No Kings: How Do You Make Good Decisions Efficiently in a Flat Organization?

That's why democracies have terms of office, and that's exactly what we do in our student organisation.

You're right: there has to be a way to get somebody out of a misfitting leadership role without completely humiliating him. Regular elections are a great way of doing so, and giving many different people the chance to lead benefits both them and the organisation.

Our teams at Tanium do something similar. There’s no pay bump to lead the group... and we have a number of folks who’ve led the a team and then went back to IC. It’s worked quite well overall; however, is a difficult position (sometimes leading north of forty folks)

Holacracy has some mechanisms for this -- work gets divided into explicit roles (most people have 3-15), and what roles you fill can be pretty fluid based on the needs of the work, etc. Some roles may come with some decision-making power, and management functions get divided among the roles in whatever way makes sense for the job -- sometimes they cluster at a single point (such as a liaison to a bunch of outside contractors), while other times they're completely distributed between roles or even delegated to a process rather than an individual. There's a lead for each "circle" (team), but it's mostly a coordinating role -- they search for the ideal role-fillers, and they manage questions of prioritization (at least by default -- basically everything is customizable in Holacracy). There's nothing particularly special or manager-y about the lead role (I actually strongly dislike the one lead role I have), and leads can't hire or fire people to the company, so even if you get moved out of a role that doesn't necessarily mean anything for your job (although if no one thinks you're a good fit for roles in any circles, that doesn't bode well).

It avoids the politics of democratic selection, while decentralizing the power dynamics enough that tyranny is more-or-less impossible.

I think that this is an excellent insight.

Explicit leadership roles also inevitably result in situations like, "well, they were the best person we had at the time" whereby a very unsuitable leader is kept in power. The worst orgs I've participated in have suffered from this kind of problem.

People only fantasize about flat orgs when theirs is led poorly.

Unfortunately, this sort of ignores that many academic departments are often crippled by overwhelming internal politics, and they rarely scale beyond a few dozen people at most. Moreover, many academicians within a department often work on completely separate projects requiring little or no coordination of any kind. This is dramatically different from what is encountered in many commercial organizations and companies.

My experience is that flat structures in large commercial enterprises usually reproduce stratified structures embedded within them, and they actually enhance the degree of stratification by obscuring who makes decisions and creating unassailable cliques.

If your organisation already has good management, then promoting the right people isn’t that hard of a problem to solve.

> Self organizing teams still have a leader, but the team selects the leader.

People who are good leaders will naturally start to show leadership aptitude, without being given any management position. If you have good management, then those people can be fostered. They can be given extra responsibility (without being given a promotion), and if it goes well, they can be promoted. If it goes poorly, they can go back to what they were doing, no harm done.

It depends on the organization. We've had leaders move back to IC roles (and are going through one of these transitions now, actually), but I think it requires the organization to be explicitly supportive of such moves. We call it finding your "zone of genius". We'd much rather keep a good person then lose them because they're best suited for another role.

The way my workplace addresses this is:

1. Transitioning to management or tech lead is not a promotion, but a change in responsibilities (and transitions back are common);

2. Whether manager or IC, we only promote people based on demonstrated ability, not potential, to avoid the Peter Principle

> The problem with explicit leadership positions is that people have no downward mobility. If you are bad, then the organization loses a great resource.

There is downward mobility if you want it.

I have seen technical people promoted to managerial positions in the 'usual' way, i.e. seniority, who ended up not liking it and not who were not very good at it, and went back to purely technical positions within the same organisation.

Nothing prevents this as long as people are sensible, talk, and don't promote battles of egos in the organisation.

It's the last place you'd expect, but the (Canadian) military actually has this in a certain measure. Unit commanders are only appointed for a limited term (usually 3 years), after which they're expected to move on to another unit.

In the Canadian Cadets program, we dont have that expectation, so corps commanders usually stay in the same corps and either mentor the new commander or take on a specific role if there's a need.

There's one "old" captain at my corps who has been mentoring many generations of officiers; I wasn't even born during his first term as commander.

That does mean that there isn't much room to grow past that, though. Luckily, teenagers (and activities offered to teenagers) change a lot through the years, so that forces us to always be innovating.

