My Elephant in the Room, Part 3

I gave an Old World postmortem at GDC 2022, which is available on YouTube:

However, I fully scripted the talk ahead of time, so I decided it would be worth taking the time to post the slides online, in three parts to have mercy on your browser.

After shipping Civ 3, one thing I heard often from the Civ 2 community was that the modding tools lacked support for something called “events” which I eventually learned meant a system of triggers and effects that modders could use to give games a narrative arc. It could create a series of chapters, for example, which pushed the story forward when the player achieved certain milestones. I tried out a series of mods to see what was possible and was surprised to see how effectively people could push the Civ engine to create something completely new. For example, here is a Civ 2 Fellowship of the Ring mod which lets you retrace Frodo’s journey from the Shire to Moria, encountering all the events of the book along the way.

So, to enable this type of narrative-focused mod, we added triggers and effects to Civ4 using python as the scripting language and, ultimately, just released the game source code itself, which led to some amazing mods, like Fall from Heaven…

…and Dune Wars, both of which completely transformed the game and proved the amazing potential for both story and modding in 4X games. However, although I had given modders all the power they needed, I hadn’t actually done the work myself on how to make narrative work in a 4X game.

I noticed that events were starting to show up in various strategy games, with increasing depth and complexity. They added real texture to the experience and, perhaps more importantly, variety.

The most interesting mixture of strategy and events was the cult classic King of Dragon Pass, a 1999 game that vanished without a trace on release and then somehow snowballed into a hit decades later as word spread of its wild mix of traditional 4X strategy, clan management, and dynamic narrative. The event system was the star, and your choices largely determined the path your game took, often in wildly unpredictable ways. Each event forces you to make difficult tradeoffs between the demands of various factions, both internal and external to your tribe, just as we wanted to do with Old World.

However interesting this was, it’s not a game I could make. First of all, the game has an actual beginning, middle, and end, and I have neither the interest nor the ability to tell a single, cohesive story. More importantly, Dragon Pass doesn’t tell you the effects of your decisions, you are meant to just intuit the results, which works for some games but not for Old World, a game where transparency is one of the most important design aesthetics. The event system might surprise you, but the direct result of each of your decisions needs to be clear. For me, the promise of a strategy game is understanding what’s going to happen each time you click a button while still not being able to predict the future.

So, because transparency was important, I turned to the world of board games for inspiration. Specifically, the dynamic narrative masterpiece Tales of the Arabian Nights, which comes with a gamebook of over 2,000 events, randomly drawn from a deck of cards, and which both react to and can change the player’s current state. The mechanics for choosing events and how they affect the player are transparent and easy to understand, which was necessary because, as with all board games, the players have to do all the work themselves.

So, for example, having the Wit and Charm trait might help you escape a Vengeful Sorceress while an unlucky player without that trait might end up Ensorcelled, which will affect further events down the road. What inspired me about this system was that it was robust – it’s not an intricate event tree where missing a node might cause a story chain to break. Instead, the events are loosely coupled as they are meant to work together regardless of which random set you draw each time you play the game.

One of the benefits of a loosely coupled system is that multiple authors could create events at the same time, without requiring close collaboration or really even any collaboration. Here are some of the authors of the over 3,000 events currently in Old World, led by our CEO and Creative Director Leyla Johnson. Many of these writers worked on the project at completely different times, creating dynamic story arcs by accident. One writer might add an event that results in your Leader becoming a Drunk while another author, years later, creates an event that only triggers if the Leader is a Drunk, and now we have the makings of a little story. Indeed, as we add more events with each bi-weekly update, the story system becomes more and more cohesive as more events are added to cover all the unusual permutations that might happen for each playthrough.

Here are just some of the possible inputs that the event system can look for and most of these can be changed by the system as well. So, the event system is a virtual deck of events where each one has a potential trigger (such as meeting a new nation), a set of requirements (like a childless leader), and possible effects (like a foreign spouse). It’s a broad, deep system that makes one look forward to each new turn to see what will happen next.

What I might be proudest of is that the multiplayer community for Old World plays with events turned ON – we had assumed that players who wanted to play competitively with each other would be put off by the randomness of the system, but they feel like the game is not complete without it. In fact, we put a lot of work into maintaining an alternate version of the game without events or characters or families or all the things that increase randomness, but I’m glad to say that it was probably a waste of time.

One of the best thing about the event system is that it adds content to the game without bloating the design, without adding new rules for the players to learn. We currently have over 3,000 events, but doubling or tripling that number will only add variety to the game without adding any more complexity. Often, with strategy games, less is more, but this is one place where more is actually more. It’s the same reason why card-based wargames like Twilight Struggle and We the People have become popular – it creates a deeper experience while keeping a slimmer ruleset.

Ultimately, the event system ensures that no two games play out the same way as there are endless possible stories as one event leads into another, changing the path of your game while your in-game choices feed back into the event system itself. I accidentally built a interactive fiction engine inside of a 4X game, and I am very excited to see where our writers – and the modding community – can take it.

In Civ 3, we introduced the bargaining table to the 4X genre, pick and choose what you want to give and to receive from all sorts of potential options. It’s become a staple of the genre.

Here it is in Galactic Civilization 3.

In Total War: Three Kingdoms.

In Total War: Warhammer.

In Stellaris.

Unfortunately, it was a big mistake. My first inkling there was a problem was after Civ3 shipped and people started to complain that the AIs all tended to have the same techs. The reason was that the AIs were using the bargaining table the same way humans did – every time they got a new technology, they would contact all of their friends, rivals, and even enemies to see what they could get in return by trading it away – which cost them nothing but could get them a little something in return.

From the human’s perspective, it looks like the AIs were part of a giant tech cartel and were selling techs to each other at bargain prices, but the AI was simply pursuing the optimal strategy. Again, we have a system where players were ruining the games for themselves because there was no cost to contacting every civ every turn and also endlessly tinkering with how to get the best deal possible. No reason not to put just one more gold piece on their side of the table until you’ve hit the AI’s maximum price for what you are trading away.

There are ways to mitigate this issue, but this is a Cursed Game Design problem as defined by Alex Jaffe in his fantastic 2019 GDC talk, which I recommend everyone should take the time to watch. There is a conflict here between the power and flexibility of the bargaining table and the give-and-take of real diplomacy where flawed personalities come into play and you can’t nickel-and-dime a rival without offending them.

Ultimately, we come back to this – there is no solution here because we are giving the player all the tools to ruin the game for themselves.

Fortunately, Old World had a system in place that could solve this problem by replacing the bargaining table with something else – the Event System! Here is one example – you married a Babylonian many years ago and now because of that connection you must choose a side in the war between Babylon and Carthage.

This system presents the player with interesting diplomatic events and choices that react to the current game state, serving up possible paths to war at a pace that is healthy for the player. Getting rid of the bargaining table was a risky decision because players expect it by now, but the end result could be so much more dynamic and interesting and free the player of the burden of trying to min-max the table. So, when you ask another nation for a truce or for a trade mission or to start an alliance, the game gathers all the events with those specific triggers, throws out the ones that are not applicable to your current situation (such as the events that require a child ruler), and then randomly picks one to present you. Angry nations are still less likely to want to trade with you, but the actual result of a trade mission will still be unexpected, making it a worthwhile gamble to take.

In this example, you can get out of a war if you captured a hostage during combat – a good example of the loosely coupled events I mentioned earlier.

In this example, based on a story from Livy of a meeting of Hannibal and Scipio Africanus after the end of the second Punic War, you are forced to choose who is the best general, either one of your own or one of theirs. You must choose between damaging your own legitimacy or angering your guests. These incidents stir the pot of diplomacy and make the game dynamic.

Here, Rome is offering you one of its unique units, a Hastatus, as a Mercenary, a good example of something we’d be afraid to put on the bargaining table because it would be abusable, but it works fine as a random event which isn’t guaranteed to appear.

Now, here is the great beast, the biggest design challenge for every 4X out there – how to bring the game to a satisfying conclusion after hundreds of turns and an untold, perhaps embarrassing, number hours in front of the computer. I view Victory Conditions as a necessary evil; for awhile, I had hoped that maybe we could do without them altogether, like the Paradox Grand Strategy games do. No one really cares about winning or losing those games, it’s all about the experience, MAN, so maybe victory conditions had become old-fashioned. Maybe I was old-fashioned! The truth is that games like Crusader Kings and Europa Universalis are really more like simulations than they are like games, and one important fact about 4X games is that they are g-a-m-e-s games. Players expect to win, or give up trying.

One thing I did not want to inherit from the Civ series was themed victories – another thing that I helped get rolling back with Civ3 which added Cultural and Diplomatic victories to the traditional Conquest and Space Race. The problem with themed victories is that, because they require a high bar, the player has to aim for them early on, which warps all the decisions made over the course of the game. Aiming for the religious victory? Make sure to always make the religious choice each time you get it as an option! Thus, I needed something more, well, generic, and I needed it quickly because we were playing Old World as a multiplayer game within the first six months of the project.

So, I went for the most boring solution possible – victory points – and it worked out surprisingly well. The reason it worked is because we tied them to city sites – which were a known quantity because we determined how many sites there were at the beginning of the game. If a 4X game without city sites had a victory condition that simply required X cities to win, then it’s no mystery what would happen – the game would have the worst case of ICS ever as players would be cramming cities everywhere. Instead, because we knew that a specific map had only 30 city sites on it, it became easy to pick some threshold which would trigger victory. 

