Testing Report #4: Testing Out Demigod Difficulty

For my fourth game, I stuck with the standard game because it was still experiencing a lot of problems I wanted to iron out. Having tried the Agricultural trait in my last game, I was now ready to experiment with the Seafaring one and so picked Spain as my civ. I wanted in particular to test out the changed Spaceship victory condition and see how some of the changes to Industrial and Modern Age units played out. I also hoped to be the first player to defeat Demigod difficulty. What followed was not quite what I had expected.

Scenario: Small Random Map
Tester: Sullla
Version: 0.31a
Civ Choice: Spain
Difficulty: Demigod
Clock: 18 hours, 39 minutes
Settings: Standard Map, 60% water continents, roaming barbarians. Temperate/standard/4 billion age world with 7 random AI opponents.
Other Rules: All standard victory conditions enabled, everything default except culturally linked starts and AI respawn off.

Result: Diplomatic Victory in 1445AD (Turn 279 out of 540)


Spain 6087
Inca 2691
Aztecs 2070
Japan 1744
Persia 1717
Ottomans 1425
Scandinavia 960
America 568

Summary of Events

I decided it would be good to play as Spain in order to try out the Seafaring trait for myself and get my own point of view. Since my last game had been an all-out warfest resulting in an extremely early finish, I wanted this to be a peaceful cultural/building game in which I could try out the new spaceship victory conditions. I set up the map with pretty much standard everything, except a little bit extra land to see how well the Demigod AI was at expanding compared to my experience with the Deity AI civs. The difficulty was set at Demigod to get a feel for how it played out compared to Deity and Emperor. The map generator gave me a pretty average start on the first roll (which was what I wanted; I would have started anew with either horrible land or cattle/wheat starts) and I settled down to play with good expectations in mind.

I started on the coast (of course, since I was seafaring) and founded on the starting tile. I noticed that the curragh only cost one shield to buid, a clear bug, so I started by building two of them to explore followed by a warrior and then a granary. I began researching The Wheel at max rate to start. Around 3500BC I had to do some fancy juggling with creating scientific specialists so that my barracks prebuild wouldn't finish before I researched Pottery and could swap it over to a granary - fun stuff. I met America to the northwest with my curragh in 3000BC and swapped Alphabet for Masonry (what a ripoff! but better than being left out of the trading loop) I met the Aztecs a few turns later in the same direction and topped off my research on The Wheel (which they already had) so that I could start researching Mathematics.

My second city wasn't founded until 2590BC, and my third not until 2190BC, but that's really not that bad when the capital city lacks any food bonuses. I also met Persia at around this time, who I had somehow missed despite being the closest civ to me, and swapped around more stuff in deals to my profit. Unfortunately the AI beat me to Mathematics (a tech they usually ignore) in 1990BC, but I was able to top off my research of it for 70g with America and trade Math @3rd to Persia for Iron Working @4th. Min science of 50 turns (!) began on Currency, in the vain hope that I could get it first. Map Making was discovered in 1375BC and traded around, again to my profit. From what I could see, the AI was expanding almost - but not quite - as fast as it usually does on Deity. The AI was also researching almost as fast. From these observations, it appeared that the difficulty was dead-on accurate, right where it should be between Emperor and Deity.

My exploring curraghs (which NONE of the other civs build, AI foolishness) met the Japanese overseas. Deals ensued which caught me up in tech and brought me everyone's world map plus tons of gold, which I used to establish embassies with everyone. Everyone loved me. Finding the even-more backwards Ottomans in 1200BC allowed me to clean out more money from the AI. While the AI civs were researching at a fairly fast rate, I was able to keep up by pulling two for one deals, like with Construction and Code of Laws in 1100BC. Even better was when super-far behind Osman discovered Monarchy first and I traded him SIX ancient age techs for it. Cha-ching! :)

