Analysis: The Opening Game Unraveled

[AndrewDJ]

For purposes of a .15d retest, instead of, or in addition to, using a WorldBuilder edited Fractal Map, you might also want to consider using one of the more esoteric map scripts. I think there's one that will generate either two or four mirror-imaged continents (meant for multiplayer, to ensure a balanced start, IIRC).

In that situation, the only random factor, I think, would be goodie hut pops and barbarian activity, and all the civs, player and AI, would have essentially the same starting plots and resources.

Of course, my memory of the mapscript in question could be totally wrong.

I should add that I started out not really having an opinion on whether or not the Ljosalfar are overpowered, but looking at your analysis, I'd be inclined to agree that they need, at minimum, serious balance tweaking.

 

[TheCowSaysMoooo]

I'm pretty familiar with creating a WBS file from scratch, and will gladly help Unser with the creating a new test map if it's needed, as well as changing the civs, etc for testing purposes. If the help is wanted, hit me up with a PM, and we'll hash out what is desired of the map.

 

[Ghostmaker]

Okay, after checking the saves and reading trough your notes, here are my post from the point of an elven defender.

I mainly compared Arendel, Khandros and Hannah. The reason for this is easy to explain, I never played Clan before and find it hard to comment on this without knowledge of the civ.
So here are my points.
All comparisons are done on the basis of the 200 AD save.

First I compared the economic output:

Arendel
84 beakers / 42 gold / 47 expenses
NOTE!! Because of low treasury you have to reduce research next turn. So the real values are:
73 beakers / 55 gold / 47 expenses

Khandros
85 beakers / 25 gold / 20 expenses

Hannah
77 beakers / 25 gold / 22 expenses


At this snapshot the elves are on par with the other civs. This is also proved by your study. However if we read your notes we found out that you went "worker first" with the elves, resulting in a 48 turn delay of cottage build for the Kazad and a 33 turn delay for the lanun.
I then went to the save of the Kazad and checked every existing cottage for them. Providing they could have started at the same time as the Elves they would have 5 towns, 6 villages and 1 hamlet as compared to 0 towns, 8 Villages 3 Hamlets and 1 cottage. If both races had started at the same time then the dwarf would have a good research bonus compared to the oh so strong elves. Also keep in mind that at the same time the dwarf were able to accumulate around 1500 gold, when the elven vault is empty. Even the lanun were able to keep up with only 9 cottages and a far from optimal use of city spots and coastal useage.

=> Result: Elves are not stronger then any other civ in economic worth.


Second comparison is the shield output:

Arendel
57 basic production
NOTE!! From this production 22 shield are out of cottaged woods.

Khandros
40 basic production
NOTE!! This is basic production. For their full vault the kazad are getting 9 shields more.

Hannah
22 basic production

Wow, this looks like a clear winner for the elves, doesn't it. Let's take a closer look. We are talking here about a difference of 8 shields (49 Khandros to 57 Arendel). These 8 shields are won out of 2 more cities and 11 more population, so this looks quite fair to me. The elves are bigger and more numerous so they have better production but they pay for it in reduced research and higher upkeep. (see first point).
Also just theoreticly the dwarfs can outproduce the elves at this point of the game with just 2 more pop points working mines. I will not say that there is no production bonus for elves in the beginning, but it becomes irrelevant by midgame latest.
As for Hannah there is not even a slight change that they become production leaders in the beginning with their playstyle. They are completly geared for trade, not for production.

=> Result: While elves have a production bonus at start it is not benefiting them in mid-/lategame. There is no production dominance because of this in later stages of the game.

Other Facts:
You are putting the elves as military leaders compared to the other races because of their higher basestrength. This is a total wrong idea. Most of your elven units don't have experience at all, your 4 Hunters are useless at the moment. On the other hand the dwarfs for example have highly trained units of warriors often with 5* and movement bonuses. Please take this into account when judging the military might. In my opinon they have about the same strength with better potential for the dwarfs, as they already know bronze working.

So much for your games for now. I will now continue with your review.