Of course, while victory points are a perfectly serviceable win condition, they hardly fire the imagination, so we needed something a little more interesting, a little more thematic. I wanted a victory condition that dynamically told the story of your dynasty and hopefully even pushed you to play a little differently. I found my inspiration for this from Jetpack Joyride, which had a mission system that encouraged you to achieve one of three random goals, often ones that forced you to play the game differently. Our initial ambition system worked just like this, three random ambitions which give a bonus upon completion and which get replaced by slightly more difficult ones.

Eventually, we made these part of the character system as the Ambitions would be attached to your leader, similar to how it works in Crusader Kings. Further, the Ambitions that come up would be related to the current game state. In this case, the leader is a Builder, so she gets Ambitions to build Wonders. However, they were not initially part of the victory system.

In fact, the initial reason I tried turing Ambitions into a Victory Condition was actually to save the original name of the game, Ten Crowns. The name initially meant that you had ten lives to play the game, ten rulers before the game would end, but that proved too hard to work out in practice as ten rulers could last 50 turns, or they could last 500. Thus, I tried to retcon the name by renaming “Ambitions” to “Crowns” and then the victory condition would be right in the title! However, the team didn’t buy it, and we of course shipped with a different name.

Speaking of which, when we ran the trademark search for Old World, we got a one-line note from our lawyer that Amazon had reserved the word New World for some upcoming video game. I didn’t give it much thought because how often does Amazon actually ship their games…

…yeah, it was fun trying to spend the end of last year explaining on Twitch that we weren’t some weird prequel to New World.

So, the game would be called Old World, and we now had a simple, thematic victory condition – fulfill ten ambitions picked by the rulers of your dynasty and win the game. However, one issue I have often seen with the themed victories of Civ is that, because they can focus on internal progress that is not visible to other players, when one player achieves a Cultural or Religious or Scientific victory, it can some as quite a surprise, and I don’t believe a player should ever be surprised by a “You Just Lost” popup after a twenty hour game. So, we made the simple decision that the AI could not win via Ambitions, they could only win with victory points, which are much more straightforward (and always visible in the upper-left corner).

This type of asymmetry is unusual in a 4X game where the AIs are ostensibly supposed to be stand-ins for human players, but that’s always been a myth. Players don’t actually want the AIs to behave like humans – as a simple example, real human players would usually all gang up on a player who is coming close to victory, regardless of previous relationships. However, if you have the AI behave that way, players will accuse it of treating the human unfairly, of not letting them win fair and square. There’s nothing inherently better about symmetrical design – indeed, I’d say much of the most interesting work in the tabletop renaissance is with deeply asymmetrical games like Root or the COIN series.

Once we broke the seal of asymmetry, it opened up the design space significantly. The AI, for example, doesn’t get events because we wanted events to have meaningful results. It’s ok for one of YOUR events to end a war because you convinced the AI’s heir to seize the throne – however, it wouldn’t be ok for that to happen to you just because the AI drew that event. Trust me, that may sound interesting theoretically, but it’s not going to go down well with players. Further, without symmetry, we could rethink difficulty levels. Instead of giving the AI bonuses, like faster research or cheaper troops, we simply start the AI with more cities than you.

Thematically, you are a new nation in an Old World, somewhat like Rome was in the fourth century BC, a small kingdom centered on one city and surrounded by larger, more ancient empires like Greece, Egypt, Persia, and Carthage. Thus, the AI in Old World plays without any bonuses or cheats at all, simply with a significant head start depending on the difficulty level. We felt players would enjoy knowing the AI was playing by the same rules as the player while also finding a way to provide a challenge for veterans.

Maybe this is a better animal themed metaphor for where I am now as a designer. I’ve gone deep on 4X games during my career. I probably know more about these types of games, what works and what doesn’t work, than is really healthy for a person. My design journey with Old World was truly about putting that knowledge to good use, to not let it go to waste, so that we could push the genre forward in an intelligent, considered way.

Often the best source of innovation is from new blood coming into the industry with new ideas, but I’d like my work now to show that it’s possible to innovate after 22 years in the industry as long as you’re willing to give an honest assessment of what parts of your games have been making players’ lives better and which parts have been making them worse. Many of the mechanics and systems from Civ that I’ve rejected with Old World are ones that I came up with myself. The bargaining table and themed victories, cultural borders and strategic resources, are ideas that I pushed for when I was 24, just out of college.

Old World is a conversation with myself as a younger designer, and I feel fortunate that I got the opportunity to have that conversation. Opportunities like this don’t come along everyday.

Thank you.

Part 1, Part 2

My Elephant in the Room, Part 2

I gave an Old World postmortem at GDC 2022, which is available on YouTube:

However, I fully scripted the talk ahead of time, so I decided it would be worth taking the time to post the slides online, in three parts to have mercy on your browser.

Now, let’s talk about the tech tree, which has always had a problem with golden paths. For example, at the start of Civ 4, you can just click on Gunpowder, and the game will show you exactly which 11 technologies to research to get there. For a game with average playtimes in the hundreds of hours, this becomes a real problem. Indeed, with the popularity today of games based on random runs, fixed tech trees are going against the grain of contemporary design.

I found a solution from deck-building games like Dominion – what if we treated all the techs like cards instead?

So, all techs currently available are in the deck, and you draw four at a time. The cards you don’t pick go into the discard pile along with any new cards that you unlock. Then, you draw four more for your next choice and don’t reshuffle until you have exhausted the deck. This doesn’t just add variety – it also makes the decisions more interesting because you know that when you don’t pick a tech, it won’t be available again until it comes all the way back through the discard pile through the draw pile and then back to your hand. Passing on Ironworking means that you might not see it again for a long time.

The system is certainly more complicated than a traditional tech tree, but it helps a lot that many players are already familiar with deck-building mechanics because of their popularity.

Turning techs into “cards” also enables the idea of bonus cards, which give an instant boost, like a Free Settler, as you see here, or a Great Scientist or a lump sum of Stone. There is a nice short-term vs. long-term tradeoff here as science points are not easy to come by, so slowing your progress for a short-term boost is a difficult choice, an interesting decision.

If you’ve played my last game, Offworld Trading Company, you know that I love stockpiles and markets and building improvements that spit out resources at different rates, so I wanted to give that a try for Old World.

One of the reasons Civ games haven’t done this is that cities can get stuck because they couldn’t build something – if you have no access to Stone, are you just not able to build buildings? What happens each turn then? We solved that problem by borrowing the dynamic open market from Offworld.

Need some iron to build a Swordsman? Just buy it from the market. If the price of iron is too high, then sell off your excess stone first.

Note that although we were making a game with a free market, it’s not a game ABOUT the free market, so instead of having a single price for all goods, there would be some friction, so we used the original system from Age of Empires where the buy price is double the sell price. Thus, while buying resources from the market is always an option for the player, it’s generally best to produce them yourself.

Civ has always had a generic Production resource, sometimes called Shields, Hammers, or just Production, but it went into building everything – Settlers, Warriors, Temples, the Pyramids, everything. I wanted to split this production into different categories so that cities could actually specialize, one might be good at building military, another at developing specialists, another at creating settlers and workers, so we split Production into three categories: Growth, Training, and Civics.

I didn’t originally know what to do with these yields when the city wasn’t producing that type of item, but the stockpiles for food, iron, wood, and stone were working so well that I decided to try that for Training and Civics. I didn’t know exactly what I would use it for but was confident that I would find a use for it over time. As you can see, we eventually found plenty of uses for the global Training and Civics stockpiles.

I put together this chart of all the non-resource yields and how they connect with the game. You can see how every yield occupies a unique place in the game, and even this is an incomplete list! We could also do it by Shrines, by Theologies, by Council, and so on.

Another benefit is that the code considers all of these yields, including food, iron, wood, and stone, the same type of thing, so it is very easy for the event system to suggest some unusual tradeoffs. Would you like to sacrifice some orders for some extra science? Maybe trade your civics to another nation for food in return?

One of Civ’s most troublesome systems is the tile/citizen model, where your food, production, and commerce is determined by which tiles your citizens work. It’s a complicated system. Here’s the Civ 1 city screen – I remember one Civ developer once quipped that it’s best feature was that if you clicked anywhere, it just went away and you were no longer afraid.

Each citizen in a city is assigned to a specific tile, and these tiles all have different yields, and the player has the freedom to move every citizen around every turn. Over time, the designers have added all sorts of automated systems to encourage players not to mess with their citizens. (And remember with workers how automation is a red flag?)

Even worse, for hardcore players, the citizen system is a temptation to lose hours and hours of time with needless micromanagement because there is no cost to moving your citizens around every turn, to eek out some small 1% benefit. These are not fun decisions. There’s a pattern to these systems which suck the player into boring micromanagement – they lack any real tradeoffs, either because they have no costs or because they are temporary. Interesting decisions come from giving something up and from making decisions that you’ll have to live with for the rest of the game. Decisions where you have to think holistically past the information horizon. You aren’t just doing math to figure out what gives you a single extra food this turn; instead, you are making an intuitive decision about what might get you more food farther down the road AND whether food will be more or less valuable to you latter on than it is right now.

Our way out of this problem is to have ALL improvements produce yields on their own, but Citizens could be permanently turned into Specialists on an improvement to boost its output by 50-100% and give multiple other benefits, like extra science. Further, these Specialists would each have a cost, both a food cost and an opportunity cost. Building a Trapper or a Poet doesn’t just cost Food, it also means your city is not building a Chariot or another Settler. More importantly, though, your decisions would be permanent, there is no opportunity to go back to micromanage and rewind your decisions. Permanent decisions are a very important tool for a designer – it gives the player the freedom to move forward and not optimize the fun out of the game.