As I read it, the point of the article is to explain how you deal with decisions when the "captain" can be anyone in a group, and not some designated hierarchical leader.

It's not democracy per se: for instance a designer can come with a proposition to give people a chance to review it. In the rafting example they're basically saying "we go right", and will do so if nobody shouts "there's a cliff on the right".

a somewhat similar phenomenon happens in basketball. when the game is not competitive, no leader is needed and everyone just plays to their ability (and hopefully to their strengths).

but when the game gets competitive, it's easy to fall apart and get beaten without a leader. on offense, that's usually the point guard (e.g., steph curry), who implicitly chooses who gets the ball, when they get it, and for how long. on defense, a forward/center (e.g., draymond green), who can see the whole floor in front of them, often has the responsibility of coordinating the defense to minimize gaps and mismatches.

That's a good point when in that situation. But it's worth remembering that in development people are rarely in that situation. For various reasons, however people (including me) often bring that kind of emergency thinking into development and it just shows things down. Rushing into the debugger instead of slowing down for a minute and thinking about it for a minute. Doing a little bit of design and thinking about what I want from the system instead of getting lost in the details of the existing implementation.

Most of the time in development we need to slow down to go much faster. It's nothing like white water rafting (it would be cool though if it could be like rafting sometimes - everyone gets sick of having to slow down all the time at some point)

Perhaps you need "situational captains"? Identify strengths/weaknesses of your own staff and when SHTF, delegate to one that can employ their strengths to resolve your situation, while keeping communication channels open to the rest of your team in case somebody has a better idea?

I believe most people would appreciate if they weren't infantilized and forced to defer to some dysfunctional hierarchy.

Look at the time, it is hair-splitting o'clock! >_>

Having a single captain can still be a democratic system; what is being replaced isn't the democratic element (where power rests with the people) - that is still in place. What is being changed is swapping a deliberative assembly (or committee or something) for a single executive for making decisions.

So what is being identified is technically something like 'deliberative assemblies are better at dividing up resources and responsibilities but single executives are better at implementing policies and achieving outcomes'.

Interestingly, this is a lesson that is implicitly very well understood by major democracies without being explicitly bought up very often.

One boat one captain!

But democracies are capable of producing such leaders when required, as England did for the Second World War.

For very slow parts a captain is probably also more helpful. I.e. if you were on the open ocean, you should probably listen to one person that has some kind of compass, instead of going in circles.

It's not just the women's movement, any postmodern "leaderphobic" group of anti-heroes will operate under this adhocratic masking of power. It's still power, but it's the power-pill that palatable to their sensibilities, suitably masked. Marxist revolutions "of the peasants" pioneered this mass delusion of flattening.

But isn't this just gaslighting people? I mean "flat" is just disempowering existing power structures, to empower a secretive cabal, that pretends to rule "benevolently" by consensus by usurping the existing "oppressive evil" but ends up just becoming worse. Prove you can actually be better, then come back and tell me how to.

This article should have been called: "Afraid of Kings. How we tried and failed to get s*it done when we need to pander to everyone's leaderphobia."

Bottom line is, this is not new hat. It's old hat. What I'd say to the author is: It's an adhorcracy. Read Down and Out in the Magic Kingdom. Then come back and tell me how to.

While I agree with something of an abstraction of what they're saying, when I read things like this, it's hard for me to see anything but a person that wants to hijack the tyranny for themselves.

A few years ago I joined a company with "self-organizing" teams, which have many similarities to "structureless".

It was an awful experience, for me at least. Many important choices were made by people trying to avoid ruffling the feathers of the least-mature team member.

My woes were likely compounded by being mildly affected by Asperger syndrome: First, I have an innate preference for clear organization and structure. Second, it took tremendous mental/emotional energy to navigate the social/interpersonal minefield tied to every technical decision.

Perhaps my experience was unusual, but I cannot understand the thinking that leads upper-management to think this approach is a good idea.

What about democracy? It is a clear rule (one person one vote) and yet it is not hierarchical.

Saving this as it makes the point I've been trying to make for years much more elegantly.

Just because you don't call out the structure doesn't mean there isn't one.