Ljosalfar:
There have been many criticisms of the Ljosalfar results, generally involving a presumed total lack of Barbarian interference. I addresses the Barbarian matter in some detail on a Page 5 message. After going through my notes, I am comfortable that the number of turns lost to barbarian activity is roughly comparable between game to game. Quite frankly, the Raging Barbarians are not so much a threat in FfH, as they are a game of Tetris. (Sorry, too good a line to not use.) Furthermore, the close AI civ also got to two city sites I wanted. Much of the criticism could be filed in this category. Only one side of a two-edged sword was used to cut.

But there are a few legitimate, tangible differences that provided Ljosalfar some small benefits. Their capitol was founded on a Hill/Plains. That situation means the city tile produces a 2nd hammer … very nice for Year 1. Also, the capitol was able to work an Oasis. 3 Food and 2 Commerce is also nice for turn 1. Also the presence of the close AI civ meant for a small advantage no one cited. Each new trade route generated +2 commerce instead of only one. These small advantages do add up and they do compound themselves over time. But OTOH the Lanun map was very nice too. Thanks to Silk in forests and Reagents near a river, they were earning 33% more than “normal” in the early game. And no one has speculated what Ljosalfar might have done on that 13 Flood Plain map. A two-edged sword cuts both ways.

Your'e kidding right, just compare the numbers of barbarian kills from the dwarfs and the elves and tell me again that their activities are comparable. You killed more then 200 barbarians as the dwarfs, as compared to maybe 50 for the elves.
Ljosalfar on Flood Plains is an easy answer, they would have done as the dwarfs did, maybe even worse, because their bonus would not be existing.

DRUMROLL

So each map came with certain advantages or disadvantages. Despite the best efforts, a lot of variability exists. Were the results anywhere within the same ballpark, I would conclude the test was inconclusive. But the results were nowhere near close. Ljosalfar blew everyone away. Blew everyone away, not in one aspect or two, but in nearly any aspect you care to mention. Military? Ljo. Number of cities? Ljo. Total population points? Development work already done and operating? Hammer production? Projected future growth? Ljo Ljo Ljo Ljo. Only Khazad beat Ljosalfar in any of the major indicators, net commercial output. Everywhere else you look, it is Ljosalfar in a landside.

There is a lot of uncertainty in this study,. But not so much to explain all the massive leads established by Ljosalfar in so many categories. The basic premise has been supported; a realm can develop rapidly by focusing on Cottages, and Ljosalfar is tremendously better at this than anyone else.

Conclusion: Ljosalfar is broken. Fatal flaws exist that must be addressed.

Whom did the Ljosalfar blow away and in what aspect?
Military:
Agreed that they are strong, because they have their hero. Without him you are perfectly on par with all other races, you are even weaker because your units are missing experience. But they are surely not out of reach for the other races, Bambur can now be trained by the Khazad and was avalaible even earlier then Gilden and he is much stronger. You have Soldiers of Kilmorph ready to build and with one more city will have access to axeman, a unit the elves even missing the technology for. The Lanun have drowns (strength 4, can be upgraded from warriors with experience). So, what else do you want. A hero unit is supposed to be strong by the time it's created otherwise he pointless. But for balancings sake, make him stronger and move him to a later tech.

Cities:
Okay they have more cities, but they pay for them heavyly (see point 1). So this is no advantage at this point of the game. I could continue this for every point you make, but I stop here.

I have only one more point regarding this sentence:
"The basic premise has been supported; a realm can develop rapidly by focusing on Cottages, and Ljosalfar is tremendously better at this than anyone else."
=> I aggree with you on the point that the Ljosalfar should be better and are better to some extend, because thats their specialty, but they are by no means tremendously better.

On to the last point, your "root causes".

ROOT CAUSES

1> Starting Army: Why is this bad? It also means no early defence warrior. It offers them a small exploration bonus early on for about 10-15 turns, till a scout is build in another civ.

2> Starting Tech: Other civs have this as well, so why is this bad? Everybody can go chant, education if they want, some even have chant, so what? It's their to give elves a running start on leaves.

3> Woods Movement/Combat: Aggreed with the point, but again why is it bad. It's their TRAIT, and they pay for it heavyly in unavailaible units.

4> Raging Barbarians Aren’t Very: Again it's their trait, like malakim moving faster in desert and hippus have faster horses. Other civs need roads for faster movement, so what about, it offer a slightly better defence early on.

