advocacy entrepreneurship games industry recruiting

How Valve runs a successful game business, hires people, and more

Here’s a long (long!) video interview with Gabe Newell, CEO of Valve (one of the biggest / most successful games companies).

(incidentally: this post is shorter than intended. Someone at WordPress considered it acceptable to DELETE your post if your login cookie timesout before you hit the save button. Completely the wrong way to build a blogging platform)

Listening to the long interview, I found him saying some very concise, pithy things about the games industry, and the roles of us working within it. Some of them are clearly at odds with the “corporate” messaging that typically comes out of the larger games companies. Personally, I have often railed against those corporate statements and shouted “don’t believe a word of it! read between the lines – this is a person with their own hidden agenda!”, so I was delighted to hear Gabe providing much more rational and intelligent messages.

I transcribed a few as I listened, as they resonated with a lot of the concepts I’ve tried to hilight on this blog and elsewhere.

Employer responsibility, and a culture of humanism

“You cant ruin people’s home lives to benefit the business

we’re not telling them to work on the weekends, but people are working on the weekends

those really are the things we worry about”

Contrast this with the issue that made me quit the IGDA:

Mike Capps (CEO of Epic Games) who claimed that: “working 60+ hours was expected at Epic, that they purposefully hired people they anticipated would work those kinds of hours, that this had nothing to do with exploitation of talent by management but was instead a part of “corporate culture,” and implied that the idea that people would work a mere 40 hours was kind of absurd.”

Even when doing a PR-interview to try and un-fuck the issue – supposedly on his best behaviour, trying to sound like a good guy – Mike Capps felt this excused his behaviour:

“My guys ask to crunch. They say, “Hey, we’re not crunching yet. What’s going on? Why isn’t everybody crunching? This is really serious!” That kind of stuff.”

No. Doesn’t stand. You can’t abrogate responsibilty – especially not when you’re an at-will employer in a country with employment law that gives employers many rights, but employees almost no rights at all.

Gabe’s language (whether or not Valve actually does this) is in the opposite, humane direction: at Valve they “worry about” this, and supposedly seek to stop the behaviour, not to work with it.

A real games “business” is self-funding, always

“we fund our own projects so I dont have to worry about how the bank or whoever feels about our business decisions … it makes it a lot simpler to run the business that way”

This is the most common recurring issue I see with good indie games companies that fail – they cannot (or “will not”) grasp the importance of the above statement.

(EDIT’d this section to be clearer; and, of course, this is all IMHO – I have no idea what Gabe/Valve thinks on this)

Read that carefully: it’s “a lot simpler to run the business”. That should be a wakeup call to all the studios that say “I’d love to work that way, but I can’t afford to”; I’d say: you can’t afford *not* to.

It’s generally accepted that *if* you get to that point in your studio lifecycle, you’ve got it “made”. In practice, that should be turned on its head: until you get to that point in your lifecycle, you’re heading towards failure.

Often they make excuses to themselves that it’s “not possible” to run this way, and accept it won’t happen, and then blithely go about their business.

Net result: their games get worse and worse, as their competitors pull away from them, and sooner or later they drop below the standard it takes to keep getting new projects, and BANG! studio goes under.

All digital products these days are an order of magnitude easier/cheaper to make than they were 15 years ago, ignoring the staff costs; service prices have plummeted (web hosting costs, software suite costs, etc). They’re at least an order cheaper/easier to launch and sell in the marketplace. If you’re a startup, you should find it trivial to get to self-funded project status – ignoring the staffing costs.

So. Compared to 15 years ago, you have two obvious routes to self-funding: get someone else to pay your staff costs, but move *very* quickly to where you don’t need their money (because otherwise you’ll have a hard time forever), or do what you can with the people you have (you, your co-founders, the goodwill you can get from ex-colleagues, etc). It’s not excusable to say “self-funding our projects is out of our reach” – this is simply not true. It may require some ingenuity – or it may simply prove that your business is non-viable (if your business plan is to out-do Zynga at their own game, for instance, you’ll probably find it’s just not possible. In that case, declaring “we’re starting off non-self-funding, and when we get our first hit game (like Zynga did), it’ll be easy from there” is just papering-over your hopeless business plan).

How to get a *good* job in the games industry

“the main characteristic we look for is the ability

  • to create something
  • develop an audience about it
  • measure the reations to something you’ve created
  • and then change what you’ve built to reflect that
  • and measure again how much of a difference you made

Sound familiar?

If you’re serious about startups, it should do – it’s the path that et al have been pushing startups along for the past 5 years. The best of the entrepreneurs are expected to live and breath this approach by now.

It’s not even rocket-science – a big part of it is nothing more or less than the Scientific Method, over a century old now, which has driven most of the world’s research. It works. It’s a pity that so many people ignore it.

If you want to be a game maker, then … make games

Partly responding to the oft-quoted fear “but how can I get experience making games, if the pre-requisite to joinging a game team is that I already have experience making games??”:

“iteration cycle with Customer Feedback is the most important characteristic for somebody to be successful right now, and ability to demonstrate that through a portfolio, through a website, through a mod

If you have learnt anything at all, if you have achieved anything, if you have any skill – then you can *always* demonstrate that, somehow. If not, then implicitly your achievement doesn’t exist – if you can’t show it, it’s not there. c.f. the section Marketing is a science, not an art, and read Sergio Zyman’s book if you need inspiration here.

Which matters more: credentials, or mindset?

Atttitude and approach wins, apparently:

“you have to actually act almost like a CEO yourself, in terms of understanding an audience, understanding a market, building a product, taking feedbakc about the product evolving the product communicating about the product

more than whether or not you go to an Ivy League school … or take CS classes … or drawing classes … that for us is the key indicator of future success

an awareness of what’s actually going on right now tends to trump a lot of previous experiences … I think it’s going to be harder and harder for people to stay current as the pace of things accelerates … get in front of instead of get behind any structural changes of an industry you’re going into

Don’t take a job you don’t want, to sneak into the one you were too crap to get

And, so important (and lied about so many times by journalists, HR departments, recruiters, et al): the worst thing to do if you want to get into a game development job is to join QA expecting it to be an “easy route in”:

“each person that we hire has to be able to do that, even if they’re just going to be in marketing … or support … or QA”

i.e. QA is no “easy path” – you’re still held to the same criteria.

But also, as *so few* execs from EA etc are willing to admit (and I pick EA, because I’ve seen their senior people HR blatantly lie (IMHO) about this on multiple occasions, following their own agenda):

“at most companies they put in all these barriers to keep people from moving out of QA or support … in some companies you can actually get fired for trying to get out of support positions into the development organization …[so instead] build a flash game; ship it; make it better … and you’ll get everybody’s attention if you’ve got talent”

14 replies on “How Valve runs a successful game business, hires people, and more”

It doesn’t surprise me to hear Gabe saying this stuff — despite their always-horribly-delayed ship dates, Valve always releases good quality games and have basically tied up the digital distribution market. You don’t make the latter move unless you’ve got a keen eye on current trends and a lot of smart people.

Steam was originally a huge bag of shit when it was officially released (For example, I distinctly remember that if you lost connection, the UI would hang indefinitely), but they iterated on it to the point where it’s almost unthinkable that we played PC games without it. I keep on reading entrepreneurs saying “ship it and start charging for it as soon as possible, even if it’s shit”, and Steam is a huge advert for that approach (though they did have a community to migrate across and dangled a patch carrot — I remember having to install it to get the last TFC patches, as was quietly phased out).

I seem to recall reading that Valve hired one of the programmers who created bittorrent, too.

Another example of being in control of your own destiny is Valve’s TF2 monetisation schemes. I personally dislike what it’s done to the game (a very simple game with well-defined roles is now a stramash of different hats, silhouettes and weapons making it more noisy and random) , but how many other ‘Triple A’ companies have the ability to release something like that and then make an absolute killing in a 3+ year old game? The polycount guys (who were receiving something like 40-50% of the sale price, if I remember correctly) raked in $39-47k each in two weeks, and that’s just for a handful of simple art assets. . The sales will obviously tail off, but still, that’s the numbers for a handful of items. Those numbers are insane. Valve maintained their community over the years via releasing scores of free updates, and now they’re hitting the jackpot.

