Fall from Heaven is a dark fantasy Civilization IV mod, built by a team headed by Derek Paxton. The first version was released on December 16th, 2005 and last month – exactly three years later – the “gold” version was uploaded. The project is the most successful Civ4 mod yet created, with hundreds of thousands of downloads and positive nods from critics like Tom Chick. I recently had an opportunity to interview Derek and his team on the final release of his the mod, which is available for download here. (Note: Fall requires a fully patch version of Beyond the Sword to play.)

Soren: One standard way to write a game post-mortems is to list things in two categories: What Went Right and What Went Wrong. I’d be interested to hear what you would put under each heading. (Sometimes, things fit in both places!)

Derek: I think we made the right game for the platform, both from the user-base interest and what the Civilization IV engine was able to deliver.  We didn’t come into this with an idea of the game we wanted to make and then tried to make the engine do that – though, at some point, that’s what it feels like.  We started fresh, played and loved the core game, and looked at what we could offer on top of that.  Because of that, new versions/expansions of Civ made our game better as well.

As for what went wrong, if I could go back, I never would have started the design with 21 civilizations.  I would have had much less, maybe as few as 9 or 10. Although people love the 21 civilizations in the game, and I feel like we did a good job differentiating each one, there is a balance between depth and breath that is a real physical limitation on how much detail we can give to each asset in the game.  At this point in the process, I can’t imagine cutting any of the civilizations, and Fall is a better game for their inclusion, but when I consider the amount of detail and polish that could have been given to a game with only 9 civilizations to focus on, it makes me envious.

Tom: This might not belong here with this question, but one of the things that is both right and wrong with Fall is the fact that Derek is running a nearly perfect ship.  He mentioned above that I’ve been involved in numerous other mods, and they all fail in comparison with Fall in many ways, but the most glaring is that they simply don’t have his leadership.  Many times we’ve had discussions that other mods are being unfairly compared to Fall, and – from first hand experience – it’s true.  Other mods simply didn’t or don’t have Derek and his drive to make a perfect mod.  After playing Fall, its unfortunate that players who aren’t involved in modding hold these other mods to unobtainable levels of quality because Derek makes Fall look so easy and polished.  The bar is set very high for Civ4 mods.

Soren: Let’s talk a little about the gameplay itself. What are some of the mechanics that you are most proud of developing in FfH? What ideas turned out to be the most fun? The most original? The hardest to get right?

Derek: Thats tough.  I’d love to hear the answers form the other team members too.  Lets see:

1. The most proud of: Orthus.  Orthus is a unique, powerful barbarian unit that spawns early on and terrorizes any civ unfortunate enough to be near him.  The unit that kills Orthus gets a special promotion called “Orthus’s Axe” that makes them more powerful.  I learned so much from Orthus.  He made games dynamic, players loved the surprise of seeing him coming into their lands for the first time, and loved hunting him down to gain his axe.  It was the first dynamic component of the mod that showed me what could be done, and he had a huge influence on later game design.

2. The most fun: I really like our vampire mechanic.  One civilization, the Calabim, are a society of vampiric aristocrats ruling over a peasant class that they treat as little more than livestock.  The actual vampire unit comes in the midgame for this civilization, and vampires have a few special powers: they can consume population to become stronger (gain experience), they can feast on weak Calabim Bloodpet units to heal themselves and regain the ability to attack, and they can gift other units in their stack with vampirism if they are above a certain level.  Those changes are relatively minor (in the scope of how much code it takes to do), but it does such a great job of marrying the flavor of the civilization with the gameplay.  It’s so easy to play as the Calabim and picture your population as a food resource for your vampire units.  We try to hit that kind of synergy with every civilization, but I think we got closest to it with the Calabim.

