DOTA UNDERLORDS
The original Auto Chess – a fan-made mod for Valve’s Dota 2 that caught fire at the beginning of 2019, amassing millions of players and essentially spawning a new genre in the process – is exactly the kind of thing you’d expect the studio to poach. After all, Valve has a proud history of hiring modding teams, from the original Team Fortress to Counter-Strike to Dota itself. And it did tr y: last February, Valve flew Drodo Studio, the five-person Chinese team behind Auto Chess, to these very offices for a chat. It’s not hard to spot parallels with the way it bought TF Software a plane ticket from Australia back in 1997 – but unlike on that occasion, the conversation with Drodo Studio ended with both parties concluding, apparently amicably, that they couldn’t work together.
A little more than a year later, Dota Underlords has just reached version 1.0. After the Drodo conversation, Valve didn’t hang around. Development began the same month, and it was a sprint from the outset. It took the newly formed team two months to pull together their first prototype – just in time for the company’s annual trip to Hawaii.
“By the time people went on holidays in April, they all had a game to play on their phones,” gameplay programmer Adrian Finol tells us. Note the last word there – Underlords is Valve’s first game to come to mobile. “When we first got it working, we were amazed it even ran,” product designer Lawrence Yang says. “The phones got really hot and it looked sort of shitty, but it was working – and that was enough to say, ‘As a proof of concept, this is something we can totally go after.’”
Around this time, Drodo was releasing Auto Chess – a version of its game with all the Dota serial numbers filed off, the same way Dota itself did years ago with all those Warcraft 3 heroes – on mobile. And again, Underlords wasn’t far behind, opening up a beta across Steam, Android and iOS in June. This couldn’t be further away from Valve’s reputation for leisurely development.
The development process doesn’t sound all that different to the ones we hear about from other teams: tr y, test, iterate. It’s just that normally it happens behind closed doors. “It’s exactly the kind of thing we would have done internally if we were tr ying new features,” Finol says. “We just decided we were okay getting egg on our face in public.” As for why this was the game to do that on, Finol’s answer is simple: “We didn’t know which kind of game we were making.”
“We were really kind of leaning on the community to help us understand,” Yang adds. “We also tried to be very experimental, to see how far we could push things,” Finol says. “And I think the only way you can properly do that is with actual people playing the game and providing feedback. Because it’s easy internally to be like, ‘We are all geniuses, this is the greatest feature’ – and then you ship it and the customers say, ‘Eh, we don’t we like it’… and you just spent three months working on this thing that you thought was great.”
The beta was a huge success. In its first week, it attracted a peak of 200,000 concurrent users on PC, according to SteamDB, and, per Sensor Tower’s figures, was downloaded 1.5 million times on mobile. Player feedback was mixed, but that was the point. Ideas were generated, pulled together as quickly as possible and then tossed out into the world. “If it didn’t work out, it was okay for us to say, ‘Just rip it out’ – there’s not a lot of ego there,” Finol says. “Let’s just tr y it. If it works, it works.”
We meet the team at the end of this year-long sprint, a couple of days after the game has left beta. They’re clearly happy with what they’ve made, and are excited to get stuck into the second season. But while launch has given Underlords’ player count a significant boost – Steam concurrents doubled from a peak of around 15,000 in January to just over 30,000 – that’s still an order of magnitude smaller than the numbers it was attracting at the beginning of the beta.
The team don’t seem too perturbed: for now their focus is on making the game as good as it can be. “It’s the same advice that we give any developer who comes to Steam and asks, ‘How do I sell more stuff?’” Yang says. “Make a good game, and the rest will just happen. The company is really good to us: there’s no external pressures to meet this quota or hit these numbers – just do your work, just do the best thing you can. And that’s what we’re doing right now.”
And perhaps it doesn’t really matter what the fate of Underlords is. That’s short-term thinking, and Valve is able to take the long view. “We create new projects and games, and every time there’s a technical step forward the rest of the company can then branch off it and use it for their own things,” Yang says. “In this case, it’s Source 2 on mobile, and learning how to work with the App Store and Google Play, and think about what cross-platform looks like for our games.” Underlords has helped Valve iron out the kinks of an entirely different gaming ecosystem, and find a new way of working, making its mistakes in public right from the beginning. And it has done it all within the space of a single year – which, in traditional Valve Time, is no time at all.