I'm a major supporter of open allocation (I coined the term) as used by Valve and Github, but I'm skeptical of "flat" organizations. It's not that I dislike the idea. I don't think it's really possible.
One of the reasons (at least, in theory) why legible middle management is a good thing is that they don't have to compete with the managed for visibility and work. When you have emergent pseudo-managers (or managerial favorites) who are still technically the same rank but far more influential, they use their influence in competition with the other workers and do a lot of damage: it's not a fair competition, because they take the sexy work and delegate the crap. Most Valley startups tell every engineer that he's reporting to the CTO... but then they get there, and there are 57 people nominally reporting to that CTO, which means that they actually report to the people who have the CTO's ear (and that changes over time).
I like the idea of a flat organization, but I think what you actually need is constitutional management. You need a strong set of principles ("employees are trusted to allocate their time to anything that provides business value", i.e. open allocation) and then you need official and legible managers who are accountable for enforcing them. It's better to know who is in power, so you can hold them accountable and remove them from power if they fail, than have a "flat organization" where power disparities still exist but are subtle. You need checks and balances, you need laws and you need cops but the cops shouldn't be making laws on a whim.
In other words, the optimal point is somewhere between the demarchy of the "flat organization" and the executive dictatorship of the traditional corporation.
Flat organizations, in general, make it harder to define accountability. Look at the case of Ellen Pao, for example. Venture capital partnerships are flatter than traditional corporate hierarchies and rarely have HR departments with any power over the partner level, but are still afflicted by infighting and retaliation. Pao's case is one where (a) she was clearly wronged, but (b) her legal case appears pretty weak (or, at least, lives in undefined territory). It's obvious that a creepy ex-boyfriend (or, more accurately, ex-few-night-stand) took vindictive and inappropriate steps to ruin the internal reputation of a talented female venture capitalist, and that this prevented her from having a successful career at KP. It's unclear whether the firm or its upper management is legally responsible. He was engaging in retaliation, but upper management wasn't; they were denying her promotion based on (faulty, maliciously provided) information about her performance, not her gender. It's unclear who is responsible; clearly, the man in question (since then, fired) is, but he probably doesn't have the money to make her whole. So should Kleiner itself be held responsible? That's really hard to answer. That's a problem with flat organizations in general; they still have politics and they still have good people ending up with bad reputations and vice versa, but the lines responsibility and accountability are less clearly drawn.
As a side note, I don't generally like anti-discrimination laws. I am proposing to allow EEOC or anti-trust to impose restrictions on specific companies on discrimination if necessary instead.
One of the reasons (at least, in theory) why legible middle management is a good thing is that they don't have to compete with the managed for visibility and work. When you have emergent pseudo-managers (or managerial favorites) who are still technically the same rank but far more influential, they use their influence in competition with the other workers and do a lot of damage: it's not a fair competition, because they take the sexy work and delegate the crap. Most Valley startups tell every engineer that he's reporting to the CTO... but then they get there, and there are 57 people nominally reporting to that CTO, which means that they actually report to the people who have the CTO's ear (and that changes over time).
I like the idea of a flat organization, but I think what you actually need is constitutional management. You need a strong set of principles ("employees are trusted to allocate their time to anything that provides business value", i.e. open allocation) and then you need official and legible managers who are accountable for enforcing them. It's better to know who is in power, so you can hold them accountable and remove them from power if they fail, than have a "flat organization" where power disparities still exist but are subtle. You need checks and balances, you need laws and you need cops but the cops shouldn't be making laws on a whim.
In other words, the optimal point is somewhere between the demarchy of the "flat organization" and the executive dictatorship of the traditional corporation.
Flat organizations, in general, make it harder to define accountability. Look at the case of Ellen Pao, for example. Venture capital partnerships are flatter than traditional corporate hierarchies and rarely have HR departments with any power over the partner level, but are still afflicted by infighting and retaliation. Pao's case is one where (a) she was clearly wronged, but (b) her legal case appears pretty weak (or, at least, lives in undefined territory). It's obvious that a creepy ex-boyfriend (or, more accurately, ex-few-night-stand) took vindictive and inappropriate steps to ruin the internal reputation of a talented female venture capitalist, and that this prevented her from having a successful career at KP. It's unclear whether the firm or its upper management is legally responsible. He was engaging in retaliation, but upper management wasn't; they were denying her promotion based on (faulty, maliciously provided) information about her performance, not her gender. It's unclear who is responsible; clearly, the man in question (since then, fired) is, but he probably doesn't have the money to make her whole. So should Kleiner itself be held responsible? That's really hard to answer. That's a problem with flat organizations in general; they still have politics and they still have good people ending up with bad reputations and vice versa, but the lines responsibility and accountability are less clearly drawn.