Hacker News Reader

Comment by jeremymcanally | Hacker News Reader

> In an ideal world, I believe that a flat structure would be best.

"Any sufficiently complicated company w/o management contains an ad hoc, informally-specified, bug-ridden, slow implementation of management." — @wycats

In my experience, that's held true nearly every time I've worked in a flat team or know someone who has (anecdotal, I know).

Point is: figure it out now and you'll thank yourselves later. Starting flat and changing it later will just lead to more pain and damage to the team cohesion and structure.

View on HN

There are two structures in a company: the management/decision structure & the communication structure. The management structure should be hierarchical. The communication structure should be flat. Make sure you have both!

Preference for flat structure is largely a product of bad management experiences

Why are bad management experiences the norm? There's a good reason for the desire for flat structures. My question is whether they can be made to work; I think they can.

Bad management is the norm, especially in Silicon Valley/Startups because:

1) No training for managers. 2) No measurement of managers against what actually makes a good manager (ie- only rewarded for results, ignoring if your team hates you and is quitting left and right). 3) First time Founders may not have had great role models or training themselves so they won't set a great example either. 4) First time managers are common because you were an early employee or around longer, so I guess you should manage. 5) Misaligned incentives: the only way to get a raise is to go into management 6) Being a manager isn't for everyone, but few companies make it safe to go back after 7) Turnover is tolerated more in the Valley than other places which can hide management problems others companies could never afford (it costs about 2/3rds of a salary to replace someone).

Re: Being able to be flat sounds great, but at scale creates a lot of challenges when you need decisions made and you don't all fit in a conference room anymore. Having someone who can make a final call (and ideally does take input from everyone involved) is more efficient than having 50+ people all moving in independent directions.

[2] Google Questions for Good Management https://getlighthouse.com/blog/google-management/ [5] [6] Why people take management jobs and shouldn't https://getlighthouse.com/blog/bad-manager/ [7] Why it costs over $65,000 to replace an employee https://www.linkedin.com/pulse/hidden-costs-replacing-employ...

I'd add to this, someone is more likely to be promoted if they are great at managing up, than managing down. Ideally you would have someone that does both well, or is stronger on managing down. I'm sure we're all seen that weaker results/contribution employee that seems to fail upwards.

I wonder if companies have tried doing promotions via a 360 selection rather than managers putting in their choice. It would be interesting to see people from lower ranks push for promotion and how this effects dynamics/results if weighted in the decision? Might work or just be a popularity contest.... I guess like everything it would be more about execution than idea.

Great added point, Gustomaximus. If you don't create a good feedback loop in your company, then people at the top making the final call for a promotion are going to go on the only info they have: their personal experience.

Deloitte got rid of performance reviews recently and one of the things they replaced it with was asking managers who they thought was ready for promotion and why. They also asked if you had to start a team over again, would you want this person on your team. I think all of those questions cut through the BS and are exactly the kinds of 360 questions you're suggesting.

Imagine a manager is up for a director role and each person they manage is anonymously asked "Would you work for them again if they were building a new team?" It could be a great way to understand how much people really like working for them.

Google actually does this with their questions every manager's teams are asked it affects compensation and promotions for managers:

1) My manager gives me actionable feedback that helps me improve my performance

2) My manager does not “micromanage” (i.e., get involved in details that should be handled at other levels).

3) My manager shows consideration for me as a person.

4) My manager regularly shares relevant information from his/her manager and senior leadership.

5) My manager has had a meaningful discussion with me about my career development in the past six months.

6) I would recommend my manager to other Googlers.

(More here: https://getlighthouse.com/blog/google-management/)

> Starting flat and changing it later will just lead to more pain and damage to the team cohesion and structure.

I totally agree with this. My company started with just 4 employees, and now that we have over 15. Three years later, we still struggle with defining management roles, because the original employees (including me!) have had many, sometimes overlapping roles dues to the small size. Now, it's difficult to really define roles and responsibilities because we are historically used to working free from any management from others.

> "Any sufficiently complicated company w/o management contains an ad hoc, informally-specified, bug-ridden, slow implementation of management."

