Blog Home

The Artillery Blog

Atlas Design Update

Sean by Sean "Day[9]" Plott
on 01 October 2015


I’ve been speaking about Atlas frequently on my show over at for the last year, but I’ve been fairly quiet blog-wise. Thusly, I’m electing to write some condensed design blogs about all things Atlas, our in-progress real-time strategy (RTS) title. We’re planning broader external testing at the end of 2015, so I want you to know what you’re getting into. Today I’m going to talk about teamplay, squads, and our playtesting process!


We value gaming as a connecting, social experience. So, we’ve designed Atlas to be a team-focused RTS, specifically 3v3, so you can go to war with your friends (or enemies).

Some of my favorite gaming memories are playing 2v2v2v2 StarCraft games on Big Game Hunters or massive 4v4s at LAN parties. The insane cross-faction combos, like scout/tank or statis/nuke, created entirely new and hilarious strategies. I’d spend months logging on every day to develop a fresh and ridiculous plan with my friends. Sometimes I’d get nearly murdered early on and have to rebuild inside an ally’s base, microing only 2 or 3 units in gigantic battles until I could stabilize. Over time my competitive itch would draw me back to 1v1, but I yearned for that element of community and teamplay-based combos.

Atlas is built to be a bridge between the competitive nature of 1v1 and the fun of teamplay. The systems in Atlas are built to account for this. Some objectives or tactics need only one player. Others require all three. Multiple simultaneous points of interest would be impossible to manage as a single player, even one with 300 actions per minute. But with teams, players can plan out how to divvy up resource collection and attacks on multiple fronts. We’ve designed units so the three players together form a logical “single race” with each player controlling ⅓ of the forces. In other words, team based combos and strategy are a core part of the game.

With teamplay as such a focus, we’ve had to rethink some core mechanics. One of the big design decisions we’ve made is how to think about factions – let us talk about squads.


In Atlas, rather than pick a full race of 10-15 units, you pick a squad consisting of one hero unit and three unit types. During a game, you may only build among those three unit types. For instance, you might choose a squad with fighters, rangers, and tanks, intending to shoot for a composition of many fighters, a few tanks, and one ranger.

3v3 team games can become pretty tough to process if every player has access to 12+ units – we couldn’t even figure out how to talk about our games with 30+ different unit types on the battlefield for each side. One hero and three unit types per player has made Atlas much more visually and strategically understandable.

Squads can also be designed with more clear, niche purposes in mind. In a typical RTS, figuring out strong compositions from a full 10-15 unit race is tough. There are so many options and, in particular, many bad options. With squads we can design for and communicate a role more easily. A freeze squad might focus on slowing enemy groups or disabling key enemy spellcasters whereas a fire squad might be focused on area-of-effect damage and space control. Together, players can form an ad-hoc “race” from their 3 squads. One player might focus on heavily defending team expansions, while the other two might harass aggressively. Different team compositions represent different strategies in the same way that, in card games, different card combinations create different deck styles.

Squads also allow us to have easier controls. In a typical RTS, if you want to cast an ability, you must first select the correct unit and then press the corresponding hotkey. With multiple abilities that need to be cast in quick succession, misclicks and frustration can happen frequently. With a pre-designed squad of a hero and three unit types we can avoid this statefulness problem, allowing you to cast any ability from any unit. With your entire squad selected, you could press “F” to launch an AoE explosion from a tier 3 spellcaster and immediately press “D” to dash your tier 1 units out of the way.

Aside from units, we chose to give each squad a hero for a few reasons. First, we think hero units are super cool, providing a powerful centerpiece to your army. Second, we can add extra-high impact abilities to the hero that we couldn’t place on a unit. Third, Atlas has on-map objectives that only the hero can interact with.

Some things don’t quite make sense in squads, such as air transports, simple scouting units, or defensive structures. To expand the scope of strategic possibility, players have access to simple, more specialized “neutral units” in every game. Over the course of a game, you choose which neutral units to combine with your squads.

In our tests, we’ve found that squads each have individual flair and help combine with allies to form interesting playstyles. Moreover, there’s a solid space of possible build orders from both squad and neutral tech trees to explore.


We playtest Atlas as often as possible. In addition to nonstop internal testing, we run external playtests twice a week and additional focus group testing about once a month.

We believe that humans fundamentally struggle at predicting what will be fun or how other humans will play their game. Think about football (soccer in USA) – I can’t think of any intuitive reason why a game where using-your-feet-and-not-your-hands should be fun, yet football is an absolute blast to play and watch.

The opposite is also true. I have heard countless fun-sounding ideas that turn out to be terrible once implemented. Worse, I have seen implemented ideas start off fun, but break once a playtester discovers a fundamental flaw. It’s difficult and, as a business, extremely risky to rely too heavily on intuition for design decisions.

Playtesting is the core of the Artillery design process.

We test Atlas with RTS veterans, absolute newbies, mid-level players, single player gamers, FPS players, and more. Our core group has been playing for over a year, and they have identified core problems with previous builds and continue to suggest improvements for future builds. We’ve run month-long tournaments with RTS professionals to see how they push our game (congrats to qxc and Sasquatch for winning the most recent 2v2 test!). Though the game might have unpolished or unfinished pieces, it’s absolutely critical to gather as much gameplay data as possible to make better decisions.

In December we’ll grow our testing group to 1,000 players to get more data. If you’d like to help us, we’d be thrilled to have you sign up here. We’re working hard to lock down the core gameplay and improve the experience for new players so we can get Atlas out to the public in 2016.

Check out the blog in a few weeks for some business announcements.

Share: Y

blog comments powered by Disqus