by Gaute Godager, Game Director for Funcom's Anarchy Online KEY TOPICS
The Foreplay
The First Trimester—Development of the Bone Structure (The Technology) The Second Trimester: The Heartbeat of the Auto Content Generator System The Last Trimester—Getting Ready to Be Born
The Birth: The Launch
Post-Launch: Infancy and Toddler Years
AUTHOR NOTE
The first time I met Funcom's co-founder and game director for Anarchy Online (AO), Gaute Godager, it was across a conference room table in Oslo, Norway. He was holding a report about AO, which my company, The Themis Group, had written for Funcom's president. He was not very happy with us; he had no idea we were writing the report until he'd been handed a copy and was wondering why we hadn't consulted with him and his team. He felt ambushed, somewhat justifiably so.
From that rocky start, we grew to appreciate each others' talents and experience and dug in on the game. Using some of the recommendations Themis made and adding in their own talents, ideas, and agile minds, he and the rest of Funcom's team performed that rarest of feats: turning around a persistent world (PW) that had become synonymous with the phrases "bad launch" and "failure," making it a successful enterprise. These days, I proudly tell my friends in the industry that Funcom may have used us as advisors, but the work—and the sweet smell of success—is all theirs.
And because AO is one of the very few examples of a turnaround in our industry, I thought it important to have Gaute write a post-mortem on what he and the rest of the Funcom crew experienced and what lessons they learned. I think you'll find this article fascinating; I did.
"Every man should get a terminal disease, and, in fact, we all will. It makes us realize how much we take for granted—sunshine—the laughter of grandchildren."
[1]
—Desmond Tutu, Archbishop and Nobel Prize winner, December 2000
[1] Speaking about his cancer in an interview with BBC World Radio, December 2000.
I would like to start out saying this is not a regular post-mortem. It does not follow the normal "form" of such a document, and we have just begun! Anarchy Online (AO) should hopefully run for several years. This is a personal chronological tale of the events that shaped the game.
Looking back on the years of work on the game that came to be AO, it leaves me now with a feeling of melancholy. It would be a bold lie pretend AO so far has become the success we, Funcom, yearned for. Be that as it may, we shall nevertheless search history to find the battles won and also try, like any pirate on the Caribbean sand, to hunt the buried treasures. We will hopefully uncover some invaluable lessons learned. I put the quote from Archbishop Tutu at the top to try to remind me that some of the lessons we learned might have seemed like stuff we take for granted.
Also, I think it is a bit early to talk about the after-death of this game, as it has only just been born. But then again, birth and death are forever intertwined. Let us thus go back in time, to the very beginning…
[ Team LiB ]
[ Team LiB ]
The Foreplay
The history of AO is also the recent history of Funcom. I need to tell you some things about Funcom, here at the very beginning. If you do not understand this, the making of AO could seem absurd. Funcom (and yes, this is the right way of spelling it) was founded in the spring of 1993 in Oslo. This was back in the "Klondyke" years of game development, or most likely in their ebb. Anyway, we quickly grew from 5 to 100 in one and a half to two years.
Developing original console games and doing "conversions" was our trade. (To "convert" a game meant basically taking a very complex game, on, for instance, an arcade platform, and reverse-engineering it—trying to make it run on a home game console [back then, one such system was the SNES] with one-tenth the processing power and memory.) I daresay it was not a very creative process on a gameplay level. It was more a matter of engineering craft—and there is a lot of creativity there—at which we were pretty good. (Have a look at the published games on our home page, www.funcom.com).
We had a young crew. The mean age was 24 or so. We quickly learned that being a developer in the game industry was like being a bitch in the food chain. I don't easily use that term. I have thought many hours about this. Of a game sold on the shelves for $50, we would normally get $1–$3. This was, of course, after a history in which we took more than half the risk ourselves, funding the development to a large extent.
Anyway, Funcom for 4–5 years was a very interesting witch's cauldron of successes and disasters. We made some decent games (Casper for Interplay and Pocahontas for Disney), we made some terrible games (Dragonheart), and we learned.
But look at what we did. We were faithfully trying to squeeze a good game out of half-digested ideas, based on creative material from totally other areas—that is, movies. The combination of the lack of any real prospect of success due to the business model, combined with the lack of creative freedom, left everyone with a hunger for something else.