3. The most original: The Armageddon Counter (AC).  Turn-based games tend to go into a late game deadlock situation, and there isn’t much reason to have wars outside of specific victory conditions.  The AC begins to have an effect from the mid-game on.  It rises as “bad” things happen in the world – cities are razed, evil religions spread, powerful demons enter the world.  It goes down as “good” things happen, powerful demons are defeated, evil holy cities are destroyed, graves and city ruins are sanctified.  If the player stays in his borders and doesn’t engage with the rest of the world, the AC may not become an issue or may begin to rise depending on what is happening in that game.  However, the important part is that the effects of the AC, powerful creatures appearing, blight and pestilences striking the world, AI players becoming more and more likely to go to war with each other, are shared by all players. So if you hide in your borders, you will still be subject to them.  Because of that, players have an incentive to come out of their shells and try to direct the outcome of the world.

They may set out on an expedition to raze the holy city of the evil religion or to defeat a powerful demon.  They may decide to go to war with a civilization thats performing a lot of AC-raising activities.  The AC keeps the player engaged in the late-game, keeps the game strategically interesting and is, as far as I know, a unique way to handle the slow late-game issues TBS games are subject to.

4. The hardest to get right: The Armageddon Counter.  Trying to balance its effects so they aren’t too punishing, but punishing enough to make them strategically significant, is very hard and still isn’t perfect.  Part of it is the difference in individual taste and part of it is just the nature of a random game.  Anytime negative effects (those that punish players) are added to a game, we have this challenge.

Tom: I think most every unique mechanic that we came up with is something of which I’m proud.  Trying to make each of the civs spicy and different without making it look like we tried too hard was challenging and fun.  From the Khazad vaults to the Kuriotate settlements.

1. The most proud of:  The spell system and filling every sphere.  Talchas did amazing work, and it’s been one of the things that really makes FfH stand out.  Also, the metal system is so brilliant that I can’t play the original game now because it isn’t there.

2. The most fun: The Sidar mechanic of turning level 6 units into wanes and then adding these as great people to your cities and creating super cities is wildly fun.  When we first introduced that mechanic, I playtested them to death, and it really made them stand out.

3. The most original: The metal system again.  Gaining +1 from bronze, iron, and then mithril is a great way to upgrade older units and adds another reason to fight over resources.

4. The hardest to get right: I’ll write Orthus here.  He used to be much more of a badass than he is now.  He used to make me enter the WB or simply restart a game.  Now, he isn’t nearly as strong.

5. Biggest disappointment that didn’t work out:  Multiple maps.  We had such a good idea originally for hell.  Sadly, hell, the underground, or other areas, simply won’t work with multiple maps.

6. Mechanic still hoping for:  The wilderness.  It’s been on the backburner for 18 months or so after the original idea was dreamed up.  I’d still love to see areas unexplorable that have a fog-of-war for long periods of the game.  Or, when you enter them, you can’t see more than the tile you are on.  Very dangerous.

Randy: The Mercurian and Infernals could probably be a fitting answer to all of those categories, I think. These two civilizations enter the game at about the half way point. The player can choose to switch and begin controlling the leader of the new civilization or stay the with their original and have a new partner or rival, respectively. Obviously, starting a civilization fresh midgame calls for some special mechanics that have to be used pretty carefully to avoid overpowering or trivializing that player, in human or AI hands. Also, they are each tied to one of the more polarizing religions as well as the AC. Quite a balancing act, and it’s taken awhile for them to evolve into their intended roles – it was disappointing at times to summon a huge demon who hides out in the arctic circle for the rest of the game, wishing he had room to settle in! At the moment, I think they are in decent shape. They probably won’t be a major factor in every game but should be able to to provide either a challenge, a decent partner, or a new civilization choice often enough.

On a much smaller scale, the Naval Crews are a pretty neat effect, allowing any ship in a city to make a small optimization by choosing a promotion that adds either strength, movement, or cargo space, but each one has a drawback as well. This can be changed later if the ship is in a city or on a Lanun civ-specific improvement. It adds some strategic variety to what is otherwise a fairly small number of naval units.

