Recognizing the Breaking Points of Management Structure

Recognizing the Breaking Points of Management Structure(tomtunguz.com)

27 points by dvdgrdll 10 years ago | 8 comments

fallingfrog 10 years ago |

I'd be willing to wager that tighter or more hierarchical management structures are correlated with lower levels of engagement for the employees at the bottom. In other words- there are two ways to get people to cooperate: to use secrecy and threats as a lever, or to have a common myth or vision that everyone believes in. Increasing secrecy and mistrust might be a signal that the common vision has deteriorated. Of course it depends on the size of the organization too. I'd like to see a study confirming or disproving that idea.

andygcook 10 years ago |

"When startups approach those employee counts, communication within the company breaks down and the startup can’t effectively coordinate its people. The telltale signs include confusion in the organization, uncoordinated efforts across teams and frustrated employees."

Can anyone who has quickly scaled headcount at a startup elaborate on some of the challenges in scaling communication + coordination and what you did to solve the issues?

bluejekyll 10 years ago |

I don't even know what to say about this article, except that it perpetuates this ridiculous idea that there are laborers at the bottom and people above those laborers who know how to communicate and effectively distribute knowledge across all other groups.

If people were computers, then what's being described is a tree of systems, and each node in the tree has full connection to its siblings and its children. But anyone who's designed large scale networks knows that this is horrible for data, because it doesn't allow for failure, and creates bottlenecks. This is why spine leaf architecture is so important.

But people are not computers. Communication is way more lossy than that. What this article fails to point out is that you actually have multiple decision trees in organizations, the architects and technical leads in a software group, vs the people managers. And combining those two roles creates horrible inefficiencies in descision making. Similarly, empowering people closest to the problem to make decisions means that the organization is much more nimble and increases agility.

I don't know what this article's point is, because it does help understand what effective leadership is.

yodon 10 years ago | |

The article is talking about the numerical scaling laws that you understand intimately from your day job but that most people who don't think about node-based communication structures on a daily basis are surprised by. Specifically, the article is pointing out that the cognitive loading placed on managers grows much faster than linearly in the number of reports.

Have you thought about how to translate spine-leaf into an organizational design and/or what sort of organizations it might be optimal for?

bluejekyll 10 years ago | | |

I'd like to think that modern network design actually could inform organizations, especially large ones, of better methods of communication.

The goal IMO is to increase collaboration between as many groups as possible. Where scrum teams, or similar, are autonomous in making decisions and the scrum masters (or managers) are responsible for inter-team communication and collaboration, but that shouldn't be the only channel. There should be other people from the scrum team responsible for interdisciplinary collaboration.

Basically, as an organization grows, I believe it becomes more important to have redundant communication channels into the scrum team. Like network design, this would be akin to have multiple uplinks from the ToR leading the the spine.

yodon 10 years ago |

Has anyone found a derivation for the formula in the article? The n(n-1) terms are clearly the cost of managing the relationships between subordinates, but I'm not understanding the motivation for the n(2^(n-1)) term.

analog31 10 years ago |

>>> n(2^(n-1) + n-1); n = # of reports

The same formula could be applied to the people above you in the management structure, or the number of stakeholders in a project.

yodon 10 years ago | |

If you mean the equation measures your boss's cognitive load, then yes. If you mean that it measures your cognitive load resulting from the organization above you, I don't think that's correct except in the most pathologically dysfunctional of organizations.

As a manager looking down in the organization, you are responsible for the relationships between your reports. As an individual looking up the organization, you are responsible for your own relationships with people but you aren't responsible for their relationships with others (and even in a mildly dysfunctional organization where you need to worry about your boss's relationship with their peers, that still only adds an O(n) term because you only care about your boss's relationships not each of your boss's peer's relationships).