GABE NEWELL
The Valve CEO and co-founder discusses planning for the future and optimising productivity
The co-founder and CEO of Valve scarcely needs an introduction. After spending 13 years at Microsoft, Gabe Newell quit to set up what would become of the most innovative, and successful, companies in not only the videogame industry, but the entire world. With Half-Life: Alyx now released, here Newell reflects on the decisions that have led to Valve being the company it is today – and offers us a tantalising glimpse of what the future might hold.
What does being president of Valve actually mean in 2020, and what do you spend your days doing? Usually I get pulled in if there’s something that’s unusual or out of the norm. If there’s some disaster, basically. But most of the time, the way the company is designed, it’s pretty good at operating without me. But my background thread, the thing I’m always thinking about and working on, separate from being pulled by teams into something that they’d like my help on, is brain-computer interfaces. That’s kind of a longer-term thing.
That seems like a long way from Valve’s origins, but the company is built on these sorts of changes of tack. What do you consider to be the inflection points that got you here?
We started off with singleplayer, because we thought there were opportunities there that were not being exploited. That was really the design impetus behind Half-Life, and we learned a huge amount and that helped inform a bunch of the decisions we made about Half-Life 2.
But at the same time as we were high-fiving ourselves over Half-Life shipping, we also started thinking about multiplayer games – and that was a point in time when there were no commercially available multiplayer games, there was no business model for them. It seemed like a radical concept, that having games that behaved more like sports and less like movies was a good way of evening out the boom-and-bust cycles for videogame development. It sort of shifted the burden from AI to ‘ meat’ intelligence, where a lot of the entertainment value is created by people all over the world rather than by entities running on your computer. That’s why we started getting into Team Fortress and Counter-Strike and things like that.
The next step after that was we went out and pitched people the ideas for Steam, saying: ‘Look, we’re a videogame developer, it would be super-helpful if somebody provided these sets of services to us’. It seemed pretty obvious that there was a much better way to provide value for customers and reduce the complexity of distribution, in a way that would also be a really powerful way of improving the development process. That ended up being Steam – but the funny thing is, we originally were just trying to get somebody else to build it, because we desperately needed something like that.
Steam has quickly become much more than just a store, though.
We were always used to thinking about games as entertainment experiences, but then we started thinking of them as productivity platforms. As a sort of proof-ofconcept I decided to be a gold farmer in World Of Warcraft for a while. I was making $20 an hour farming gold. I was making what was a spectacular wage for most people in most parts of the world.
That’s when we started focusing heavily on things like the Workshop, and trying to think of everybody as a content creator. There’s this story of the parents that called us up because they thought we were selling their kid drugs. What happened was PayPal pinged the parents and said, ‘Your kid is exceeding our limits of how much money they can put into PayPal per month. They’re probably selling stolen goods or drugs, because there’s no other explanation.’ So the parents called us up and I said: “He makes items on the Team Fortress Workshop. He’s making $500,000 a year.” That to us was an indication that this was a helpful way of thinking of games – as platforms – and it’s informed all of our decisions about multiplayer games subsequently.
All of these points represent certain milestone challenges for the company, and the next big one was thinking about how we could increase our design reach to include hardware components. Steam Machine was an early attempt in that space, which was not successful.
Steam Machine was a big focus of yours when we named Valve the best developer on the planet seven years ago (E250). Why didn’t it work out?
We should have been selling those things ourselves. I think we would have been more successful if we had been manufacturing and delivering them. And the other thing was something that happens to people a lot, where you fall in love with your own business plan and then you get annoyed that customers don’t understand what would be beneficial to that business plan.
It seemed really clear to me that everybody, including Valve, would benefit if we were moving towards more open-standard hardware. But the hardware we were pushing for was super-incomplete at the time. I thought, ‘This is clearly where we all want to end up, and this is a point along the pathway to getting us there’. And people were like, ‘Yes, but you’re asking me to pay you money for the privilege of being on your roadmap, and I’m not really sure what I’m getting out of this at this time’. We needed to be a lot further along in terms of delivering
polished consumer experiences before we were trying to get people to actually pay money for those things.
