A Spontaneous Order Firm

Valve’s Chief Economist tries to explain the Valve Model:

A corporation that tries to function as a type of ‘spontaneous order’ (i.e. without an internal system of command/hierarchy) seems like a contradiction in terms. Smith’s and Hayek’s spontaneous orders turn on price signals. As Coase et al explained in the previous section, the whole point about a corporation is that its internal organisation cannot turn on price signals (for if it could, it would not exist as a corporation but would, instead, contract out all the goods and services internally produced). So, if Valve’s own spontaneous order does not turn on price signals, what does it turn on?

The answer is: on time and team allocations. Each employee chooses (a) her partners (or team with which she wants to work) and (b) how much time she wants to devote to various competing projects. In making this decision, each Valve employee takes into account not only the attractiveness of projects and teams competing for their time but, also, the decisions of others. The reason is that, especially when insufficiently informed about projects and teams (e.g. when an employee has recently joined Valve), an employee can gather much useful information about projects and teams simple by observing how popular different projects and teams are (a) with others in general, (b) with others whose interests/talents are closer to their own.

Just like in a marketplace, everything in Valve is in flux. People move about (making use of their desk’s wheels), new teams are formed, new projects are concocted. All this information is observable by the naked eye (one notices an empty spot where David’s desk used to be, and then finds out that David moved to the 4th floor to work with Tom, Dick and Harriet), on the company’s intranet, in cross-team meetings where teams inform each other on what they are working on). People learn constantly, both by observing and by doing, the value to them of different projects and teams. These subjective values keep changing, as the time and team formation signals that are emitted by everyone else are updated.

The idea here is that, through this ever-evolving process, people’s capacities, talents and ideas are given the best chance possible to develop and produce synergies that promote the Common Good. It is as if an invisible hand guides Valve’s individual members to decisions that both unleash each person’s potential and serve the company’s collective interest (which does not necessarily coincide with profit maximisation).

It’s an interesting post, much longer than I have quoted here, although no evidence is given that the time allocation system works anything like an invisible hand–a few good games do not a social revolution make.

Coase’s islands of conscious decision are also often misunderstood. The islands are not cut off from the market sea but are permeated by the sea. Everything that goes on within the firm does so in light of the shadow prices projected from outside. Absent those prices the firm fails into socialist miscalculation.

Still, there is something to be said for how modern technology can coordinate mass action. The phenomena is perhaps most evident in the way that distributed computing coordinates the actions of thousands of computers to solve various problems, each day and each hour drawing on a different set of computers. Coordinating people in this way allows them to quickly participate in joint actions, such as a flash mob. (See also anonymous). Indeed, silicon Valley writ large is not that different from internal Valve, people end and form new firms all the time.

Capitalism allows within itself many alternative social arrangements, to think, however, that one particular such arrangement is the one that must govern the whole is badly to miss the point.


One thing this model has going for it. It is an incredible marketing tool for amazing talent. Valve now has a hardware division and have assembled a team of electrical designers with a lot of background in the make community. Very impressive resumes and very good at getting stuff built.

Where are they based? I was trying to explain a model for collaborative venture capital funding of films rather like this to a friend in the 1980s. I also pitched articles to magazines in the 1990s saying that the future of the internet was volunteer manufacturing, making things as a hobby business and even a form of leisure activity once the collaborative networks were up.

A Forbes editor was interested in the manufacturing-as-volunteer-leisure story idea in 2000, but in the end took a piece from me about adults being better at learning languages than children instead.

I think they're based in Seattle, my cousin just applied to work there

socialist miscalculation? yes, that was a point before an internet era.

but now, there are papers, which show - price auctions ( emulation of market ala Lange, but orders of magnitude faster) could produce adequate ( market like ) prices even if there is no market ( think of amazon which is presumably is the only selling organization in the world, but with auctioning of prices between planners in organization ( there were always bidders in socialist enterprises which negotiated terms of exchange in given bounds ) and final customers - the right prices will certainly be found in a matter of days and then will keep adjusting ).

This is fine for larger projects of at least a few days in duration, but I'm not convinced that this will work with, say, tech support. Ideally I could ask for a price to respond to a bug report. But sometimes seemingly simple bugs require fiendish sleuthing. What if I underbid?

Also, sometimes the best tech support response is "we don't have a fix for that and don't plan one any time soon." But how do I know that's the correct with response without someone (or a market) making that decision.

My incentive will be to close bug reports as quickly as possible, but that doesn't necessarily mean I will fix all those bugs.

Shades of my education past, Sergey! Thoughtful post.

Yet, who will own Google?

This model only works in specialized situations of course. I'm not sure I agree with "Still, there is something to be said for how modern technology can coordinate mass action" as being applicable here. One could imagine the Valve of 1950 being teams of highly driven engineers collaborating on various car design projects in a similar fashion.

