AGAW2: Warfare Test

Sullla AGAW2: Warfare Test
Zulus - Shaka (Aggressive/Organized)
Standard Pangaea, Monarch, Normal Speed
v1

My initial game with Warlords was designed to be a peaceful builder game, one which would give me a quick look at some of the economic aspects of Civ4. But with many of the changes to the expansion dealing with combat, it was important that I also take a closer look at warfare in greater detail. The goal for this game, therefore, was to make war from start to finish as much as possible. I selected a Pangaea map to cut down on the logistics of pursuing a Conquest/Domination victory condition, and since Prince is apparently "too easy" (judging by the comments made in my last report thread), I kicked the difficulty up to Monarch. I'll return to Emperor soon enough, but I wasn't trying to kill myself here on the challenge.

I took the first map the game threw at me, and here was the starting position:

I drew about what I'd consider an average starting position here, on the eastern edge of the Pangaea. With Shaka again sharing the Spain colors/starting techs, I rain my "Spain on a lake" (TM) opening start to grab Hinduism very quickly while building a warrior. After that, it was time for a worker while researching to Animal Husbandry for my cows. I popped a scout from a hut early on (as you can see in the above picture - I have a scout before I've researched Hunting!) and that gave me three units to explore the nearby area. This wouldn't be useful ordinarily, but here in Warlords I found that ANY experience gained added to the Great General bar - even that from killing barb animals! My scouts and warriors quickly began farming barbs for experience, leading to some quick results on the bar:

This is from 3240BC. I'm already 1/3 of the way to a Great General! Obviously this is quite silly and doesn't make much sense. Victoria (with that Imperialist +100% Great General production rate) even popped a Great General in 2560BC!

Craziness. Incidentally, I already had 11XP from farming barbs by this point too. I'd have more to say on the issue, except that Alex has already mentioned that barbs will not provide points towards Great Generals in version 2. That's a much needed step in the right direction, because this was reminiscent of the barb "training farms" from way back when in testing that Fried loved so much. And I don't we want to go THERE again.

It was pretty obvious that experience from fighting barbarians shouldn't count towards producing Great Generals; I'm not really sure how this made it through multiple test builds.

By the way - we need to make sure that the barbs don't pop Great Generals period. Not just that YOU can't get Great Generals from barbs - we also need to cut out the barb production of these guys. It was quite humorous seeing this, in other words:

The barbs have apparently produced a Great General here, and yet no unit appears because I guess there's no programming in place for the barbs to have Great People (?) Anyway, the whole thing is very strange. If you weren't already planning to do so, make sure that the barbs don't produce Great Generals under any circumstances.

I got hit again by the inability of work boats to enter ocean tiles at all, even in your borders. Argh! That's been reported to death, of course, so I'm sure you guys know about it already... What WAS working correctly was the F7 Religious Advisor and the endgame replay, both of which had been non-functional in my first game. Patching from 1.52 to 1.61 and then reinstalling Warlords fixed both of these issues.

As far as the game itself here, I was squeezed for land badly early on - but what else would you expect on a Pangaea with these settings? I pushed my first two cities towards the AI civs early on, but still was boxed in pretty badly. Here's the map from 1000BC:

I caught a bad break with the appearance of that barb city in my back lines, because I had no copper or horses and thus no real way to attack it for ages. That barb city delayed the founding of my fourth and fifth cities by a good 30-40 turns, and then ultimately cost me getting the spot I wanted on the island to the north of me. To make matters worse, I've invested heavily in culture here (founding two religions and building Stonehenge) to try and flip either Medina or Djenne. Yet neither one came close, and Medina even built the Parthenon SOMEHOW (I have no clue what was going on with that one, and it wasn't a Great Engineer). Thus I got out to one of my slowest starts in ages, turns just crawling by with no progress in the Ancient Age. Far from my best stuff here.

My barb farming did pay off with a Great General as early as 850BC:

I of course used him for a Military Academy in my capital, which I already knew would be a shield powerhouse down the road. My goal was to see just how overpowered the +50% for military could be this early in the game. I don't want to spoil the story, but... let's just say that scaling things back to +25% in version 2 seems like a good idea. A VERY good idea.

In this test build, Great Generals could construct a military academy for 50% more production on military units, immediately, which was absolutely as broken as it sounds. This ability was moved all the way back to the optional Military Science tech later on where it rarely sees much use.

For a lot of turns, very little happened. I researched very slowly to Iron Working, hoping I would have the resource, then laughed out loud when I discovered the tech and found there was some UNDER my capital. Heh. That allowed to finally build a unit better than archers and raze the barb city in the east. I replaced it with two cities in better positions. Using that Military Academy in the capital, I cranked a few swords while researching Alphabet (tech trading still as useful as ever; I got 5 techs for Alphabet). By 50AD, it was time for the first war:

I declared war on Saladin and took Medina (being suffocated by my culture) pretty easily. Notice the lack of defenders in this city. Despite the fact that this was an Arabian core city, with a wonder inside, it was barely guarded at all. Very odd to see that. This would begin a recurring pattern throughout this game of the AI leaving its cities relatively lightly guarded. Certainly much more lightly guarded than I remember from Civ4's release version. Did we alter the AI's programming on that in some way? I'll have more pictures exploring that phenomenon in a bit.

So I captured Mecca as well, and with my own stack largely depleted, and an AI stack headed for my own largely unguarded city, it was a good time to make peace.

Mecca was a particularly useful catch, because Saladin had built the Temple of Artemis there. Since there was also marble located at the city, I would eventually build the Great Library in Mecca too, and put an Academy there to make it my top science city. Of course that was still in the future; in 500AD, I still only had seven cities, two of them recently captured, and was in the middle of the AI pack in score, still trying to recover from my slow start:

Building the Jewish shrine revealed that the completion screen is still bugged:

Anyway, while my capital did nothing by crank military units endlessly, my other cities worked on infrastructure. This proved to work surprisingly well, especially since I had ivory on hand and could build uberphants pretty much every turn in Zimbabwe with Military Academy (+50%) and Heroic Epic (+100%) speeding along unit construction. This only snowballed further when I discovered Civil Service and added Bureaucracy to the mix... Here's a picture from a little later, when I was building maces every turn:

This was... powerful. Definitely too much so. For fun, I merged all additional Great Generals into Zimbabwe for extra experience points (eventually getting up to crazy proportions), but it would have been more effective to build additional military academies all over the place. The +50% was clearly too much. +25% seems more reasonable, but I'll have to look at it and re-evaluate things. Popping out a maceman every single turn (on Normal speed!) is disgustingly effective.

Using my elephants and some cats (the eternal pairing that always proves so effective), I went to war with Saladin again and grabbed some more of his cities. I probably would have finished him off entirely, except that Mansa Musa declared war on me, and I had to respond to that. I had to fight off swarms of Malinese elephants despite the fact that Mansa had no ivory in his territory. What gives?

Qin is sending him ivory! Argh. I really wish the AI civs wouldn't casually throw resources around like that. They almost never trade strategic resources; I think the problem is that they apparently aren't considering ivory as a strategic resource - which it clearly is. Maybe tweak that a little bit? Seeing Mansa get ivory for dyes seems like an unfair deal.

Here's where I stood after annexing most of Arabia and preparing to begin my crusade against Mali:

Notice my science is down to 50% from the warring that I was doing and the maintenance costs I had imposed. As a result, I began a major push for courthouses at this point - and for the Zulu unique barracks, the Ikhanda! To put it bluntly, this building is too strong. Anything that reduces maintenance is already extremely strong - so it's crazy that we're attaching a 25% maintenance reduction to a barracks, one of the cheapest buildings in the game. And wait - it gets even better! The Zulus are Aggressive, so they get DOUBLE SPEED barracks! AND they're Organized too, so they get double speed courthouses too! In short, no one fights maintenance costs better in the entire game than Shaka, and he can easily get -75% maintenance from two extremely cheap buildings. That's a bad, bad idea. We need to brainstorm a new function for the Ikhanda, because - while this was one was cool in theory - it's majorly overpowered in practice.

This was another broken combination at the time: Shaka had Aggressive / Organized leader traits which gave him half price barracks and half price courthouses, with his unique Zulu ikhanda also providing another 25% maintenance reduction. All of this was severely nerfed, with Shaka moving to Aggressive / Expansive traits to lose the courthouse benefits and then the ikhanda increased in cost while also being downgraded to 20% maintenance reduction. It didn't make sense to have Shaka as this fantastic economic leader so these were solid changes.

While we're on the subject of maintenance, I saw a weird bug on the F2 Finances screen:

The number of cities maintenance is the same here as the overall maintenance cost! Were the distance costs just not being counted? What's going on here? Take a look if you can and find out what's going on.

