Valve's Gabe Newell on, well, everything

I think that that's one of the factors that you look at in your design is what's the footprint and we need to do a good job of enabling that both for ourselves and for people who are relying on us to solve those kinds of problems for them.

I think our problem right now is we almost always come down on the side of making things bigger and more fabulous rather than making things smaller and faster to download, and that has an impact both on our footprint and on our schedules, which always... you know, we've never met something cool that we didn't end up deciding to put in the game as opposed to hitting a ship date. So, it would be a really good exercise for us. Just talking to somebody who has to run a company, it would be a good exercise just to go off and try to do these even smaller and more focused projects. I think that that would be good for us and a nice change of pace.

We've had a couple of internal proposals to do things like that and I think it's a smart thing to do. I think anybody that has a big sort of tadpole game would really benefit from building those much smaller experiences, especially if you can leverage the assets that are already on the person's machine. So it'd be nice if it were small, but it would be nicer if the download were even smaller taking advantage of the fact they've already got Half-Life 2 or Unreal or Doom III installed on their machine.

I think the question I'm personally wrestling with right now is the bigger entertainment experience that people want to have. I think that we have these accidents of production technology - you know like 'I know how to make games' or 'I know how to make comics' or 'I know how to make action figures', and I think we're missing the boat in terms of what customers really want to have, which is a more comprehensive entertainment experience, so that's what I'm really scratching my head about a lot right now. Does that make sense?

Well when you look at what people are responding to really well right now... If I'm a fan of Pokémon, right, I go to one place to get my Pokémon DVDs, I go to another place to get my Pokémon cards, I go to a third place to get my Pokémon games, and the development of those things occurs in completely separate groups with completely different notions of quality and so on, and clearly people want to have a more integrated experience than that, and we're just not doing a good job of providing for that. Figuring out the right thing to do there is something I'm scratching my head about a lot right now.

I worked in a bunch of different kinds of organisations, and I think they're better suited for different kinds of problems. Very hierarchical, very command-and-control organisations work really well for doing the same thing over and over and over, where removing defects is your main goal. It's like if you're making cars or you're doing product support, repeatability and defect-detection are critical aspects and you need to build an organisation that works well, but I think that the challenges that we have right now in the games industry, specifically and more broadly in the entertainment industry are about inventing new things, about seeing things that occur between disciplines.

So a lot of time the people who are most successful doing that here at Valve are people who are engineers who have fine arts background, and our goal is not to make them increasingly narrow and increasingly specialised but instead push them to be broader in their perspective and see the opportunities for what hasn't been done rather than to do what's been done in the past more cheaply or with greater quality. So the organisation that supports that is one that minimises the boundaries between, not emphasises the restrictions and values specialisation, so that's why we don't want people thinking of themselves as increasingly specialised and narrow in their focus; we want people to say, 'you know what, I was thinking about this problem and it's not really an engineering problem, it's really this kind of problem; it's a production problem, it's a tools problem, it's an art problem, and here I can show you because I think broadly enough'. So to that extent titles are not particularly helpful to us in terms of getting people to be more successful.

The thing we talk about most publicly in terms of how we do things is the cabal approach - the collaborative, iterative approach. We work very, very hard to keep people here because so much of what's necessary when you're doing this is experience with the other people around you. Like, we [gestures to Doug Lombardi] have been working together for ten years and it really helps that we have that shared history to draw on when we feel like... It's much easier for us to take risks, like, 'oh, let's do this Orange Box thing,' which... We had enough of a history to say these are the positives, these are the negatives, and have confidence in each other's judgement and know where each other's coming from. So it would be much harder to do that if we didn't have that shared history.

That's true for lots of things. I've been working with everybody here for a long time, and we get better. It's one of those things that when you ship a product you learn a bunch of stuff about how to do that better, and it would be a tragedy to lose that. So that's another thing that contributes to our ability to do this is just the longevity of our shared experience. People change roles a lot inside of the company.