I work for a "flat" company. I've certainly observed some of the troubling characteristics of this organizational approach described above and elsewhere in comments.

There are some positive characteristics [in our implementation at least] as well, a few of which are :

1) Decisions with broad effect [i.e., beyond the decider's personal work domain] are handled with the "advice process," wherein all affected parties are consulted and their advice weighed before the decision is taken. This can be cumbersome, but on the other hand, it is a check against "cowboy" and "act first, consider later" types.

2) Each person is assigned an "area of responsibility / AOR" -- a work or problem domain[s] in which they are acknowledged to hold more expertise than those not assigned that 'AOR.' This prevents [in practice] people [say] in the shipping department making decisions about software development.

3] The org is not _totally_ flat. The CEO and COO are acknowledged to have veto power and the 'AOR' of big decisions that have a potentially existential bearing on the company.

Having always worked at rigidly hierarchical companies before this, I'll say I've more or less gotten used to it and can see the benefits.

One thing I especially like about it is that there is no chance of an "Office Space" situation where a boss, or bosses, continually interrupt work to pester about things that are wastes of time -- each worker is more or less left to determine their own task prioritization and has the power to discontinue tasks that are pointless. We use the kanban system for task / bug / feature tracking.

Most flat org I've seen are not flat. The people naturally organised in tribes that are only known by themself, and that are in no organigram.

Each tribe have an implicit leader that bears no title yet influences all decisions just by regular social interractions.

Tribes communicate, negociate even, through the implicit leaders even if not directly.

Some kind of overuling bdfl also oversees all tribes, like a king, in case of conflict or emergency. Often the CEO.

It's not "flat org", it's just good ol' human society.

Rules are less clear, social skills are more required and if you get there in the early days you get a big advantages over the newcomers.

But if you get a nice tribe, things rock.

Flat does not mean structureless. Jo Freeman's piece correctly points out that groups without structure inevitably grow shadow organigrams. My point here is that flat structures should lack hierarchy but they should strive to make their organization as transparent as possible.

I disagree. I oppose concentrations of power. Will there be a certain power disparity and hegemony of certain particiants in their little groups? Sure, but as long as they are prevented or discouraged by various mechanisms from joining together into still larger groups, then everyone still has a fighting chance of opposing them or switching groups. The trouble arises when you have concentrations of power. Just look at the Web! It disrupted AOL. And Wikipedia disrupted Britannica.

strongly seconding this essay - it remains one of the most influential things i've ever read, in terms of completely and immediately changing my views on a topic.

It’s hard not to see this as purely a response to an article title. It bears no relation to the actual article content, which was thoughtful and interesting and really deserved a more on-topic critique.

I should copy and paste this into must discussions about how Bitcoin is awesome because it's decentralized and not controlled by a government.

The point of a "flat" organization is not that there is no hierarchy or structure, but that it is not officially prescribed, and instead emerges organically. This allows the structure to change as the needs and members of the team change, whereas in a traditional organization, the leadership and structure often long outlive their usefulness.

When you are talking about complex dynamic systems like human group behavior you can always find counter examples to anything anyone points at.

Here's an example of from Niall Ferguson's book The Square and the Tower - "Formally, the directors of the East India Company (EIC) in London controlled a substantial part of the trade between India and western Europe. In reality, as the records of over 4,500 voyages by Company traders show, ship’s captains often made illicit side trips, buying and selling on their own account. By the late eighteenth century the number of ports in the resulting trade network was more than a hundred, ranging from open emporia such as Madras to regulated markets such as Canton (Guangzhou). In effect, private trading provided the weak links that knitted together otherwise disconnected regional clusters. This network had a life of its own that the Company’s directors in London simply did not control. Indeed, that was one of the keys to the success of the EIC: it was more a network than a hierarchy. Significantly, its Dutch rival banned private trade by its employees. This may help explain why it ended up being superseded"

What works and what doesn't varies a lot from org to org.

To me flat orgs default to lord of the flies leadership.

You can have structureless orgs on boards for significant organizations but people need to be accountable for a well-understood scope.

My little league has a president at 20 board members. There is a president and 4-5 committee chairs. (IIRC finance, concession, baseball and grounds) Somebody is responsible for the snack bar, but any board member may be on duty running it at any time.