Finally, let’s talk about Culture, which as I mentioned before, no longer determines borders. Instead, Culture is now measured in four discrete steps that determine the internal advancement of your cities. (Brian Reynolds actually suggested I rename Culture to “Civilization”… but that was too cheeky for me.) Having four separate culture levels per city was a new orthogonal way to measure progress outside of the tech tree. Thus, Wonders are not unlocked by techs but by CULTURE, which also allows us to rotate the ones available each game.

We could also tie your nation’s unique units to a city’s Culture level, and also many urban buildings, so Courthouses would require cities with Developing Culture, Ministries would require Strong cities, and Palaces Legendary ones.

Now that we felt good about the gameplay, we felt it was safe to start exposing publishers to our ideas. This is the first slide of our pitch deck. Note that it used to be called Ten Crowns, and the simple elevator pitch is right there in the first slide: The Fun of Civilization plus the Drama of Crusader Kings. Expect many things to change about your design, including the title, but it’s important to have a core vision that doesn’t change. Indeed, most of the reviews for Old World describe the game exactly this way, as a hybrid of Civilization and Crusader Kings.

The initial impetus for characters originally came from the popularity of Crusader Kings – we could see how much players latch onto real characters who are born, grow, age, decline, and die. It’s not just CK, though, you could see this trend from XCOM through the Total War series up to new games like Wildermyth. People want to care about the characters in their games.

However, the bigger question is WHY would a game like Civ benefit from characters. First off, we should mention that it’s actually impossible to put flesh-and-blood people into Civ because the game covers 6,000 years of history. That’s why we limited our timeframe to just Classical Antiquity, so that the game could plausibly last a few generations. Nonetheless, ignoring the thematic issues, how would characters change the GAMEPLAY of Civ?

To answer that question, let’s talk about another long-standing problem with Civ known as ECS – the Eternal China Syndrome. It means that over time, Civs become more stable, less dynamic, and less interesting to manage. As they add more and more buildings and wonders and laws and technologies, the internal problems get less and less interesting. A bonus that was interesting 30 turns ago now just fades into the background. The only real pressure exerted on the player is from external forces – enemy players.

Character, however, provide a way out of the ECS problem. Buildings and technologies never get old and die, but character sure do. If we attach powers to characters, the game will shift as different leaders take and leave the throne – not to mention new courtiers, heirs, councilors, spouses, and so on. The dynastic landscape is constantly changing. Civ might be a lot more interesting if, say, the map changed every so many turns. Unfortunately, that just doesn’t make any sense thematically even if it would be good for gameplay.

However, characters changing, growing old, and dying, doesn’t just make sense, players EXPECT it to happen. I can’t overstate how significant that is – taking powers away from the player in Civilization is basically a non-starter, yet here is a situation where players would be upset if they DIDN’T lose these powers.

Indeed, we can add powers to characters that would normally be impossible to add to the game if they were accretive in the way Civ usually works – meaning that once you unlock the power, it never goes away. Each character in Old World is one of 10 archetypes, each with special powers if that archetype sits on the throne. Because players only have partial control of archetypes, we can add significant, game-changing abilities here, knowing they will be active roughly 10% of the time. Hero leaders can Launch Offensives (which allows units to attack twice a turn), Orators can hire tribal units as Mercenaries, Tacticians can Stun enemy units, and so on.

Dynamic characters are also a huge boon for diplomacy, an area where sudden changes are always tricky to pull off. In Civ games, if a friendly ally suddenly attacks you, it’s often describe as “random” or “unpredictable” AI, even though it’s crucial for the flow of the game that the AI’s ARE willing to change their opinion of you (or else the Eternal China Syndrome applies to them as well). With characters, however, it’s expected that a nation will change its opinion of you when a new ruler takes the throne. We didn’t design this – it just flows naturally from adding real people into the game. Thus, a problem that has bedeviled Civ for decades was solved in a way that feels natural to the player.

Part 1, Part 3

My Elephant in the Room, Part 1

I gave an Old World postmortem at GDC 2022, which is available on YouTube:

However, I fully scripted the talk ahead of time, so I decided it would be worth taking the time to post the slides online, in three parts to have mercy on your browser.

Welcome to My Elephant in the Room: An Old World Design Postmortem.

Here are the games that I’ve worked on. Spoiler alert: Civilization 3 and 4 are going to come up a lot in the presentation…

I also do a podcast where I interview game designers about why they make games, so check it out if you have time for 4-hour interviews.

Old World is a 4X set in classical antiquity, with a focus on characters. The simplest pitch is imagine a game like Civ but you are actually Alexander the Great, each turn you get a year older, and you will eventually die, and one of your children will take over.

I’ve already mentioned the world Civilization a number of times, so let’s just talk about the elephant in the room. There is not a single preview or review of Old World that does not mention Civ somewhere in the opening paragraph. Every Twitch stream, you’ll always see a “is this the new Civ?” somewhere in the comments.

More personally, here is my own elephant in the room, Civ3 and Civ4. So, why should I go back to make a historical, tile-based 4X game? A Civ-like, so to speak.

Indeed, Civ4 was a best case scenario for a young game designer. It’s the #18 PC game of all-time on Metacritic – and the number one strategy game this century.

It somehow even won a Grammy award, for Baby Yetu, by my college roommate Christopher Tin, which was definitely not even on the radar when we developed the game. At any rate, Civ 4 was the first game I was the lead designer of. Frankly, it’s all downhill from here.

(Editor’s Note: Later in 2022, Old World actually was nominated for a Grammy!)

So, before I explain why I made Old World, let’s talk a little about what the game is and how it was developed.

Here is the very first pitch screen for Old World. It was a lot simpler at the start – in this image, it was conceived of as a tablet game, with the orders system and a resource market but without characters.

Here are some screen’s from the game’s prototyping phase. The earliest playable version, in the upper-right, was multiplayer only, just as I had done with Civ 4, and you can see that by 2018, we had started to add characters to the game.

Here are shots leading up to release. There’s a lot to digest here, but I just wanted to give a sense of how the look and feel of the game changed, year-by-year.

So, now let’s talk about why I made Old World. Why go back to revisit games that stand well on their own and, moreover, are part of a franchise that is continuing to succeed and grow.

One good reason is to know your inheritance, to reexamine it, to look for ideas which were baked into the very earliest version of the game and see if changing them could transform the experience. A good example is what I’ll call “Every Unit Moves” which is how Civ has always worked. Every unit gets to move once per turn.

Civ inherited this mechanic directly from Empire, a game from the 80s which had much of the same tile-based, turn-based combat as Civ but without the scope of all human history.

And Empire got Every Unit Moves from wargames, which you can see here in the form of a monster version of the Eastern front. The conventional way to play a hex-based wargame was for every counter to move once every turn.

The problem with Every Unit Moves is that you are not actually making any tradeoffs, not deciding between your military and your infrastructure. It’s Guns AND Butter. There is no reason NOT to take an action. This is one of the main sources of the late-game slog in Civ games – if you’ve built 100 units, then that means you have 100 decisions to make each turn, and frequent decisions are rarely interesting ones. Indeed, worker automation has become a standard, expected feature in latter Civ games because players know that these constant, low-stake decisions become boring quickly. Because creating more mines and farms has no cost to the player, there is no reason NOT to do so. Of course, players don’t want to have to do the busywork themselves – and it’s busywork because these decisions have no cost, no tradeoffs. Generally speaking, anytime you have to add automation to a game, it’s a red flag, a good time to evaluate whether the thing you are automating is well designed.

This is a slide from our original pitch deck explaining how Orders would work. On the left you can see Every Unit Moves and on the right you can see how it works with Old World. It’s up to the player to decide how to spend their Orders.

Orders didn’t come out of the blue. Instead, it came from a number of sources, including some odd ones like Facebook games. Here is Frontierville, by Brian Reynolds, the designer of Civ2, which took Farmville and added an Energy mechanic that limited the player’s actions.

This was an odd little moment in time, by the way. Three former Civ designers, myself, Bruce Shelley, and Brian, were all working for Zynga, and even Sid was making Civ for Facebook.

Note that the point of the Energy system in Frontierville was to ration out progress (so you don’t burn out on the game) which creates friction and, thus, a potential microtransaction. I didn’t have any interest in that – indeed, I’d say I had anti-interest in it – but I did like how it made you think about what to do with your actions. Suddenly, interesting decisions emerged!

Here’s another example from that time period – Hero Academy from Robot Entertainment. This game gave the player five actions each turn which could be used for moves or attacks however the player wanted. In fact, this game went even further than Old World as units could attack multiple times per turn.

Not all wargames used Every Unit Moves. Eric Lee Smith’s Civil War game used an initiative mechanic where you alternated moves with the other side, which could allow you to move one army many, many hexes in one season while other troops were left in place. Coincidentally, this was the first wargame I discovered as a child, so I do feel like its ruleset has always been in the back of my mind.

Worker placement games are also an influence as they are all about giving the player a lot of options but a limited number of actions. Choosing to do X means you can’t do Y.

The thing all these systems have in common is forcing the player to choose one thing and NOT to choose another. In other words, Guns or Butter.

The Orders system was always part of the design, and it gave me a reason to return to making a Civ-like game as this one simple change would radically transform the game. There were many other new things I wanted to try, but this one change was the reason to get started.