5> Early Worker Build: And your point is? Many civs have Education as their second invention, for some it can be their first. Also I can't see the defence problem you see, a cottage on open land right next to your city can easily be defended, let the enemy move onto it, clubber him, done.

6> Elven Workers: Trait.

7> COTTAGES: no Elven problem

8> FOREST COTTAGES: Again, I don't see your problems here, what is so damn good about a normal cottage with +1 Hammer bonus. 3-4 mines can put them of for an entire city full of them.

9> GOD KING: No idea here, I never ever used God King for more then 50 turns max. It's simply to pricey, for little effect.

10> TRAIL TO LEAVES: So does every other civ going for leaves?

11> ARCHERY: Every other civ going for leaves can research archery afterwards and usally do? For the elves the byproduct is their hero. Why not they are the kings of the archer aren't they?

12> ANCIENT FORESTS: no Elven problem again


I will not reply to any of your so called recommendations. I will however post some of my own.

Elven cottages : To slow down the early elven cottages a bit increase the build time of them a bit (like when building in deserts). This will slow down early builds a bit, but will not hamper mid- / lategame. This can also be done to farms.
Also a longer build time for all elven builds as Kael suggests should be okay.

Gilden Silveric: I like him as he is but as an idea, move him to a late tech (bowyers or precision) and make him adequately stronger. Then he can help with lategame, which is still a bit weak and he is removed as an early "imbalance".


a final note to Kael and team: You did terrific work with this mod so far and with the changes of 0.15 the Ljosalfar can now hold their own in mid- and lategame and win even under pressure. Please don't break them again.

Now of to bed it's late.
Greets Ghostmaker

 

[Bad Player]

The"endpoint" was turn 200. Each realm was given the same amount of time to develop.. The realm that develops faster will be bigger at the end of the same fixed time period. That's why hostilties were avoided. The intent was to measure how the economies would do when they are not constrained. When maximum performance is known, you know in acual play, something less than maximum performance will be achieved.

The intent wwas to measure each Civs economic development in the early stages. You objection to the test is your perogative, of course, but really, you are not adding anything that was not already addressed in the introductions. And ultimately, you are of course quite free to run your own study.

Honestly, at this point, semantical counterarguments carry little weight with me. Critics are under the gun to supply their own contra-indicatiors, be they test games or at least step-by-step analysis of the game mechanics (e.g. the God King discussions.) Just saying, I would have designed my test differently, so nothing can possibly be learned from yours... sorry, no sale.

So as you say, your experiment appears to show that the Ljosalfar will probably have a better economy by the year 200 than other civs you tested.

However that does not mean the Ljosalfar are overpowered if you define (correctly IMO) overpowered to mean significantly more likely to win than other civs.

 

[AndrewDJ]

So as you say, your experiment appears to show that the Ljosalfar will probably have a better economy by the year 200 than other civs you tested.

However that does not mean the Ljosalfar are overpowered if you define (correctly IMO) overpowered to mean significantly more likely to win than other civs.

Unser appears to be asserting exactly [EDIT] what you say in your first sentence[/EDIT]. I think that the question might be how significant that advantage is.

I also think that this experiment needs to be repeated in .15d, with, as has been suggested, a world-builder map set up so that each civ tested starts in the same location, with the same resources, and same competitors. It should be possible to give the test civs a start location that is neither too handicapping nor too perfect.

 

[Sureshot]

These tests do not appropriately model optimality for many civs. Unser obviously feels it is not on purpose, so one can only assume he doesn't yet know what would be optimal for the other civs, and as such things are more optimal for Ljosalfar.

I'm making a map that is to have optimal startup locations and placement for 9 cities total for each of the following civs: Khazad, Lanun, Ljosalfar, Malakim (First three as economic "powerhouses", Malakim as controlled civ, using best economic traits from vanilla).

Another problem is the use of fractal map, its a toss of the dice. It's one of the most random map settings. And then the selection of "good" maps are done by Unser.
Random selection followed by biased decisions leads to clouded results (everyone is biased, its called being human, but I don't think changes should be made because one person performs tests based on their biases, especially when polls have shown the majority favour things to remain as they are).

