I'm super sick of these articles patting Valve on the back for its super-innovative business structure.
The fact is, you can afford to have a goofy office like this when you've already achieved a comfortable level of success. Valve has been free and clear ever since Steam became the go-to place for digital game distribution, and Github is pretty comfortable in its place as the top source code host.
Emphasizing these companies kooky structure conflates correlation and causation. It's not that they're profitable or interesting because of their management structure: instead, they can afford to play around because they were early movers in a very profitable space.
I'm curious how you think it's possible for a structure like this to emerge after some level of success? You think that any company sets up a traditional management structure, accomplishes success and then fires all the managers and tells everyone to work differently? Or do you think it might be more likely that thinking differently early on about how to treat employees and how to focus creative energy somehow helps to achieve that success?
GitHub has always run this way, since it was 4 people. When we tell people how we operate, they have always responded that it will never scale - people have told us that since we were 4. We're now at nearly 100 employees and it's still working great. Valve is 300 and Gore has several thousand, so we're pretty sure we can keep going this way. We are slightly different than Valve, but their handbook really resonated with the way we do things. The main difference between companies like Valve or GitHub and many other companies is that we don't look at other places and ask ourselves how we can copy their structures, how we can cargo cult their success, instead we look at our problems and ask ourselves how we can address them in the best way possible. It appears that Valve has done the same thing and we've come to some similar conclusions.
I think this is a big reason why we've been successful - that we ask ourselves this when approaching product too. You say we're sitting pretty at the top, but when we started there were tons of source hosts. Valve was just as bad - starting in another industry that was totally saturated. The reason we were able to break through I believe was largely due to the fact that we thought about problem solving differently than the industry leaders that we started up against.
The real point you should take from the article is not that Gore or Valve or GitHub are lucky, but that we approach problem solving in a very different way and that might be an interesting thing to take a look at. If all you read in an article like this is that we're 'goofy' or 'kooky' then you're missing everything that's important and it's a waste of time for you to read the article at all. In fact, if that's what you take and then you try to copy us you will fail horribly in what you do, because you're blindly copying the least important of the many symptoms of this approach.
Approach problems from first principles. Figure out what the best possible experience would be for the person using your product (not just the person buying it). Make that experience a reality. Don't copy anyone. Do that same thing internally. Now what does your company and your product look like? That should be what you take from this article.
> The real point you should take from the article is not that Gore or Valve or GitHub are lucky, but that we approach problem solving in a very different way and that might be an interesting thing to take a look at. If all you read in an article like this is that we're 'goofy' or 'kooky' then you're missing everything that's important and it's a waste of time for you to read the article at all. In fact, if that's what you take and then you try to copy us you will fail horribly in what you do, because you're blindly copying the least important of the many symptoms of this approach.
Well said, sir. Reminded me of this article on cargo cult management [1]
I don't know if it's possible to figure out which is the chicken and which is the egg in these cases.
Semco, a Brazilian manufacturing company, has been using a "boss-less" style of management since the 1980s. From what I understand they were not exceptionally successful before they implemented this management style.
I think it's more complicated then that. The fact is that there are multiple ways to do things. These different ways have different tradeoffs.
You know, if you are just starting to found a company with 2-3 friends, chances are you jump into eachothers' roles as needed. There may be an official hierarchy but chances are pretty good that this hierarchy exists mostly on paper. As you hire employees, typically you want to maintain control and this is where management structures come in.
I think that there are all sorts of possible management structures that could work and these have different advantages and disadvantages.
I forwarded Valve's employee handbook to an economist friend a few months ago, and he suggested that whoever set up the company was actively experimenting with public choice theory, and that from that perspective the good and bad things mentioned in the handbook were not that difficult to suss out from the structure.
Many popular web-centric software companies operate this way, and it's basically how open source software development operates.
Anyone with even an ounce of management skill and education knows that when you have skilled people working for you, you get out the way and let them do their thing. This isn't exactly news.
That doesn't sound like the Fog Creek I've read about, but maybe I'm just projecting from their well-publicized salary/seniority ladder? The dribs and drabs I've read about Joel's companies suggest that they're well-run classically-organized teams with a heavy dose of _Peopleware_ thinking (which is a great book).
"Anyone with even an ounce of management skill and education knows that when you have skilled people working for you, you get out the way and let them do their thing. This isn't exactly news."
I don't know about the internal organization of the teams
The whole notion that everyone is responsible for product is pretty mainstream at this point, increasingly so since the whole "lean startup" concept gained traction.