Funcom was ripe for making something on its own. The whole company wanted badly to give birth to something where we took an even bigger chance, took even bigger risks, but had greater chances of doing it big!
And we were ripe with misfortune. The publishers we contracted with went bankrupt. Publishers would buy the rights to our games simply to delay them since they had an in-house competitor (which they launched like a bombshell when our game unexpectedly stayed in the polishing phase for half a year more than expected). Unfaithful engineers would steal the most promising piece of technology Funcom had ever made and start their own company—shortly after selling it for major money to some other greedy publisher who did not, six months earlier, want to buy it from us. That particular incident led to a 1,000,000-kroner out-of-court settlement, which, by the way, may be a Norwegian record for stolen technology. We had people from other game developer companies standing outside our entrance trying to recruit key personnel. It was absurd.
Thus, making The Longest Journey was like kindling a flame of hope in the darkness to us. First off, we became our own publishers. It was also our first major 100% in-house production! It became a great game, albeit within a fading genre. (Please head over to
www.thelongestjourney.com and check out Ragnar Tứrnquist and Didrik Tollefsen's great game.) Being a "normal" PC game, it still had slim chances of making Funcom a successful company. I think I've heard somewhere that only 50% of all PC games started make it to market, and that 2% of the market eats 90% of the revenue. It might be hearsay, but it contains something very true: we needed another platform—a platform of gold and honey.
The Conception
It is easy to name the father of AO. It was Marius Kjeldahl. He was at that time our vice president of new technology. I say he was a typical father in this sense, with the "mothers" in development doing the job after conception. But the seed, the kindling, cannot be taken away from him.
AO was conceived from 1995–1997. The very first movements generated by Marius resulted in a game concept in 1995. The concept
was never used, but the platform and technology were intriguing. After throwing the idea back and forth for some time, it was the base technology platform that started taking shape within our Research and Development department in winter 1996. The lead man back then was Martin Amor, who followed the game all the way to launch.
The focus the first year and a half was basically server technology. We needed a platform on which to run any and all games. So, AO did not start as an idea or story; all that was added later. The motivation for this was, of course, to have a totally different revenue model.
The Internet could give us all that: a place to effectively market our games, a place to sell our games, a place to gain income from our games, and a place where people could gather around our games. It was a bypass of the whole normal revenue model, in which the creative people were standing on the lowest rung, begging for morsels.
[ Team LiB ]
[ Team LiB ]
The First Trimester—Development of the Bone Structure (The Technology)
It was quickly decided that the server technology would be UNIX-based. I do not believe Linux gained wind as the platform for the finished server until halfway through development. Although the servers were developed on that operating system, the engineers believed for a long time we had to use "real UNIX" on the live servers. The projections for "Intel"-based servers at the very beginning were 5–10 simultaneously playing customers per PC. We obviously needed other numbers to make money. The story of this first projected number of simultaneous players is a great joke with us these days.
One of the things I believe we accomplished well was to actually develop the core server technology ourselves. During the long development, many nay-sayers said the technology was never going to perform, that buying third-party tools would be much better, and that the 5–10 people per server was probably dead on.
Our early plans, which didn't end well, were the projections to make a universal online game engine. The idea was to develop an independent core engine, with many independent layers, that could be used for any online game. We wanted to use this engine not only in more RPG games, but also any type of game—from shooters to backgammon. As time passed, several "hacks" in between those engines (from the game code to the bottom layers) were made, and the layers now live more or less in symbiosis.
It is hard to say today if we should have scaled back ambitions and made the core engine smaller and more easily manageable. What I can say is that making these quick fix methods cost us more in development than if they had been better planned for. But then again, thinking reuse is critical, and it is very hard to find the optimum level.
Some things that did not work well from the very start were the game data development tools. From previous experience, we knew that getting the tools used to develop the game up and running as soon as possible was critical. (I am talking world designer tools, item tools, monster tools, and so on.) The first iteration of these game tools was almost totally useless, though. We shifted some personnel, made some changes, and the tools became usable. I guess there are many reasons for this, but the 8–12 months already spent left us with a bitter feeling.