Of course, like all design, just because you have an idea, doesn’t mean you know how to implement it. There are probably 100 ways to make an Orders system work. I know because I tried 99 of them. Here, in the very earliest, multiplayer-only, version of the game, you could actually buy and sell orders freely, just like the other resources. It was super interesting strategically but was perhaps TOO interesting as it warped the whole game towards who could buy the most orders. 

Here are some of the different versions of the orders system we tried – one on a real-time clock, one where you could stockpile unused orders between turns, one with a hard cap, and so on. We eventually settled on a fatigue system where most units could move three times a turn but could move farther via a Forced March which had a steep cost.

However, most of these other ideas didn’t get tossed away. Instead, we hid them away behind late-game laws – Coin Debasement unlocks buying Orders, for example, and Elites unlock Stockpiling Orders. That’s a useful trick if you ever try something which is interesting but just too powerful. Don’t cut it right away – instead, shuffle it away to the late game where it can stay interesting but rare.

I was worried about the ramifications of Orders for the AI but, in the end, it worked out for the best because it enables the AI to actually attack on their own initiative, instead of coming at the player slowly, turn-by-turn, allowing itself to get picked off. Multiplayer sessions, on the other hand, were fascinating – victorious teams were usually the ones who saved Orders for their economy as roads became hugely important in order to make units more Orders efficient.

The strength of the Orders system was the huge possibility space it created each turn because there were so many different ways to spend your Orders. This vast, sometimes intimidating, space led to the Undo feature which kept players from feeling overwhelmed – players could try out multiple ways to launch an attack and then just change their mind and do something else.

It was initially intended just to help with misclicks but became a pillar of the game (and was awesome for debugging too).

Indeed, undo turned out to be one of our most popular features – it was often one of the 3 or 4 bullet points listed in the pro column for reviews. It’s an unexplored area – a number of games like Invisible Inc and Into the Breach have experimented with limited undos but we didn’t find that necessary. Players appreciate being able to play the game however they want to play it. (It helps, of course, that we have deterministic combat. Games with a lot of output randomness would not be a good match for an undo button.)

While we are mentioning quality of life features like Undo, I also have to mention our help system as our tooltips have tooltips. As is often the case with strategy games, the tooltips refer to something that you need more information on, which usually means taking a trip to the manual or the wiki or the in-game encyclopedia. In Old World, you can either middle-click or shift-click to…

…just open up a new tooltip off of the previous one. Now, we might want to know what Rancher means, so we just…

…open up another tooltip. Hmm, I wonder what Civics are…

…and you can go as far as you want. We have Infinite tooltips!

I definitely want to mention that the designer who pioneered infinite tooltips is Jon Shafer, who worked with me on Civ 4 and was the lead designer of Civ 5. Old World was the second game to implement them, and I expect them to become a standard convention for turn-based games in coming years.

I wrote this line in a column entitled Water Finds a Crack back in 2011, and it is now permanently the most popular post on my blog. It’s sort of taken on a life of its own, showing up in random videos and other GDC talks, and it is also super applicable to ICS. What is ICS, you might ask?

ICS means Infinite City Sprawl, the bane of Civ designers. Basically, players learned early on that the most optimal way to play was to squeeze as many cities onto the map as possible. Each iteration of the game tried a different method to fix the issue: Civ 3 had Corruption, Civ 4  had Maintenance, Civ 5 had Global Unhappiness, but they all are unfun bandaids.

Interestingly, this is a solved problem for space 4X games – Master of Orion, traditionally considered the first space 4X, didn’t have this issue because “cities” are equivalent to planets, so one city per planet. Thus, a fixed number of cities each game.

Endless Legend and Humankind do something similar with regions, cutting the board up into territories at the start of the game and then allowing only one city per territory. We tried something like this with Civ 3, but I was unsatisfied with it because I felt that your choices should determine the borders between cities, not the designer’s hand before the game begins. 

I found a middle ground between limited sites and dynamic growth by putting preset city sites on the map but then all the border growth after that would be determined by the player’s actions. New borders would come from tile specialists and urban improvement. Thus, we rejected one more piece of our inheritance as borders have always come, at least since Civ 3, from a city’s culture.

Finally, city sites also gave us a natural place for tribal camps, which helps drive conflict and ensures that the player needs to balance expansion with military and not just spam settlers.

Civ 5 introduced one unit per tile, a big change for the series. Definitely another elephant in the room, although in this case not my elephant. At any rate, the popularity of Civ 5 and 6 meant that a lot of players would expect one unit per tile. 

Debates rage over Civ4’s stacks-of-doom vs Civ5’s carpets-of-doom. However, making a big change to combat was needed for Civ 5 as each iteration needs to shake things up. Just as Old World needs to justify its existence, each version of Civ does too.

We started with one unit per tile simply because it was the easiest thing to code, but I thought I had a clever solution for a better stacking system – simply put, the player could stack units, but if that tile was ever attacked, each unit on that tile would be hit equally. Therefor, because Old World has no counter-attacks, stacking a bunch of units on the same tile would be very dangerous. So, we were naturally encouraging players not to stack their units, but they could be tempted to if stacking an extra unit on a tile got them a kill.

However, before finalizing anything, we decided to try classic one unit per tile combat just to see how it felt, and surprisingly, it felt great. It works because Old World doesn’t have counter-attacks like all of the other Civ games. Combat is essentially split across multiple turns, with units trading blows until one dies or retreat. Each separate attack is without risk – only the target is damaged. The Orders system made this necessary because allowing defenders to damage attackers is akin to giving them a free Order-less attack, and defenders almost always have an advantage in these types of games anyway.

Ultimately, however, I wanted to reward attacking because attacking is more fun. Pictured here is a very rare version of Risk – only 1000 copies were ever made. It’s the one Rob Daviau worked on before designing Risk Legacy, and it made one huge, crucial change to the game.

It added objectives, which are crucially important because you win the game not by conquering the world but by simply being the first player to achieve 3 objectives. They were so important to achieve that they changed how people would play the game. Instead of playing defensively and turtling up in Australia, players would want to go on offensives each turn to grab these objectives, which go to the first player to achieve them. You would overextend yourself to try to actually control Asia just because you wanted that objective even though you knew you would lose Asia the next turn. It made for a more dynamic game because attacking is fun.

Generally speaking, taking actions in games is more fun than making it harder for other players to take actions. Conservative counterplay is less fun for you, and it’s definitely less fun for your opponent.

Civ players are used to beating the AI by allowing it to kill itself against your heavily fortified units, so taking away counter-damage is a big change, but again, I wanted to reward attacking because attacking is more fun.

However, allowing players to overload a combat front via stacking to get a kill took away the opportunity for the defender to counterattack, so one unit per tile was actually a very important piece of the puzzle to making our combat system work.

Really, you have to look at it holistically, as the three systems support each other. The biggest problem with one unit per tile in Civ is that it leads to traffic jams where units clog up the tiles between cities.

City Sites alleviate this problem because we can enforce a very high minimum distance between cities, much higher than in a Civ game. The Orders system, on the other hand, ensures that units don’t block each other while moving through tight passages because every unit can make multiple moves per turn if necessary. And, as I mentioned, one unit per tile balances some of the extremities of the Orders system by making it impossible to form stacks to kill a unit in a single turn, which removes the ability to counter-attack. So, the three systems fit together nicely, buttressing one another.

Part 2, Part 3

Designer Notes 67: Old World

In this episode, Tom Chick and Bruce Geryk interview Soren and Leyla Johnson about their new historical 4X, Old World. They discuss whether the Old World timeline really makes sense, how they approach Early Access differently, and why Soren didn’t want to give up on the name 10 Crowns.

Games discussed: Old World, Civilization series, FrontierVille, Neptune’s Pride, Crusader Kings series, King of Dragon Pass, XCOM, Empires of the Middle Ages, Alpha Centauri, Seven Cities of Gold, Colonization, Power Grid

Old World Designer Notes #11: The End

The following is an excerpt from the Designer Notes for Old World. The game, a historical 4X set in classical antiquity, released on July 1, 2021, and is available for purchase here.

The hardest part of a 4X to design is the ending. Truthfully, no one has ever done this well, and during development I often reminded myself of the old joke about how you don’t need to be fast to avoid getting eaten by a bear – you just need to be faster than the person next to you. I don’t think it’s possible for a 4X game to consistently provide a satisfying, climactic ending; the genre’s strengths come from the vast possibility space, the lack of guardrails, and a commitment to reward smart play fairly. All of these factors make a good ending harder to achieve consistently. Once the turn count reaches well into the hundreds, it’s almost impossible to predict what the game state will be, so where to draw the finish line is, at best, a guess.

Indeed, I generally view victory conditions as a necessary evil. Without them, a 4X game suddenly becomes an aimless simulation where the players set their own goals. Simulations can be great games, but they tend not to be a good format for turn-based games with transparent game mechanics. They work better in real-time with their systems hidden inside black boxes; the lack of control is key to keeping the player on the edge. (Thus, Paradox grand strategy games follow this model.) Old World is a game, not a simulation, so it needs victory conditions; however, I have grown disenchanted with the increasing specialization of victory conditions in 4X games, which provide a path to victory by just focusing on one aspect of the game, such as culture, religion, or diplomacy. The problem with these victory conditions is that they are so specialized that the player needs to aim for them from the beginning of the game, which turns their play into an exercise in predetermination. Want the religious victory? Then make sure to always choose the religious option!

