Hacker News new | past | comments | ask | show | jobs | submit login

Valve has an amazing culture, but I'm not sure it would scale to a 50,000 person company. Valve seems to operate as a tribe and there's a limit on how huge those can get.

Microsoft seems to be run like a prison camp.




> Valve has an amazing culture, but I'm not sure it would scale to a 50,000 person company

This reminds me of just a few years ago when I was arguing with someone about this, when 37signals was in their infancy and were kinda trucking along with the same 'bossless' philosophy. I was trying to show that it could be successful and the person hacked me off at the knees by saying "come talk to me when they get to 100 employees" ... well Github and Valve seems to be doing just fine with that philosophy ... but now the goalposts have suddenly shifted.

I wonder if they get to 50,000 people (how many companies even have this many people anyway?) if somebody isn't going to pop up and go "50k people? pffffft ... come get me when they have 200,000 employees!"


There's a number of companies with that many, particularly if you do manufacturing in addition to tech.

Intel - 100,100

Dell - 110,000

HP - 349,600

Google - 54,604

Microsoft - 90,000

Apple - 60,400 (I don't know what fraction is retail sales)

Numbers from Wikipedia, which I assume are from the annual reports these companies put out.


IBM - 433,362


I think if you're very picky about having just the right people at the "bottom", and just the right people/person at the "top" (because even in Valve's supposedly flat hierarchy there's still someone on top: Gabe) then it can go up quite aways. But yes, it's probably inevitable that it eventually fails and devolves into department, fiefdoms, politics, etc. You can push it out but can't put if off forever. But you know what? That's not necessarily a bad thing. There are plenty of companies that can be viable and successful with 5-100 employees. Not everything has to be a megacorp.


According to http://en.wikipedia.org/wiki/Dunbars_number Valve should already be two tribes.

You have probably heard about Gore-Tex solution—build plants with 150 workplaces.



> I'm not sure it would scale to a 50,000 person company.

Split the company into a bunch of mini-companies, don't get beyond "the limit".

Microsoft used to be a pretty flat organization back in the days, depth of the management layer is one of the things MS employees mocked at IBM in the 80s and 90s. It was already a pretty big company back then.

BTW, MS's headcount is closer to 90k than 50k these days.


It's a hell of a lot bigger than that when you include vendors...


The vendors are intentionally not a part of MS, because they're lower margin operations.


Then you need a special division to coordinate the anticompetitive behavior.


http://en.wikipedia.org/wiki/Maverick_(book)

"Among many 'radical' policies, Semler let his employees set their own hours, design their workplace, choose their own IT, share all information and have no secrets. Every six months bosses are evaluated by their subordinates and the results are posted. Semco has a policy of complete internal financial openness, even teaching factory workers how to read accounts so they can understand the company's books. Salaries are public information unless the employee requests they not be published. In addition, all employees can set their own salary."

As of 2003, they had 3000 employees and was growing at 40% a year. No idea how they're doing now.


But what's the advantage of a company with 50k employees? Why not have several smaller companies that collaborate?

Microsoft could easily be broken into several smaller companies.


I've often thought this was a good idea. The "interfaces" between the smaller companies would be far more strict and simple then interfaces between departments in a single large company. It's like the difference between buying a hamburger from a fast food chain which is extremely easy and asking a co-worker to sell you a burger which gets complicated very quickly.

Overhead in terms of office space etc would increase. However the simplicity of the necessary interactions between the smaller companies could turn the entire arrangement into a net gain for the parent company.


Yeah, they are already bouncing around above their Dunbar number (http://en.wikipedia.org/wiki/Dunbars_number). Methinks changes are afoot (either no longer growing or no longer operating the same way).


Valve seems to operate as a tribe and there's a limit on how huge those can get.

Genghis Khan would disagree: http://en.wikipedia.org/wiki/Mongol_Empire


Like Gabe he had a rather large collection of weapons. The difference is he wouldn't hesitate to use them on those that disagree with his leadership.


Even ignoring that there just are going to be a lot more 300 person companies than 50,000 person companies, I'm not sure we actually need that many 50,000 person companies.

I personally think we're better off if businesses are large enough to operate reasonably efficiently at a single or small set of tasks/products, and no bigger. For one, you begin having weird things happen like a product being canceled or retired not because it was unprofitable to make, but because the company thought there were more profitable products they could be making. (Anyone remember HP's decision to stop making desktop computers?)

For another, you end up with a single point of failure with respect to management. With a hundred companies with a hundred CEOs, some of them of them will prosper and some of them will fail. If those hundred companies merge into a single company with a single CEO, the consequences of terrible management are now a hundred times worse.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: