The Trouble With Consensus

Imagine a method of governance which works as follows: Every citizen has a vote and every citizen has a veto (interesting that these are anagrams for each other). How much could be accomplished?

Not much, I'd guess, since in a population of any size there very, very rarely will be unanimity, or even consensus. Thus, it is necessary to resort to majority rule.

What has this to do with team-based cultures? Plenty! There is some size for an organization beyond which consensus politics just can't scale effectively, so we rely on majority rule with institutionalized protection for minority interests. In the name of consensus, however, some organizations find themselves paralyzed as any management person can cast what is a virtual veto by refusing to buy in to a group consensus.

Such was the fate of an outfit I used to work for, which required that any effort be either initiated secretly ("it's easier to beg forgiveness than to ask for permission"), or lobbied to the point of exhaustion. So many managers and directors and vice presidents were involved in even low-level matters that the disagreement of any one of them could effectively scuttle any decision on which consensus was required. Engineers spent little time actually in design and a great deal of time working up business cases to justify design activity. It wasn't until months after I left this job that it occurred to me that in their zeal to become a consensus-based organization, there was no distinction made as to when consensus should apply. And yet, ironically enough, in this very same place, some decisions were made by fiat, decreed by some vice president and implemented, despite the fact that there was nearly universal disagreement regarding the policy.

Has anyone ever seen this in their experience? How does one identify the point beyond which consensus decisions will not occur? In our small teams, consensus works beautifully, allowing a range of disagreement but always able to bring in all participants to a solution. Enlarge the scope, even one or two people, and that all might change.

Is it just personalities? Or is there some phenomenon of organizational psychology at work?

DonOlson (in a flu-induced delirium) 95/10/22


I grew up attending Society of Friends meetings. The Society of Friends runs meetings for business by consensus; you can't do anything unless it is "the sense of the meeting". This has two major consequences:

Consensus is not unanimity.

Under most circumstances, if Fred says "A", I say "B", and the conversation appears not to be resolvable, one of us is expected to say "Well, I guess A isn't so bad after all." In short, consensus doesn't mean "everybody is happy" or even "everybody agrees", but "everybody is willing to accept this decision".It is very bad manners to continue to argue your position after it is clear that most of the Meeting disagrees. Sometimes, it is appropriately bad manners; some famous Friends continued arguing their position for years (John Woolman against slavery) before the Meeting finally adopted it. But, in general, people are expected to respond to the absence of authority by submitting themselves to an unspoken authority.

I think that you can't "manage by consensus" effectively unless all participants understand that it is rarely appropriate to continue debate indefinitely, and that it is often appropriate to concede your point even when you're still not convinced.

If any active members of the Religious Society of Friends want to correct me, please do.

--BetsyHanesPerry


I think I've run across a similar situation to Don's. We had a boss here that insisted that all designs were to be voted on (informally) by the team, and in the case of roughly equal splits, she'd end up making the decision. It was bad enough being forced into a "design by committee", but then to have a manager make technical decisions! Argh!!

Now she's gone, but since the systems we are creating all interact we end up in "design committees" again, with the VP acting as referee. It's hard to perform in a strong Architect role in such cases.

--DavidHooker


It might help to formalize a face-saving way of reaching consensus. A way to proceed with a proposal whilst logging the objecting opinions. Some people will be more willing to go ahead with a plan if the fact that they dislike it is on file. --DaveHarris


Intruiging notion, "There is some size for an organization beyond which consensus politics just can't scale effectively." I would believe the statement without justification except I keep hearing of Japanese companies running this way, and I have seen Lucent work this way. Both exceed the scale threshold. Do someone know more of a large company that succeeds in working this way? I suspect that manners count for a whole lot in the consensus business. --AlistairCockburn


The DecisionGame is one way of getting consensus organizations over the tendency to postpone decisions. The basic idea is for someone to present a proposal, then everybody gets to ask for clarifications, after which the presenter presents a modified proposal. Anybody can then veto the proposal, at which point it goes back on the shelf until a presenter wants to try again. The HillsideGroup developed the DecisionGame as an alternative to disbanding.


RomanEvaluation is the consensus-based decision technique that I've used at several companies. It succeeds, in part, by offering an alternative to a strict yes/no vote. --DaveSmith


Perhaps, if you have trouble reaching design consensus, people aren't doing enough listening, enough experimenting, effective-enough communication. Disagreements in design often come from unshared priorities. Really hearing each other tends to result in proposed solutions that are clearly delineated and subject to quick experimental determination.

And perhaps the project needs a little more delegation of authority: the person with the task gets to decide how to do it ... and is expected to decide well, based on understanding the alternative approaches. --RonJeffries


OffTopic:

Amusing that Don originally misspelled manager manger. More common is the misspelling mangler, in fact I hear that term used in derogatory fashion in Dilbert organizations everywhere! --JimDensmore

Spelling (typo) corrected! As tempting as it was to enter "manglers," well... --DonOlson

You know, on reflection, "manglers" is a much better term. --DonOlson

Interesting that in two years of preferring "manglers" as a term for "managers" no one has seen fit to respond.


51 percent vote to kill the other 49 percent.


My understanding of the "consensus process" may differ from the viewpoints expressed here. In my experience, the value of the consensus process is that it strongly encourages the group to explore new approaches. When Fred says "A" and Betsy says "B", my understanding is that Fred, Betsy and the group then seek to find new alternatives -- perhaps incorporating parts of A and B. The group seeks a solution that allows both Fred and Betsy to say "Yes, that addresses my concerns." Yes, of course the process can be sabotaged -- that's why it's important that the group be able to discuss process as well as content. If the group, after an extended period of sincere work, is genuinely deadlocked, then it's probably time to take a break, draft a statement outlining both sides of the deadlock, sketch the agreement on everything else, and defer the deadlocked decision -- at least for a day or so.


JimMcCarthy's SoftwareForYourHead protocols for Decider/Resolution are interesting ways to make proposal handling more manageable. (Haven't actually used them myself, though...) --BillSeitz


Imagine a method ... How much could be accomplished?

Is the quantity of things accomplished really the correct measure for success? One must take care not to confuse activity with progress.

Where is the harm in addressing the concerns of a single individual? If he is correct, then his concern should be addressed; it may be resolved, minimized, or merely accepted as is, but the concern has been recognized and addressed. If he is incorrect, then that individual should be able to walk away with a better understanding of the issues and not merely ignored.

The reason for having more than one person involved in making a decision is to apply more than one viewpoint. Different people will have different perspectives on the issue and have different knowledge of the issue. Consensus serves two purposes, it ensures that all perspectives have been addressed and it helps ensure that knowledge has been more evenly distributed.

There are very few individual critical decisions. It is far better to take the time to reach consensus and broaden understanding than to rush off and accomplish something, especially when we don't know why the action succeeded or failed after we are done.

-- WayneMack


Is the WikiWikiWeb a giant consensus finder? After all, every viewer has a 'vote' and a 'veto' (and a veto of the veto, and a veto of the veto of the veto, and so on). They can change whatever they want. Consensus is achieved when an issue on a page becomes "resistant" to changes by the community (i.e. being reverted constantly). -- JimmyCerra


C.f. MeasuringConsensus


CategoryInteraction


EditText of this page (last edited April 20, 2006) or FindPage with title or text search