Gamasutra: The Art & Business of Making Gamesspacer
Postmortem: Unknown Worlds Entertainment's Natural Selection 2
View All     RSS
October 26, 2014
arrowPress Releases
October 26, 2014
PR Newswire
View All

If you enjoy reading this site, you might also want to check out these UBM Tech sites:

Postmortem: Unknown Worlds Entertainment's Natural Selection 2

February 26, 2013 Article Start Page 1 of 4 Next

On Halloween of 2002, Cory Strader and I released a free Half-Life mod called Natural Selection. It blurred the lines between shooter and RTS, in a wildly asymmetric, harsh and fast-paced sci-fi environment. It became the most popular independent Half-Life mod, and I became forever hooked on independent game development. I lived off my $40,000 of savings during its development and I became highly motivated to turn this into a living.

I moved to San Francisco to find investors and teammates, and to start a new life: one that I hoped would include a sequel to Natural Selection and enough income for me to become self-sustaining. After a few false starts, I started Unknown Worlds, and convinced my new partner Max McGuire to go in for half. At the end of 2006, we started working on NS2 in earnest. We lived off our own savings until we got some angel funding in 2008 and then carefully started growing the team.

This past Halloween, in 2012 -- 10 years after the release of the original Half-Life mod, and after almost going out of business multiple times -- we released Natural Selection 2 using our own "Spark" engine on Steam. It went right to Number One and has since sold around 300,000 copies. This article hopes to summarize what we learned during this epic period of toil.

What Went Right

1. Unstructured Development

We had no producer, no schedule, no milestones, and no meetings. We also had no design document, no technical plan and no business plan. Anarchy? Not exactly.

Our core team was seven people in a room, but we also had many critical remote contractors and part-time community contributors. Every time we tried daily meetings, we ended up deflated and demoralized. It made no sense why spending 10 minutes a day talking about our work made us so unhappy, but at least we were smart enough to stop. No status meetings naturally extended to no meetings at all.

We did try to have a schedule, but we found it was inaccurate and out of date instantly. It didn't seem worth maintaining something that was perpetually wrong, so we stopped doing it. After writing a business plan, I realized that I learned almost nothing writing it. A year or two later I realized I never looked at it, either. If I wanted to describe "our plan" to someone, it was easier to send along a couple paragraphs by e-mail. Very few people will get through a 40-page business plan anyways.

What we did do, though, is have lunch together every day. Anything important came up then, and the time there was naturally bounded, so talking didn't go on too long. The daily ritual of breaking bread was a powerful one for our team cohesion and happiness. And not having structure meant we could adapt to changes very quickly, and that we could spend all of our time working on the game, instead of planning it.

2. Empowered Design

One reason this unstructured approach worked for us was due to our design process. Instead of a design document, we used four "pillars" to describe the game. Each pillar is the name of a feature, described in specifics by a short paragraph. Imagine the bullet points on the back of your future game's box (even if it won't have one). All decisions about high-level features were compared to these four pillars.

This let everyone on the team have a set of "beacons" for the game direction, which allowed everyone to vet new ideas, have discussions and cut long conversations short if they weren't advancing one of these high-level goals forward. Having the pillars be high-level allowed us to be as flexible as possible, while still keeping the "spirit" of the game intact. You can see that these pillars are the main topics on our game page.

At a lower level than features, we kept artists, level designers, programmers, and animators working together cohesively through a set of design principles. These principles are "rules" that I implicitly followed when making decisions about balance and goals for the game. Examples include "No hard counters," "Public vs. competitive play," and "Non-obsolescence."

For each, I included a paragraph or two of reasoning and examples. I felt bogged down explaining my reasoning behind balance and design decisions, so I needed a way to quickly get everyone on the same page.

For instance, the marine team can research generic weapons upgrades, which let them do more damage. The aliens have no analog (following the "Two Unique Sides" design pillar) -- so in the late game, aliens can be ripped to shreds by even starting marine weapons.

Someone made the suggestion that aliens automatically have their armored shell able to take more damage, the more hives their team has (this was present in the original Natural Selection). But the design principles include "No Magic Numbers" -- in other words, hidden modifiers that are not readily apparent to the player shouldn't be in the game. This principle immediately changes the conversation away from any variant of hidden scaling armor, and towards adding new specific traits that could be used to compensate (extra speed, cloaking, etc.)

The specific goals in the document don't matter, but they shouldn't change much. I wrote these design principles in one sitting and they stayed throughout the rest of development. Embodying these cultural design principles in a public document empowered not only our employees, but our community members, and allowed everyone to contribute their art, sound, code or design that "fit."

Article Start Page 1 of 4 Next

Related Jobs

Digital Extremes
Digital Extremes — London, Ontario, Canada

Character Artist
Digital Extremes
Digital Extremes — London, Ontario, Canada

Sound Designer
Disruptor Beam, Inc.
Disruptor Beam, Inc. — Framingham, Massachusetts, United States

Lead 3D Artist
Red 5 Studios
Red 5 Studios — Orange County, California, United States

Graphics Programmer


Geoffrey Rowland
profile image
I remember when some of the original NS team came to the biggest Tribes LAN - UVALAN in Chantilly, VA to show off Natural Selection. Even though this game doesn't have vehicles or skiing or definitely reminds me a bit of Tribes at times (mainly the teamwork aspect)

Charlie Cleveland
profile image
Tribes was a huge inspiration, Geoffrey! It blew my mind.