I've worked on good sized project teams where there was a sponsor (the king), the PM (task guy) and 10-20 contributors. It most mostly democratic as long as the milestones were hit.

I've worked in "flat orgs" where there are 100 employees, a director/vp/king, and a big band of equals. Just like in Animal Farm, some of the animals are more equal than others.

Structurelessness must be distinguished from established structures with a flat hierarchy. Codifying a non-hierarchical structure with explicit rules and norms counters the above situation where a few people start to become dominant through more informal means, and everyone else is left confused.

Flat hierarchies are not necessarily structureless and need not suffer from the drawbacks described above.

It isn't just Freeman's article. There's a whole academic/industrial literature about this stuff, a literature which seems to have been fairly well known to the broader public about 25 years ago but to be completely unknown to it today: https://news.ycombinator.com/item?id=19468090#19468699 .

There was a good podcast with a famous FBI negotiator, Chris Voss, who recommends if you want to sway someone, frame it as, "are you opposed to doing X?" rather than "should we do X?". People are more likely to agree with the former.

And for kids under two, you give two options and place the one you prefer second and emphasize it. Most of the kids agree. And then disagree. and then agree. and then disagree. and then you distract them with a snack and they happily go with whichever!

We used to have an engineer who used this maneuver. He abused it because it was effective. We do have an issue on our team with engineers being indifferent to new ideas, and our manager loves consensus. Regardless, this engineer thought his ideas were always "good ideas." They weren't. He was eventually chewed out by our manager because he was becoming a passive-aggressive tyrant. He left the company soon after that.

Nevertheless it goes to show that fortune does favor the bold, but that fortune might be at another company.

Oh man. This is one of those things that's great when done well and terrible when done poorly. Miss the head-up memo until the following day? Too late, the author's already a day into the work and you'd be a jerk to make them back all that out. If it happens often, you have to constantly monitor communication channels or risk missing an important "heads-up," since that's how serious decisions are being made now. In the middle of your own work, but see a 10-page design doc go by in slack or email? Better drop everything and read it well enough to form an opinion because every minute that goes by it becomes harder to object to that change.

But, as you say, it can be a great way to overcome bikeshedding. It's a good way to speed up a ponderous decision-making culture. It works.

Like a lot of organizational tricks, the line between use and abuse is something that simply requires judgement.

Another way to put this is "bias for action"

Indeed, like any management strategy, it can be done poorly, inadvertently incentivizing people to form shadow organizations, or it can be done well, encouraging people to engage in good faith and reaping the promised benefits.

In practice, we get lumpy mix of good and bad implementations within organizations. It takes a lot, however, to overcome the near-term, personal-empire-building incentive. "This way is better for everyone" isn't going to cut it for a significant proportion of folks. If you bake it into the bones of your recruitment process and in your management training, there's a chance it'll be an authentic value of the org. And even then it's hard to shake the suspicion that this strategy only further rewards showhorses and marginalizes in-the-trenches doing-thankless-work types.

If you get an extraordinary manager, she can mitigate that, but your culture isn't robust if it relies on the extraordinary.

Regarding Greenspun I have to think about a colleague. He is a studied mathematician. So in the past 15 years he was working on a tool in Perl. Now he switched teams and is tasked to work on a Python problem. You don't hear about him for weeks besides he need to setup is development environment. Then he comes back and starts to code. It seems like he doesn't need to spend any time learning Python. You look into his code and also have no idea what's happening or why it's working.

What he basically did was implement his own programming language out of lists, dicts (hashtables), and functions, with class hierarchies, type checking, inheritance, etc. And the time he spent "setting up his system" was the time he needed to basically code his VM engine in Python instead of Perl, and then adapting his standard library files to the quirks of the underlying interpreter.

It was weird for me at first, but actually makes sense if you look at it mathematically. A math lover starts with defining zero, one and plus, right? So this is the only logical way to program for him. He doesn't even know what a VM or standard library is.

The article makes an early mistake in describing the IETF as "membership based". Nope. You cannot join or leave the IETF, it has no members, everybody who is interested can and should participate. It likewise confuses itself by insisting on calling the hum a "vote" even though it acknowledges that humming mustn't be used to try to enumerate support which is exactly the point of voting.