My min science run on Currency actually worked, and I entered the Middle Ages first in 670BC. Currency was traded for Republic, which I immediately swapped to, of course. Then disaster struck. Suddenly all my trade routes snapped in 650BC, cutting off the flow of luxuries to ME (not vice-versa) and I was left with a bunch of angry AI faces staring at me. What I later determined what happened was that the Ottomans had gone bankrupt, and unable to pay their expenses had a harbor disbanded in one of their cities. Or maybe Osman sold it off, I don't know. In any case, that snapped the trade route to the Ottomans and my reputation was destroyed forever. If this makes no sense to you, you're dead-on right. I was thus forced to play the Middle Ages without the ability to make use of gold per turn payments - ugh! Yeah, because I really "broke" a deal, right? Please. :rolleyes:

I noticed the plague was still around in 570BC - more on that later. Vikings declared war in 410BC when I refused to give them furs, a conflict which saw no battles whatsoever on either side. I completed my Forbidden Palace in 290BC by hand, another very productive move which jumped me up to #3 overall in productivity on F11. And... this was when the atrociously poor AI worker routines began to show. The fact that they mined all tiles really didn't hurt them in the ancient age, but with the despotic tile penalty removed they began to falter - badly - in the Middle Ages. More on this later. I declared war on the crippled Americans in 270AD to grab a couple cities on my border which were bugging me. I made peace again in 350AD with three cities gained.

At this point I began to do my own research, starting on Music Theory. I wasn't able to get that tech first, or Banking which was my next effort, but in each case I was able to research the techs myself without having to pay the AI civs anything. I was also able to send the techs I discovered to the backwards civs in exchange for more gold, fueling my research efforts while making them dependent on my efforts. Around 500AD, the Persians and Aztecs declared war on America again to finish off their one tiny remaining tundra city, prompting a huge stampede through my territory. You can see this amusing (and somewhat nervous) event below:

Discovering Democracy in 540AD first got me Physics, Chemistry, and over 2000g. I next got Free Artistry so that I could build Shakespeare's and test it out, which seems to be a useful wonder in the right circumstances. Certainly having a size 13+ city in the Middle Ages was a significant boost in building the early Industrial Age wonders. I entered the Industrial Age in 660AD by researching Magnetism, and by this point was the clear tech leader. I discovered Steam Power first in 740AD, and timed my factory building with a late golden age triggered by a conquistador against the tiny one-city remnants of America in 750AD. From this point out, I blew past the AI civs in research as they self-destructed under Fascism/Communism and continuous wars triggered by the "insane frigates" bug. I noted in 840AD that the breakdown of the AI was almost complete; the Inca were building Suffage in a city stuck at size 6, every tile mined, pulling 11 shields/turn. I was building it in a size 15 city pulling 85 shields/turn with factory, coal plant, and golden age. And the worst part: the Incas were doing the BEST of all the AI civs. Ouch.

Of course I got hit by the "insane frigates" bug too. Japan randomly declared war on me in 930AD, 1130AD, and 1255AD due to this, as well as the Aztecs (950AD) and Ottomans (1300AD). The Aztecs were the only one of these on my continent, so I calmly took four of their cities to send them a message and then made peace. All the other wars were inconsequential. I built Theory of Evolution in 1020AD, Hoover's in 1110AD and grabbed two scientific great leaders as well (Scientific Method and Combustion). By 1200AD I was six techs up on the nearest AI civ. By 1400AD I was TEN techs up on the closest AI, and their performace was like watching a train wreck in slow motion. There was no point in continuing to a meaningless space race, I would take the earliest win possible, the diplomatic one.

Just compare my cities to the Persian ones in the above picture and you'll understand what I was talking about. To amuse myself, I researched Espionage and build the Intelligence Agency, then used my massive amounts of gold to propaganize cities over to my side. It was quite easy to get size 1 and 2 Persian cities to flip to me, due to my vastly superior culture. But, of course, it was not going to be that easy. The Aztecs sneak-attacked in 1365AD and captured two of my lightly-defended propagandized cities in the north. Oh boy, it is ON now! :) I was a small nation, but not a weak one by any means. Over the following dozen turns my 30+ artillery pounded the Aztec armies into the dust while my bombers (operating completely unopposed since no one else was even remotely close to Flight) used their lethal bombardment to devastate cities. My cavs walked into city after city rendered undefended by the combination of artillery plus bombers. I discovered Motorized Transporation and tanks on the second turn of the war, and it only got more ugly from there. If infantry had required rubber (as they should), it would have been even worse. I made peace in 1405AD after taking about 2/3 of the Aztec nation, because I was bored of fighting such an easy war.

