache issue afoot indeed.)

Anyhow, I don't mind this extra expense for many turns into the future, it is an annual expense I chose myself, and the point of it is that it should be almost like a permanent expense (though not entirely permanent, just perceived like one in the moment). If the inflation were to be a lot stronger but last for far less time, i.e 5 turns rather than 5→infinite then we could just as well remove inflation and make hurry production far more expensive in itself. If it were to last only 5 turns then even if one goes bankrupt one won't get many turns of STRIKE like one should because the problem self correct in 1-4 turns anyway.

Oh dear. This time, the inflation isn't double what it was before. It's not triple, or even ten times what it was before. It isn't stablising at all, it just goes up and up forever. This is the flaw I'm talking about.

This is what I consider the strength of it, that it is harder to predict and plan around, if expenses unexpectedly shoot up for some reason 50 turns down the line, inflation from back then can come to bite you in the @ss as inflation impact is derived from all other expenses... And it doesn't just go up and up forever, one just have to stop hurrying production if one need more money in the treasury or more money per turn than is currently the case, I really don't see the problem.

Anyhow, I'm signing off this conversation for now as I'm going to go hunting for that bug I found in income and expenses calculation/caching. The inflation cost should really not be quadratic like you found it to be.

 

[pi4t]

And it doesn't just go up and up forever, one just have to stop hurrying production if one need more money in the treasury or more money per turn than is currently the case, I really don't see the problem.

The problem is that if you ever let inflation go back down to 0, you will discover that you've hurried exactly one item for every five (or whatever) turns. It doesn't matter how strong your economy is, whether you spaced them out evenly or rushed a bunch of things in one go and then waited for inflation to go away. Therefore, if you ever intend to let inflation drop to 0 again, there is no reason to let inflation rise at all, even if your gold generation can take the hit. You end up building things at the same rate in the long term, but lose more money to inflation. It's strictly worse than just rushing one item every time inflation drops to 0 and nothing else.

I would like to emphasise that I am not saying that in my opinion, the latter strategy seems better. I am claiming that it can be mathematically proved that rushing one item every 5 turns will land you in a better spot than any other strategy for rushing (unless you need to build a bunch of stuff really urgently).


I'm genuinely not sure why I'm failing to communicate this to you. I'm not sure whether it's the technical details, or just that we're starting with fundamentally different ideas about what rushing is for. So let me sincerely ask you the following question: If I could mathematically prove that you will always end up better off (except in urgent situations) if you never click the rush button until inflation has died away completely, would you say that was a problem that needed fixing? Or would you say that it's working as intended?

 

[philipschall]

why were seasonal/fishing camps disabled? there's no replacement or alternative means of resource discovery.... Will an alternative be implemented soon?

 

[raxo2222]

why were seasonal/fishing camps disabled? there's no replacement or alternative means of resource discovery.... Will an alternative be implemented soon?

These were removed.
Instead of it any worked tile has chance to discover resource. Resources least present on map have highest chance of discovery.
It is separate thing from improvement discovering resource. In this vanilla feature improvements can discover resources, that are relevant for them.

 

[strategyonly]

Q: started a new game, and i am at turn 363 just barely into Ancient era, and then one civ has just entered the CLassical era, that means i am 39, again 39 techs behind, thats insane!!!!!!!!!

 

[raxo2222]

Q: started a new game, and i am at turn 363 just barely into Ancient era, and then one civ has just entered the CLassical era, that means i am 39, again 39 techs behind, thats insane!!!!!!!!!

AI got better. So if you want to play more causal game you can play below Noble.
Tech leader should reach Classical era in 1200 BC.
So looks like AI is bit faster than calendar, which correspond % of turns passed - much more important metric here.

Also indeed dark blue text is unreadable.

 

[strategyonly]

AI got better. So if you want to play more causal game you can play below Noble.
Tech leader should reach Classical era in 1200 BC.
So looks like AI is bit faster than calendar, which correspond % of turns passed - much more important metric here.

Also indeed dark blue text is unreadable.

i know about the AI its kicking my ass .. . with that damn 1 city crap next to my civ all around me,, and i am playing on warlord level . ..

 

[Toffer90]

also can we pls change back to were i can read what it says when taking over a city the BLUE i cant read, sorry. . . .

Better?

 

[Somebody613]

i know about the AI its kicking my ass .. . with that damn 1 city crap next to my civ all around me,, and i am playing on warlord level . ..

Which is why:
Always go straight for Neanderthal units, if and when possible.
Then go for at least Spiked Clubmen.
Then take over (or destroy) at least several nearby cities (capture is limited due to penalties, razing is not).
I'm now playing an "easy" (option-wise) game on Pit's map, and have Jerusalem at size 6, while I already killed every other Middle Eastern civ (all cities were size 1, so it auto-razed them).
This got me 4 Captive_Civilians (random luck-based result upon razing a city, could've been more or less), which gave me +4 city size (and +1 more from Venus).
This way, I get no penalties from multiple cities, while getting a good boost to my capital (including more available citizens for Priests).
Note that this requires playing with Merging, but I see no reason why anyone (except Pit) NOT use that option, lol.