I think those mistakes, and the lessons we learned as a result, got us to where we are today, which is our ‘kid in Middle America making half a million a year on the Workshop’ moment for hardware. The combination of Index and Half-Life: Alyx, to my mind, is where we were always hoping we would get to – which is the ability to be designing hardware and software in concert with each other, attacking a problem which would be really difficult to attack in a piecemeal fashion: how do we move that immersive singleplayer experience forward?
I’d love to sit down with Miyamoto and say, ‘This is our best work, this is our best thinking, what’s your reaction to it?’ And to some parts of it he’d say, ‘Well of course you have to design the controller at the same time you’re designing experiences, because that’s what we’ve been doing forever’. But the other pieces of it, hopefully, he would step away from and say, ‘Yes, this is a this is a really good step forward in terms of this genre’ – which is what I feel we’ve traditionally done with Half-Life.
When you talk about pushing the genre forward, do you mean firstperson shooters or are you talking about VR games more broadly?
No, I mean immersive firstperson entertainment experiences, whether they’re shooters or not. We happen to think that firstperson is one of the best ways to take advantage of that and I think we’re making the case why VR is another critical piece in moving those kinds of experiences forward.
But it sounds like you’re already looking beyond the current phase of VR, and to the next evolution of technology: brain-computer interfaces. Can you talk us through that future as you see it, and where Valve fits in?
I think they’re coming way faster than people realise: there’ll be some really interesting announcements that happen this year. And in some ways, they’re terrifying, because they change just about everything about not just videogames but about entertainment experiences, and a whole bunch of other stuff. But I’d much rather be ahead of the curve than behind the curve on that transition.
I mean, it sounds like science fiction, but I think it’ll be surprisingly easy to start interacting with brains in a way that feels real. I have to be a little bit careful about what I talk about, but it’ll be full read/write to your brain. The range of applications that will fall out of that will be tremendous, but a lot of people are going to want entertainment experiences that work by directly reading and writing from areas of your brain.
These kinds of opportunities you’re talking about – is this why Valve is so concerned about owning the whole in terms of software and hardware?
If you asked any game designer, ‘Do you think your life would be better if you could also change the input method?’, I think every one of them would say, ‘Oh my god, there’s a whole bunch of interesting stuff we could do, even just having a custom mouse for our shooter’. So if we could make hardware more like software in terms of how elastic it is in the design process, everybody would benefit a lot from that.
As it is, you sort of end up with these local minimums. Like, we’re still stuck with keyboard and mouse. I mean, I love keyboard and mouse, obviously, but we can’t even explore the space in the neighbourhood of that, because of the cost. If I came up with this cool new input device and I had no software support, no software developer is going to re-engineer their experience – which is what they’d have to do to take advantage of your device – because you’ve only sold 2,000 of them. So you end up with the chickenand-egg problem.
In order to start moving forward in any of these dimensions, you need to be able to deliver value to customers. But now we have that capability, it’s starting to impact everybody here at Valve – you know, it’s not this special project over here, it’s a tool that we now have in our toolkit. You want the Counter-Strike team to be thinking, ‘How would you create a better Counter
“IT SOUNDS LIKE SCIENCE FICTION, BUT I THINK IT’LL BE SURPRISINGLY EASY TO START INTERACTING WITH BRAINS”
Strike experience if you were also able to start modifying or replacing people’s hardware components?’
And on the BCI side, a lot of the stuff that we’re doing as a company makes it a lot easier to be involved as those technologies move out of the theoretical, and start looking more and more like actual products.
So Alyx is an example right now where you have full control over both sides, and it seems like you’re happy with it…
Oh yeah, I’m really proud of it.
The VR installed base is comparatively small, though. What does success for this project look like, in terms of it actually finding an audience?
We’re going to be super-interested in what reviewers have to say, and we can engage with a lot of our peers in the industry and see what they’re responding and reacting to. And there’s a reason that my email address is public, right? If people know that they can speak to you directly, you get pretty unvarnished and accurate signals back from them.
