Long read: Who is qualified to make a world?

In search of the magic of maps.

If you click on a link and make a purchase we may receive a small commission. Read our editorial policy.

How "bossless" Valve makes decisions

The Half-Life maker on its "organic gravity well".

Valve has explained how it's able to make decisions despite being "bossless".

Earlier this year Valve's employee handbook revealed a somewhat utopian structure at the Half-Life maker, which currently employs around 300 people. There are no bosses in the traditional sense; instead people are encouraged to interact with projects that may not suit their skills, and decision-making is based on the interest projects gather.

During a Q&A following his keynote presentation at the Develop conference in Brighton this morning, business development chief Jason Holtman was asked whether he thought the quality of games such as Portal 2, Team Fortress 2 and Half-Life 2 was the result of its unstructured system, or whether the success it's enjoyed enables this freedom.

"I think there's a link," Holtman responded. "I think there's a really strong link. It's both of those things, but probably the first one is the more true. The quality of the games and the quality of the things we make comes from this unstructured-ness. In other words, it comes from localised decision-making. It comes from the closeness to the customers.

"What happens is, as you succeed, you start feeding back on that loop, right. Obviously the success lets you do it more and lets you take risks, but I think it's both."

Holtman was then asked how decisions get made at Valve given its bossless structure. What happens when two people simply can't agree on something?

"Thunderdome," Holtman joked. "It's pretty awesome. No, Dota. We've actually found that works really well."

"We rely on the structure of being peers, and a bunch of people making the decisions themselves," Holtman said. "But you're like, that's all well and good if it's like one or two people and they're cowboys and they're really good at something.

"What we actually rely on is an organic gravity well. And we encourage it. So we say, look, if there's a project going on, say the game's being worked on for two years, we figure out whether or not that game gets shipped and gets continued to be worked on by the amount of people that are around it.

"The amount of people that are around it aren't around it because of a P&L [profit and loss]. We don't reward people on, okay, you got attracted to making Portal 2 and I know Portal 2 is going to sell. We instead say, people are getting around the gravity well because it's interesting, because it's strategic, and that's what they're going to build."

Holtman said this organic gravity well helps Valve decide what to ship, but also what not to ship.

"It's a really good indicator of what's good or not. If that gravity well of folks knows they're going to be responsible for that game and get to make decisions about that game or project, they do a really good job. It's a good feedback mechanism.

"They can look around and go, I keep trying to find an engineer to do this for me, and the engineers keep saying they're too busy and have other stuff to do. What we do is say, okay, go and have that discussion. And if they keep saying that and everybody has the right information, it probably means the thing you want done isn't the right thing to do, and maybe you should help them."

Of course, there are problems associated with this structure, and Holtman singled out one that goes some way to explaining the concept of Valve Time, that is, the length of time it takes the much-loved developer to create games.

"It's a little hard," Holtman said. "The negative of this kind of structure is we do lose the quick mandate structure. There probably are people who would be really good about doing, if we did this in the next six months we would be at X, and that can have advantages. We don't have that.

"But we're trusting a bunch of people to organise around X and do it and have those conversations. But it's a hard problem."