Hence, themed victory conditions were out for Old World, but we needed something in place, especially since we were playing MP within a few months of early prototyping. The simplest solution was just good old-fashioned victory points, which have the advantage of supporting multiple play styles while also providing a faster victory for a dominant player. We attached points to cities and to wonders (and later to culture levels and techs at the end of the tree), so that the game would instantly end when a player passed the threshold determined at the start of the game. It worked extremely well from a mechanical perspective, which is all we needed for early MP testing before the game even had characters, families, and events. We were surprised that victory points had never been tried in a Civ game but realized that the city site system enabled it to work. Because we knew the exact number of sites on the map at the beginning of the game, coming up with a point threshold for victory was a fairly straightforward calculation; in a game with unlimited cities, victory points would just be one more mechanic pushing players to squeeze in as many cities as possible.

Of course, while victory points were a perfectly functional victory condition and might work fine in multiplayer, they are not particularly interesting, especially for a game where the player is purportedly creating the story of their dynasty. I wanted a victory condition that would dynamically adapt to the path the player’s game actually took; the player would need to achieve many tasks to win, but these tasks could be chosen on the fly, and the player could change directions, turning away from a militaristic victory to a more peaceful one. The direct inspiration came from Jetpack Joyride, which showed the benefits of giving the player a random selection from a large and diverse set of missions. In this infinite runner mobile game, players are always given three “missions” which encourage them to play the game in an unusual way (“Have a near miss with a missile” or “Rub your head against the roof for 500 meters”), which is a surprisingly effective way to change the feel of a game that one might play over and over again hundreds of times. By redefining the player’s goal from just surviving as long as possible to something fresh and unusual, the game itself was constantly mutating for the player, making it much more replayable. A set of varied goals is now standard practice in mobile gaming, and if it can work for a game played with a thumb, it can also work for a 4X game. In Old World, missions could push the player to try out new paths and strategies – if a mission requires me to build eight shrines, for example, maybe I’ll try out pagan religion for the first time?

The first implementation of ambitions functioned similarly to the missions of Jetpack Joyride – the player drew three random missions at a time which rewarded the player with Legitimacy, a value representing the achievements of your dynasty. The other sources of Legitimacy are the cognomens of your current and former leaders, with diminishing values each time there is a new succession, and from event choices, giving the writers an interesting lever for short-term decisions that damage your standing. (At their best, these events can highlight the cultural gulf between Classical Antiquity and today – that your people should always look to you for their answers, not to some foreign discovery or faith.) Legitimacy provides two effects that are key to making the game work – each point of Legitimacy gives a tenth of an Order per turn and +1 opinion to all of your families. The former was important as a simple way to slowly increase the Orders pool as the number of units grew, although certainly at less than a linear rate. The latter was important as a consistent way to counterbalance negative family opinion from city Discontent, suggesting a simple equation where the player needs to accomplish greater and greater tasks to make up for the natural tendency of each family to want more and more cities and favors and gifts.

The ambitions themselves automatically become harder and harder as the player fulfills old ones and draws new ones. Each ambition has a tier range so that they are not part of the ambition deck until you have, for example, completed more than two ambitions but less than five. Thus, the ambitions keep pace with the player’s progress in a very natural manner, assuming the tier ranges are correct. Drawing inspiration from the ambition system in Crusader Kings, we eventually attached the ambitions not to the nation but to the leader, each ambition would be a major goal that they try to accomplish in their lifetime (which then led to legacies, which are ambitions successors can finish for a limited time after the leader’s death).

We now had a dynamic and robust ambition system which fed the leader increasingly difficult goals to achieve. The last step was to tie it to a victory condition, and the answer turned out to be right in front of our faces. The original pitch for the game was that one game session lasted literally ten lives, the first ten rulers of a dynasty. Hence, the working title of the game was 10 Crowns as a metaphor for ten lives (and using numerical digits because I had maniacal plans to make sure the game was listed first in any alphabetical game library). However, I never revealed publicly why the game was called 10 Crowns (although Allen Cook guessed correctly on the Gamers with Jobs podcast shortly after the announcement). I kept it a secret because I wasn’t sure playing ten rulers would actually work, and it turned out I was right. Good old-fashioned randomness meant that ten rulers could last 50 turns, or they could last 500 turns. There’s just no way to balance a game with that much variability, and I was afraid to put my thumb on the scale and start covertly killing off rulers who had overstayed their time on the throne. However, I now had a simple solution to make the name 10 Crowns work – the “Crowns” would now be the ten ambitions you needed for victory, not the ten rulers of your dynasty! Problem solved –  I took the rest of the day off!

Unfortunately, the team didn’t buy it and staged another mutiny where they expressed their very reasonable concerns that ambitions were not “Crowns” and vice-versa. I suggested renaming ambitions to crowns, but it was too late. We needed a new name, and we needed it soon as we were shipping as an Early Access title in a couple months. I considered calling it The Great, but there was apparently some Hulu show about Catherine the Great in production (which turned out to be amazing, by the way). Someone threw out the name Old World, which was stupidly close to Offworld and yet also the perfect name for a game from this era – simple, accurate, and evocative. I did the research and was surprised that no one had ever made a game called Old World although there was apparently some MMO being made by Amazon called New World? That seemed hardly worth considering because Amazon always conveniently cancels their project before shipping (or, just to mix it up, sometimes after shipping), so I shrugged that one off!

So, we had our new name, Old World, but in the process we had accidentally hit upon a victory condition that was varied, dynamic, and thematic – simply complete ten ambitions and achieve victory. We weren’t quite done yet, however, because there wasn’t actually a way to lose the game. Obviously, the AI could still wipe the player out, but the real problem would be if the player falls behind and gets stuck in purgatory where the AI is dominant and the higher-tier ambitions are too difficult to achieve. I spent a lot of time waving my hands and talking about how the Paradox games didn’t even have victory conditions! We were being so old-fashioned! Gamers today set their own goals! Kids, go outside and make your own fun!

Of course, one might ask, wouldn’t the player lose if the AI fulfills their ten ambitions first? Well, they would if the AI actually got ambitions, but I knew that would be a mistake. One of the other problems with specialized victory conditions is that because they either measure internal progress (cultural or scientific victory) or something orthogonal to most of the game (religious or even diplomatic victory), an AI victory can come as a sudden surprise. I was sure of one thing above all – a surprise ending to a 20-hour 4X game is not a good ending. Without ambition victories, the AI would clearly be playing a different game from the human, which meant that Old World would be an asymmetrical game. In reality, it just meant that we were admitting that the game was asymmetrical because there is no such thing as a symmetrical 4X game. The genre likes to pretend that it’s symmetrical, like chess or most board games, but a single-player 4X always orbits the human.

The most obvious place that proves this fact is diplomacy. I’ve already discussed how tech trading needs to be controlled to make sure the AIs don’t play like humans, but the same pattern is true across all diplomatic decisions; humans will not accept AIs which behave like humans. For example, in a free-for-all multiplayer game, no one would find it unusual for the losing players to gang up on the leader (including by former allies) to stop that player from winning. On the other hand, most people hate it when the AIs automatically declare war on the human as the player gets close to winning. (I say most people because there is a small minority of players that prefer AIs to play like humans, which is why we added the Ruthless option.) The only difference between these two scenarios is whether the player is facing other humans or their own machine, so the AIs are coded to play like NPCs in the human’s playground, to follow their personalities and remember their friends/enemies but not to try to win at all cost. This arrangement between humans and AIs is inherently asymmetrical, and there are plenty of other ways humans play that AIs should never replicate, such as leaving their cities empty of defensive units because they know exactly when their enemies might or might not attack (or that a war declaration could be easily undone with a recent auto-save). Furthermore, there is nothing inherently superior about symmetrical games versus asymmetrical ones. In fact, even though board games are usually thought of as symmetrical by default, much of the most interesting design work in the current tabletop renaissance is with deeply assymetrical games.

The trick is knowing where symmetry is important. Combat, for example, needs to maintain perfect symmetry as the human interacts directly with the AI using the same pieces on the same tiles with the same rules – the more the player needs to understand the AI’s potential actions, the more important symmetry becomes. By stepping away from the myth of human-AI symmetry, we could also create a competitive game without giving the AI any per-turn production or research advantages, which is the normal way to strengthen the AI in 4X games. Instead, in Old World, the AI simply starts with more cities than the human does as the difficulty levels increase but otherwise plays by the same rules for city growth, unit production, tech research, and so on. The situation was even thematic for the era as the player is basically in the same situation as Rome at its founding – a single city surrounded by much older and much more established empires. You are leading a new people in an old world.

Although the AI could not have ambition victory, they still need a way to win to make sure Old World is a game and not a simulation. Fortunately, the answer was already available via the original victory point system we implemented for multiplayer. By simply leaving the victory point system on, the game could handle situations where an AI starts to dominate. At competitive difficulty levels, VPs serve as a dynamic game timer that the player is racing against to fulfill their ten ambitions. Victory points also handle the opposite situation, where the player starts to dominate, via double victory, which triggers when the player doubles the score of the next closest AI. Double victory helps Old World avoid the worst of a common problem once victory is certain but the player still needs to push End Turn fifty more times.