20% time is not nearly the same thing as "we hire you, you figure out a team to go work on, leave those teams at any reasonable point, and feel free to start an entirely new full-time project as long as you can get buy-in for it, for there are no managers or bosses to tell you what to do".
Anyone with even an ounce of critical skill who has educated themselves on the literature of the firm knows that a hierarchical management structure does not suddenly become a flat structure simply because management has decided to "get out of the way." Not to mention the absence of literature on what it means exactly for management to "get out of the way."
Valve is pretty unique in that it was founded by Microsoft millionaires. They had their own personal fortunes to play around with (much like Curt Shilling, except they also had backgrounds in shipping very successful software) while they figured things out.
However, while Steam is certainly a very big deal now I think it is silly not to mention Half-Life as a very important part of Valve's early success. Half-Life was very successful and the fact that it, its sequels and the Source engine begat so many related franchises like Counter-Strike is a big part of the reason Steam was able to leverage a huge built-in audience and succeed in its early days while so many would-be-competitors failed.
One thing about Microsoft though is that they leverage things like email lists internally to provide many opportunities for participation one might think of as primarily characteristic of these flat organizational models. For example, a lot of the work that was being done on competing with Linux when I was there was taking place on the basis of this sort of flat model.
The point I am making is that these wouldn't have occurred to the Valve founders in a vacuum. They were built on top of what they already knew worked. So it's not a matter of playing around but rather trying to copy what you think works well in your past experience and ditching the rest.
You might not have a manager , but you always have a Boss to a certain extent. There is always someone who is named as the companies MD/CEO , someone who has the authority to legally hire you / fire you. So therefor their opinion will always be more important.
Also the problem with having the team evaluate someones performance is that teams are vulnerable to cliqueness and groupthink and you may therfor often end up with decisions that are based more on popularity than anything else.
In theory at least a good manager should be able to step back from the inter-employee politics and make judgements with some degree of perspective.
There's always someone more invested. Management can pay lip service to flat hierarchies and all that, but at the end of the day structure emerges more or less by consensus, the same way everybody realizes the guy who always farts or something. The common-if-not-genetic "desire to be led" causes people to produce hierarchies for themselves, to produce their leaders. This isn't robotic, but it comes naturally to a lot of people, even if only to confer authority upon that person.
But the difference is that in a structure like Valves the inertia of those structures is much lower. So institutions don't stagnate into structures concerned with preserving their internal structure. Some unit starts to become irrelevant exactly when the people who are the top producers turn their attention elsewhere. The dead weight doesn't exert power to keep itself afloat.
Here's the question I don't get - how do you fill "support" roles? Like how do admins fit into this structure? Or HR? Are there really people, who given a job at a video game company and no defined role, would choose to set up meetings and answer phones?
Or similarly, how do you determine that some devs build awesome video games while others maintain a corporate intranet or do some much more mundane task?
Clearly, it works. And in a theoretical sense, I can see why when everyone has somewhat similar skillsets. Can someone fill in the blanks for me when there are specialists and support people involved?
Actually, as I approach retirement age, I'd love a basic wage job working with clever people and smoothing the edges a little; thinking ahead and setting stuff up.
Note to startups: Your local 50+ and 60+ people may like the involvement. You could get a lot of experience/support for not much money depending on how local pensions work.
I imagine you'd advertise for a support specialist in the traditional sense, but once hired, they wouldn't have a job title. Their skills would be in customer support though so that's what they would naturally do. If they tried to work on development, but had no skills, then the group would remove them from the project. They'd then have to go back to support or find something else they can work on with their skill set.
OK, so who cleans the toilets and empties the garbage? Contractors, presumably -- easier to get non-employees to do this than to incorporate menial staff into your oh-so-flat structure. Reminds me of the way ancient Rome ensured that all citizens were equal by creating a huge subclass of non-citizens.
What's my point? Just that it's easy to over-fetishize a non-heirarchical structure, but it only works (a) under certain very specialized circumstances and (b) with a little bit of smoke and mirrors.
Funny enough, we're a small company here (8 employees at this location), and I simply told them we couldn't afford a cleaning crew [a], that if they wanted to live nearly a third of their life in a dirty, disorganized environment, that was their choice. The bathrooms stay clean, the kitchen has the dishes washed every day, and every Friday they spend an hour scrubbing the place. No one is forced to do it, yet everyone takes pride in their workplace and wouldn't stand for a customer or someone else to judge them.