1405AD was the same turn I discovered Radio and entered the modern age. The most advanced AI civ, the Inca, were researching Steel and Atomic Theory at the time... Seeing that there were six civs left, and I would need four votes to win, I attacked the one-city remnants of the Japanese in 1435AD and destroyed them. Tokugawa would never have voted for me, so this gave me a shot to win. I needed 3 votes to win; my own was guaranteed and Persia, my long-time ally, was locked up as well. The Inca would be my opponent, since they controlled the entire other continent, and the Aztecs would never vote for me after my war with them. That meant I needed the Ottoman vote, so I allied with them against the Inca (they were already fighting) and gifted Osman a ton of techs to get him to polite. Fission was discovered in 1440AD and rushed with one of my scienfific leaders. The next turn I built the UN and won the vote, 3-1. I controlled 31% of the world's area and 46% of the world population; the Inca were tied with me for land area and far, far behind (30%) in population. No AI civ had a single city over size 12.


This was without a doubt for me the most broken version of the testing process I've played yet. My 0.29a game ended too soon for me to see the full effects of poor worker automation coding on the Industrial Age, but I saw it here and it was nothing less than virulent. The numerous bugs and poor AI essentially made the game unplayable after the Middle Ages were over. I'll go through the various issues I had in no particular order.

Curraghs only costing one shield allowed me to reap major exploration benefits that should only be confined to expansionist civs. The AI also did not seem to make any use of them at all. I'd go on and say more about this, except that curraghs have already been fixed in the latest patch by making them cost 10 shields and removing their transport capacity. Good fix, which should close out some of the exploitative possibilities inherent with a 0-shield cost unit.

The Seafaring trait I found to be very useful in this game. The bonus food was nicely canceled out early on by the despotic tile penalty, and the bonus shield really helped out coastal cities which are usually very low in shield production. Seafaring civs get exactly the kind of bonus they should: the ability to have very strong coastal cities and the strongest "fishing villages" I've ever seen. I honestly think that this trait should remain the way that it is. The problem is not with the agricultural or seafaring traits - the only problems comes when they are combined together in the form of the Dutch civ. Don't change these traits around just because they are bad when combined together! A better solution has to exist than that. But I see that this has already been changed in the newer build of the game. Suggested Fix: Change the Seafaring trait back and work on solving the problem of combined agriculture/seafaring! The latest change is to change from a bonus food and shield in the center tile to a bonus commerce in the center tile. This is a flat-out horrible idea, since commercial civs get this in ALL cities, not just coastal ones, plus two extra commerce in cities size 7-12 and three extra commerce in cities 13+. It's a huge blow weakening the power of seafaring civs, and to make up for it all seafaring civs get is +1 movement and a decreased chance of their ships sinking. So the whole point of the seafaring trait is now to help out suicide galley runs? :weed: Not good changes, and regardless of the intent, THAT is the real effect of these moves. This is to say nothing of the negative effect that this has on the Great Lighthouse for Seafaring civs. I strongly recommend switching the seafaring trait back to the way it was and dealing with the real problem of the Dutch food bomb.