Thus, ambition victory centers the game on the achievements of the player’s dynasty. Indeed, we hope to push this further in future updates by using the event system. The player will get opportunities for new ambitions not by picking a random ambition for each family but via handwritten events customized to the current situation. Maybe the Romans kill the unit your queen is leading in the field, and the game gives you a new ambition to get revenge by killing Roman units or capturing their cities. Ambitions and events can be a powerful framework to help the player build their story of victory. When the system works together as a cohesive whole, the game truly comes alive. An unanticipated strategy has emerged in multiplayer that shows the system at its best. Negative cognomens, such as the Unready or the Bloody, are possible if a ruler loses enough units and cities to outweigh their positive accomplishments, and a negative cognomen will reduce the player’s Legitimacy, which can be devastating. Thus, players discovered that the best time to attack is when there is a new ruler on the opponent’s throne who hasn’t achieved anything yet, saddling them with a negative cognomen from the war losses. It’s a cunning strategy, and one reflected in real history where wars were often started to take advantage of a new ruler with a shaky hold on the throne. As a designer, this mirroring is difficult to achieve, and I’ve fallen short of it many times, intentionally or not, but I don’t mind celebrating these moments when they do happen. Opportunities like this don’t come along everyday.

Old World Designer Notes #10: Diplomacy

The following is an excerpt from the Designer Notes for Old World. The game, a historical 4X set in classical antiquity, released on July 1, 2021, and is available for purchase here.

Some mistakes are inevitable because the idea behind them is so hard to resist. When tasked with improving diplomacy in Civ 3 – and not just in comparison to Civ 2 but to Alpha Centauri as well – the most obvious decision was to add a bargaining table, where players can customize their deals as much as desired. Instead of relying on the AI to suggest interesting deals, the player could pick and choose whatever they liked – gold and iron for a city, or a luxury for a technology, or some gold each turn for open borders and a map, or any other combination. It was one of the standout features of Civ 3 and became a standard feature of the 4X genre. Unfortunately, it was a big mistake.

The first issue that appeared was how quickly the AI traded technologies with each other. I had allowed the AI to trade techs freely because the human was now free to do so via the bargaining table. The problem was that unhindered tech trading will always be problematic because the player that “gives away” the tech is not really giving away anything. The only downside is that a rival is getting access to a tech, but the usual rationale is that if you don’t give the Babylonians Ironworking, then someone else will do so, and you’ll just miss out on whatever you would have gotten from them. Even worse, if you didn’t play the tech trading game, the AIs would make the trades anyway, and it would appear to the human like you were playing against a cartel of AIs that all seemed to be at the same tech level. According to the rules of the game, the AIs were just playing optimally; the problem was that the rules were bad.

With Civ 4, I made a number of changes to improve the bargaining table. Tech trading was severely limited by making the AI unwilling to trade techs with players they disliked (or even just didn’t like enough). They would also simply refuse to trade techs with players who were in the lead, and as these rules applied to humans and AIs equally, the worst excesses were avoided. The system also prevented trading lump-sum items for per-turn ones, so no more taking a giant loan from a neighbor and then declaring war to get out of the payments. Players could now ask the AI to fill in either side of a deal so that, in order to play optimally, you weren’t forced to add 1 gold to either side to see if the deal crossed the magic threshold.

However, the system was still fundamentally flawed because it gave the player too much control over diplomacy. The best strategy remained contacting the AIs as frequently as possible to extract whatever deal was most advantageous. Diplomacy did not feel like diplomacy; it felt like walking up to a vending machine and picking the option that suited you best. Players were ruining the game for themselves by playing optimally, however boring that might be. Diplomacy had become simply another part of your economy, a reliable source of gold or technology or resources, depending on what you needed. Ideally, diplomacy would be unpredictable at times, would be outside of the player’s control although still reactive to the game state, and would allow the player to be opportunistic but also force them to face difficult decisions.

Fortunately, Old World had a system in place that could do all of those things easily – namely, the event system, which is both inherently unpredictable yet also reactive to the current game state. The event system became more flexible and robust the longer we worked on it, and we already used it for both the tutorial system (teaching you about harvesting the first time your Scout steps onto a neutral resource) and the mission system (every assassination has a small chance of triggering an event which can spin the game off in unpredictable directions). Using events to manage all diplomatic interactions between the human and the AI was perhaps a risky decision – players had gotten so used to the bargaining table that leaving it out could come across as a step backwards – but it would make diplomacy more dynamic and also force us to expand the event system to include more diplomatic triggers, requirements, and effects, which could pay off in the long run for scenarios.

The first step was handling all diplomatic changes (war, truce, peace, alliance) via the mission system, which would then trigger an event when the mission was complete. Thus, the event system would need to know about the proximity of different nations, their relative strengths, and the current state of any war (who is winning and by how much). The event options would need to be able to set diplomatic states, handle trade and tribute, and add memories and relationships which modify opinions. Each of these elements might not have been part of the event system if it wasn’t responsible for diplomacy, but once they were added, they could be used for any event, making the whole system more varied and robust. Once all the diplomatic elements were in place, the event system could do all the work of a traditional 4X diplomatic system but with the upside of all the other features of events, such as testing to see if your spouse was from their nation or if you share a religion or if one of your cities used to belong to them.

Because diplomacy was now less predictable, it was important that the player couldn’t ask for something too frequently, or else the player would be encouraged to keep trying until they got the event they wanted (exchanging a vending machine for a slot machine). To avoid this problem, we gave diplomatic relations (such as asking for peace or a trade deal) a cost in Civics or Training, and we also limited diplomacy to missions that could be conducted by your Ambassador. Not only did this bring another character to the forefront, always a good thing in Old World, but it meant that only one diplomatic mission could be pursued at a time and, more importantly, that missions would take multiple years to produce a result. The latter was important so that players were not tempted to use the undo system to reverse unsuccessful diplomatic missions and maybe offer tribute to end a war instead (in other words, trying to protect the player from themselves). Of course, the most unpredictable aspect of diplomacy is if your ambassador dies in the middle of a mission (although we do warn players before they send a sick ambassador to ask for a truce).

I was actually expecting more pushback from players for the lack of a bargaining table (the inability to found cities anywhere has been more controversial), so I hope that players appreciate how much the event system makes diplomacy more dynamic and unpredictable. Unlike many other systems I’ve designed, the health of the diplomatic system rests not just on the quality of the underlying algorithms but on the simple quantity of diplomatic events our writers have created, and by this metric the system will only get better as our writers add new events with each weekly update.

Old World Designer Notes #9: Events

The following is an excerpt from the Designer Notes for Old World. The game, a historical 4X set in classical antiquity, released on July 1, 2021, and is available for purchase here.

After finishing Civ 3, I spent a long period of time engaged with the community to learn what they did and didn’t like about the game, with an eye towards the patches and a potential sequel down the road. In the modding community, a common complaint was that our editor lacked something called “events” from Civ 2, which I learned meant a system for connecting triggers and effects that could give games a narrative arc. I didn’t immediately grasp the potential of such a system until I tried out the community scenarios. I remember two that stood out to me because my actions pushed the story forward – one recreated the journey to Moria in Fellowship of the Ring and the other was a retelling of Odysseus’s Mediterranean wanderings in The Odyssey. Neither was particularly replayable, but they both were interesting because they had functional stories built from only a layer of events sitting on top of the base game.

For awhile, I wasn’t sure what to do with this discovery because I didn’t have a clear vision for how events could make a Civ game better, but I did ensure that our Python layer for Civ 4 had both triggers (calls from C++ to stub Python functions) and effects (the stub functions could change the game state). This system led to many interesting Civ 4 mods and scenarios as well as allowing the team to write a series of events for Beyond the Sword, many of which focused on natural disasters. While these events were a first for the series, it also represented a bit of an evolutionary dead end as they were not carried forward into Civ 5.

Nonetheless, I believed that events could play an interesting role in 4X gaming, and indeed noticed that many other strategy games, including the Galactic Civilization series, the Crusader Kings series, and Stellaris, were using them more and more frequently. However, my biggest inspiration came from the most text-heavy genre of them all – interactive fiction. The genre was experiencing a renaissance in the UK, led by Inkle (80 Days, the Sorcery series) and Failbetter (Fallen London, Sunless Sea), and I began to quietly haunt the GDC Narrative Summit while also interviewing the writer/designers for my podcast to see what I could learn. I also finally found a physical copy of King of Dragon Pass, a cult hit from the 90s that was still unavailable online, perhaps akin to finding a disc of The Velvet Underground’s Loaded before mp3s. The game was a wild mix of traditional 4X strategy, clan management simulation, and dynamic narrative built around random events that triggered based on hidden factors and which had unknown effects (but which the player would slowly learn to anticipate).

I began to appreciate how important narrative could be in a video game, how it could pull players into a game world much more effectively than by simply making numbers go up, a trick that was perhaps starting to get old in 4X gaming. To be clear, I had no interest in writing a story with imaginary characters and a beginning, middle, and end (Who did I think I was? A writer?!?), and I was also fairly uncomfortable with opaque triggers and effects which kept players in the dark, forcing them to play by feel. Transparency is an important part of my design aesthetic, and while it could be violated for effect occasionally, I didn’t want to build an entire system around it.

Perhaps unsurprisingly, inspiration came to me via board games, specifically Tales of the Arabian Nights, a brilliant adventure game built around the ancient folktales of the Middle East. The game comes with a giant storybook of over 2,000 events (some by Paul Murphy, writer on Civ 3, in fact) which are randomly drawn from a deck of cards and then cross-referenced with the character’s location, the player’s decision, and the roll of a dice. Although another player would read the event to create suspense and hide the outcome of the choices, the exact mechanics of how the game chose each event was transparent by necessity as the players did the work of finding each event themselves. The readers would see how the events were constructed and what would have happened if the active character possessed a certain trait, so that they in turn would anticipate the range of possibilities on their own turn.