(Mind you, the engineers who are used to having someone clean up after them have taken longer to pitch in - but they all eventually join the party. =)
It's worth noting, that we're not "flat" as far as structure goes, there's an operations manager as a foreman, and there are two of us partners at the top who set clear directives. However, we let anyone do anything they find interesting - the key is that they must -do- and not -talk-. If someone has an idea, they can get it done as long as they meaningfully contribute to it and shepherd the process - produce mechanical designs, circuit design, software, or simply make new jigs to make an assembly process more efficient. If they can show me they tried to make something and failed because tooling was inadequate, I buy a new machine for them.
But, yes, I agree it does only work under certain specialized circumstances: it works best at small companies that don't have a lot of interaction with enterprises. =) Trying to sell or work for an enterprise largely requires you to behave like an enterprise as well.
Smoke and mirrors? At the end of the day, even in the most lauded cases, someone has to write the check or pay the consequences. They always have a veto.
[a] - Originally, I had offered anyone the same rate as a cleaning crew to take on the responsibility. No one bit.
There's no argument here - just sarcasm and unsupported claims. You invoke janitors to imply that non-hierarchical organization can't possibly work. But you don't know that; no one does. People have barely begun to experiment with these forms. How much imagination has gone into figuring out ways to accommodate menial labor so far? It takes time.
If we can figure out how to build software without bosses, I'm confident we can figure out how to clean toilets.
Indeed. One possibility (though I'm sure there are many others) is to divide the work that no one wants to do. There's even a handy word for that: chores.
Not sure I get you. Valve doesn't have to be either of those things.
Work democracy and self-organization are old ideas. They go back at least to late 19th century anarchists. At that time they were so far out that only utopians and revolutionaries took them seriously. That we now plausibly discuss them is already notable in the history of human hierarchies.
there are all sorts of ways that could be done/ For example you could contract with a company to provide janitorial services. You could also have a sub-organization that did things like janitorial services.
The areas I would worry about more would be areas where you have direct customer deliverables at stake, and where you need a team of people to ensure this. What happens when people switch teams? How quickly can you get a replacement? The article said that hiring often takes a while....
To me, setups like this -- along with employee owned models like the one at the Mondragon Group -- are the future. It's always struck me as strange that we value democracy so much in our government, but not in our businesses.
To help decide pay, employees rank their peers—but not themselves—voting on who they think creates the most value.
Considering how larger society often 'ranks' and rewards people (e.g. celebrities, local politicians), I wonder if these schemes could be a breeding ground for contempt or cronyism.. or, more likely, people who are happy to work at such companies rank compensation as a high priority and mismatches aren't that important?
The idea of worker autonomy is implied a lot here. Should that go hand in hand with workers having a direct influence and role in determining their compensation with whoever holds the pursestrings?
While I don't disagree with your thesis, politicians and celebrities are usually unknown, or at least only known through small appearances in the media. You work with your colleagues so I would expect a fairly different dynamic when it comes to making these judgements.
I came here to comment on that sentence. I think it's a great idea, at least compared to the alternative of evaluating yourself. That's how it's done where I work. I mean talk about the LEAST objective person you could ask.
I applaud Valve's visionary business organization. As a developer I'm sick of having to work under liars and frauds who have misrepresented their technical ability and knowledge to get into an 'engineering management position' they are not fit for. The costs of job changes is very high to me, and if there were more companies like valve, tech would be a much better place to work.
This is fairly common (liars and frauds...), though there are some great managers out there: they're just as rare as great developers, and worth their weight in gold when you find them.
In the US, you have protections if you have a union contract. Otherwise, you pretty much need to be considered a protected class (varies by state), and your sacking was related to your class status.
Forget about the legal risks; how does that even work? If I don't think my coworker is doing a good job, am I supposed to secretly whip up enough votes to have them fired?
In my experience, it generally becomes clear to everyone if someone is really not working out, and it's already been the subject of hushed conversation.
If you need to "whip up votes", this will say more about you than the person you think isn't doing a good job.
Borderline cases are trickier, but they are for managers too.
The only time that a "single boss" would know better than the team would be if the single boss was as knowledgeable about the development process as the team was, and knew that the team was, for some reason, cooking up a conspiracy to fire a certain employee who didn't deserve it. Probably pretty rare, especially compared to the number of people who get let go because they just didn't click with their supervisor.