To correct this, I'm placing all four test civs on the same map, on their own ideallic continents. I will play four games, one as each, and likely won't make contact. There is no tech trading, and no interference, and each has equal optimality. I gave each city spot 2 resources, and picked them as best as I could. I gave each city a full fat-cross of area to work without clashes (in the case of Lanun this is mostly water, in the case of others it is all land).

I chose Quick gamespeed, since all our multiplayer games have been thus (and if civs should be balanced, it should be for multiplayer, not single player, which needs AI improvement not random axes), since even after 6 hours its only typically at mid to late game if one specializes heavily. I picked noble difficulty, since there will be little to no interaction, and I want a nice quick easy startup and I will be comparing AI's scores to AI's scores, and my scores to my scores.


My scenario is not typical, and I wouldn't say the game is balanced for this type, because anything can happen with the map generators, and I'm not allowing easy war, or contact, or tech trading. All their cities are optimal, they don't have to worry about someone stealing their land. It's not a true civ game. It's merely a test to compare economies given equal optimality tailored to the civ. Their only challenge is to overcome the raging barbarians and build an economy.

As it stands though, this is necessary, since Unser's methods are highly uncircumspect and I feel they cloud the subject (in case I need state it again, I make no claims Unser does this on purpose, there are however large inadequacies in the results due to his lack of knowledge of the optimal settings for Lanun at the least, and possibly Khazad).

I don't believe Ljosalfar is any better than Lanun or Khazad, or even a Financial leader like Varn Gosam, based purely on economy, and having played Deity settings and survived with civs like Kurioates, I don't feel Ljosalfar have an advantage over the norm, in fact I would place them at a distinct disadvantage to every other civ due to the lack of siege equipment and the million other things I've said before, but will forgo for now.

 

[AndrewDJ]

To Sureshot: I'd be interested to see both your results and your scenario, perhaps even play it. I do think you raise a valid point in that single-player balance != multi-player balance.

On reflection, you should probably make your scenario available here: the more people who play it, the more fluctuations in the data should even out.

I wonder if it would be possible to set up a test game similar to the Warlords barbarian scenario for the purpose of evaluating the AI?

 

[Sureshot]

It's about 62.5% done (3rd civ land done, but not its resources), it might be a good scenario except that only the Lanun's ideal city placements have coastal squares, so the others would have trouble lol.

but i might still make it a multiplayer scenario by making a canal into one city for the other civs.

 

[Ghostmaker]

Sweet! A slower elven workrate is a perfect way to ease into this. (That's what they get for insisting that 25% of the construction budget be devoted to roadside art.) I 100% support the notion of making small changes first.

Ahh, I fiddled with Worldbulder a small, small amount soon after starting Game 2. But I didn't figure out how to switch the settler nationality. But since it is do-able, that's a good idea.

There's ways to lessen the pain of a retest. Don't have to go all the way to turn 200, for one. 150 would probably be OK. And if I take standardized notes directly into an Excel fil I would save literally 5 work-days just transfering data and proofreading. :wallbang:

And the reports could be much more streamlined.

But first, I am gonna play me an actuall GAME of 0.15. I hear it came out recently.

I cannot stress this point enough. Please don't judge from the first 200 or 150 turns. This is neither accurate nor does it proof anything in term of strength.

Sidenote to Sureshot: I'd be interessted in the scenario as well.

Greets Ghostmaker

 

[Halancar]

I cannot stress this point enough. Please don't judge from the first 200 or 150 turns. This is neither accurate nor does it proof anything in term of strength.

I would have said do not judge from economy alone (there is also the military side, the diplomatic, etc...). On the other hand, I would think that being significantly ahead after 150 or 200 turns is a major advantage.

 

[dreiche2]

sure, but the problem is, how to find out how major that advantage has to be so that it cannot be countered by mid-game or warmonger strengths of civs...

btw, I suggest to somehow incorporate the epic war of the giftzwerge vs. the elves into the history of FfH , or maybe a giftzwerg unit or something....

hm I just found a candidate for a dwarven assassin unit it seems.... with a bonus against elves...

btw., wouldn't it much more appropiate to call "unser" "giftzwerg" instead? After all, it's just "our poissondwarf" in german, so being called "our" should be somewhat confusing...