Eli: I’m most proud of the way the Balseraph (evil clown) civilization turned out. In my opinion, it has the best overall feeling of completeness in artwork, backstories, and mechanics. For example, the Balseraph Freaks are units that can be built very early on, without requiring a building in the city (in FfH nearly all units have a building req), and starts mutated. In other words, the Freaks all start with slightly different stats and abilities, sometimes good and sometimes bad. Cool by itself, but what makes the Freaks really interesting is the synergy between them and the other Balseraph mechanics. After building a freak, if it has potential to be a good unit, then you can send it to fight in the Arena (Balseraph UB). If it wins the fight (50% chance), then it gets bonus XP and should then be upgraded to a swordsman or hunter. If it dies, then you just build another freak. Or, if the freak’s mutation makes it too weak for combat, then you can use it to build a “freak show” in its city, increasing happiness and culture. Either way, plenty of strategy involved, and that’s just one facet of the Balseraphs.

Derek: I have a question for you Soren, understanding that we had the luxury of taking everything you did and then spend an additional three years on our own ideas and changes, what do you think doesn’t work well in Fall?  What things would you change?

Soren: First of all, I need to start by saying how impressed I have been by Fall from Heaven. Besides the remarkable level of polish – from the graphics to the sound to the lore – I’ve really appreciated how your team has subverted many of the core Civ conventions to make a compelling fantasy game that is unique from Civ yet still feels familiar to long-time fans of the series. It’s hard to stop playing just to see what other new wrinkles the game is going to reveal. There is a tangible sense of discovery to the game, which is hard for a designer to achieve. Your team should be very proud of what they have accomplished.

As for what I would change, I think my criticisms fall into three major categories: communication, pacing, and density. Let’s start with the first issue – as I’m sure you’re aware, many of Fall‘s features are not fully documented, either in-game or within the Civilopedia. However, the biggest problem is that of the new choices and objects that one encounters during the game, quite a few are not even partially explained by the pop-up help. For example, it’s great that the Pool of Tears is identified with a label on the map, but mousing-over it doesn’t tell me that it makes nearby cities happy or that it cures disease, plague, and poison. Similarly, the Fire I promotion allows Blaze, but the help doesn’t tell me what the Blaze spell does. What about the global March of the Trees spell? I see that I can only cast it once per game, but what is the effect?

For some of these more unique items, I’m surprised that you didn’t use the <Help> tag, which I believe is available by default for all of the XML-based game elements. This tag lets you add an extra descriptor to any item if the normal popup help is insignificant. Because our goal with Civ4 was to have a completely dynamic help system, we rarely used these tags. Instead, I envisioned that they would be most useful for mods and scenarios. On the other hand, if you really want to follow through on making Fall a platform instead of just a game, you need to extend our dynamic help system yourself to cover all the features that you have exposed for modders. I see that you have already done some work along those lines as I saw new, unfamiliar pop-up help while playing but much seems left to be done.

The next problem is pacing – on default, the game is tuned to last 690 turns. That’s a lot of turns! Fall has a plethora of divergent play strategies, which is great, but I would rather play 3 short 300-turn games with a different civilization each time than one epic games in which I’m waiting and waiting and waiting to see if my choices are going to pay off or not. I think the best answer, though, is not just to speed up the game by reducing the costs of techs and buildings and whatnot. Fall simply has too much Civ-like infrastructure busywork – which is not Fall‘s strength – and all that settler/worker/citizen/building-management really slows things down. To solve this, I would fill at least half of the map with wasteland-type terrain that might hold treasures and dungeons and monsters and mana but could not be settled by civilizations (or, perhaps, by specific civilizations). This change would allow for a faster game with more exploration and more variety. The victory conditions also seem quite difficult to achieve – sixteen different mana sources are needed for the Tower of Mastery, correct? I’d prefer easier to achieve end goals that allow for more of a race-to-the-finish instead of a slog.