Scientific great leaders are certainly a good idea in concept, and I do like getting them, but they are badly flawed in practice. The fact that a scientific great leader can rush any wonder represents a serious problem, as several multiplayer games have highlighted. It was incredibly illogical to have an elite combat victory produce a unit capable of rushing the Pyramids - now we have routine research creating a unit that can do the same thing? This is not a step in the right direction. Now I believe that the idea behind scientific great leaders was going to be that they could only rush certain wonders, scientific wonders, which I don't have a problem with and I think makes good sense. So what's the deal with them being able to rush anything under the sun? And why have military great leaders not been toned down either? I thought that both of these things were going to be implemented long ago, but the latest patch makes no mention of it either. Instead, we're left in limbo here with the scientific great leaders causing more problems than they solve at the moment. Suggested Fix: Make sure that militaristic great leaders can only rush militaristic wonders, and scientific great leaders can only rush scientific ones. A couple of the game's strongest wonders (Pyramids and Great Library come to mind) should not be able to be rushed with ANY great leader. This was proposed a while ago, and it's far overdue to be added to the game. I sincerely hope that it has not been dropped from the to-do list, because it is needed very badly to help reduce some of the luck elements from Civ3. Decreasing the likelihood of scientific great leaders appearing is NOT the solution to this problem. (I should also mention in passing that the "increase scientific output" option did NOT work with scientific great leaders in the 0.31a version.)

I will mention again that the reputation issue with regards to trade routes getting broken really should be fixed if at all possible. Why does sending a resource to a civ that gets kill destroy your reputation? Why did the Ottomans going bankrupt destroy my ability to make gold per turn deals? Why does two civs on the other side of the world going to war make me the pariah of the planet? I can come up with endless anecdotal stories to argue my point, but the fact remains that it represents a serious and completely illogical flaw in the trading engine and consistently causes problems in games. Soren e-mailed me and said that there was virtually no way to correct this problem, but I have to make the point again. PLEASE, if there is any way to solve this, take care of this problem. It really can destroy a well-played game through no fault of the player. I will be eternally grateful to any programmer who can sort this out. :D

Another issue that's still bugging me is the continued presence of the plague in random map games. Yep, it was still there in this game from time to time, bugging me out of my mind with its idiocy. The incidence rate has gone down, and the amount of time it lasts has been greatly reduced, but plague is still there. And I emphatically don't want it there. Suggested Fix: Well this one is simple: get rid of plague in random map games! It serves NO point now and should be removed immediately. The only place where plague has a reasonable purpose is in the Middle Ages scenario to simulate the black death. In fact, by cutting the plague down to a shadow of its former self, the Break Away folks are AGREEING with me that its complete randomness is bad and should be toned down. So if you see my point, then why is it still in there at all? I know the answer, of course: to sell copies of this game by using the plague as eye candy to put on the display boxes. I respectfully submit my opinion that plague is horrible feature and will only frustrate players. If you insist on having it, then by all means give us a switch to turn it off if desired. And I maintain that plague is far more likely to irritate the average player than entertain them. As I said in my last MTR, It's one of those things that rapidly goes from "wow, cool!" the first time you see it, to "this is kind of irritating" the second time you see it, to "I hate this frigging thing" by the third or fourth iteration. Get rid of it!

A few words on the new governments are also in order. Without fail, all of the new ones were used extensively by the AI civs and all of them were absoutely atrocious in performance. Fascism in particular was the government in which the AI "went to die", as I started labeling it in my head. But Feudalism was also horribly weak throughout the Middle Ages; the civs that dropped into it inevitably fell apart and were carved to pieces by their neighbors while those civs who stayed in Monarchy did better and those who stayed as Republics trashed the competition. Only myself and the Ottomans stayed in Republics for the whole period, and we went from the two last place civs to competing for the tech lead in a span of about 80 turns. Communism was as bad as always, and Fascism was even worse. It was almost like a game for me, using propaganda to flip the size 1 and size 2 cities in the AI civs which had fallen in population as a result of fascism's population penalty. And when you kick in the endless whipping that the AI does under all these governments... well, it got ugly in a hurry. With all the tiles mined, the AI civs couldn't even rebuild their population fast after whipping. Every single civ except for myself literally self-destructed in the Industrial Age after they went Fascist/Communist. This does not speak highly for the new governments as viable choices. Suggested Fix: Well, none at the moment since all of the new governments have been thoroughly overhauled in the latest patch. I'll have to test them under the 0.32a version to see how they now stack up against the older ones, as well as checking out if communism can ever compete with a democracy. Soren himself described Fascism in 0.31a this way: "Fascism with the current penalties would not be worth ever having." My comment in response was "Yep." ;)

