I'm curious to hear opinions of the "up or out" philosophy at startups. I can conceive of both pros and cons (mostly cons) but a forced philosophy sounds flawed.
I'm not sure it would work at most startups. Firms that use "up or out" policies offer a few things to potential hires:
1) A high salary (to make up for potentially short tenure)
2) Potential for equity shares in profits and control (even though McKinsey is technically a corporation, its run as a partnership with partners being the shareholders)
3) Training backed by a brand they can take to their next job
When someone takes an offer at McKinsey or Cravath or Goldman, they know that they'll be compensated at the top of their field, they'll get a resume line they can take to their next job when they get "or out-ed", and that if they prove to be a super-star they'll get a real stake in the company.
None of these things are true for a startup. That is not to say that startups don't offer their own benefits (I think working at a startup is the best training you can get as a software engineer, but it stings that future employers don't really put much stock in it unless you get lucky and the startup turns into a brand name), but in adopting an "up or out" model you adopt all of the downsides of that model without offering any of the upsides.
Up or out in a growing organization is a lot harder.
It also works best in a professional services firm where your intake is mainly undifferentiated college/professional school graduates, with a clear career progression.
Basically, neither of these apply to (successful) startups. Startups are more likely to rapidly promote people when things are going well, and then to turn into "evaporative cooling hell" once things go badly -- all the good people jump ship to better opportunities, leaving only those who can't or won't leave, accelerating the decline.
In a startup you're much more likely to see roles growing in responsibility or scale faster than some people might be able to adapt, so someone who is hired as head of development in a 3 person team might not be able to be VP Engineering if the company grows to 50-100 engineers. At that point, you have another problem to try to solve -- do you just replace him with another person, or keep him in a smaller role and hire someone above, etc.
A variation is used in many financial services sales teams, where the bottom 10-20% each year are fired, or if sales don't increase. Similarly was/is used at IBM to remove the bottom 5-10% of under-performing staff each year.
Zuckerburg talked about this at a startup school event years ago. No clue if it's actually implemented.
Up or out mirrors natural social behavior- the highest status individuals tend to evaporate to become lower status people in higher status groups, while the lowest status people tend to drop off to become higher status people in lower status groups.