You can’t do it if you’re on console (or at least, not without MS/Sony demanding a large cut), it’s easier to do on PC and it’s simplest if you own your own publishing platform where the majority of users have already purchased something, so getting them ‘over the line’ is easier.

I take my hat off to them for the way they do things, even if I don’t like the fact that one party is dominating digital distribution. Still, better Valve than EA or Microsoft. Games for windows live? Really? Just yet another example of MS totally missing the boat and the point.

Valve got started with publishing and financial support from Sierra On-line, so while the conservative approach of growing off of revenues is definitely a good one, the notion of a 100% being the only viable method of starting a business is a bit too idealistic.


Yeah, sorry I wasn’t clear enough (I’ll go back and edit it to make more sense).

Basically … I’m saying it’s *fine* to start as e.g. work-for-hire – but you have to recognize at a fundamental level in your business plan that this is *not* an OK way to run as a business long-term – that you have to be using this as a temporary stepping stone.

Especailly: it is NOT acceptable to use it as a holding-pattern; if you’re still not able to self-fund projects – say – 3 full project cycles later, you’ve sleepwalked into AO purgatory: too poor to move forwards, constantly sinking further and further behind.


Yeah, me too – Steam is a terrible user experience, and often considerably inferior to what it’s replacing (retail). But, with a business hat on, I’d say they’ve made a great business there. I think that’s doubly impressive given how crap it is compared to what digital distribution ought to be :)

What specifically about it do you dislike nowadays, Adam? I find it to be pretty low-friction these days with a couple of exceptions.

Steam-is-down-you-can’t-play. This is bad; I understand maintenance is required and that downtime is always going to be peak time _somewhere_ but…

They have a client on your PC, yet only post downtime announcements in a forum sticky (that nobody reads) — a ridiculous state of affairs in my opinion. I’ve had many nights of gaming ruined by the steam service going down while my friends and I were part way through a game of Left4Dead. Once it goes down, nobody can join the server you’re on, so your game just dwindles away to nothing and you can’t find another until Steam comes back up. If folk knew it was going down, they would just go and do something else until the maintenance was finished. Instead, we get mystery surprise and ruination.

Connectivity bugs. It’s hard to know where Steam ends and Valve’s game code begins sometimes, but L4D1/2 in particular have some horrendous bugs with their steam-integrated lobby systems. “Session no longer available”, “Steam ticket expired”, “Steam validation rejected”, joining a friend for a game only to find you’re placed in a different server for no reason and so on. Quite frustrating.

I don’t have any real problems with the usability of the program, and purchasing, downloading and installing games has always worked fine.

It’s the small things that make Steam crap – and the big things that only happen when you’re unlucky (c.f. the comment recently on this blog from someone who “gave up” trying to install Civ5 with Steam — it was (allegedly) quicker for them to pirate it, and bittorrent it, than download from Valve’s servers. That’s ridiculous!)

e.g. – “maintenance” on a game that I already own? WTF? WTF-ingF?

There is no such thing as “maintenance” on a retail game – it’s not strictly necessary even on persistent-world MMOG’s (it usually happens there because the MMO server was written badly in the first place, or because the team is trying to cut corners and save costs).

That’s the kind of BS I’d expect if someone like Ubisoft were to build a DD system – Valve should (and does, IMHO) know a lot better. I’m sure they’re well aware how much of a rip-off that is for consumers, unless they give substantial cash discounts for buying the “inferior” Steam version of retail titles.

(and it does backfire. e.g. Firaxis prevented me from buying Civ5 for so long that the *FREE* community-mods for Civ4 are now so much better than Civ5 I have lost the desire to buy Civ5. Doh! Civ4 works 100% of the time, and will always work – it’s not dependent on flakey Steam servers – and the mods are (did I mention this yet?) FREE! Firaxis is shooting themselves in the foot here…)

Unfortunately, Valve demand you need to be a coder or artist to work for them, they won’t hire people who can’t “make the content”. To me, this is an indictment of their tools, not good hiring practices.

The best level designer I know works for a company who use an advanced visual scripting engine. He’ll be there for the entirety of his career, when companies like Valve should be queueing up to hire him, *because* he can’t code a line.