The key thing here is that designing and building can overlap significantly in game development, and be done (in some stages) with relatively few people. I'd be surprised if there is a single project at Valve (Half Life 3 anybody) that is structured exactly this way once it involves more than 50 people. You need additional coordination mechanisms at scale.

Having said that, it is amazing what happens when you give a bunch of highly talented driven people funding and allow them to discover a common purpose. John Carmack's QuakeCon talk reinforced this - he (and presumably his team) cover an enormous amount of technological ground very quickly, in pursuit of a technology that can deliver their vision.

I do wonder why this model isn't used more often in the rest of business. Is it that its hard to find a deeply talented people who are passionate and have a shared vision about a new type of derivative or toothpaste?

This is to an extent how large research labs are structured. The best PIs have a range of projects running, and graduate students will gravitate to those that they are interested in.

I suspect the model isn't more popular because it can be damaging managers' egos to watch talent flock away from their projects. These sorts of markets are damaging to monopolies on power.

In most organizations "managers" and "executives" have a role beyond the technical execution of a project. They must sell work to clients, find funding, structure financing, decide on overall strategy, etc. These are often different skills than pure execution. You can have a someone who is great at selling work to clients but bad at mentoring junior staff, for example. Or someone smart and great at bug picture thinking, but disorganized and hard to work with. This makes it hard for some organizations to give up on the "top-down" allocation of resources. But yeah, sometimes it works, and I do hope more firms will at least try it.

A few questions:
1. Who makes sure that all these people get paid on time? How about the office rent?
2. If somebody is off on his/her own, not interacting with any of the teams, who reviews that person's productivity?
3. Who's tracking revenue?

While the employees are self organizing when it comes to product creation, the company does have owners. They do not try to direct what people work on, but ultimately it is their responsibility to get the company working.

Most other iasues are solved by social contract: You can't sit around on a side project that nobody believis in without serious risk. People can get fired still. Models like this can only work with pretty extreme company cultures that have a tough view on freeloaders and disrupting agents.

Just like socialism does not scale, the Valve model does not scale either. But if you do not need 70000 individuals in your community, a model like this works just fine.

Following on from Mark-s point, does anyone here remember the Volvo experiment in doing away with production lines for some years at some part of their plant in ....perhaps Gotenberg? in the seventies and eighties? There was a huge rise in employee satisfaction, small groups of workers built engines together, work-in-progress was wheeled around between worker groups at their initiative. Absenteeism and sick leave fell, all sorts of good things happened....

....but, as I recall the sad story, Volvo stopped the experiment after some years. It wasn-t profitable.

Can anyone remember more details?

Taking into account that the goal of Valve is to attract talent in the CS and other engineering areas, the culture they are describing here resonates a lot with this type of creative, curious and innovative people. My boyfriend is a PhD student in Control and Simulation and his dream job is at Valve. So this is a very effective PR strategy. A bit like Google's spot from a couple of years ago "You're brilliant? We're hiring"

From a gamer standpoint I put Valve on a higher level than other companies. Creating Half Life, and specially Portal, is not creating just a couple of good games. These are awesome experiences, that can only be created when you have an incredible pool of talent and you give them freedom to create

"A few good games" is an understatement. Valve has released dozens of games in over a decade, and very, very few of them weren't met with near-universal praise. They also run Steam, which I'd say is the best virtual goods delivery system that currently exists on Earth. They're more profitable per-employee than Google or Apple.

Their organizational structure works because they hire the best people in the industry, though it may be adaptable to firms that don't have that luxury.

And exactly how is Valve going to help us out of the economic morass that the US finds itself in right now? While these stories are intriguing from a micro perspective, they do little to address the real problems. At some point in time, all the gamers have to go out and find real jobs.

Small contributions relative to our national problems, but they are pioneering digital distribution services in international markets while combatting digital piracy by offering worthy alternatives.

These are real jobs

I was obviously referring to the people who play the games. While Valve may be doing some very interesting things, their overall contribution to the economy is slight compared to a large manufacturing company.

Still, small firms employ a lot of people. Their model seems to be working for them. Rinse and repeat.

It is worth noting that the owner of Valve can credibly joke about acquiring Electronic Arts (a $4B company). While Valve itself is a small company, Steam effectively routes payments to many thousands of game development companies. Those are real jobs (compare to the ad business - which piggy backs off TV consumption).