The game had an interesting system for linking events via loose connections based on skills, traits, or treasures the character picked up over the course of the game. A character might become Ensorcelled early in the game, have the result of a later event change based on being Ensorcelled, and finally have the opportunity to remove the trait with a further event. None of these events were necessarily going to happen during a single playthrough and, often, potential narrative arcs were left dangling without resolution. Nonetheless, when a cohesive series of randomly chosen events do come together to tell a real story, it’s a magical feeling. What appealed to me about this system was that it was robust; there was no intricate event tree that could fail if one node was changed or stopped working. Furthermore, loosely coupled events could be written by a large group of authors who could work in parallel without close coordination. One writer might add an event which results in your leader becoming a Drunk, and a different writer could add an event that requires the leader to be a Drunk (or, if so, forces the player to take a certain option), and these two writers have now created a little story without having discussed anything. Indeed, they could have been working years apart and perhaps not even know each other. I am very excited for the potential of community event packs which can co-exist with our 3,000+ events (and with each other) to create new narrative possibilities.

Thus, I had the basic blueprint for an event system – it would be a virtual deck of event cards which each had a potential trigger (such as meeting a new nation), a set of requirements (a childless leader), and possible effects (a foreign spouse). When a trigger occurs, the game finds all events in the deck valid for the current game state and then chooses one randomly based on the weight, probability, and priority values of each event. The backbone of the event system are the subjects, which are a set of game objects randomly chosen for each potential event. Subjects can be anything from a character to a city to a family and even to a law or technology. Each subject can have multiple tests to find the perfect one – an adult child of the leader who is NOT the heir but IS a bloodthirsty schemer is a good example of a very specific subject that might mean bad things for the current heir. Further, the system can test for relationships between subjects, such as two nations that are at war with each other, the religion of your spouse, or a character who is vengeful against another. The event options can affect any of the subjects and can also be unlocked based on the current ratings or traits of the leader.

There are many other wrinkles to the system – for example, it is possible to link subjects together for multi-step events such as a duel – but this basic architecture enabled the writing team, led by Leyla Johnson, to create a wide variety of events that fundamentally change the flow of the game. Dynamic random events disrupt the steady flow of a 4X game, which can often devolve into deciding which bucket should fill up and at what speed. Perhaps an event provides a sudden burst of Orders that allows the player to move enough units to defend a city that was just about to fall. Or, perhaps sending one of your children out to explore the world leads to the founding of a new world religion. Or, perhaps a severe combat injury means that Alexander will need to abandon the field and become the Learned instead. A common theme of Old World’s design is avoiding predictable and boring games where the same actions lead to the same results, and the event system is an important tool to ensure no two games are ever the same.

Old World Designer Notes #8: Opinion

The following is an excerpt from the Designer Notes for Old World. The game, a historical 4X set in classical antiquity, released on July 1, 2021, and is available for purchase here.

One of the defining features of 4X games with an Eternal China Syndrome is that the conflict primarily comes from external sources, from barbarians and rival nations. With certain random maps or diplomatic situations, players can often be in a position where they are largely unchallenged, and without any other pressure, the game slowly slides into auto-pilot. Old World addresses this problem by adding internal pressure from families and religions, each of which have their own opinion of you, just like a foreign power.

Making internal opinions matter requires a pair of mechanics – how the opinion is determined and how it affects the game. Designing the latter was the easier task as family opinions could simply affect cities and units if they each had a specific family type. Thus, each new city would be assigned a family, and units produced by that city would also be attached to that family. Then, each city and unit could get various bonuses and penalties depending on the opinion of the family itself. Cites belonging to a friendly family have reduced Maintenance, units of an angry family have a combat penalty, and so on. Perhaps most importantly for family opinions thematically, an unhappy family has a chance of producing rebel units which are capable of capturing cities if not defeated.

Attaching units to a family was a much debated topic as doing so made it even more difficult to differentiate units of each nation and tribe. There are only so many team colors available, and using a secondary color for families was only partially effective. Thus, we added a distinctive banner shape for each of the ten family classes and also did not show the family of opposing nations to simplify the mix of colors. Communicating family type is still difficult, but without assigning units to families, it would be impossible for family opinion to affect units, which risked dulling the entire system.

The trickier question is what should determine family opinion. Some modifiers were easy to add – families prefer having more cities, like having their cities closer together, like having members in the royal succession, and dislike having cities with high discontent. However, to keep things understandable, we didn’t want to add too many modifiers, so we split them up between the different family classes, which also added to their flavor: Champions prefer having the largest army, Clerics dislike cities without a religion, Patrons like having cities with Wonders, etc. A diversity of family opinion modifiers also makes it more likely that each family will have a different opinion of you, which makes for more interesting gameplay as angry families could become jealous of pleased ones, fertile ground for the game’s dynamic events.

However, although families now had opinions with inputs and outputs, the system felt very abstract; it’s harder to relate to a family than to a specific character. Further, we had a separate problem that although characters had opinions of you, the opinions didn’t seem to matter all that much unless they were in the court. We addressed both problems by creating family heads whose opinions were directly applied to their family’s opinion – if the head had a +100 opinion of you, then his or her family’s opinion would be modified by +100. Now, if a player wanted to change relations with a family, all game systems that involved a character’s opinion could now apply; for example, the player could improve relations with a family by conducting an Influence mission with the family’s head. Conversely, the event system could give an option that might offend the family’s head which would then reduce the family’s opinion.

Another important vector for family opinion would be religion. Characters adopt religions local to their family’s cities, and once enough family members follow the same one, a family officially adopts that religion. After that happens, the religion’s opinion is applied directly to the family’s opinion, so now all missions and events affecting a religion could also potentially affect a family. I added religion to Civ 4 primarily to create a reason why one rival nation would like you and another one wouldn’t. Religion serves a similar purpose in Old World, except that it now applies to tribes and families as well. Religions also have heads that work similarly to the heads of families; the opinion of the head would be applied directly to the opinion of the religion, which would then be applied directly to the opinion of nations, tribes, and families that follow that religion. Thus, religion heads are very important characters that touch multiple levels of the world, a new way that a character’s opinion could matter.

I began to use a river network as a metaphor to describe how opinion flowed throughout the game. More specifically, opinion only ever flows in one direction, a fact I discovered when the game crashed after a character’s opinion boosted their religion’s opinion which then boosted their family’s opinion which then affected the original character and continued in an infinite loop. Thus, character opinion flows into nations, tribes, families, and religions, and religion opinion flows into nations, tribes, and families, but the opinions never flow in the opposite direction. Understanding this flow is key to learning who to favor and who to ignore, which is important for keeping families happy. Further, putting characters at the origin of the river keeps the opinion system inherently fluid and dynamic as fortune’s wheel has its way with the people of Old World.

Old World Designer Notes #7: Characters

The following is an excerpt from the Designer Notes for Old World. The game, a historical 4X set in classical antiquity, released on July 1, 2021, and is available for purchase here.

Veterans of Civ communities will recognize the acronym ICS (Infinite City Sleaze) as it has haunted the series since the beginning. However, there is another acronym that is less recognized but just as big a problem and, frankly, a lot harder to solve – ECS, which stands for Eternal China Syndrome. The term refers to the tendency of nations that have gotten over the hump of early expansion to maintain a level of internal stability that is both ahistorical and, more importantly, not much fun. (Of course, students of Chinese history will know that the term is an exaggeration of the country’s actual internal stability.) Most of the pressure applied to the player in Civ comes from external sources, meaning other nations, and the internal pressures (including unhappiness) are really just different flavors of taxes. Furthermore, although new abilities and powers are unlocked throughout the game (via laws or techs or Wonders), they are accretive, meaning that they are only given to the player, never taken away. Strategy games are built from players adapting to their current situation and making difficult decisions along the way, but in most 4X games, these decisions are front-loaded to the initial exploration and expansion phases. Once stability sets in, the path ahead for the player becomes very predictable, which is an important reason why 4X games become such slogs – the path to the victory (or defeat) gets more and more predictable the longer a game continues.

Characters were not introduced to Old World to alleviate ECS; they were added simply because more and more games (Crusader Kings being the obvious example) were adding characters in meaningful ways and, in doing so, appealing to larger and larger audiences. Turns out that people like playing games that are about people, and a game that lasts 6,000 years is more about gods than humans. The benefits of adding characters to Old World could be spun out into multiple new articles, but to some extent, the lessons learned are not particularly interesting as the benefits were largely free, a simple result of human empathy and vengeance, of our sympathy and our avarice. Adding flesh-and-blood humans to a game is somewhat akin to adding realistic physics; it adds instant depth, but the depth is going to be the same across all games that do a good job representing the human condition. I’ll do my best to avoid getting carried away here and not end up quoting Anna Karenina and simply move on to how adding real characters improves the core 4X gameplay.

Simply put, characters add a dynamism to Old World that prevents it from reaching ECS, the usual fate of most 4X games. The most obvious way characters disrupt the game’s stability is via diplomacy. Simply having foreign leaders actually change – from death or abdication or even deposition – over the course of the game makes a huge difference. Perhaps you have a great relationship with Phillip of Greece but not so much with his heir, Alexander, because you offended him at a dinner years before? The latter’s eventual ascension (unless, say, some unfortunate accident might come to pass) will mean that your diplomatic status with Greece could go from good to bad. The amazing thing about this outcome is that it flows completely naturally from having real characters who age and die; players aren’t shocked when relations change and, indeed, expect them to change.