The information flow is good – completely separate from sales, because VR obviously constrains the number of units that you can sell enormously. Even if there’s a huge uptick in VR devices as a result of Alyx, it’s going to be a fraction of what’s possible for a non-VR title. We can look at growth rates and attach rates and things like that to get a sense for how we did, but a lot of it will just be the zeitgeist, and what people say after they play it.
We have a track record with Half-Life. The expectations for Half-Life games are incredibly high, both internally and externally. And if the response to Alyx doesn’t look like that, then that’s going to tell us a lot.
We’ve talked about the history of Valve and how you developed from the original Half-Life. But you’ve moved progressively further away from singleplayer games. Now Alyx is here, but you still maintain a number of multiplayer service titles. What’s the balance going to be like moving forward?
So you can think of entertainment experiences as this massively distributed computation problem, right? If you think of lots and lots of people on the Internet as a way of generating entertainment experiences, there was a point in time where it became easier to connect meat than it was artificial intelligence.
We’re starting to head towards a period where that’s going to reverse again, driven by what’s happening with AI. Right now, the OpenAI bots are better than 99-point-some per cent of all the Dota players in the world. That’s actually a surprisingly narrow challenge for artificial intelligence. Beating humans is easier than entertaining humans. But over the next several years – and if you ask me, my little spreadsheet calculation is it’s about nine years – we’ll have artificial general intelligence that can do anything a smart person can do.
It’ll probably initially take something like a billion dollars to build one of these silicon humans, but then they’ll just keep getting cheaper, and it’ll get cheaper really quickly, and eventually reach the point where you have ten or a hundred people living in your computer all the time. And harnessing that will mean singleplayer games get a lot more interesting.
Where does that leave multiplayer, and in particular esports, which Valve has got into in a big way over the past ten years?
The way we think of esports competitors is as content creators. They happen to create really great entertainment experiences in the same way the creator of a mod or a level or a cosmetic would. I think that will continue to be super-entertaining. What may change is how their value is captured. The way that [pro Dota player] Puppey generates value right now is by creating spectacle, but he will always be a scarce commodity. Right now he monetises himself through tournaments. But there’ll probably end up being different ways of generating revenues for these kinds of superfreak game players.
We think esports is great, we love it, it’s growing super-fast – and these things take a long time. People have seen us sort of step away from immersive singleplayer experiences for a while, and a lot of that was that we saw more tractable opportunities elsewhere. I mean, a typical gaming company would just keep cranking out sequel after sequel – but the reason people value Valve is that we’re supposed to be the ones picking interesting problems and solving them. It’s superexciting to be back in that space.
If you could build a singleplayer game that just never ended, where I could play 20 hours a week and it just keeps growing and getting richer, and I’d be having as much fun 400 hours into this experience as I was in the first 20 hours… I think that is a way more likely scenario
“EVENTUALLY [WE’LL] REACH THE POINT WHERE YOU HAVE TEN OR A HUNDRED PEOPLE LIVING IN YOUR COMPUTER”
looking forward five years than it would have been looking forward five years ago. That’s going to be a tectonic shift in the industry, with AI becoming way more useful, and it shifts the value-optimisation inflection point between multiplayer and singleplayer games.
But also, I don’t want to be too prescriptive. One of the things we try really hard to avoid is saying that ‘Steam is good for x’, because then we’ll be tempted to put our thumbs on the scales. Like, if you’d asked me to predict, I would’ve said that Steam was going to be hugely beneficial to independent games – smaller games that traditionally had a lot of friction were going to benefit – but you don’t want to predict that, because you become fond of your own predictions and create self-fulfilling prophecies.
How do you ensure you keep developers happy? The smallest changes to Steam can have a huge impact on their livelihoods.
We spend a lot of time talking to all [kinds of] game developers. We’re always saying, ‘ What can we do better?’ And a lot of the decisions we make are steered by the input we get from developers. We could just be authoritative and say, ‘This is our platform and this is where we’re going’, but Steam is at a point where it’s way better to ask, ‘What should we do next week?’ People will say, ‘Here’s a problem that I have’, and we just go and solve that.
