Valve is one of the most successful video game companies in the world. The firm's online game distribution and multi-player platform Steam has 65 million users. At next week's CES conference, the company will announce hardware partners for one of its most ambitious undertakings so far: a line of gaming console alternatives running on the company's linux-based Steam OS.
What makes Valve so successful? In November, I sat down with Valve CEO and co-founder Gabe Newell in the gaming company's Bellevue office for a feature story. Newell argues that attracting and retaining talented programmers and designers is key to the firm's success, and explained the company's strategy for doing that. This interview, the first of a two-part series, has been lightly edited for length and clarity. The second half is available here.
Andrea Peterson: Can you tell me about the company and why it works why it works the way it does?
Gabe Newell: The company grew out of trying to think about how to design for the specific business challenges we would have. It’s sort of a good news/bad news situation that our industry is changing so quickly. If you look at the requirements for just one piece, like art, from one generation of games to the next, it will change radically. You need people who are adaptable because the thing that makes you the best in the world in one generation of games is going to be totally useless in the next. So specialization in gaming is sort of the enemy of the future. We had to think about if we’re going to be in a business that’s changing that quickly, how do we avoid institutionalizing one set of production methods in such a way that we can’t adapt to what’s going to be coming next.
Also, it was pretty clear that there were very large differences in productivity between people who were good and people who were great. So we needed to figure out why those people would come here rather than working some place there or starting their own company. So we had to have a clear model for how Doug [a long-time Valve employee who handles much of the company's press and was in the room during the interview] was going to be better because he was here rather than going off and starting his own gig.
The set of those requirements led to decisions about not having titles, not having organization structures, and things like that because as useful as they are in the short-term in the long-term they really end up hurting you a lot.
For people that started off here they sort of get used to how it works, but it's sometimes a bit of a culture shock when people are coming in from other industries. So you need to sort of signal to those people that this approach is actually a useful set of methods for doing what we do.
Can you give me some specific examples?
For example, we don’t track vacation time or sick time – we just tell people we trust you to make all of these other decisions, of course we are going to trust you to manage your own time. It’s actually a pretty minor issue in terms of how much time people actually spend on vacation or sick leave. But it’s a really important issue for someone who is say, coming out of Hollywood. When you tell them that -- and it’s really true -- it seems to be useful in getting them to start to realize that there is a rationale behind how the company works. There’s sort of the flashy public things like desks on wheels, but it really is intended to create a better environment for a highly technical set of tasks that vary fairly quickly over time.
So, if somebody becomes the group manager of X, they’re going to really resist it when X is not what you want to do in the next round of games. You don’t want them to sort of burrow into that – you want them to recognize that being really good at Half-Life level design is not as nearly as valued as thinking of how to design social multi-player experiences. You've had them feel like they have an organization and title tied up to something when the key is to just continue to follow where the customers are leading.
You mentioned Doug earlier. He mentioned a really interesting anecdote earlier about how he came to work with Valve. He said he worked as a Brand Manager at Sierra Online and was going to be moving to Microsoft. But then you took him to lunch, went into an office together for three hours, and he came out with a job offer. It seems like that might be indicative of Valve’s approach to trying to bring in that individual talent. What did you tell Doug?
I don’t think I was trying to tell Doug, so much as show him, what it would be like. The kind of people who are successful here like to work and get stuff done, and like to feel like they’re collaborating and that they’re going to be working with people who are going to help them be better at what they do. So I think more than trying to create some sort of coherent argument with him, I was just trying to say “this is what it would be like if you were here, is this the kind of experience you want to have? Do you see me as being someone who is going to help you do what you want to do – only more so.” So I think that that’s probably more of what I was focusing on than trying to tell him about the benefits package, right?
People need to trust the promise that you’re making, whether it’s people who work here or your customers. Everybody understands that you’re supposed to say “our employees are our most valuable asset” to the point where even if it’s really true they’re not going to really trust you until you’ve earned that – same with customers. [Employees] really look to the integrity of interactions over time to decide whether or not they can trust that you really are trying to do best by them or help them be more valuable, that this is a good place for them and their families for the next ten years and those kind of things.
It all sounds really good, but until they start to experience it a couple of times, they’re not going to really believe there’s any sort of reality behind that promise. So we’re really conscious about that. You really need to be clear about the promises that you are making to Doug, to your customers, to your partners, and recognize that they’ll test you on that.
How have you organized the company to help retain good talent?