This leads to the larger issue of AI worker mismanagement. This is well known by now, of course, but was almost frightening to see it in action. In the ancient age, the AI civs were able to perform at a very competitive level, since most tiles are generally mined in the ancient age anyway. My impression was that Demigod was dead-on accurate in terms of difficulty at this point. Then after the first government swap in the late ancient and early medieval periods the excessive mining really started to catch up to them; despite being one of the smallest empires on the planet, I reached the tops in productivity halfway through this period. On a difficulty just a little bit shy of Deity, it should have taken until the Industrial Age for that to happen. I was able to first catchup to the other civs with my own research in the late middle ages, and then utterly blow by them in the industrial age, as my control of rails and factories shattered the cost factor benefits applied to the AI civs. It was really a combination of three factors which caused the implosion of the AI civs in the industrial age: 1) The enormously bad AI workers mining everything kept their cities small in size and weak in both commerce and production, allowing my small in territory but large in population cities to race past them in every aspect. 2) The "insane frigates" bug whereby the AI civs would go to war over nothing and stay in war non-stop throughout the rest of the game was a real problem. Since the AI civs get frigates with the tech Magnetism, it's no coincidence that this was about the time that they stopped researching and growing altogether. When they fight wars nonstop, the AI civs stop trading luxuries and techs back and forth, and they drop into 3) Fascism and Communism, which decimated the AI further by cutting off the extra commerce from Republic/Democracy and changing rushing to population-based. With all their tiles mined, population recovery from pop-rushing was very slow and componded the problem. The result was a viscious cycle in which the AI civs spiraled further and further out of control, slipping into a vindictive downturn from which there was no possible recovery. Let's not forget, I got TEN (!) techs up on the AI civs and was stretching that out further when the game ended. It was only getting MORE ugly, not less as the game progressed. Doing that in Play the World Deity would be unthinkable, and Demigod should not be far off from that. Suggested Fix: The first thing that has to be solved is the worker automation routine. Nothing can be done if that is not solved first, and I have hopes that the next patch will be able to do that. Secondly, the insane frigates bug which puts the AI into an "Always War" situation has to be resolved as well. I have heard that this was addressed in the 0.32a patch; we will see. Finally, the AI has to learn to avoid these destructive governments and pop-rush their cities into the ground. With the governments being changed around, I expect (hope) that the case will improve. But as far as the 0.31a AI is concerned, it still plays like a retarded child.

Concerning upgrade costs... they're really pretty good where there are now at 3 gold per shield. This is already 50% more expensive than the old system, and it closes out the mass upgrade strategy pretty well in my eyes. 400 gold will only allow you to upgrade 6 warriors into swordsmen now (at 60 gold each) rather than 10, and that makes a BIG difference. While the mass upgrade strategy isn't ever going to go completely away, changing upgrade costs to 3 gold per shield significantly reduces its appeal as a tactic. In my eyes, the problem was solved by this change. So now when I see upgrades going up to FOUR gold per shield, at DOUBLE the old rate, I have to believe that it's over the top. Upgrades are now so expensive that it virtually rules them out altogether, unless the player makes an extraordinary effort to come up with more gold for upgrades. I mean, 40g to turn a spearman into a pike? 120g to turn a pike into a musketman? This is simply prohibitally expensive for most games, and it really is not moving in the right direction. Jsmith downplayed the change by saying this "Now 4 gold per shield - use tax collectors and wealth to get the money to upgrade..." OK, it's true that tax collectors and weath now provide double their old value, but can't you see that these changes FORCE you into changing your playing style? I mean, a good game is all about making choices between different sets of options; upping the upgrade costs again just seems like punishing the players and forcing them into a certain formulaic way of playing. Sure, you COULD get the money from using wealth... but what if you don't want to? This is limiting the options of players, not expanding them. And I don't think that the fans are going to like this one bit. Suggested Fix: Use 3 gold per shield for upgrades. In this game I just played, I found it to be quite appropriate without being excessive. Upping the rate by another 50% is just going too far; is the goal to outlaw upgrading at all? I don't think so, but that's the practical effect of increasing the costs further.