That include companies with management as well. I have yet to work a place where the formal and informal management structure matched. They are usually correlated, sure, but they don't match. You have the manager that have raised to his level of incompetence that is generally ignored, people with a lot more to say than their position should indicate and people working across the company hierarchy in ways that the formal management structure does not suggest. I would in fact say that it is a requirement for a successful company to allow that to happen to some degree.

Listen to this guy and @wycats. This has been my experience also. Flat structures are horrible.

Perfectly flat structures are horrible, but on the flip side, if you have too many layers between the people making the decisions and the people implementing the decisions you're playing one large game of telephone.

That's why I like mid-stage startups. Typically, everyone reports directly to a VP, who reports directly to the CEO. The CEO has a clear picture of the company, the VP has the clear picture of a single specific product and the engineers have clear direction.

Are flat structures horrible, or are horrible flat structures horrible? Which systems have you experienced?

Non-horrible flat structures very quickly become neither. There's an old essay available online called "The Tyranny of Structurelessness" that talks about this.

I've read the essay, but regardless of its merits, it talks about unstructured organizations, not flat structures; hierarchy is just one of the possible structures you can use to avoid the problems she describes.

Flat structures are really project based orgs. They work fine if you are in fact working on projects.

When the org grows and operational/day to day work and process start to appear, flat orgs falter more often than not. Someone needs to be accountable for critical things, and whomever ends up in charge of the most critical thing ends up being "more equal" than everyone else.

Why are flat structures necessarily project based?

I'm sorry to appear argumentative, but this is an area where we as tech workers have a great opportunity to explore alternatives to the dominant system of command-and-control, since we have small companies which are already tasked with trying new stuff, and what I'm seeing is blanket rejection based on a few (often misguided and naive) tries.

I'm not an expert either, but I've seen many small organizations as a grew up (specifically, theatre groups), and while they had project (plays) based distribution of tasks, some also managed the other stuff (like finances and requests of grants) without creating an hierarchy or putting those people in any way in charge.

IMO, the project-based nature of a flat org is essential because the project is a unifying mission with regular updates and well-understood outcomes (project milestones).

Small orgs are different because motivated people can self-organize. A small theatre company or neighborhood block party committee can self organize because the core cadre of a half dozen people take care of the finances, grants, etc. Even in those cases, someone is at least nominally in charge -- an individual needs to sign checks, apply for grants, hold the cash, etc.

I've lived this at least a dozen times, in a mid/late stage startup, F500 company, and sprawling government bureaucracy. The story is always the same. New business units emerge with a guy in charge and a nimble team of <10 people. They achieve success and grow. Problems start around 10-15 people because folks are focused and cannot productively communicate across the group. Leaders emerge and work out differences. The next point of resistance in around 25-30 people -- the lack of explicit authority undermines the emergent leaders. Everything breaks at 40-50 people unless there is an established chain of command.

There may be a way to tweak these things and make it work, but it takes special skills/knowledge/instincts that aren't obvious to people. Special things are hard to repeat, and business relies on repeatable process.

IMO, there are a couple of ways that flat orgs can work. The most obvious is where the work is delivering unkilled/skilled tasks. A plumbing company has a dispatcher who sends out plumbers. A consulting company has salesmen called Vice Presidents, Project Managers and dispatchers. No hierarchy.

I think the other type of org is a co-op type arrangement. Coops have some hierarchy, but the power structure is elected vs. appointed which changes the dynamic.

Flat as in completely amorphous teams are a management anti-pattern because (as the poster above describes) cliques will crystallize anyway and a balkanised, feudal pecking order emerges.

Flat as in shallow hierarchy, though, that can work well. My preferred team size is five, with a maximum of seven, because communication patterns grow factorially. Beyond seven, the leader is overburdened by admin.

Seven groups of five makes a 35-strong engineering team and can be stunningly agile and productive, especially if the duties of maintenance and support are rotated periodically.

Ideally the maintenance and support are part of each team's duties. You'll never be more motivated to do things right than when you have to maintain and support your own work.

Reply on news.ycombinator.com