Andrea Peterson has a fascinating article on The Switch about Valve software. Valve is an incredibly successful software firm that has both pioneered genre-busting games (Half-Life) and created the dominant platform for distributing PC games. What’s interesting is how the company organizes its workplace.
The company is flat. … [Valve co-founder Gabe] Newell describes Valve as “a bit of a culture shock when people are coming in from other industries.” In fact, a leaked New Employee Handbook mainly advises new hires on “how not to freak out now.” And even that document is a team effort. … The guide book is half training material, half manifesto for Valve’s management style: action-oriented democracy. “We want innovators, and that means maintaining an environment where they’ll flourish,” it explains. The best way to do that, Valve believes, is by giving its employees the chance to do whatever they thinks is best. So, Valve has no institutionalized management structure.
The new employee handbook makes it clear just how radical this commitment to openness is:
we don’t have any management, and nobody “reports to” anybody else. We do have a founder/president, but even he isn’t your manager. This company is yours to steer — toward opportunities and away from risks. You have the power to green-light projects. You have the power to ship products.
There are some constraints. Employees get regular evaluations. They also get “stack ranked,” so that their compensation depends on how much they contribute to the company. But both of these depend on peer review, too.
As Peterson suggests, Valve works through a kind of action-oriented democracy. Without a formal hierarchy, decisions are made through argument and persuasion among peers. You try to make a case for a project, or a new feature, and you succeed if you attract enough people to build a team to start working on the idea. It’s likely that some peers are better at persuading their colleagues than others (if they have a good track record, I imagine it’ll surely count in their favor). But nobody has formal power to order anyone else to do this or that.
This hands-on approach emerged for practical reasons (see here for a useful short history of how the “cabal” process of cross-company decision-making came into being). The people at Valve are not interested in abstract theory but in making good software. Newell describes Valve’s philosophy in terms of economics (possibly, he’s building on conversations with Yanis Varoufakis, who spent some time as Valve’s economist in residence). But we political scientists sometimes fancy that we have some specific insights about democracy that economists (even awesome ones like Yanis) don’t always appreciate.
Some of these insights are described in a paper that Cosma Shalizi and I have written on the cognitive advantages of democracy. I can recommend this paper in good faith because it tries to synthesize other people’s ideas more than to showcase our own. Briefly, there’s good reason to think that democracy is better suited than either hierarchies or markets to solving certain kinds of complex problems. To solve these problems, you need to bring people with many diverse points of view (each of which can capture different facets of the problem) into contact with one another so that they can argue about solutions and determine the better ones.
Hierarchies are bad at representing different points of view (the views of the bosses prevail over the views of the underlings, even if the underlings know more), while markets are bad at bringing people into conversation with one another (the price mechanism is a near miracle, as economists like Hayek have argued, but it’s a poorish medium for subtle argument that can change people’s minds). Democracy creates equality — allowing different people to bring their various perspectives to bear. It also thrives on the kind of argument that can lead to better solutions. A new social science literature is beginning to emerge, focusing on these pragmatic benefits of democracy, and how they can best be harnessed.
Valve has clearly come to similar conclusions, but on the basis of practical experience (perhaps a better guide to what works and what doesn’t than academic theorizing). As described in the Valve employee manual:
There are lots of stories about how Gabe [Newell] has made important decisions by himself, e.g., hiring the whole Portal 1 team on the spot after only half of a meeting. Although there are examples, like that one, where this kind of decision making has been successful, it’s not the norm for Valve. If it were, we’d be only as smart as Gabe or management types, and they’d make our important decisions for us. Gabe is the first to say that he can’t be right nearly often enough for us to operate that way. His decisions and requests are subject to just as much scrutiny and skepticism as anyone else’s. (So if he tells you to put a favorite custom knife design into Counter-Strike, you can just say no.)
Relying on the collective smarts of your employees, rather than on only the smartness of managers, has paid off extraordinarily well for Valve.
So, why don’t more firms do this? Again, the new employee manual has some interesting political insights. Running a flat organization is hard and requires constant commitment from everyone involved. Put differently, it takes a lot of decisions that are usually made by managers in traditional organizations and makes them the subject of direct and observable politics. This can pose a lot of strain (I suspect it is one of the reasons that Valve is so careful in its hiring process because even a few crazies or jerks could ruin everything if they were sufficiently persistent).
it requires a great deal of freedom from outside pressure — being self-funded was key. And having a founder who was confident enough to build this kind of place is rare, indeed.
In other words, the American business environment makes it difficult to experiment with the kind of radically flat organization that Valve is pioneering. Firms usually turn to private funders or to capital markets when they reach a certain point in their life cycle. Both kinds of investors are likely to be impatient with arrangements that benefit employees if they don’t provide measurable short-term payoffs, even if they have long-term economic advantages.
More subtly, as Washington University political scientist Gary Miller argues in his book Managerial Dilemmas, too much emphasis on “shareholder value” makes it hard to make long-term commitments to employees. Miller argues that firms are really successful when they can create trust-based reciprocity, in which the business promises to look after their employees over the longer run in return for their employees making extra efforts that are difficult to measure but hugely beneficial for the firm as a whole. If employees suspect that in a crunch their employer will side with anonymous shareholders rather than with them, it’ll be difficult to create this kind of relationship. That’s why it is likely much easier for self-funded firms like Valve to create strong bonds of trust with their employees than it is for businesses that rely heavily on capital markets.