Jack Nilssen
profile image
Was the press even really all that important, then? Seems to me that the success of this game is riding squarely on the shoulders of a dedicated fanbase that's made sure that others know about good the experience is.

Raymond Ortgiesen
profile image
That was a great read, lots of advice and useful information. I'm so glad you guys pulled out okay on the other end, nice work! My favorite part:

"It was more work and stress than we had originally anticipated, but everything worked out fine, and no one died or got sued"

That's all you can hope for right?

Nick Allen
profile image
"We have a completely unstructured environment, without task lists..."

What about the Pivotal Tracker that was public during development?

Christopher Casey
profile image
Speaking as someone who followed that tracker for a little while, it seems that they experimented with using that but ultimately discarded it.

Hugh Jeremy
profile image
Pivotal served to track specific programming tasks for a time. But broader tasks, and the goals that those specific programming tasks were part of, were not tracked or defined.

Daniel Tarantino
profile image
"Maybe your first business won't be a success, but at least you weren't in school, where there's not even a chance of success."

This is very inspirational to an aspiring entrepreneur like myself.

Chris Clogg
profile image
Great read! Played the original NS a lot haha :)

The last page was really interesting; we ran into a similar problem with our iPad game last month... was very hard to get press (our first game release and no publisher I guess), but even with the blessing of Apple featuring us, our sales have dropped pretty quickly. It's nice to hear you still had solid sales even with the press issues. The app store is a bit of a "wild west" scenario... I think Steam has somewhat better content curation (or maybe just less stuff lol).

By the way was there a reason you didn't end up using the Source engine? Seems like it provides a lot in the form of network code and its UI/system/editor/feel is familiar to most PC gamers.

Charlie Cleveland
profile image

Our sales drop off pretty quickly in general - it's all about the sales and promotions. But even when they drop off, our long tail is still profitable so we're in good shape. I wonder if sales and updates are useful for driving sales on IOS as well?

We actually started development in Source - for one full year. Dynamic infestation was the first thing we added and it caused enough scary changes that we immediately felt like we were fighting the engine.

Chris Clogg
profile image
Ah yeah, interesting. I feel like the Source engine was intended by Valve to be a big juggernaut in terms of a license-able engine, but didn't get very far compared to others like Unreal. Still a lot of great mods though.

For iOS, it certainly helps to have sales and updates, but you're already starting from a low point of a few dollars. This also makes it difficult to advertise because if you convert someone from a click or impression, you're only making $1.99 or w/e. As well, many mobile ad networks aren't so hot on apps that cost money. If you're a free or F2P app though, then you can use the ad networks to buy users and then get up on Apple's charts (but you're going against $100k-$million of others' ad budgets).

Long tail is tough on iOS (or it seems so far) because as soon as you're done being featured, it's up to chart position and word of mouth (or ad budget)... and you're in the fray of 800,000 apps. 2 years ago it wasn't so saturated though haha.

Edit: Ps just bought NS2, going to play it later tonight :) Good thing your blog post reminded me of it haha!

John Trauger
profile image
I loved "Nobody died or got sued"

Patrick Mullen
profile image
It's "premature optimization" not "preoptimization". The real idea is that while a system is in flux, optimization can be very bad. Issue #1 is that that entire system may not even stay in the larger product if the direction changes. Taking time to make sure it performs as well as it can is a wasted effort if the entire section of code ends up being removed. Secondly, and more important, optimized code is harder to read and understand, and if things aren't finalized will slow down your process and may even limit you from seeing what really needs to be done. Sometimes this is in an isolated system, but it can happen in integration as well. For example, a slow pathfinding routine that it turns out the game only has to execute occasionally - making the routine itself faster is not as big of an improvement as making the game call the routine less often. But yeah, waiting until the entire game is done to make sure it runs well is overmature optimization :)

Steven An
profile image
Nobody said programming was easy... :P

Charlie Cleveland
profile image
Regarding task lists or lack thereof: people had their OWN task lists, but we didn't have a schedule or list of things for other people to do. Certainly people kept track of their own tasks their own way.

Gareth Eckley
profile image
Having been one of Charles minions in the past, I'm going to have to point out that his extremely easy going and reasonable approach to everything has been the major factor in the success of the Natural Selection brand. He's a modest guy, but he does draw the best possible work out of those around him.

Brian Ortiz
profile image
What about Zen of Sudoku? Was that time well spent?

Marcus Dublin
profile image
Fantastic postmortem Charlie, and very inspirational. The Art Bully team and I are extremely grateful to have taken part in the games development. I personally can't wait to see what you guys do next.


Nathan Ware
profile image
If you were to expand on what it meant to everyone in the company to have an unstructured management environment, I think there are a lot of developers here who would love to read an article like that.

Virginia Martin
profile image
For a six tough years of development you have only 3 things that went wrong. Not look a bad statistic!