top of page



Originally published here as part of an open discussion.

MMO cities (I’m going to use the word cities to encompass all of the node development stages, as it is more natural and will tie into a theoretical framework I bring up shortly) have been central to my personal interests and the research I have performed over my career so far. I consider myself an explorer type player, and in every game I play I savor that feeling being of being lost in a new world, before the inevitable familiarity and mastery sets in. On my mage, I can recognize and be anywhere in Azeroth in less than 20 minutes (I’ve dabbled with a variety of MMOs but spent the most time in WoW, and thus my examples will generally come from there.) 


Which brings me to the question – what are the primary functions of an MMO city? In reality, cities emerge as a consequence of people gathering and building up useful infrastructure. In the game world, the consequences are much the same but the causality is reversed – cities already exist in a location, and players gather to make use of the infrastructure and to meet other players. In WoW Classic, the functional elements of the capitals include high level class trainers, the auction house and bank, transportation nodes, specialized vendors, and they are where you get the head drop buffs. In Ashes, we are not quite as sure the full list of purely functional elements, but include many of the same things – warehouses, special processing and crafting stations, government services, marketplaces, housing, raid access, crossroads, and so on.

MMO cities provide additional services besides the purely infrastructural, as well. Players gather to see and to be seen. We’ve already heard that high level equipment and legendary gear is going to have highly unique appearances, and so we can safely assume this game’s Scarab Lords and wielders of Thunderfury will similarly be spending a fair amount of time standing around in public spaces. The promenade is a key element of any successful city. Finally, a good MMO city has breathing room for players to spread out, whether for spaces to roleplay or simply to be away from the crowd.


Knowing what players are going to cities for, we can start to reflect on what makes cities ‘good.’ Here is where a useful comparison to real world urban design theory is useful. Kevin Lynch’s Image of the City explores how people actually conceptualize their environment, rather than the ‘bird’s eye view’ we often associate with mapping. Instead, Lynch talks about the way we break down out cities into:

  • Paths – the routes by which you move around

  • Edges – thresholds between things, like a wall, overpass, change in ground texture

  • Districts – large regions of distinct character, like neighborhoods

  • Nodes – area you enter into, such as a plaza or square

  • Landmarks – big things you see but not enter, like a tower or mountain


So, when navigating a city, the more fluidly we can break it down into these elements – building an image – the more pleasant the experience of moving through and occupying it is. Similarly, when encountering a new environment, we try to build up this image as quickly as possible to achieve that level of navigational confidence. For example, think about the last time you traveled to a new city – you likely found a series of landmarks and pathways that’d connect you back to your starting point, and as you continued to explore, you attached the new destinations to your existing network. This is why, in an unfamiliar place, we often take longer routes to get home if it takes us along parts of the city with which we are familiar.

These characteristics also apply to MMO cities, and if you keep these attributes in mind throughout the design process, you can create a city more pleasant to occupy. So, remembering Kevin Lynch’s elements and the reasons we established people go to MMO cities, I’m going to use Ironforge as the example of the ideal permutation.

  • The most critical city services – the bank and auction house – are spaced on opposite ends of a public space. In order to make use of both, players have to run across the plaza.

  • Nearly all means of travel into and out of Ironforge pass through the plaza, guaranteeing traffic. The front gate, the inn, the flight master, the tram, and the mage trainer are all orbiting around the center, and so to get from your arrival point to an exit you will likely pass through the plaza.

  • The central public space is well scaled, providing enough room for players to stand around and look badass on their cool mount or with their gear, but not being so wide as to get lost in the inconvenient vastness.

  • The central public space makes use of a neutral and flat ground texture, allowing player models to not get lost in the textural noise.

  • Sprawling outwards from the center are the less critical capital services – trainers, professions, vendors - housed in a variety of distinct environments – the spooky cave, the gnome district, the lava forge, etc.

  • Spaced throughout the entire city is a comfortable amount of breathing room and empty buildings, allowing players space to wander off and be alone or to have a spot for a roleplayed encounter.


It is worth noting, though, that not every city needs to be legible and pleasant. A key counterexample is that of the Undercity: it is nigh unnavigable. Everyone who has played Horde will have a story of getting lost in the Undercity – it is a right of passage. To this day, I consider myself fairly spatially fluent, my main character remains a Forsaken, and I still have to run loops around the upper tier to find the inn. But this very complexity of travel makes the Undercity memorable. It is not easy to use, but functionality is not always the most critical design aspect.