And of course there always is a single boss even if this fact is hidden; in the case of Valve it's presumably gaben, who owns the majority of the company and must presumably be the final arbiter of everything that happens.
I can only assume that the actual firing process involves a lot of hushed talking followed by an approach to gaben followed by a discussion followed by a "sorry, bye".
One could in fact suggest that the flat management structure only works when there is exactly one clear owner of the company. Everyone can pretend that everyone gets a say, when it's clear to everyone who really has the final say.
I get the sarcasm, but Gaben said that is exactly why. Nobody came forward with some great idea to make the game really good, so they never started it, and instead worked on things like Portal 2.
So isn't this a fairly huge flaw showing up in this management style, if the cats can't even herd themselves well enough to produce a long-promised sequel to the company's biggest franchise?
It doesn't matter that much, of course, if Valve never finishes the Half Life series; they'll no doubt continue to make other projects which rake in the bucks. But for most companies you can't just abandon a cash-cow project like that. What would happen to Toyota if all the engineers decided they really wanted to work on sports cars and nobody could be bothered to work on the next iteration of the Camry? What happens if all the employees for the water company decide to design a new dam and nobody stays on the maintenance team to stop the existing dam from bursting?
It's inspiring that a company has the courage to let a franchise rest when it has artistically run its course (you need only to look to Hollywood for infinite counter-examples to this -- see http://www.pastemagazine.com/blogs/lists/2009/10/-worst-sequ... ). The fact that Valve can still be successful, even without pressuring their employees to develop more "sure thing" games (read: sequels) is pretty awesome if you ask me (I say this as a long-time HL fan).
But as long as Valve is making enough money, the employees can work on whatever they want and Gabe won't care.
This is why many of the "unsexy" parts of Valve are in utter disarray. The Steam chat network and steamcommunity.com goes down at least once a week. And their security isn't very good. They lost credit card information earlier this year and they think encrypting your password with Javascript RSA ontop of SSL is going to help.
So isn't this a fairly huge flaw showing up in this management style, if the cats can't even herd themselves well enough to produce a long-promised sequel to the company's biggest franchise?
I'm struggling to see why producing a chain a hit games and making the company millions and millions of $ in the process instead of working on a sequel that the developers didn't have a strong vision for is a flaw personally.
Sounds like the system is working. When and if some of the developers stand up and say, "you know this is how HL3 should be" and convince the others to follow them, then we'll get HL3.
This is a huge flaw, but it is even more common to have the opposite huge flaw, that is, to only do the safe cash-cow things, which has killed many big companies. I believe that this is the core problem of any business with hit-miss dynamics, huge gains for single hits, (small) losses for every miss. Once you get a hit you can often monetize it for quite a while, like Disney does with all the stuff they sell related to the movies.
The dilemma in other words is: If you don't have a hit, it is hard to create one. Once you have a hit, it is easy to monetize, but will not last forever. In other words you need to both milk the cash-cow and produce new hits. It seems Valve is doing the latter excellently, but maybe they should build a separate organization for doing the former, since their inner workings is not optimal for that. Milking the cash-cow is often much less demanding, and more tolerant to mismanagement, so it should be a prime candidate for outsourcing.
I just switched from a horrible boss/manager to a not that small private software company where the developers are very respected and don't have any managers.
And... while it is a lot better than a bad boss, no management is not ideal in my experience. Ideally what you want is a good manager.
I'd like to see someone look at the mechanics of how boring, menial, repetitive grunt work is performed under this model. No doubt Valve/Gore/etc has a lot of that to do, just like any other business. And obviously it gets done. But I'm interested in how that looks up close. Does this kind of work tend to find its way to those in the company who don't have a problem rolling up their sleeves and gutting it out? And if so, do they ever end up resenting it?
In my current (traditional) workplace, you're responsible for your own grunt work, there's no pushing it off on someone else. That is enforced by management, however, and if that traditional management disappeared, I imagine the distribution of grunt work is one of the first things that would change.
This is the only true way a boss-less organization can work. And I assume you mean voting equity. Vesting requirements would be OK, but stuff like options or Google's non-voting shares wouldn't be.
The fact is, you can afford to have a goofy office like this when you've already achieved a comfortable level of success. Valve has been free and clear ever since Steam became the go-to place for digital game distribution, and Github is pretty comfortable in its place as the top source code host.
Emphasizing these companies kooky structure conflates correlation and causation. It's not that they're profitable or interesting because of their management structure: instead, they can afford to play around because they were early movers in a very profitable space.