Occasionally there’ll be some larger-scale thing around, like, ‘ We need to provide better tools for spectation’. And that tends to be more speculative, so we have to work for longer periods of time before we’re delivering that to our partners. But a lot of times it’s just them saying, ‘For the next sale, we need these three features in our dashboard’.
The accumulation of all those changes over time means it’s relatively unlikely… The dangerous thing is when you have to make a big leap, when you have to guess, two to three years in advance, about something where we’re going to have no way of knowing if we’re doing the right thing until we get there. With Steam right now, we do have those kinds of things, but a lot of times, it’s much safer to create value for our partners by simply saying, ‘Let us know what would make your life better’.
If somebody calls us up and says, ‘You just fucked me’, we’re going to pay a lot of attention to that really quickly. That gets everybody’s attention, because it’s very much a service-oriented… you know, we’re the concierge at the
“COMPETITION IN GAME STORES IS AWESOME… IT KEEPS US HONEST, IT KEEPS EVERYBODY ELSE HONEST”
hotel. Somebody says something’s gone wrong, or ‘You made a change and my game has stopped selling overnight’ – everybody jumps on that, and there’s usually a post-mortem. Let’s fix the problem first, but let’s also figure out how we ended up doing something that was unintentionally negatively impacting people.
You make the concierge comparison. But now, with the Epic Game Store, there’s another concierge across the lobby, shouting, ‘Hey, come and speak to me instead’. Does that affect the way that you approach Steam and build this relationship with developers?
First of all, competition is great. That’s why we love the PC platform, everybody has to compete on a wide variety of dimensions. Competition in game stores is awesome for everybody. It keeps us honest, it keeps everybody else honest.
But it’s ugly in the short term. You’re like, ‘Argh, they’re yelling, they’re making us look bad’ – but in the long term, everybody benefits from the discipline and the thoughtfulness it means you have to have about your business by having people come in and challenge you. When you’re in the service business, your partners may come to you and say, ‘We have some additional ideas’, and then you say ‘Okay’, but it doesn’t really change the underlying loop very much.
We get a lot more freaked out not by competition, but by people trying to preclude competition. If you ask us which is scarier, it’s people who are falling in love with Apple’s model of controlling everything and having faceless bureaucrats who get to keep your product from entering the market if they don’t want it to, or designing a store in a way that minimises software’s value-add to experience and stuff like that. That’s way scarier to us than competitors. In one case, somebody is challenging you to do a better job. And in the other case somebody is not letting you do your job at all, and that’s more unnerving.
You mention corporate facelessness – in many ways you are the face of Valve as a company. How does that sit with you personally?
I mean, I like gamers, I hang out with gamers all the time. The community we’re part of, it occasionally has its rough edges – like, ‘hey, pig fucker’ emails, those kinds of things. The thing that’s a little bit weirder to me is… I’m fine with people coming up and talking to me. But there’s this thing that happens, where somebody will come up and say, ‘can I take a picture with you’ and I’m like ‘sure’ – and then I put my arm on them and they’re shaking and that bothers me way more. I like our community. I like our customers. I like the gaming world. And so interacting with them is fine, but I’m less certain about the whole cult-like aspects of it occasionally.
That’s what happens when you transcend humanity and become a meme, we suppose.
Yes. I’d much rather be an email address than a meme [laughs].
But the cult of personality around you exists because you’re the public face of a company built on, as you say, solving these big, difficult problems. People expect you to make these big, mad bets. How do you deal with the pressure?
Betting that you’re right about something that’s two to three years – or, god forbid, five years – in the future is very stressful. But I’m more willing to do it than most people in the company, simply because over the course of my history I’ve made both interesting positive and negative bets, so I’m a little less terrified to go years without being able to test my assumptions. And that’s one of the things that I do that’s useful for the company.
Like, I’m the only person in the company who can go off and think about brain-computer interfaces, because if somebody else did, they’d all just laugh at them and say, ‘Why are you wasting your time on that science fiction?’ But with me it’s like, well, I was right about Steam, I was right about the connected economy! And then they get to say back: ‘Yeah, and you did Steam Machines too’ [laughs].