So, what does this mean for MMOs? Many contemporary MMOs are intending to use procedural generation to populate the world. Typical theme park MMOs might have as many as half a dozen large cities, but they are static and established beforehand. A procedural MMO, however, must account for any number of potential locations and variations. As such, procedural generation does seem to be the most logical choice, rather than trying to hire a veritable army of level designers.

In the process of setting the parameters for the procedural generation, then, I think the above qualities should be present in just about every permutation. At this early point in thinking about designing the generation, it’s exactly the right time to be setting precedents that will make all of the cities that appear in Ashes as interesting, memorable, and successful as the very best theme park MMO examples.
Recapping the most critical elements of a successful MMO city:

  • Thoughtful paths, edges, districts, nodes, and landmarks

  • A well scaled, neutrally textured, central public space

  • Key city services arranged around the central space

  • Less trafficked city services arranged around the central

  • ‘Breathing room’ for exploration and roleplay


Procedural generation is an opportunity to go further than just best practices for cities, though. We know that the appearances of cities will be varied in some seemingly large part derived from the racial makeup. We have already seen some of the various different types of buildings rendered in a race’s specific style. I think there is a chance to extend the variation by race further than simply different assets run through procedural placement – extending to the urban design, and thus spatial character.

I imagine each race can have a spatial typology unique to themselves. Each variation still adheres to the guidelines I set up earlier – the elements of good MMO cities – but subtly varies based on the race whose individual buildings are present. Such urban spatial types might include a centralized point type, a circular type, a linear type, or a decentralized type.


The result of altering the procedural generation in this way would be that each racial city not only looks different, but feels different to occupy. Instead of simply swapping out the assets that run through the generator, the entire essence of the city can be altered. The way you move through and do business in a Vek city will feel entirely different than a Kaelar city.


Moving past alterations and tweaks to the procedural generation, I think a real opportunity for comes in the manual curation of the procedural generation and the utilization of extreme sites. My biggest fear – given the current level of information we have now - is that the node generation system is going to produce a huge variety of interesting and unique cities that all appear in a convenient open field.

Instead, I am hopeful that these cities have sites and locations as varied as the best theme park games. Cities tucked in canyons, cities in the shadow of towering trees, cities clinging to the side of a mountain, cities spread over a chain of islands, and so on. The richest factor in determining the uniqueness of a city is the location, to the point where I would want to see site variation with raw asset swapping well over racial variation.

Based on my limited understanding of the procedural generation, though, this would likely be challenging. To that end, I hope Intrepid finds a way to synthesize the generator with the deliberate touch of a designer – working in the form of brushes? Volumes? Generative iteration followed by manual cleanup? Who knows.

In any case, I think variation by location must be the single most important factor in creating the cities of Ashes. Without that, I worry that we will have ‘The orc metropolis of ZOI 72” and the slightly varied “orc metropolis of ZOI 24” without much to distinguish them. As was the case in the early days of No Man’s Sky, having infinite variation does not mean a whole lot when the variation is evenly and uniformly distributed. Instead, I hope to have “Yoloti, Jewel of the Mountainside” and “The Canyon Observatory, Tecolotl.” Infinitely varied cities in a field does not feel memorable – slightly more similar cities in epic locations will.

Now, losing to a siege does not only mean losing your hard work, but it also means losing a one-of-a-kind environment, potentially never to be seen again. I want a game where people speak of their cities in awed tones, with the losses and gain being every bit as real as losing a work of history and art in reality.


Based on my experience with procedural generation, this last thought moves to the edge of impossibility, but I will include it anyway for the sake of evoking imagination and eliciting discussion. I would love to see the history of a node fully embodied in its condition. That is to say, if a city falls to a siege and is razed, but is later rebuilt, it would be incredible to see ruins of the first city still present in the new one. This creates a sense of time, serves as a living embodiment of history for players that were there and a mystery to be explored for new ones, and shows the changing demographics in the form of potentially varied racial architecture (and spatialization!) The implausibility of such an approach stems from requiring the generator to have a memory of what was and a means of interfacing with that knowledge, but still, the idea is appealing.


  • In general, the procedural generator should be calibrated to adhere to urban design principles

  • Every city needs careful attention paid to its dominant public space

  • I’d like to see not only racial variation in aesthetic, but also spatial typology

  • That said, I think the single most important city design factor should be a unique and evocative location

  • Finding a way to include ruins (a temporal element) would be awe inspiring, if implausibly difficult to automate

bottom of page