 

[strategyonly]

Which is why:
Always go straight for Neanderthal units, if and when possible.
Then go for at least Spiked Clubmen.
Then take over (or destroy) at least several nearby cities (capture is limited due to penalties, razing is not).
I'm now playing an "easy" (option-wise) game on Pit's map, and have Jerusalem at size 6, while I already killed every other Middle Eastern civ (all cities were size 1, so it auto-razed them).
This got me 4 Captive_Civilians (random luck-based result upon razing a city, could've been more or less), which gave me +4 city size (and +1 more from Venus).
This way, I get no penalties from multiple cities, while getting a good boost to my capital (including more available citizens for Priests).
Note that this requires playing with Merging, but I see no reason why anyone (except Pit) NOT use that option, lol.

always do that anyways, but to no avail most of the time. . .

 

[Somebody613]

always do that anyways, but to no avail most of the time. . .

How exactly is AI beating you in the tech race?
Oh, waaait, Jerusalem, yeah - the "Dead Sea scrolls" cheat (not really a cheat), loool.
And Stone literally at the door now, right.
Though that's a bit debatable, since it's unrelated to units getting Free Promos and anything else that usually makes the difference.
What are your usual goalposts, if you have any specific ones that you always try going for?

 

[JosEPh_II]

i know about the AI its kicking my ass .. . with that damn 1 city crap next to my civ all around me,, and i am playing on warlord level . ..

So you have the Game Option Realistic Culture Spread On? the North/South and Eas/West axis is something you really have to think about before placing any city while also using the 1 City Start Option. I thought is was "cool" the 1st couple of Test games. Now... I'm beginning to question if the old way for RCS for what tiles of the next ring (8 surrounding tiles) was not more "Realistic". With RCS before you did not automatically get all 8 surrounding tiles. As you do when you have RCS off.

And don't blame 1 City Tile Start for the AI's Enhanced desire to cut you off as fast as it can! That was a separate piece of coding Toffer did. You get neighbors Much faster Now that you ever did with or without RCS On. On using 1 City Tile vs Not using it. 1 City Tile does not deserve being called the culprit in that at all.

Blazenclaw has refined his Culture changes somewhat now. And with RCS Off city placement is a bit more "relaxed" on where to place the city( especially your starting city). But even so, the "new code for AI City Placement", the AI and NPCs are going to try to hem you (the Player) in as fast as they can now. Every game I start I have to fight a barb city asap or get cut off Until I can get a tech like Rafts to get me around them. And the Number of Mil Unit the barb or Neanderthal city can produce is like Wow! I counted 65 units just in one stack! (it had multiple stacks) When I started a new game the other day on Emperor Difficulty. You must go to War almost from the getgo now.

 

[Blazenclaw]

Now... I'm beginning to question if the old way for RCS for what tiles of the next ring (8 surrounding tiles) was not more "Realistic".

What was the old way? Before I did my recent changes, I'm pretty sure RCS would get all 8 tiles on first expansion if 1TS was off.

And yes, 1TS+RCS is meant to be a much harsher experience; RCS by itself is more forgiving.
Edit: Next SVN has this altered a bit. RCS with or without 1TS is harsh. RCS+Min City Border (and 1TS off) is what gives a middle ground, where you quickly get 1st/2nd ring but slow past that.

Also: For next SVN (savebreak), I organized the gameoptions, which required some minor/moderate renaming in places; hopefully it won't be too much of an issue as it should be much more clear now, but just something to be ready for.

 

[strategyonly]

Every game I start I have to fight a barb city asap or get cut off Until I can get a tech like Rafts to get me around them. And the Number of Mil Unit the barb or Neanderthal city can produce is like Wow! I counted 65 units just in one stack! (it had multiple stacks) When I started a new game the other day on Emperor Difficulty. You must go to War almost from the getgo now.

i dont play with any oceans, just dryland. . .

 

[JosEPh_II]

I'm pretty sure RCS would get all 8 tiles on first expansion if 1TS was off.

Only if you had the option Minimum City Border On. Minimum Borders and 1TS are opposites of each other. Minimum Borders guaranteed that with RCS On, you get all 8 tiles around the start/main city tile.
I have not looked lately to see if someone removed Minimum Borders Option or not. I never used it. But we had players that complained long ago that, even with RCS On, a city Must get all 8 tiles around the Main tile. The Minumum Borders Option was borrowed from AND2.

 

[JosEPh_II]

i dont play with any oceans, just dryland. . .

So you are saying that using just land you don't get cut off from expanding your borders? Yet you have neighboring AI?

 

[Thunderbrd]

So you are saying that using just land you don't get cut off from expanding your borders? Yet you have neighboring AI?

A little background here - Koshling made the AI super aggressive about this strategy but at the time he did it, they wouldn't consider settling into adj spots. Still, it was his player level settling weight that gives so much strength to this strategy - and it's largely because he himself was VERY good at it as you and I both saw (me twice because the first game he stuffed me into a corner on our continent QUICK). I personally like to grow more like a blooming flower from the center out and don't even LIKE rushing border conflicts if I can avoid it (but I reserve lots of land per city). In a perfect future, we should have different leaders using these differing varieties of strategies I think.

 

[JosEPh_II]

@Thunderbrd
Almost every new game I have started over this winter I have had to fight a Barb or Neanderthal city 1st to get a path to the main body of the landmass I am on. While the AI is rapidly sucking up the territory I'm being blocked from. Makes for a tense Preh and Ancient eras struggle. You would be very happy with this.