There are a bunch of things, like the people who are really talented often don't fit into rigid boxes. It's part of why they're good. For example, Ken Birdwell was one of the first people here and he has a bachelors in Fine Arts -- he's an artist -- and he also happens to be able to program really well. At any other company he would be sort of forced to fit within an existing structure but if you look at the class of problems he's able to solve since he can animate, model, and program he's able to invent solutions that other people can't.
He can have an insight into these classes of problems and say you need to design it this way so you can solve implementation problems here, or here, and have them be completely different from the sort of solutions other people had. So you see that over and over especially in space that's evolving rapidly: The old way of doing it is not right way to do it next time. People who are constantly looking for the opportunity to do something new are also people who are not going to be helped by having job titles -- job titles create expectations of specialization and focus which don't map really well to creating the best possible experience for your customers. Even if they do, they're only going to work for one point in time.
So yeah, there's advantages to be able to attach a tag to somebody's name and say they do X -- but we've found that there are disadvantages to that in the long run and the people who are most successful at creating value for our customers are the people who respond best to being able to contribute in a bunch of different ways.
Our best people at talking to customers are not people who would ever be called "marketing communications." Often time they spend most of their time writing code or creating art -- they also just happen to be really talented people in this other dimension. For them it makes it more fun and rewarding to be able to move between different roles.
What kind of incentives do you give employees to stay at Valve?
We spend a lot of time trying to help people understand how the company does compensation -- to make it transparent. They don't want to spend all of their time trying to figure out how to game the system, they just want to trust that if they're creating value there's a reasonable, fair, and transparent way that turns into dollars for them and for their family.
A lot of times also we just ask people. We say "we want you to be here ten years from now, so what do we need to do for it to make sense for you to be here?" A lot of times those answers are idiosyncratic. Right now, the answer is surprisingly often, "I have an ailing family and I need to go away for a while and be with that person." And they'd assume that that means that they have to quit in order to do that -- we just say, "No, everybody is going to have that problem at some point. Let's figure out a way that you can do that."
It's always going to be changing. So the way that you get really talented people to stay here is just respecting their own unique choices and dealing with each of those on a one by one basis. You can institutionalize a lot about helping them understand how they're actually creating more value here than they would at other places, having transparency around a lot of the decision making, the freedom to make choices as far out, and to all of the individuals that are here but a lot of it also just comes back to "what is going to work for you personally?" because each person is going to have a different set of priorities.
We're also always trying new stuff. The latest sort of new thing that we're doing is that we have onsite trainers -- and for a percentage of people here that's incredibly valuable, to have trainers that show up. So far that seems to be one of the most valuable things that we can do. So it's like, "okay, great, people value that highly and we're going to do a lot more of that" but it's also about experimenting and trying to find more of those things.
Another thing that also turns out to be super valuable is that we have a company trip every year. And it turns out that you actually get a huge amount of work done. It's supposedly a vacation, but all anybody talks about is work in a different environment where they're interacting with a bunch of other people right? So you have the usual group of people you are hanging out and working with, but now you get to hang out with a bunch of other people -- and it tuns out we all like each other a lot.
But one of the most valuable things about the trip is that people take their parents. And it's amazing the number of times parents have come up to me and said "I'm telling my daughter, I'm telling my son how lucky they are to work here and how proud of them I am." I'm like "Yes, that person is going to work here for another five years!"
We're always looking for ways to make this a better place and just be cognizant of people who have families. My wife, who worked here at the beginning when she was pregnant, is super annoyed about how most companies make it really difficult for their female employees to deal with raising kids so we're sort of hyper vigilant about making that as easy as possible. She feels like a lot of women get forced out of the workforce because of the trade offs they have to make and it tends to be this fairly large gap. She now runs this organization dedicated to helping from birth to around three years, so she sees how hard it is for most families to keep the mother engaged with her career during that period. A lot of times, after three years they've just sort of fallen out. And that's just another instance of a class of difficulties that we all have.
I'm sure we'll have new ones as our workforce ages and we go through different sort of demographic challenges. And we'll also have people who come in and say, as we've had, "I"m a member of the US national ultimate frisbee league team and we're going to nationals and I'm going to be gone for six weeks." That's awesome -- let's figure out how you can go do that. And I don't think that will be a problem that happens frequently.
It seems like a lot of this involves a lot individual experience for employees and you've figured out a way to make that really work with the current size of your organization. Are you at all worried about scaling that if you continue to grow?