Finally, I noticed that now that I had beaten on Saladin a lot, he was willing to become my vassal for a few techs:

He's willing to become my eternal slave for a couple of cheap techs? That seems... off. I thought about whether to accept this or not, and ultimately declined because the Arabian borders were in the way and it would be easier just to finish Saladin off. Later on, however, I would get the chance to play around a bit more with the vassalage concept. I'll have more to say on that in the next section.

Earlier, I had had the option to turn Saladin into my vassal, but his borders were in the way and he had some resources that I wanted, so I resolved just to finish him off instead. While I was doing that, I observed something very strange: a galley on a freshwater lake:

What's going on? How did that even get there? I can't even begin to figure that one out. Hopefully it's an isolated incident, but if someone spots it again, maybe we can begin to track the bug down. I had to kill off Saladin's civ to make it disappear, since there was no way to attack the unit or reach it.

A half-dozen turns later, Saladin was destroyed:

By the way, the new Saladin is Spiritual/Protective; I must say that the Protective trait doesn't seem to be all that helpful. The free Drill I promotion isn't that bad, but the Drill line of promotions is generally weak until you get to Drill III. Drill I by itself only adds the CHANCE for 1 first strike (doesn't even give you one!) so that's not really anything to write home about. I think it's pretty firmly established that adding to combat strength is almost always preferable to adding more first strikes, so Aggressive's free Combat I certainly looks a lot stronger there. The other trait attribute is double speed walls and castles, which (let's be honest) is pretty much as useless as the double speed broadcast tower from Charismatic. Does anyone actually build lots of walls and castles? (I sure don't.) I will reserve judgement until playing a Protective civ straight through a game, but my initial impression is that this is a very weak trait, probably the weakest one in the game right now. Something probably needs to be done to strengthen it.

Believe it or not, this version of Protective was even worse than the finished trait which is widely viewed as the weakest in Civ4. This Protective trait only granted free Drill I promotions along with double speed walls and castles, not even having the City Garrison promotions which are the most famous aspect of Protective. How Alexman thought that this was in any way comparable with traits like Financial or Expansive is beyond me.

I was still at war with Victoria, who had declared on me seemingly at random, but never saw more than one or two of her units. We eventually made peace with very little action transpiring. I took just enough time to move my units down to Mali's border and then declared war on Mansa Musa. Using the newly acquired Arabian territory, I moved my forces around to the back side of his civ and went after his capital first. There, I once again encountered the phenomenon of poor AI city defense:

This is the Malinese capital. Where are all the units? Never seen an AI civ leave itself this unguarded before, at least not since way before release. It would be one thing if that was the only city and this was an isolated incident, but it was a recurring pattern throughout the game. Here's Kumbi Saleh a few turns later:

Kumbi Saleh has the Pyramids and Sankore University inside, so it's a highly desirable target. And there are a total of... three defenders? That's some fine weed the AI's smoking there. Thing is, I could understand it if Mansa Musa was just weak and didn't have any soldiers. But he DID have units - a decent army, in fact. He just didn't have a lot of units guarding his cities, something which I definitely do NOT remember from Civ4's release version. Did we changed the AI programming in some way here? This is a lot different from what I'm used to seeing, and it happened way too often in this game to be brushed off as a coincidence.

Alexman was mucking around in the AI programming and it was apparent he had broken something along the way. I believe Soren came back to fix this up a little later.

Wang Kon declared on me in 1505. Two opponents at once... OK. Then a few turns later, Qin jumped into the party as well:

Just to be clear: at various points in time, Mansa Musa, Wang Kon, Victoria, and Qin have all declared on me. Brennus and Saladin are the only ones who didn't. This wasn't a game with Aggressive AIs checked either. What was going on was that I had accidentally put myself in a diplomatic quagmire. Look at those religious icons - all of the other civs were sharing Buddhism! They all LOVED one another, even Vicky (who had been Buddhist and later switched to Christianity.) The AI civs were all sharing techs, resources, etc. under Pleased or Friendly relations while I was locked out in the cold. I thus had to fight ALL the military forces of each AI civ, with no opportunity to pit them against one another. That made this game a full difficulty level harder, in my estimate. Saladin might have been my ally, since we shared Judaism, but he settled right on my borders, and I attacked him early on. So much for that idea. It was my own fault for not focusing more on diplomacy, but man did it make for a harder game!

After capturing the Pyramids, I swapped into Representation (which fit well with the Mercantilism I was running.) I was amused to find that the Great General super-specialists get the beaker boost as well:

This is not a bug; they should get those three beakers - but it was still funny to see.