Of course a small company is going to need multi-talented people (but even then, I’m currently working for a MMO startup, and I’m a designer/writer, editing XML files for their data-driven engine), but as a larger company? I do wonder what they could do tapping more of the talent pool out there.

Also, I detest Steam. I buy from Impulse, if the game is available from there, in preference. Two of the single player games I have on Steam are unplayable for me because of bugs in their patches. If I could simply install the patches I needed, rather than being forced to “update” to the latest…

I happen to agree with that sentiment(*), but … I don’t recall Gabe ever mentioning it during this interview, which is interesting. I wonder if that omission is significant?

(*) – if someone refuses to draw, code, or similar, fair enough … in return, I’d generally refuse to hire them. The chances that someone genuinely “can’t” do one of those are measured in small fractions of a percentage point – i.e. very very unlikely.

And, on the flipside, the people who are crap at their job in the games industry but skilled at getting jobs and sitting there doing nothing (or wrecking projects with their incompetence) always (IME) refuse to make content – they know that would expose them too clearly. So, in addition to the other benefits, I know that for some studios it’s become a fast way of filtering out most of those people.


I can script in at least a dozen languages, including several C-based, and know my way around several data-driven engines. I get lost at “pointers”, in coding, though.

Of course you’re entitled not to hire me… I’m a designer of the sort Valve don’t want, since I can’t *code*. As I’ve said in the past I also believe strongly that most designer tools suck (including Unreal and Hammer, oh yes) and that visual scripting is amazing for faster iteration times…

Well, it’s working pretty well for Valve, it has to be said.

They often hire modders and graduates who have proved they can take an idea from just that — an idea — through to something tangible. Portal was a great example of that; Narbacular Drop was created by students and Valve hired the team, they didn’t just cherry-pick the idea and develop it in-house. It’s pretty cool that they still do this; it shows their decision to hire John Cook & Robin Walker (QWTF developers) wasn’t a flash in the pan — they obviously gained a lot from it.

Not only that, but they don’t pay lip-service to the whole learn/iterate thing. I know a few amateur level designers who were flown over for interviews despite having no professional experience; they had released strong TF2 maps and some other stuff, but had no shipped titles. The fact that their maps stood alone and were played and enjoyed by a lot of people was enough to secure an interview. There wasn’t a standard “sorry, not enough experience” HR reject. One of those guys had been hard-rejected by a load of local (small, not particularly successful) companies, yet Valve flew him half way around the world and spent the best part of a day interviewing, showing him around and whatnot. I’m not holding up their hiring process as some amazingly perfect model (I’m sure they’ll have their own failures and problems), but they make fun games and seem to focus on certain things that other companies filter out.

Also, Valve hired various writers and other more ‘wooly’ types, so it’s not strictly true.


OK, we’re using different definitions. Pointers have nothing to do with programming per se – they’re a single syntactical feature of a single language.

I’m sure you could handle Pointers very quickly if you met someone who was willing to explain them clearly – the concept is trivial, but the syntax used by C and C++ is insanely, mind-blowingly, maddeningly shit. It isn’t even logical. Most C/C++ coders have simply memorized it and moved on, because you see it thousands of times a day if you write in those languages.

…but the same is true for several core pieces of C-syntax – badly-designed, and mostly ignored by the rest of the world. C-derivative languages tend to either remove / fix these bad bits, or else don’t see much mainstream usage.

adam – Well, it’s an example. I really don’t understand the structure of coding languages, or have an interest in what they do per-se. I’m a functional user of scripting languages, which usually /are/ logical in their nature.

I struggled heavily with modding, for example, Supreme Commander – which used Lua, but in such a way it was effectively laid out as a coding language, not a scripting one.

In my experience, that’s not really acceptable to Valve, so…

@Mark: Of course, Valve has benefited tremendously from the mod community. Counterstrike helped get Valve to where they are, so it isn’t surprising they are willing to keep looking at the mod community for the next thing / employees.

Plus they’ve got the money to fly people out due to their arguably near-monopolistic hold on the PC DD market. Game development is probably not Valve’s core business any more (and given they can wait years between releasing even ‘episodes’ of their flagship title Half Life, it would appear to rank down the list somewhat) compared to their digital distribution operations.

Comments are closed.