The final issue is density, by which I mean the sheer number of options available to the player at any one time. At one point, I had 18 different technologies to choose from! The design provides a great number of paths which the player can pursue – many more than in standard Civ4 – but showing them all to the player at once can be overwhelming. I also have a sense that the game has simply too much stuff in it – over 150 buildings and wonders, 200 units, 160 spells, 56 heroes, 21 very unique civilizations. As you mentioned above, with fewer game elements, the polish could have been much higher. I wouldn’t necessarily start by cutting stuff, but I think it would be interesting to have more civ- and style-specific techs that lock away a significant chunk of the game so that the player is never drowning in options and also so that each playthrough would feel significantly different. Your design heads down this path in many ways – by tying buildings and spells to flavors of mana, for example – but I wish your team would go farther.

One thing I find very inspiring about Fall is how many small tweaks were made to the core Civ rules that led to instantly interesting, new gameplay. For example:

• Buildings are required in each city to unlock most units (such as an Archery Range to build Archers)
• Some units cannot be built but only upgraded from other, high-level units (such as High Priests from Priests of level 6 or higher)
• A few Wonders have simple, one-time effects (such as the Pact of Nilhorn which gives three Hill Giants)
• Hero units can only be built once (including Guybrush Threepwood!) and share their own extra sub-tree of promotions
• Spell-casting units receive free XP each turn, letting them access higher-level spells as the game progresses
• Hybrid units, like Acolytes, which can act like Missionaries (spreading religion) or Great Artists (spreading culture) but can also still fight

Ultimately, many of these changes were natural extensions of our data-driven design that eschewed hard-coding, enabling a lot of variety without having to touch the code-base itself. If I’m not mistaken, Hero units were actually possible in vanilla Civ4 just by using the same mechanism that allowed buildings to be Wonders of the World. We never actually used this feature, but the code supported it because we generalized the concept of “single-build items” across the game.

I find it interesting how the design of mods are always, to a certain extent, a reflection of the core technology decisions made by the original development team. For Civ2, the emphasis was on a flexible trigger-and-event system, which enabled some interesting, story-focused mods, based on works like The Odyssey or Fellowship of the Ring. With Civ3, we punted on events but worked to remove all hard-coding from the game. Theoretically, the game/AI engine could handle any number and assortment of game elements. Thus, the result was the popular Double Your Pleasure mod, which doubled (or tripled) the number of units, buildings, and techs in the core game. For Civ4, we opened up the game’s algorithms and interface, which finally allowed modders to create their own, brand-new systems, of with Fall‘s magic system is one of the best examples.

Soren: I am curious how you hit upon the innovation of giving spell-casting units a steady drip of XP each turn. The mechanic neatly solves the problem of how players can explore the spell tree without having to fight – an important question as these units are not always strong enough to win enough battle to reach higher levels. Did you try other XP systems first?

Derek: The initial mechanic idea was that units would have to study in cities in order to gain spells and levels.  But having to leave the units in cities isn’t fun.  Players don’t want to have to not use a unit for it to become effective, so we removed that requirement and allowed them to gain the xp regardless of what they were doing.

Soren: I also would like to know how much you changed the barbarian code of the core game. Fall definitely has more of a PvE feel than base Civ4, with all the lizardmen and goblins roaming the countryside, not to mention the dungeons scattered across the countryside. This extension of Civ‘s barbarian tradition really helps keep Fall combat focused without requiring full-scale war. What are the underlying mechanics here?

Derek: We really wanted that initial exploration to differentiate Fall from Civ.  From that first trip out into the wilderness, we wanted the player to realize he was in a different world, and it was a much more dangerous one.  Civ was always about quick growth, filling your area and firming up your borders against your neighbors before you began to consider wars or other options.  There wasn’t much of a downside to early growth outside of maintenance costs (which typically didn’t become overbearing until you have started taking over enemy civilizations).

We wanted to disincentive early growth.  Early threats did that for us, but it was a pretty difficult balancing act.  The player has limited resources in the beginning, and he needs to decide how to split his earily hammer on defense, growth, exploration or infrastructure.  And as you mentioned Fall‘s focus is moved down a bit from Civ‘s – a bit more focus on the individual units than at the empire level.  If the focus was going to be on the units, we needed fun things for the units to do even outside of war.  The barbarians filled that role.