The IETF process works because of _what it's for_ and could not usefully be replicated in contentious political environments. Rough consensus and running code builds SMTP but it doesn't end the Troubles in Northern Ireland.

Because the IETF deliberately has no actual power, in the end the worst that can happen is you didn't agree anything and have to roll your own. You are "at" the IETF (many people never attend a physical meeting, and that's fine) to reach an agreement, and if you can't agree then too bad but it cannot stop you pressing on anyway.

This part of the model also worked for CA/B. CA/B initially exists as a meeting between two groups that both want something, the Certificate Authorities want a sales promotion tool and the Browser Vendors want better validation. But the CA/B meeting lived on long after that discussion bore fruit, because it turns out that _agreeing_ among yourselves is much better for everybody than if you all just do whatever you want and hope it works out.

That seems a like a great place to work.

Did you read the article? The whole point is not to seek agreement, but a few proponents (authors) and no fundamental flaws.

Decision making is fraught with lack of knowledge and uncertainty. I favor rough consensus. Face it - you're never going to get the best or "right" solution in most cases or practically ever. It's good enough to get something that most people involved think will work. I emphasize "involved" because there are always individuals who don't really care about the decision at inception. They sometimes complain if the idea doesn't work out, but then you can point back to the discussion and tell them they had a time and place to give their input.

I feel exercise of decision making is mostly about having stakeholders be aware that a topic is being discussed and making a space for open discussion. The outcome of the decision is largely irrelevant.

As per the article. Rough consensus means everyone states "I can live with this solution". This is contrasted against the majority stating "I have no objections against this solution".

The idea is to not get bogged down in 'I think this is fine but X is better' discussions, and only deal with 'I think this is actually bad' discussions. Because any discussion is very long, and the first kind of objection is not severe enough to warrant a lot of discussion. See also the concept of 'bikeshed-ing'

Yeah. I've never seen "flat leadership" work. Someone always becomes the leader just without the title.

Personally, I strongly prefer formalized structures over informal ones because then everyone is on the same page and the right person can be chosen to lead instead of it just being the one with the loudest voice.

An issue with that approach is you end up with leaders forced to approve/champion proposals they don't have enough expertise to fully grasp.

The 'designated' part is a blessing and a curse, I understand trying different systems with different failure modes.

This happens in organizations that aren't flat. They're dysfunctional, but they aren't flat.

It is said, "Success has many parents. Failure is an orphan."

Before you get into cage matches, engage with your colleagues respectfully and maturely, politely declining suggestions you disagree with, before rushing to cuss them out.

This can be headed off by not rewarding bad behavior in peer reviews/manager reviews. In addition, such behavior could be used to lay off the offending person.

Not that I'm a fan of completely flat organizations.

It should be noted that the right way to be productive in a flattish organization is to influence people - this is no different in a big company from a small company.

Would it work better if those "more equal" members were explicitly above the rest of you?

There was a recent HN thread about this:

https://news.ycombinator.com/item?id=20099261

In short: power is held by insiders. It's really not all that different from a traditional hierarchical company, except without any job titles to tell newcomers who holds the power, and without the ability to hold managers responsible for the performance of their reports.

This is the best description of how Valve's "flat" structure works in practice:

>Some of Geldreich's comments are similar to comments Ellsworth made after being let go by Valve. "...There is actually a hidden layer of powerful management structure in the company and it felt a lot like high school," Ellsworth said in 2013. "There are popular kids that have acquired power in the company, then there's the trouble makers, and everyone in between."

>Geldreich similarly describes 'barons' who are in with the executive arm of the company in question, and a culture in which employees must curry favor with influential 'sponsors' to enjoy stability.

>At self-organizing firms you might be placed into a huge open office and given massive monitors. This is to normalize all communications and for more effective surveillance. Everything will be monitored either directly by a corporate arm employee, one of their barons or friends.July 16, 2018

Rest: https://www.pcgamer.com/ex-valve-employee-describes-ruthless...

You're using some lingo I never saw before, what's a SAFe and a PI?