I have another, somewhat nebulous, issue to raise here that I'd like to get an answer on, if possible. While playing this game, early on I noticed that I was getting some incredibly lucky swings on the combat rolls. As in, I won something like 10 battles in a row at even odds or worse, all within two or three turns. This was a bit suspicious, but I ignored it and kept going, passing it off as an isolated incident. Well, I kept noticing that the combat results felt - for lack of a better word - "streaky" all throughout this game. Anyone who played Civ3 a lot will remember the streaky combat results from that game. This felt very similar. Here's a demonstration:

This was from a battle where I had about 93% odds to win. I not only lost, but lost every single round of combat! A bit surprising there. Now please, don't quote odds at me here. I've got a degree in actuarial statistics myself, and I know full well that there's a good chance I'm going to lose a fair amount of the time even at 95% odds. That in itself is not really all that improbable at all. However, the consitently streaky nature of the results I was seeing - going one way for noticeable stretches and then swinging the other way - seemed to defy probability. Such as the spot where I lost 99.3%, 99.0%, and 98.9% battles all on the same turn. Of course that's possible - but extraordinarily improbable. On the order of a million to one, in fact.

So what I want to know is - was the combat random number general changed in any way? I've never seen anything like this in the release version of Civ4 before. Please tell me I'm paranoid and there's nothing to this. But I have to report what I encountered, and this game was streaky in combat from start to finish. More data will of course be helpful in generating data...

Something was definitely off here as Civ4's combat engine simply doesn't function this way. Unfortunately this kind of thing characterized much of my time doing the Warlords testing: desperately trying to stop an inexperienced designer from breaking the various systems that had been so carefully crafted prior to Civ4's release. My biggest achievement in Warlords testing wasn't adding anything fun or new to the gameplay, it was preventing some really bad things from happening which would have sabotaged Civ4's gameplay. More on this in the next testing report.

Continuing on from before, I beat up Mansa Musa until he was reduced to an island city off the coast. He refused to sign a Vassal agreement while he was at war with me, but once we signed peace, I discovered that he WOULD in fact agree to become my vassal. That seems a bit odd, but ok... In 1675AD, Mansa agreed to become my eternal slave in return for a single tech, Scientific Method:

Let me put it bluntly: this is WAY too easy. The AI should NOT be signing itself over to the human for pennies on the dollar. As far as I can tell, getting an AI to agree to sign a vassalage pact is based upon two things: 1) the player must be rated much higher on the power graph than the AI civ 2) the player cannot be the "worse enemy" of the AI civ. That's it! If you can fulfill those criteria, the AI will sign itself over to the human for a paltry sum, as little as one or two techs. Pfeffersack even got an AI to become his vassal for something like 400 gold. That's insane! Relations are utterly irrelevant; Mansa was at least -10 with me, if not lower, at the time we entered into this vassalage agreement.

In another somewhat odd part of the deal, signing a vassalage agreement immediately ends any conflicts that the vassal civ happened to be in:

Brennus and Qin had both declared war on Mansa once his power was vastly reduced by me, attempting to jackal some goodies off his corpse. When Mansa agreed to become my vassal, suddenly those wars just... ended? Huh? Saved me the effort of having to protect Mansa, but that doesn't seem right. Why did those wars end? Because I personally wasn't at war with those civs? This isn't a permanent alliance - our civs don't merge together - so why would those conflicts stop? Strange.

Even more disturbing, Mansa Musa still had some techs that I did not. Although we had been bitter enemies the entire game, the vassalage agreement forced Mansa Musa into perma-Friendly relations with me. As a result, I asked him "could you spare this tech for a friend?" and he GAVE HIS TECHS TO ME FOR FREE!!!

Here Mansa is handing over Divine Right to me free of charge. He also gave me Consitution just because I asked him to. Do I even need to spill out the implications of this for warmongers? Just conquer away, turn your enemies into vassals, then have them give you the techs you don't have. This is a huge disaster waiting to happen.

To summarize:

1) AI civs enter vassalage agreements much too easily. Once the player reaches a certain point on the power graph above theirs, they will happily sell themselves into slavery for mere pennies. Relations are utterly irrelevant to this process.

2) Any wars the vassal might be in at the time simply disappear once they enter into the pact. (They also have to join any wars the player's civ is in, which makes more sense, but logically I'm not sure that must necessarily be so either.)

3) Once another civ becomes a vassal, they will literally give the player's civ techs for NOTHING. This shatters any kind of balance we had between warring and peaceful building.

In a nutshell, vassal states are seriously messed up. We can tweak this back and forth, but I really think the best option is to make this a non-default option like Permanent Alliances. There's only so much we can do with the time we have left...

