Hacker News Reader

Comment by dreamsofdragons | Hacker News Reader

If you set up the culture right, this isn't as big of a problem as you might think. On our team, it doesn't matter if you're the manager or not, the decisions for the product are made by consensus of the team. Sure management will still do approvals for time off and deal with the skinned knees and runny noses. But for day to day operations, it's peers working for a common cause.

View on HN

I've found consensus decision making to be frustrating in the environments I've encountered it.

A lot of flip flopping and lack of follow through.

Who made that decision? We did. Can we give up on that and try something else? Of course.

I kinda like to have at least a shepherd for the day to day decisions, if not a commander.

In practice in my current environment it's been the occasional PM that takes some ownership that fills that role, but in many groups I see the floundering without leadership.

Internally I think they think they're making progress because everyone on the team is always happy with their decisions, I may just be the old curmudgeon that's frustrated with what I see as wasted effort.

In my utterly-average agile environment, it's not hard.

Your peers form an agile group to deliberate. You have non-management product owners and stakeholders to guide you (shepherd you).

You have a supervisor who deals with HR business like time off, but is unrelated to your product decisions.

While it can work, "consensus" usually means ad hoc, no accountability, no followthru.

This essay was illuminating:

Tyranny of Structurelessness

https://en.wikipedia.org/wiki/The_Tyranny_of_Structurelessne...

Yes, <3 that.

"Contrary to what we would like to believe, there is no such thing as a structureless group."

The same essay was linked to me a previous time I made a similar public complaint :)

While I've always thought that essay made good points, it's somewhat dangerous to overinterpret the thoughts of a Stalinist on groups without structure. Her solution is (obviously) to just have hierarchical structure, but this leaves you with the same problems as structureless, there's now just a channel by which people who play the game well can pretend to be accountable. Stalin himself proves that hierarchical groups aren't more accountable, even if they nominally have mechanisms to be so.

To illustrate how consensus-decision-making can be a tricky balance, I've been in a situation before where I was managing a team of a few other developers (about 50% hands-on myself) and so tried to stay away from being "the decider" since there were others closer to the code.

The feedback my own manager shared with me that had been collected from my reports was that they appreciated what I was trying to do, but that they also wanted me to have a more active technical leadership/decision-making roles.

What I'd failed to do was talk enough specifically with the rest of the team around what they wanted the roles to be, vs just assuming they'd want maximum autonomy. This will vary depending on composition of team, familiarity of the devs on the team with the problem area and stack, etc.

It will also depend on if managers are expected to be almost-purely people-managers or if they're also expected to be domain knowledge leaders as well.

There's no one-size-fits-all approach, but the thing I've seen that separates a good manager from a bad manager is being able to adjust their approach in things like that based on the situation vs sticking to a script.

The question is who decides the annual bonus and raise?

More importantly, who decides who's manager next quarter/year? I've never seen non-managers exercise much leverage on this front. Being able to determine who works with whom is a very powerful position.

Actually, I planned on writing a blog post on this at some point. I work at Hired, and we had an interview process to determine who (internally) would become the next manager. This involved three interviews with engineers, and a presentation with management.

That sounds humiliating for candidates who don't make it. Especially having to go back to work with the engineers that rejected them.

In the "traditional" method of selecting a manager, most engineers don't get selected either. However, the process is less transparent that way.

I'm saying this as a candidate that didn't get selected. I got great actionable feedback on how to be a better leader, and I'll keep my fingers crossed when there is another opportunity available.

Product decisions can never be solely by consensus. At some point, barring massive groupthink, there will be a disagreement on some product decision (what color should this icon be?) but the icon will ultimately be assigned a color. There is some process by which that decision is made.

Usually A/B testing one would think

Yeah I'm not sure why people think consensus is a good strategy. The consensus established by a group should be about who you trust to make a decision, not the content of the decision itself.

i.e. when it comes down to the icon, your group should agree to trust the designer, and then trust the designer. The benefits are numerous: 1) People get to make the decisions within their realm of expertise, 2) No aimless debate, products are built faster, 3) If there was a bad decision, there is one person who made the decision, but the whole team should share accountability since they all agreed to trust the person who made that decision.

The process by which that decision is made is the process by which you select a person to make that decision. Ideally, that happens at the hiring level, so once, for example, a designer is assigned to your team, you already know that s/he is good enough to pick the right color. You don't have to engage in office politics to discover whether or not you can trust him/her, and you don't have to engage in office politics to allocate praise/blame in the event of particularly good/bad decisions.

Do you really work for a genuine co-operative where all decisions are made by consensus of the workers? I'd be surprised and interested to learn more if that's the case.

There's a Brazillian company (Semco) that's like this supposedly, but I have to wonder how they're doing today with so much of the company involved with energy.

there also is the mondragon cooperative: https://en.wikipedia.org/wiki/Mondragon_Corporation

there also is a ted talk by semco ceo ricardo semler: https://www.youtube.com/watch?v=k4vzhweOefs

> Sure management will still do approvals for time off

Not a peer.

"decisions for the product are made by consensus of the team"

I favor democratic decision making processes.

Use voting systems appropriate for the task. Roman eval for new hires, go/no go. Approval voting for triage, new features. JADs for "visions". Creative writing exercise for post-mortems. Etc.

"Democratic" does not mean winner-takes-all, which is suboptimal.

"Democratic" does not mean consensus.

"Use voting systems appropriate for the task. "

I rather don't agree with this one.

'Voting' should almost never, ever be used.

For so many reasons.

Rarely should everyone have an 'equal say'. You are not 'equal citizens' in this. Second - is that it leads to politics - 'I helped you last time, you help me this time' etc. Unless you get into crazy things like 'hidden ballots'.

There is no substitute for leadership.

Often, a mediocre decision, executed upon well, will lead to better outcomes than 'great decisions' wherein their is fudging, lack of insight etc..

In situations that require broad input, a decent moderator should be able to assess the 'consensus' and go with that.

In many design situations, architects and designers should basically set the direction, absent issues brought up by team members.

Things like 'should we all use the same IDE' or 'what IDE's can we use' - obviously require a lot of input from team members and maximum flexibility.

But things like API design ... that needs to be really well curated and it's best left to the experts - with strong feedback from team members.

The word you're looking for is neoreactionary.

Who decides who are the experts, and who checks their work?

>There is no substitute for leadership.

Second this, true in my experience as well.

Reply on news.ycombinator.com