It's really hard to evaluate Valve's model without knowing a lot more about how it works in practice (even the author of the post, being relatively new at Valve, is a bit fuzzy on some aspects). I suspect that there is rather more bossing and managing happening behind the scenes than they acknowledge. As Mark suggests upthread, it's unlikely that anyone will "gravitate" or "self-organize" into payroll processing or accounts payable (even if you outsource those functions, someone has to oversee the arrangement with the vendor). There is probably some sort of incentive scheme in place to help ensure that the developers' interests align with what is likely to sell in the outside world. At the very least, somebody has to make some decisions about how many people to hire and who those people should be. This last point is critical because if you aren't going to tell people what to do, then what kind of output you get depends critically on who you hire, more so than in a traditionally managed company. A company where the bosses tell employees what to work on might be able to hire smart and motivated people from related, or even unrelated, fields and get them to produce. If Valve, on the other hand, had hired a bunch of space scientists and turned them loose, then instead of a bunch of video games, they probably would have gotten a Mars rover, which, although cool, probably wouldn't have been much of a money maker.

It would be interesting to learn more about exactly how Valve's (non-)management works, as it would shed some light on what management functions can't be dispensed with, even under an optimal situation of the best people in the industry working on a bunch of cool projects. In the end, however, I think the strategy might be limited to businesses where the developers have a personal interest in the product. It's hard to imagine the developers at TurboTax getting quite as personally committed to their product.

RPL, you can get some start by reading our new employee handbook:


It's a good overview, although a bit fluffy. Feel free to email us if you have specific questions, though.


That's the new employee handbook from Valve, which is a very enlightening and interesting read.

I'd also like to note that Valve created and maintains Steam, which is a legitimate social revolution in the video game world. It's on the short list for the most influential pieces of software in the video gaming world ever.

Also here's a first hand report from an employee: http://blogs.valvesoftware.com/abrash/valve-how-i-got-here-what-its-like-and-what-im-doing-2/

Valve as in the Valve persistently mocked for not being able to count to three? (Portal, Portal 2, Half-Life, Half-Life 2, HL2 Episode One, HL2 Episode Two....Left for Dead, Left for Dead 2....)

I'm not saying they're not great engineers and they are incredibly data driven regarding their steam platform, essentially the first app store. It remains to be seen if they will be competitive when Microsoft releases it's app store as the major publihers all seem inclined to publish games through their own online distribution platform, eg Origin.

One of the points made in one of the articles is that copying isn't what's really valued in the software market now. Since there are huge network effects, the first-mover advantage is very large. Steam already has the advantage of a massive amount of users and games, but an even bigger advantage is how many people use it as a social platform as well. Groups by game and interest, friends lists, private servers, etc. are all connected through Steam. People might buy individual games from other stores, but if they want to migrate away entirely, they have to give all of that up along with their existing library of games. There isn't just product lock-in, but a social one as well (see Google+).

Steam is more than an app store: It is also a rather sturdy social network for games. Origin and such cannot win precisely because they segregate and give major advantages to their products: In a natural monopoly like this, such barriers kill a product.

There is also how Steam seems to undetstand sales better than anyonr. Prices keep dropping until people actually buy the games. Compare it to XBox Live, where games remain overpriced for years. Dowload from MS for 40, or purchase from Steam at 5

Team fortress two was a month from going gold in 1999 when they pushed it back to, eventually, 2007. I wouldn't be surprised to see a part three of something come in the next few years.

And I should say, valve sales are the only kind of sale that grabs me, which is annoying, as I largely stopped gaming five or six years ago.

At least it's only about $15 a year wasted.

Finishing something like the Half-Life series seems to involve so much pressure Valve has trouble focusing it's efforts. Also, it's an older series without the draw of new game mechanics to pull in the interest of developers. I wonder to what extent this structure described makes getting the needed manpower and talent impossible.

Can we stop talking about HL? It's the worst of the Orange Box games. I'd be about 100000 times happier to see a third Portal or L4D. HL is just...an old, boring game.

HL is not a orange box game. HL2 is. HL was revolutionary when it came out. There was nothing like it back then.

HL2 is still a very solid shooter.

This article should have the heading "Job Titles to Ponder: Valve's Chief Economist." Am I the only one that struck as nuts/cool?

What's even cooler is why they hired him. What he does is analyze how trade works between the virtual economies of TF2 and Dota 2.

If it weren't for the artificial limits of copyright there would be no practical reason to have a larger structure like Valve to contain all these separate projects under one roof. So it seems to me that the whole innovation is just a way of holding off more fundamental change for now with a model that's more efficient than the old firm but not as efficient as the future freelance model. Both piracy and decreasing overhead costs make the freelance model inevitable though.

If you think Valve is just a 'few good games' you completely don't understand them or the business they are in. They are the leader in digital distribution for video games. Video game market is already bigger than movies or music and will surpass books and dvds soon. They are the same leagues as itunes and amazon. In 5 years, buying video games on optical discs will seem quaint.

They don't even really make video games anymore, and when they do let their interns do a summer project, it ends up winning Game of the Year (Portal).

Comments for this post are closed