Soren: I have seen a number of groups try Fall from Heaven as a multiplayer game, often experiencing frustration from technical errors and out-of-syncs. I’m not surprised by this problem – creating a synchronous game is a very difficult challenge as all calculations must operate on all machines in the exact same order. Further, versioning control for mods is under the user’s control, which throws another possible wrench into the works. However, assuming the technical issues can be solved, what do you think the potential is for Fall as a multiplayer game?

Derek: Huge. Out of Sync errors were #1 on our list going into the latest version.  We added a logging utility that writes out the game state on all the computers on the game specifically so we could diagnose and resolve OOS issues.  I’m happy to report that it worked and, as of the latest version, we aren’t getting any more reports.

The reason I love Fall for a multiplayer game is that the core concept of the game – dramatically different factions – lends itself so well to interesting combinations.  A Sheaim player could be trying to bring on Armageddon while a Ljosalfar player tries to cover the world with forests.  A Hippus player could be raiding all nearby opponents while a Khazad player holes up in strongly defended fortresses.  How each civilizations strengths and weaknesses effect each other, along with the decisions of religion and mana sources, makes for very interesting games.
One thing I dislike in other TBS and RTS games is that since the factions are so similar, players begin to adopt a fixed strategy they always employ.  Often specific enough to have the same build orders and such.  Because Fall is so based on the combination of elements and variety a single strategy becomes impossible and players are forced to react to the specifics of the game.

Soren: Reminds me a little of Alpha Centauri, which also had good multiplayer potential because the factions were so different. However, as with AC, one of the most common criticisms of Fall is that, while the game mechanics are a lot of fun, the AI has little chance at handling them. Do you have any thoughts on this topic? What kind of progress has the team made on that front?

Derek: Some – it’s pretty heady stuff.  In the beginning, we intentionally ignored the AI as it didn’t make much sense to invest a lot of time teaching the AI how to play the game when we were still working out how the game would play.  At best, we would have to redo it later, and at worst, we would make the AI less effective.  During the conversion to the Beyond the Sword code base, the entire spell system was rewritten to be easier for the AI to understand.  It made a significant difference, and in the last few months, we have begun to go through the last major hurdles of the AI and resolve them.  We rewrote the play strategy hash to remove any of the logic that didn’t apply (like waiting for anti-aircraft units before blitzing an enemy with fast moving attackers) and added in a bit of our own where it made since.  Fall has building requirements for a lot of its units, and we spent time modifying the AI behavior to appropriately construct those buildings so it wouldn’t get trapped with the obsolete units.

We still have work to do, and the nature of AI is that it will probably never compare to a real human (one of the reasons that multiplayer is so important to us), but we have come a long way.

Soren: I enjoyed trying out Somnium, the card-based, diplomatic mini-game you recently added. Because the game encourages thinking about probability, I would like to try a version of it where all drawn cards are shown, including the ones underneath the top “banked” cards. Considering that humans are so bad at estimating probability, I don’t think that revealing more information would hurt the game. (To mix things up, you could also remove five randoms cards each game.) Why did you decide to build Somnium, and how was it designed?

Derek: Fall is a huge mixture of options that tie together in a variety of ways.  I wanted Somnium to be the opposite, only one decision, but a lot of complexity in it.  I didn’t want there to be a right or wrong answer but instead to simplify the risk/reward mechanic down to one yes-or-no choice.

I thought about showing the “banked” cards but, in the end, it got cut because I wanted the interface to stay simple, and I wanted the player to be running off of instinct and laboring over the choice.  If we let all the banked cards show, the player who drew the 3 of swords could see all the other swords cards banked and would know that he should draw again.  I wanted to avoid that certainty.

As to why it was designed, we like doing things no one else has done, things people don’t expect from modders.  Functionally, it offers a nice break in long games if you want to jump in and play a few games against the AI or play while your waiting for other players in a multiplayer game.

Soren: The final step of the Fall release process is a selection of scenarios. Would you care to talk about them a little bit? What design ideas are you exploring here that you can’t do with the standard game. What surprises are in store for players?