It is hard to articulate how significant a departure this is from a tension that has always bedeviled Civ games – that players expect diplomacy to be predictable, but predictable diplomacy inevitably becomes boring. Players will frequently rant over “unpredictable” or “random” AI leaders who suddenly go from being a friend to an enemy. These shifts are necessary for games to not slowly calcify from their earliest diplomatic states, but there are few ways to make these changes thematically palatable when the leaders never change. Civ games have experimented with all sorts of opinion modifiers that give a reason why a leader might change their opinion of you, but the most natural reason is that there is now simply a new leader who has a new set of relationships, memories, and opinions.

However, the biggest gains for dynamism are not external (like diplomacy) but internal, changing how your own nation works. As mentioned previously, one problem with unlocking powers over the course of a 4X game is that they tend to be accretive, a nation slowly adds new and better abilities over the course of the game. Players don’t like losing their powers, and Civ has only dabbled with this, such as the Civ 4 civics system where a player might give up one power but only to unlock a better one. When powers are accretive, designers have to be careful not to make them too strong, or else they could dominate. Give the player a giant hammer too early, and the rest of the game is a nail.

Instead, what if powers were attached to leaders via their unique archetypes, and these powers disappear when the leader dies? Then, the powers can change how the game works significantly but not permanently – for example, Builder leaders can add new Urban tiles to cities, Orators can hire Tribal troops as Mercenaries with Legitimacy, Heroes can Launch Offensives to allow units to attack twice, and Tactician Leaders can Stun their targets as Generals. Each of these powers fundamentally changes how the game feels, but attaching them to the Leader’s archetype means that each power is mutually exclusive and will be active less than 10% of the time. (There are ten archetypes, and young leaders don’t always even have archetypes.) Further, because these powers are attached to characters, players don’t have complete control over when these powers are turned on and off. If they were attached to Laws, for example, players might abuse the ability to switch between them whenever desired. Instead, players have some, but not total, control over the archetype of their heirs and have to navigate the natural flow of their dynasty. They can still make long-term plans for when their current Builder leader is succeeded by his Hero daughter, but they can’t pick the same pattern, game after game.

Perhaps the best thing about all of these new dynamic elements that flow from characters is that they are simply a natural extension of human nature and regular lifespans, of which all players bring an understanding to the game. For example, if a game spanning 6,000 years tried to implement our archetype system, it would need to tie itself into knots justifying why these powers are constantly changing, why the player doesn’t always have control of them, and why they are all available and viable at both the beginning and the end of the game. A game’s theme has its own gravity which puts limits on where the design can reach, and games about people provide natural affordances for an environment that is constantly changing, always a good thing for a strategy game.

Old World Designer Notes #6: Citizens and Specialists

The following is an excerpt from the Designer Notes for Old World. The game, a historical 4X set in classical antiquity, released on July 1, 2021, and is available for purchase here.

The first Civilization game was like a game design thunderbolt, sent from the heavens and marking everything it touched. Although the game does have antecedents, it is not unreasonable to split a history of digital strategy games into those made before Civ and those made afterwards. Working to extend the franchise across multiple games was an intimidating task because the original cast such a large shadow. “Whatever I do,” thought many a Civ designer, including myself, “I better not screw this thing up!” It took many iterations of the game before designers realized that there wasn’t a single magic number or formula inside the game without which the whole thing would unravel. Indeed, the basic thematic gameplay of Civ has been remarkably resilient across six major iterations with countless interesting detours, cul-de-sacs, and dead ends. Almost every aspect of the original has been taken apart and put back together again, and the franchise keeps on ticking, easily the longest-thriving strategy franchise of all time.

With that preamble done and due praise given, it is all the more remarkable that one bad game mechanic has been preserved from version to version, tweaked perhaps, adjusted at times, but still as clumsy, fiddly, and unnecessary in 2021 as it was in 1991 – the system of placing citizens on tiles. In Civilization, each tile inside a city’s territory is completely useless without placing a citizen on that tile. This is not a particularly interesting or engaging choice – it is usually obvious which tiles are optimal, and the player can at best get maybe a little extra growth or production or money if they spend some time micromanaging the tiles. Indeed, “micromanaging” is perhaps too generous a term – tweaking or fiddling or even futzing with them is more accurate. There is no in-game cost to moving each citizen in each of your cities every turn to eek out a tiny 1% efficiency gain; the only cost is the player’s time and patience. Within the overall design, citizen management is completely unrivaled for the least amount of benefit for the most amount of time spent (which is a simple shorthand for being the least amount of fun).

One can tell that Civ’s designers have never been enthralled with the system. Even in the original, the citizen system is buried deep away within the opaque and intimidating city screen. (A fellow Civ developer once quipped to me that the best feature of the original city screen is that it disappeared almost anywhere you clicked on it.) One could finish (and win) an entire game of Civ without ever engaging with or even understanding it. In later versions, the designers encouraged the player to use the city governors to set “priorities” that determined how the AI would place the citizens for you – a telltale sign of poor game design is the inevitable addition of an AI to play the game for you.

Furthermore, if the system was bad for casual players, it was actually worse for the hardcore. By Civ 3, veterans had mastered the art of rearranging citizens each turn to avoid wasting a drop of growth or production. If a Warrior cost 10 hammers, and a city produced 7 hammers a turn, then 4 hammers would get wasted every turn the city produced a Warrior (because 7 + 7 = 14, and the 4 extra hammers were thrown away). Thus, the optimal player would remember to visit their city every turn to switch the citizens around to alternate between 7 hammers and 3 hammers with more growth or money. Civ 4’s citizen management code minimized this type of micromanagement; instead of throwing away the extra hammers, the game applied them to the next item being produced. Of course, new forms of micromanagement sprung up in different places, all in the service of a system which had never delivered much enjoyment in the first place.

I was already not a fan of citizens by the time of Civ 4, and the earliest members of the Frankenstein testing group may remember an odd version of the game where there was no citizen placement. Instead, cities got yields from all tiles within their borders and higher yields still if the tiles had improvements like Mines and Farms. The change removed the awkwardness of citizen placement, but it couldn’t work without other major changes to the game. The rest of Civ’s basic plumbing (how cities grow, how units are produced, where money comes from, etc.) was built to work with the citizen system, and so it could not just be removed without causing the rest of the game to break. At a minimum, Civ 4 would need a resource stockpile system so that, for example, cities full of mines would send all the excess output to an iron stockpile instead of spitting out a new unit each turn.

However, designing Old World meant that I definitely would get a chance to build the system from scratch – this time without putting citizens on tiles! Because rural improvements like farms and mines always contributed to the national stockpile, I didn’t need to worry about a city that was overloaded with a specific type of improvement. I just needed to make sure the improvements couldn’t be built so quickly that the math became exponential. The Orders system helped out here in multiple ways. First, although players could perhaps build a ton of Workers to mass produce improvements, they might not have the Orders necessary to keep them all working, especially if wars flare up, either with tribes or other nations. Further, because the number of moves in an average turn of Old World is more consistent than in Civ (with more Orders than units early and less Orders than units later), the game has less early “dead” turns with little activity. Thus, the total number of turns in a game of Old World is much fewer than in a game of Civ, making exponential math much less dangerous. Finally, although we experimented with having cities produce improvements directly on the map, we found that tying them to Workers, which build them by spending one Order a turn over multiple turns, both made guns-or-butter work and also delayed improvement construction enough to prevent a runaway economy.

On the other hand, one original piece of logic behind citizen tile placement was that it required cities to maintain a balance between improvements and growth – building more improvements than citizens had no value because no one was there to work the new Mines and Farms. This part of the system had merit to it; balancing multiple vectors that limit each other is solid game design. I felt we could recreate this dynamic without the fussiness of citizen management by allowing cities to upgrade citizens into permanent specialists for an improvement that exists on a specific tile. We would maintain the basic idea behind citizen tile placement (put an unused citizen onto a specific tile) but as a permanent decision with a one-time cost. In other words, the player makes an intentional choice to boost a tile’s output with a citizen, but because the choice is not reversible, the micromanagement is now gone.

The tricky part was determining how MUCH a specialist should boost an improvement. Initially, I thought that we should largely duplicate the basic economy of Civ, so that an empty improvement would barely produce anything, but it was hard to get the game to scale well that way. The player would start with so little production early on that we had to lower the costs of everything, which became a big problem later on when players turned the corner and started producing far more than they could spend. The system would only work with inflating prices, which would add a heavy dose of black box complexity.

Instead, we increased base improvement production and reduced the boost from specialists to either 50% or 100%. This change got the math right for empty improvements but now specialists were no longer worth the investment (when compared with just adding new improvements). We added some Science output to each specialist (which is a good thematic match), but it still wasn’t quite enough to make specialists worth producing.

I used this opportunity to fix a couple other problems with the design. Players had complained that cities were too similar to each other, which was certainly a problem with yields coming less from the terrain itself, as with Civ. Thus, we attached four important city yields to each of the four core rural specialists – Farmers would produce Growth, Miners produce Training, Stonecutters produce Civics, and Woodcutters produce Science. (Cutting down trees produces Science because of, um, paper? Not all at because I had an extra yield in search of an extra specialist.) Now, cities would be differentiated by their specialists which were a result of their improvements which would largely be determined by their terrain, getting us back to how terrain leads to city specialization in Civ. It took some time to isolate the positive features of citizen placement and then port them back into Old World, but it’s nice to know that maybe citizens did serve a purpose back in 1991 after all.