Developing a Central Nervous System—A Brain
After the technology had been in development for some months, the game engine left the research and development phase and had a team assigned to it. After 12 months (we are now in 1998–99), the design was still completely incoherent; there was no direction, and many of the team members seemed to be working toward their own vision, not a central one.
Looking back at this, it is not hard to understand why the first months were botched. First of all, if Marius was the father, the womb into which he sowed his seed was a very hostile environment. The organization was not motivated; we did not want to make an MMP online game. The core development people of Funcom simply did not understand why players would think MMP online games would be fun (talk about people set in their ways). There were many, many feet-dragging exercises being performed throughout the organization.
"What did we learn from this," you ask? We learned that the game concept should be present as early as possible to be used as a sell-in. It is not always easy to sell only a business model and a piece of technology to an organization with other ideas in mind. When people saw the game and the possibilities to interact, battle, show off, socialize, and have fun together—they wanted to participate. Many seeds fell on stones, but when a seed found earth, it grew and blossomed with a speed not often seen before.
Sadly though, the above-mentioned half-heartedness seemed to have found its way all the way to the core of the team, in my personal opinion. Suddenly, after 12 months, the original producer of AO left. He had been starting his own company on the side, at least mentally, for some time.
Just to be precise, I believe the core technology team, headed by Martin, the lead programmer, was working steadily and well toward its goal. The lack of development was mostly on the "brain" side—not in the muscles or bone structure. The problems were design and
concept, not technology.
This was when Tommy Strand, Tor Andre Wigmostad, and I were added to the project. This was a tough but very fruitful cooperation, I think. Tommy's genius comes from his understanding of technology and complete enthusiasm in trying to make it work in the most general way possible. Many of the core game systems in use were designed by him (and me) in that period. Tor was partly also working as a designer, but mostly producing and acting as project manager to track the tasks. I must add that the cooperation with the main game programmer was really good, and that many of the ideas bounced back were better looking than before this point.
Hands and Feet—The Item System
One of the things we really wanted to do was make a completely flexible item system—a system where object-oriented code and thinking almost magically made everything interact in a logical way (*cough*). Well, at least the end result was pretty good. The idea of an item system where you can attach any script on any event is still there at the core. That is why an item in AO today can pretty much do anything to the objects around it: animate, make sound, change the weather, change abilities on characters, move you around, and so on. The only thing it does not interact with in the way we wanted is the GUI.
This item system, I think, was a good investment of time, and its use has been constantly expanded on. Many of the items in the game, including nanos, implants, armor, weapons, utilities, doors, chests, and so on, are various types of similar items.
The Interpolation System
The AO design team was understaffed. The end result was that there was little time to create the content. We needed a dynamic way of creating a lot of content very quickly. The end result was the interpolation system. Basically, with this system, we can cluster as many
"templates" (item instances) as we want, making them one item. The defining difference is their "level," or quality level, as it is called in the game. The system enables us to handcraft 2 items and put 200 into the game. We make one template at Level 1 and one at Level 200; then you can ask the system for an item anywhere in between those levels. It mixes those two, making a unique item for the level you request.
I think this is similar to the system actually seen in quite a few other games, just not yet in an MMORPG until AO. With the system, items scale with the characters. The system is also applied to monsters, chests—everything but nanotechnology items. We wanted the nanos to be unique, defining the profession, and I think Andrew Griffin did a great job here. He hand-assembled more than 4,000 nanos. (A nano is our equivalent of a magic spell.)
The good thing about the interpolation system is that it is very easy to make loads of content. The bad thing about it is that things feel similar—they are similar. If one chooses to go with a system like this, it is the drawback one has to live with. What we have learned from this is that in some ways, quantity is quality in an MMORPG—the more, the merrier. It is not enough, though. You need handcrafted, rare objects for the up-market players.
The Categorization System
One of the other things that I believe distinguishes AO from the other MMORPGs so far is its categorization system. We call it the
"Categorytree." In this, we can group objects into meaningful categories. Every object can belong to many categories, and we can request random objects from any category; for example, the fictitious weapon "Brutalis-Shotgun" can be added to the shotgun category, which in turn belongs to the explosive weapons category, which in turn belongs to the distance weapon category, which belongs to items you may carry around, which belongs to man-made objects, and so on. This mimics the way people group objects in the real world, so it is quite easy to understand. The trick here is that you can, at any one level, request any object.