Derek: The epic random game will always be our bread and butter.  Nonetheless, I wanted the scenarios to give players a unique way to play in the Fall universe.  There are three major series of scenarios following three different characters in a shared world.  Originally I wanted different team members to handle each path, so we could really give them different voices and explore different areas.  Unfortunately, one of the team members wasn’t able to commit to the time required, which is considerable, so I designed two of the series and Randy, one of the team writers, designed the other.

Randy took a different approach than I did, which gives us a lot of variety.  In his series of scenarios, which follow a leader named Decius, the players have the choice to turn good or evil.  They can be good, and become a Malakim leader, or become evil, and become a leader of the Calabim.  Although the scenarios they face are similar, they will be facing them from dramatically different perspectives depending on their choices.  Randy focused on tightly scripted sequences to keep the player very engaged in the story with significant character dialog and quests to perform to get to the next stage.

On the other side I opted to feature unique gameplay mechanics, and the storyline is really just a backdrop to the mechanics.  One scenario, The Momus, has the player fighting in a battle with 6 other players.  An AI player who isn’t in the battle, the Momus, sets all the war declarations and randomly decides to have everyone attack the most powerful player, have everyone attack a random player or have everyone attack everyone every 50 turns.

In the Barbarian Assault scenario, the player starts at a random location in a poor jungle environment with a lot of other very basic civilizations against significant barbarian forces.  The end goal of the scenario is to defeat the only truly settled civilization on the map, the Clan of Embers, who are at peace with the barbarians.  To make it even more difficult, every 50 turns the lowest ranked player is removed from the game until only 5 remain.  Forcing the player to not only survive, but to weaken neighbors to stay out of the lowest rank seat.

In Mulcarn Reborn, the player starts as one member of large team against an equally large AI team.  The AI starts with a much better position, but the real challenge is that the player is always switched into control of the weakest member of his team.  Thus, the player has to be careful to protect all team members and will struggle throughout the game since he will always be playing from the most challenging position.

From an overall perspective, we wanted the scenarios to be able to feature unique mechanics like those described above, all running on the same mod.  We also wanted to be able to store information about goals and decisions made during scenarios so that they could impact later events.  For example, you don’t have to play the Barbarian Assault scenario, but if you do and you beat it the barbarians will be weakened in all the other scenarios.  Likewise, a player that defeats Amelanchier, an elven leader, in the Splintered Court scenario and chooses to give him to a werewolf leader in exchange for their alliance in that scenario will find that they face a werewolf version of Amelanchier in a later scenario.  Allies can be made to help out in later scenarios if quests are accomplished, bonus units can be unlocked, and some units carry over experience from one scenario to another.  The story lines along the three series of scenarios intertwine but so do these quests and events.

Soren: So, the final release of Fall from Heaven is scheduled for December. It’s always hard to draw such a thick line in the ground, so is this really the end? What does the future hold for Fall? Do you expect other modders to pick up your codebase and take it to new places, even if this is the final official version? What are your own personal plans? Will the team stick together for future projects?

Derek: We will continue to support Fall, so we will do patch and art additions after December.  But my plan is to be “done” with the mod itself.  No new features, no major changes.  I’m not going to walk away by any means, but not thinking about new features means that I can spend time on documentation, create some videos to show off some features, and maybe work through my game backlog a bit (I hear good things about this Spore game…).

The team and I have talked about whats next.  We have all been interested in working together again, and we have been really fortunate to find such a diverse and talanted group of guys.  We have received offers to help produce a collectible card game or board game based on Fall from Heaven that we have been thinking about, but as of yet, nothing firm has been planned.  Personally, I don’t know what I am going to do yet, but if it’s something that multiple people can work on, the team will be invited to join.

As for what I expect from modders, I hope that, a year from now, there is an even better version of Fall that everyone is playing.  We have some interesting dungeon-level scenarios in the final version, and I’d love to see someone really take that concept and run with it.  A unit-level, turn-based tactical game would be very interesting, especially with all the cool models the team has made.

See also Part I and Part II…