All of this is apparently normal behavior under the vassal state mechanic, or in other words nothing significant changed before these things went out with the expansion's release. Thankfully my lobbying did manage to get an option to turn off the vassal states which were otherwise going to be an unchangeable part of the gameplay. The vassal states are so poorly designed that I don't think I've ever started one of my own games with them turned on, only experiencing them once or twice in community games that someone else generated. Blah!

Moving on with the story... after swallowing up Mali, my civ turned and went after Brennus' Celtic Empire next. When I declared war on him, Mansa Musa was forced to declare war as well. (Is this the gameplay we wanted? Just checking as to whether or not it's intended. Vassals are pretty much nothing less than slaves, to be honest...) Guess what I found in Celtica? Another recurrance of the "AI civs not defending their cities" trend!

Again, Brennus has a decent-sized army. I expected them to be in the core of his civ, but instead he had a dozen knights running around at the bottom of the map in those tundra cities down there (if you can see the red splotches on the minimap). I had a devil of a time dealing with them, since I only had minimal forces there myself, but while Brennus was causing a stir in the south, his core was falling apart at home. Why aren't the AI civs defending themselves? They build units, lots of them, but they aren't in their cities any more! What's going on?

While ripping apart Brennus, I noticed this weird graphical bug:

This reappeared upon reloading (and restarting my computer), so I don't know what's up with it. Furthermore, when I was negotiating with Brennus on the diplo screen, I asked him to accept a vassal agreement and got a crash to desktop! Possibly related to this graphical bug. I took a save and I'll upload it here for someone to take a look at. Just talk to Brennus, ask him to become a vassal, and the game will crash. I was able to get around this crashing bug by first signing peace with Brennus, THEN asking him to become my vassal. He too agreed for virtual pennies:

Yet another slave, and all it cost me was Nationalism tech. Please see above for all of the many things wrong with this... Brennus became perm-Friendly as well; he was still -8 on relations, but now Friendly with my civ. That's definitely messed up.

Brennus' territory took me over 60%, but I was still a little short of Domination. Therefore, I declared war on Wang Kon... and also found myself at war with Qin as well. I guess they had a defensive pact (?) Wang certainly wasn't a vassal of Qin. Anyway, I had missed that detail, so the warfare just got a little more interesting. My own vassals of Mansa and Brennus got dragged along for the ride, and the war declarations flew around like crazy on that first turn of war. Mansa was safe on his island hideaway, but Brennus was highly vulnerable to Qin and his huge stacks of cavalry. Indeed, soon enough Brennus lost one of his cities, then he lost his capital and that was all she wrote for the Celts:

I suffered no penalties of any kind for letting my vassal die (that is, aside from his land not counting towards my Domination percentage). I basically just let him go and be killed. Better Qin focus his cavalry on Brennus than me, right? I guess that Brennus was killed before he lost "50% of his territory", so he never stopped being my vassal. Needless to say, this feels wrong. (The whole vassalage concept feels wrong to me, but this even more so.) We can probably do more work here.

I was able to take the Chinese city of Nanjing a couple of turns later, which had been putting enormous cultural pressure on many of my cities. With it out of the picture, I jumped up a full 5% in territory and passed over the Domination limit. Victory followed on the next turn in 1816AD.

The final score was 64181 points, 5326 in-game. I discovered that the endgame replay was now showing up just fine, since I patched over to 1.61. The English generated a Great General in 2600BC, and the barbs got one in 2000BC; after that, the only civ to generate any Great Generals was me. We will have to see how the changes to the system play out in version 2, because clearly the ratio of Great Generals to other Great People was seriously out of balance once again in this game. The first 3 Great People were all Generals, then the AI civs never generated another one throughout the whole game.

Here are the game stats. By the way, that clock STILL is not right. Do we want to fix it now, or just leave it wrong for all time? I definitely spent more than 13 hours on this one... For that matter, barb cities still don't appear on the endgame replay. We really should fix that one too, have them show up in black. It's always strange to see "Cherokee has been founded!" and nothing pop up on the replay.

Barb cities did eventually get added to the endgame replay though I think the clock still remains buggy to this day.

I killed 411 units and lost 91 of my own, a pretty good ratio. Lots of fighting in this game. It was fun to have 17XP infantry being produced every turn in my capital (Military Academy + Heroic Epic + West Point + barracks + theocracy + 4 Great General specialists). It's rare you get to have units taking Combat V or Commando right out of the gate! I'll be away this weekend, but I look forward to more testing when I get back.