+1, I love RO. The Holacracy book is also a great read for an implementation of what a built-out Teal (though not quite flat) system can look like (and how it responds to objections like the ones seen in a lot of these comments). Of course, you have to unlearn all the stuff you learned about Holacracy from bad business mag articles....

And in a company someone has the power to fire or promote you so there is explicit power in play. You can try to distribute it but people are social and clever so they’ll work around any blockers.

I agree with your point but in the end the article was more about efficient decision making in engineering teams and less about flat organizations. Up-voted both the article and your comment :)

Reminds me about the stuff I've read about Valve. In theory, and maybe if you have strong healthy de facto leadership it could work. In practice it seems more like an excuse to dodge responsibility, leading to pretty toxic places to work - but maybe there's another side to it.

https://twitter.com/PastelChum/status/1136470238265040896

There are shadow power structures, but I don't think you are correct that they are necessarily hierarchical.

In my professional experience, many fiefdoms get created in the shadow hierarchy of flat orgs.

Align/get yourself in one that works and is healthy for your development and you can thrive. If not, then leave as quickly as you can. Just MHO.

In my experience, these shadow hierarchies exist in both flat and structured organizations. However, in structured organizations, there's the added challenge of figuring out how to navigate both the formal and shadow hierarchies simultaneously.

Exactly!

Business schools use a framework of formal power (job title, authority, etc) and informal power (influence and likability).

The most effective organizations are the ones where the formal power structures are very close to informal ones. People know how to make decisions. Flat organizations end up being run by the loudest voices in the room.

Ironically, flat organizations with largely informal power structures are often slower to make decisions because one person can become a huge bottleneck on decision making without realizing it. It can also be culturally corrosive: informal structures encourage Machiavellian behavior and the most Machiavellian people tend to rise to the top — which creates a terrible work environment for everyone else.

A slightly radical theory: flat organizations are set up to foster that intentionally.

Suppose your company has the idea that they want to identify and reward natural leaders and/or those who test limits and push through boundaries.

One way to do it is to tell people your organization is flat, then watch for the people who don't buy that and instead see an opportunity to take charge.

You create a situation with a little bit of an artificial and completely intentional power vacuum, then you watch who takes it upon themselves to step in and fill it. By holding a bit of a vacuum, you encourage these people to show themselves. (And I'm convinced management is actively watching to see who takes that bait.)

I'd say these are the most common responses: (1) never reach awareness that it's happening and continue to take flat at face value, (2) see it as opportunity and take advantage of it, and (3) see it as unnecessary game-playing and dislike it.

Whether 2 or 3 is better is an interesting question. It might separate the practical people from the idealists, or maybe it separates the self-interested from the more community-minded. I can see some of both, personally.

EDIT: I don't necessarily want to be completely dismissive of the idea. There's something to be said for creating situations that allow for people to stretch and try out new roles. That's how people develop. This is informal and stealthy, but it is a way to do that.

There's a distinction between the Sociocracy consent process and Holacracy's process with objection validity, but yeah on the continuum of similar processes I'd put chaos consensus (which is what a lot of people who attack self-management/flat are actually critiquing) at one end and Holacracy on the other, with this and Sociocracy about 3/4 toward the Holacracy end.

It sounds like fun when you describe it that way!

It's more complicated than a hierarchy. Informal power isn't transitive, so you can have arbitrary directed graphs with loops.

It'd be an interesting experiment to build a tech startup with an overt hierarchy, where decisions were made by the person in the room with the most stripes on their shoulders. The conventional wisdom is that it would suck and you couldn't hire smart people, but sometimes the big wins are to be found in such places.

The problem is bikeshedding. You can burn through a lot of hours arguing about decisions that do not affect the end goal.

It takes really little effort to use "they" instead of "he" to include all people.

In my company (which runs on Holacracy), we have a process for that rather than a person, with defined rules around how that happens and for what sizes of spend. It's not perfect, but it manages to provide firm rules that aren't in any single person's hands and which everyone has equal access to using. There are clear guidelines where clear guidelines are possible, and then process around determining what good judgment is for the areas where clear guidelines are tough.

I'm not defending "flat" (which Holacracy isn't), but there's a lot more nuance in the serious Teal/self-management community than the public debate in threads like this appears to know about.