Not everything has worked out, but you’ve built a company that has not only been tremendously successful but has also been structured, since the
beginning, in a very different way. It’s easy to look back now and think, sure, the flat, free-wheeling company design has really worked out. But what led you to know it was the right thing to do in the first place?
Way back at the very beginning of my career, The Mythical Man-Month [book by Frederick Brooks] was pretty interesting because it really talked about the design and scale of organisations from an engineering perspective. It used to be that [1980s US computer retailer] Businessland would take 53 per cent of the gross revenue from the sale of a product. Nowadays, retail distribution is like 6 per cent in some cases. That whole progression of various lines of business functions contributing less and less value, and representing less and less of the value of delivering products to customers, was going to have profound impacts on what optimal company design was going to look like.
So the bottom line was that this transformation in how you communicate, market, support and distribute to customers would mean that the design and production of the products was going to matter way more. You were going to go from an era where 10 per cent of the company was R&D to where 95 per cent of the company was R&D. If you took the world’s best programmer and you put them into 1984 General Motors, it wouldn’t make any difference. Their output to customers would be identical, and very, very low.
Somebody like Yahn [Bernier], who has on many occasions written 4,000 lines of code in a day – and they’re really, really good lines of code – means that you really want to optimise for all these super-highproductivity, brainy people. And when you try to figure out how to maximise the productivity of those people, you end up with something that looks very different than a lot of traditional organisations.
That led to this theory of, how do you optimise for the productivity of people like Yahn or Jay [Stelly] or Robin [Walker]? You’re as flat as you can be, there’s no siloing, all productivity occurs on the boundaries of skillsets, not within silos of skillsets. Robin’s great because he’s a game designer and a programmer and he can talk to the press. Ken Birdwell could build a level, animate the creatures, skin the creatures and write all the code that the creatures used – and that’s why the tentacle sequence in Half-Life could even exist. He could solve the different problems by saying, ‘Oh, I can fix that problem in code, oh, I’ll just change the level, I have no idea how to get the code to do this so I’ll just change the level or the monster design’. And at the time, those were kind of radical concepts, certainly coming from Microsoft, which was not optimising or designing its organisation to save money.
The fundamental problem always came down to: how do you attract – and keep – the brightest, most talented people from all over the world? Even at the beginning, we were hiring people from Russia and the UK and Florida, and other really remote foreign countries. Most of those people are still here all these years later, and the productivity has remained high. Near as I can tell, in terms of revenue per employee and profitability per employee, we’re in the top ten of all companies in the world. Ever. And that’s not to say that we’re financialmetric-driven, but that is a sign that our design of the company is working the way it should.
What challenges come with that company design?
I mean, we have to deal with some interesting things. Like one, we’re getting older. Our biggest competitors are actually parents. People don’t leave Valve because there’s a place where they feel like they’re going to be more productive or have more personal satisfaction; people leave because their parents have advanced Alzheimer’s and they need to go home. We haven’t quite figured out how to compete with Alzheimer’s yet.
You’ve built one of the most successful companies on the planet. You don’t need to come to work any more. What keeps you going?
Personally, for me, I could have retired a long time ago. So the competitor for my time is going and fucking off and doing whatever I want to do… Well, that’s what I do. I am doing exactly what is the most fun and most rewarding and most stimulating.
I love working with Robin. It’s a blast that Kaci [Aitchison Boyle] is here. I’ll come up with excuses to sit in meetings with Erik Wolpaw, because they’re some of the most fun I could have right now. Meeting with Wolpaw beats 99 per cent of the comedies that come out of Hollywood. And that’s what I think it’s like for most of the people who work here. This is a great place. It’s a lot of fun. It’s a good work environment, not for two years, not for a game project, but for 20 years or for their foreseeable future.
“I COULD HAVE RETIRED A LONG TIME AGO… I AM DOING EXACTLY WHAT IS THE MOST FUN AND MOST REWARDING”