I must admit that I grew rather disappointed with Sid Meier’s Civilization V once the initial enthusiasm wore off. The simplicity of the original Jon Shafer design looked appealing at first, but on closer examination was neither very elegant nor as conducive to a strong AI opponent as one might have hoped. After many erratic attempts to fix this situation, the final expansion and patches reintroduced annoying busywork while making the AI even weaker.
But hope springs eternal, and the upcoming Civilization VI by lead designer Ed Beach who headed the Civ5 expansions promises to keep the previous game’s good parts while fundamentally revising its game systems to make them actually work this time around. While the official website provides very little information, a number of previews and interviews have already appeared that provide a fair amount of detail on the new design. The most comprehensive article series is Dan Stapleton’s at IGN:
- Three Ways Sid Meier’s Civilization 6 Radically Reinvents Itself: interview with Ed Beach on city-building, science, and diplomacy
- How Firaxis Will Redefine Civilization’s Art Style in Civ 6: interview with art director Brian Busatti
- Civilization 6’s New Game-Changer Features: another interview with lead designer Ed Beach
- And by Matt Peckham in Time, 6 Reasons Civilization 6 Sounds Totally Different From Past Games
I’m going through the notable game design changes first, and then take a look at how the AI is supposed to improve over Civ5.
Fully-Featured Initial Release — Comparable to Civ5 with all expansions, including previous innovations such as city states, religion, and trade caravans. A wise decision, as nobody wants to go through that awkward year-long patching and expansion progress again to get a complete Civilization game.
Multi-Tile Cities — I seem to recall this feature mentioned in early Civ 5 previews but it never materialized. This time it’s for real: 12 different color-coded districts can be attached to one city, each occupying one hexagon. The number of districts that a city can support is limited by its population, and of course by available terrain which also provides district-specific bonuses. Each Wonder of the World takes up one tile, too, so you can no longer have a single “wonder city.”
No More City View — The annoying separate screen to show a city’s buildings and production status is gone entirely, as all this information is now visible directly on the world map, overlayed on the new multi-tile layout with a mouse click.
Happiness Per City — Civ6 returns to the Civ4 design of determining happiness per city, so you can have both happily productive and riotous cities in one civilization, rather than one unhappy city dragging everyone down. This should make expanding your empire much easier, which in Civ5 was limited by how much total unhappiness a civilization could tolerate.
Expendable Instant Builders — Rather than shuffle one worker unit around the map, wait until it has built an improvement, and then forget what you planned to do next, Civ6 has workers build improvements instantly – but only a few times until the worker is expended and disappears. This also makes the old worker-stealing trick less appealing, should help late-game congestion from excess worker units, and permitted Firaxis to entirely remove the anti-feature of worker automation.
Unit Stacking — The one-unit-per-tile rule introduced by Civ5 generally prevails, but with important modifications that both reduce busywork and make it easier on the AI. Designated support units can stack with other units, e.g. siege units with regular military units. One military unit can also stack with one civilian in a permanent protective formation, and late-game technologies allow combining 2–3 military units of the same type into stronger corps and armies.
Technology Quests — Rather than passively waiting for a new technology to be researched, players can complete a related “quest” that gives an immediate 50% research bonus. These quests consist of activities that naturally match a certain research path, so playing militaristically will itself speed military research, building naval units will improve naval research, and so on.
Card-Based Civics — Civics now unlock color-coded cards “which can be mixed and matched into a government type’s various slots. A simple starting government has one military slot and one diplomatic slot; a more advanced form might have three military, one diplomatic, two economic, and one wildcard (which can take either any of the three types or a unique wildcard card type). Different forms of government, such as democracy or communism, are distinguished by both their individual locked-in bonuses and their distribution of these customizable slots.” (Dan Stapleton)
Ed Beach states in Time that “we’ve rewritten the A.I. from the ground up, learning all the lessons that we had from Civilization V, so we know how best to solve some of these problems in military combat and so forth.” Specifically,
It’s a much more goal-oriented A.I. than in the past. The A.I. in Civilization V was very much based on a flavored weighting system, and this one is much more able to follow all the chains of logic, like ‘I need these tech boosts to get to this part of the tree.’ So it’s a very different approach.
To understand the importance of this change, consider Jon Shafer’s reflections on how his (then new) Civ5 AI was designed:
Another problem with my AI was the randomness, which is something I’ve already talked about at length. The computer opponents were weighted towards a variety of possibilities, with a healthy serving of RNG (random number generator) on the side. This meant they floated from one “strategy” to another without any real cohesion behind those decisions. This approach is nice in theory, but if you want a strong AI there are times when you need to force it to behave in very specific manner.
Note that Ed Beach’s goal-oriented AI is really the standard approach that prevailed before Civ5. Jon Shafer’s weighted AI was an interesting experiment but sadly a failure, resulting in seemingly insane AI opponents that behaved erratically and needed massive cheating to provide a challenge. Regarding combat in particular, the above-mentioned reintroduction of limited unit stacking should also greatly help the AI organize its armies in a somewhat intelligent manner. Again Jon Shafer on Civ5:
One of the biggest challenges unearthed by 1UPT was writing a competent combat AI. I wasn’t the one who developed this particular AI subsystem, and the member of the team who was tasked with this did a great job of making lemonade out of the design lemons I’d given him. Needless to say, programming an AI which can effectively maneuver dozens of units around in extremely tactically-confined spaces is incredibly difficult.
Concerning diplomacy, AI leaders are supposed to have unique history-based “agendas” that determine their behavior. However, this has been promised for every Civilization version for as long as I can remember, and usually just resulted in leaders that were more or less warlike (until they declared war on you anyway). A new ingredient are random “secret agendas to the leaders that you have to uncover through espionage,” so perhaps this will prove a winning combination.