So you may say to the system: "Give me a shotgun, or give me a distance weapon." This makes the whole item system extremely versatile. Every object with a meaning in the whole game is found at least once in this system: missions, monsters, nanos, doors, traps, This document was created by an unregistered ChmMagic, please go to http://www.bisenter.com to register it. Thanks.
Truth be told, it was a difficult system to get running. The first iteration of this data system was very unstable and crashed the servers all the time. I can still remember being shouted at by two red-faced programmers because their code depended on this data working, and it didn't.
Anyway, now it works fine, and it is really integral to the heart of our game system.
[ Team LiB ]
[ Team LiB ]
The Second Trimester: The Heartbeat of the Auto Content Generator System
This is something truly unique to the MMORPGs so far (though you can see this to some degree in Diablo and Daggerfall): the ability of the player to request content on-demand. Central to this for AO is the mission system. Here a player can use a terminal anyplace in the world and request a mission just for him/her. I think this was a very important selling point for many players wanting to migrate from other similar games. People were tired of waiting—camping as it is called—and downtime. I think my idea here was basically to enable people to have a single-player or limited-number-of-players experience in a multiplayer world. Today in AO, it is used all the time. People spend as much time in the auto content generator (ACG) areas as outside in the "static" world.
For example, let's say you ask for a mission, such as finding an item. The mission system then assembles a bunch of rooms, more or less like auto-assembling a puzzle. Next, these rooms are populated with monsters, chests, traps, and so on using the category system and interpolation system. The creation of these areas is done in a "hands-off" way, with designers simply pulling strings, setting down rules, and making building blocks. It is like setting the rules of DNA and letting evolution take its course.
I have no regrets about this system, save one. We invested too little resources in it. The major part of the system was made by only three people: a programmer, a world designer, and myself. Only toward the very end, when it was appearing to everyone how big it might become, did we add more people. The problem here was really convincing people that something made so hands-off would be fun. We had only three designers on the game at that time, and it was completely impossible for us to deliver "static" dungeons and play areas for everyone. I wish we had more people on this part, though; the data in the system is too repetitive, basically. The system itself can handle much, much more, and so it will! Just wait and see.
The Skin, Hair, and Eyes: The Client
Few things in AO have gone through as many steps of development as the client and control system. First, when we started
development, we decided to go with a system of prerendered two-dimensional (2D) characters, such as those found in Diablo and Ultima Online (UO). The point of view was supposed to be overhead isometric, and the control indirect (point, click, and run).
The first thing that became apparent was that we weren't happy with the quality of the models in the game. The graphic artists wanted real 3D, polygon characters. We stumbled about for some time on this issue.
One of the more funny things we investigated for the client was actually a totally weird technology called spheroids. We thought of, and used energy on investigating, having the characters rendered as a series of blobs stacked on top of each other. Happily, we only used four months to find out that a spheroid character was rendered slower than a polygon one. This was just at the time when 3D cards exploded onto the market. I'm glad they did—those spheroids sure looked ugly.
One of the things we wanted to do was make the game more social than previous MMORPGs—at least more social-looking. Luckily, Tommy really believed in motion capture and a great animation system. We invested in MOCAP equipment and he worked like a maniac creating the MOCAP studio. He did everything from working as a carpenter, building the stage, to setting up all the computers and equipment and actually starring as the actor in many moves. He is the male in the game. Every time I see a solitus, nano, or opifex male running around, it is him!
The vision was to give people the ability to express themselves with their bodies as well as with written text. The end resulting quality of animation is still some of the very best on the market today, well after AO's launch. I especially remember "the emotes." Having worked at Funcom for nine years, those hours when we made that list of emotes were like a dream—what being a game designer was all about.
We stayed late into the night, climbed all over the chairs and tables in the meeting room, ate pizza, and played out emotes. We wanted emotes that interacted with the world, emotes that changed with time and breed, emotes you would have to do to—and with—another character (by the way, we didn't enact everything). We wanted the whole package. The end result was pretty good and quite fun, if not quite as lofty as creative minds can become late at night.
This document was created by an unregistered ChmMagic, please go to http://www.bisenter.com to register it. Thanks.