Finally, another issue that needs to be dealt with is lethal bombardment for airplanes. It's just excessive and extremely overpowering in the hands of a human player against an inhuman AI. When the Aztecs sneak-attacked me, I had just built my first bomber that very turn and had no air force to speak of. 10 turns later, even with just a dozen bombers I was able to redline and then KILL every defender in an Aztec city, which my cavalry then proceeded to walk into untouched. What could I have done with 20 bombers - or 50? I could have killed all the defenders in every city and let my units walk into them one by one without ever fighting a battle. Combining non-lethal artillery with lethal bombers is even worse, which I was able to do routinely as well. I don't even want to THINK about what radar artillery plus stealth bombers could do. Suggested Fix: No planes should have lethal land bombardment. Sea bombardment yes, but never land bombardment. This was what I said when I first read the desciption of that change, and my testing has only proven it out. Lethal land bombardment of any kind is simply too overpowering for any units to have it. Maybe that makes you unhappy - deal with it, it IS too powerful. The fact that the range for air units has been greatly increased already massively increases their power. Adding lethal land bombardment on top is way, WAY too much.


- Curragh bug, already reported and fixed.

- Reputation problems still exist when it comes to trading, which I described above and would love to see fixed if at all possible.

- Religious civs are still getting 0-turn anarchy periods. This IS a bug, right, and not a design change? I'll submit another entry on it, although I think it has already been reported.

- The Civilopedia reports that 8 cities are needed to build the Forbidden Palace on a standard-sized map. However, since the optimal city number has since been changed it actually now requires 10 cities. I am submitting an entry that this be changed.

- Plague is still occuring in random map games. It's not a bug, but it should be. :p

- For no apparent reason, luxuries refused to flow through harbors for a number of turns and then suddenly began doing so again. Since I have no idea what was going on here, nor could I even begin to guess it, I'll just submit the savegame as evidence.

- Insane frigates continue to trigger wars for no reason. This has already been massively reported and hopefully corrected.

- Clicking on any stack larger than about 15 units causes a crash to desktop which is EXTREMELY irritating. This is also well known and reported, and I hope has been fixed already.

- Size 0 cities were appearing on the map, presumably from Fascism hitting size 1 cities and dropping their population. Trying to investigate or use propaganda on them caused an instant crash to desktop. I have a save game for this and hopefully it can be dealt with.

Final Word

There were massive AI problems in this version of the game. While I'm glad to be the first player to win on Demigod (sorry Lucky, but Regicide, tiny map, 80% water archipelago games really don't count :p) there was essentially no challenge after the ancient age on this map. To test the AI more fully, I would like to play a Sid "conquest-ONLY" game next and see if I can win. We will see how that turns out.

Impact and Later Changes

Curraghs have been fixed and now occupy a nice niche as an early scouting unit. I do not like the current version of the Seafaring trait and will have more to say on this later in another section. Scientific great leaders are still completely random, but since being the first to discover a new technology is virtually impossible on the higher difficulties, it is not going to affect high-level games much. Could prove interesting on Monarch and below, however. The new governments have been heavily boosted, especially Fascism, to make them viable. I'm still sticking with Republic/Democracy for the moment, however. AI workers and the "insane frigates" bug were both fixed (thank goodness) in later patches. Upgrade costs are now 3 gold/shield; players unhappy with this should remember that BreakAway wanted to set it to FOUR gold/shield. All planes still have lethal land bombardment, and air power is as a result massively more important in Conquests. I think they are overpowered, but at least now there are anti-aircraft units to shoot them down with. One thing is certain: neglecting air power is now a suicidal move in the expansion.

Oh, and yes, I was the first player ever to win a standard game on Demigod. :)