No. One of the nice things about having pretty distributed decision-making in the company is that it tends to scale really well. You can trust that lots of good decisions are being made all the time. That's one of the things that we try to explain when we're bring somebody in -- you guys know there are no monthly reports here? There's no "all of the information has to flow to Gabe." It's like, if I need to know something I'll figure out who is involved with it and find it because I'm just like somebody else. Nobody's going to put together a report for me so I can have a giant file of reports laying around that I never get around to reading. What that means is that a lot of bandwidth internally in the company frees up because you're not just constantly tracking a whole bunch of stuff so decision making is really distributed.
Where that bites you is if somebody makes a bad decision, like Diretide. I found out that we were doing something stupid when one of our customers mailed me and said "you're doing something stupid." I was like "really?" And I go and find out that yes, in fact, we're doing something stupid. That isn't a fault, it's just one of those trade offs. But we definitely in a sense have an army of customers who are always helping us stay honest. That's way better. We've essentially crowd-sourced supervision of a lot of these decisions to our customers and it works way better than almost any other system we could design. They're rabid, they're passionate, and there are a lot of them.
Part of me as somebody with a computer software background thinks most systems that have trouble scaling it's because of internal communications bandwidth problems. If you look at the old way of doing things called master-slave computing systems, they tended to clog up around I/O if you really want to scale something to huge levels what you have to do is have as much local data as possible and minimize the amount of shared state between all of the different processors. That's a description of about how some engineer at Google who is thinking about how they do search, but it applies to organizations just as much as describing very, very large computing problems.
I actually think that it's testable, and we'll find out. I think our structure will work better than most of the older command and control type hierarchical systems that require a huge amount of shared state between everybody, and move very ponderously, and have to throw away huge amounts of data because otherwise the person at the top can't possibly know what every single person at the organization is doing. To me it always seems like a lot of those organizations, the person at the top seems incredibly ignorant of the actual experiences of their customers. And it's really that if everything has to go through that one person they have to trivialize and simplify everything down to stuff that is really sort of a caricature of what is really going on in their organization or for their customers.
Whereas, for us, it's not really surprising that a person who is working on a specific piece of art can see ten thousand people's reactions to that fifteen minutes after they've done it and make four sets of changes without having to go and talk to anybody else. That's going to give you a much more productive and more responsive development process than what looks like a more traditional development process.
There's this idea that Valve has a tendency to go after people who have shown themselves to be capable of creating an interesting body of work regardless of what their credentialing are—for instance, you tend to hire a fair number of people who produced well received mods.
Well, the traditional credentialing really doesn't have a lot of predictive value to whether people will be successful. One of the things you have to do to be successful in our business is to be responsive to reactions that people have. You can give ten people the same set of forum posts and only one of them will actually take it in a productive direction. So the fact that somebody has been able to build something and ship it and not get sort of bogged down and give up and then deal with the gush of responses you get, filter through that in a useful and productive way and iterate is really the core of product design and development in our world. So when you see somebody who has already done that, especially if nobody was teaching or leading them to do that it's a really good sign that they're going to be successful.
Somebody having a PhD from an Ivy League school tells you nothing about whether they're going to be successful in our space. Grades don't tell you anything. Most people who end up being successful have good grades, but it's orthogonal -- there's no extra information than if they put together a website and have bunch of fans who love coming and seeing what they're doing.
That's true if they're writers like Erik or Chet, or programmers like Steve Bond, or artists like Johnny Guthrie. They all have shown that ability to engage and entertain and respond to an audience which is the critical characteristic for people here.
How has this decentralized decision-making process evolved over the the years? How would you compare how it runs now to when you had only 20 employees?
I think there's just more much more of history of it now so people are willing to trust it more and be less worried that it's going to go in some poor direction. I just think that we're more confident now and we've got that experience in lots of things. When we started out we were a single-player video game company that could have been really successful just doing Half-Life sequel after Half-Life sequel, but we collectively said let's try to make multi-player games even though there's never been a commercial successful multi-player game.
Then we tried to do Steam. There were a bunch of people internally who thought Steam was a really bad idea, but what they didn't think was that they would tell the people who were working on Steam what to do with their time. They were like "that's what you want to do wit your time, that's fine, but we're going to spend our time working on Half-Life 2. We think you're kind of wasting your time, but it's your time to waste."
In retrospect, it was a great idea, right? So the key thing was that people bear the consequences of their own choices, so if I spend my time on it the only persons time I'm wasting is mine. Over time, I think people sort of recognize how useful it is for people to vote with their time. There is a huge amount of wisdom in people's decisions about what they personally want to work on next.