 

[Ghostmaker]

btw., wouldn't it much more appropiate to call "unser" "giftzwerg" instead? After all, it's just "our poissondwarf" in german, so being called "our" should be somewhat confusing...

Not to be a nitpicker but ...

"Unser Giftzwerg" means "our poisonous dwarf" not "our poisoned dwarf" that would be "unser vergifteter Zwerg".

Greets Ghostmaker

 

[Chalid]

I am very interested how this turns out, sureshot.

But doesn't your map setting contradict the idea to test how the civs compare.
In my opinion by doing tailored continents of what you assume as optimal land (it might be or might not be) you bring in your own prejudices which makes the test as valid as Giftzwergs.

To really get compareable results you should provide each civ with exactly the same land. This is even more true when you speak of multiplayer games where you cannot restart until you get "optimal" conditions.

 

[dreiche2]

the problem is there is no "standard" map whatsoever. Taking a map setting with more sea will help the lanun, the climate might help or hurt the elves, etc. etc.

Raging barbarian on/off, victory conditions, aggresive AI, difficulty setting, all creates different situations.

I think because of that perfect balance is not even defined well.

And of course, every realization of a map can play out very differently. After all, that's a part of what keeps you playing civ over and over, no two games are alike even if the settings are the same in the beginning.


Judging the performance of a civ by playing 200 turns of one map is not very safe in my opinion. But, anyway, now that were at it, I'd like to see the results of different people in different settings at least out of curiosity...

Not to be a nitpicker but ...

"Unser Giftzwerg" means "our poisonous dwarf" not "our poisoned dwarf" that would be "unser vergifteter Zwerg".

Greets Ghostmaker

I think you misread it, I wrote poissondwarf, not poissoneddwarf....

oh, and to *be* nitpicky, it's "poisson( )dwarf", not "our poisonous dwarf", the latter would be "unser giftiger zwerg", not "unser giftzwerg"

 

[Mavy]

while you're at it: "Giftzwerg" is more of a figure of speech, so both of your translations are kinda wrong.

mfg
Mavy

 

[AndrewDJ]

To Ghostmaker: it seems like there's a fundamental difference between you and Unser Giftzwerg (more than just the elf thing). He asserts that conclusions drawn from an evaluation of civs' power and position at turn 200 can be extrapolated to the endgame, and you appear to disagree with this.

The obvious answer, of course, is to do more experiments. Let one person play a set of games with the test civs to turn 200, then hand them off to another person to play out the games to the end, and see if standing at turn 200 really can predict final strength/outcome.

EDIT: I should add that I think both Unser Giftzwerg's original test and Sureshot's proposed new test contain flaws: the first tested each civ on random fractal maps, and the second uses 'optimized' start locations for each civ. My opinion, for what it's worth, is that the best test would be to test each test civ on the exact same spot on the exact same map. But, YMMV.

 

[Unser Giftzwerg]

So as you say, your experiment appears to show that the Ljosalfar will probably have a better economy by the year 200 than other civs you tested.

However that does not mean the Ljosalfar are overpowered if you define (correctly IMO) overpowered to mean significantly more likely to win than other civs.

The first sentance is a goos summary. I like the use of the word probability. It is just all too probable that Ljosalfar will have a significantly faster and, stronger start then the other civs.

I do not believe I have used the word overpowered anywhere. I do wonder, jowever, how people think other realms will catch up. Ljo has more cities up, each of them bigger, more buildings already up and running, more military units up and operating, at least twice as many Cottages, and over twice the hammer production per turn. When Ljo is ahead in every category, and is growing in every category faster than you are, please, explain exactly how and when these massive advantages are going to blow away.

 

[Unser Giftzwerg]

These tests do not appropriately model optimality for many civs. Unser obviously feels it is not on purpose, so one can only assume he doesn't yet know what would be optimal for the other civs, and as such things are more optimal for Ljosalfar.

I'm making a map that is to have optimal startup locations and placement for 9 cities total for each of the following civs: Khazad, Lanun, Ljosalfar, Malakim (First three as economic "powerhouses", Malakim as controlled civ, using best economic traits from vanilla).

Another problem is the use of fractal map, its a toss of the dice. It's one of the most random map settings. And then the selection of "good" maps are done by Unser.
Random selection followed by biased decisions leads to clouded results (everyone is biased, its called being human, but I don't think changes should be made because one person performs tests based on their biases, especially when polls have shown the majority favour things to remain as they are).

To correct this, I'm placing all four test civs on the same map, on their own ideallic continents. I will play four games, one as each, and likely won't make contact. There is no tech trading, and no interference, and each has equal optimality. I gave each city spot 2 resources, and picked them as best as I could. I gave each city a full fat-cross of area to work without clashes (in the case of Lanun this is mostly water, in the case of others it is all land).

I chose Quick gamespeed, since all our multiplayer games have been thus (and if civs should be balanced, it should be for multiplayer, not single player, which needs AI improvement not random axes), since even after 6 hours its only typically at mid to late game if one specializes heavily. I picked noble difficulty, since there will be little to no interaction, and I want a nice quick easy startup and I will be comparing AI's scores to AI's scores, and my scores to my scores.


My scenario is not typical, and I wouldn't say the game is balanced for this type, because anything can happen with the map generators, and I'm not allowing easy war, or contact, or tech trading. All their cities are optimal, they don't have to worry about someone stealing their land. It's not a true civ game. It's merely a test to compare economies given equal optimality tailored to the civ. Their only challenge is to overcome the raging barbarians and build an economy.

As it stands though, this is necessary, since Unser's methods are highly uncircumspect and I feel they cloud the subject (in case I need state it again, I make no claims Unser does this on purpose, there are however large inadequacies in the results due to his lack of knowledge of the optimal settings for Lanun at the least, and possibly Khazad).

I don't believe Ljosalfar is any better than Lanun or Khazad, or even a Financial leader like Varn Gosam, based purely on economy, and having played Deity settings and survived with civs like Kurioates, I don't feel Ljosalfar have an advantage over the norm, in fact I would place them at a distinct disadvantage to every other civ due to the lack of siege equipment and the million other things I've said before, but will forgo for now.

The maps were never intended to be optimal "end game" maps. They were intended to be maps upon which each realm could grow according to its potential.

I am glad you are conducting your own study.

 

[Unser Giftzwerg]

I would have said do not judge from economy alone (there is also the military side, the diplomatic, etc...). On the other hand, I would think that being significantly ahead after 150 or 200 turns is a major advantage.

I said I was out to measure the economy. I think you will grow quite old looking for a quote from me saying only the economy need be considered.

If Kael has good information on how the game mechanigs synergize, he can make informed decisions on how to balance out economic issues in the otehr part of the game.

To everyone who thinks it's fine to have Ljosalfar's economy run away. Do you really want to see that balanced out in the end game? If their economy allows them to build twice as many units twice as fast, that means we will have a Ljosalfar feilding tie0r-4 units with STR in the 4-6 range. Is that the game you want Kael working towards?

In games of exponential growth, growth rates must be kept within the same ballpark. It is the same principle with earning money through investments. Go up to a banker and insist there is no meaningful difference between a 5% simple interest account, and a 10% money market compounded quarterly, and he will call the Men in Black on you assuming you to be an alien.

 

[Unser Giftzwerg]

I would have said do not judge from economy alone (there is also the military side, the diplomatic, etc...). On the other hand, I would think that being significantly ahead after 150 or 200 turns is a major advantage.

I said I was out to measure the economy. I think you will grow quite looking for a quote from me saying only the economy need be considered.

Not so much in reply to Halancar, as he understands the implications from turn 200 on...

If Kael has good information on how the game mechanics synergize, he can make informed decisions on how to balance out economic issues in the otehr part of the game. If he does not have that information, he must rely on luck.

To everyone who thinks it's fine to have Ljosalfar's economy run away. Do you really want to see that balanced out in the end game? If their economy allows them to build twice as many units twice as fast, that means we will have a Ljosalfar feilding tier-4 units with STR in the 4-6 range. Is that the game you want Kael working towards?

In games of exponential growth, growth rates must be kept within the same ballpark. It is the same principle with earning money through investments. Go up to a banker and insist there is no meaningful difference between a 5% simple interest account, and a 10% money market compounded quarterly, and he will look at you like you're some sort of bug.