On social software and consensus

Have you read David Freedman’s recent piece in Inc. Magazine’s September 2006 edition? You should. It’s been doing the rounds in some of the microconversations I participate in, and I’m glad it came across my radar screen.
Freedman headlines his article What’s Next: The Idiocy of Crowds. OK, that got my attention.

He then starts the article with:

Collaboration is the hottest buzzword in business today. Too bad it doesn’t work.

Now I wasn’t just attentive. I was committed. Very intrigued. Very very intrigued.

Here’s some more quotes to intrigue you:

  • As James Surowiecki nicely puts it in the title of his best-selling book, it’s “the wisdom of crowds,” and it’s a glorious thing. Or it would be, if it weren’t for just one little problem: The effectiveness of groups, teamwork, collaboration, and consensus is largely a myth. In many cases, individuals do much better on their own. Our bias toward groups is counterproductive. And the technology of ubiquitous connectedness is making the problem worse.
  • What he glosses over, however, is the often spectacular way groups fail in the context of organizations.
  • Things only get worse when a team is charged with actually making a decision.
  • What’s more, these electronic group decisions can be even more brain-dead than in-person meetings. The biggest problem: the fear of dissenting is magnified in a Web, e-mail, or instant messaging exchange, because participants know their comments can be saved and widely distributed. Instead of briefly offending six people at a meeting, you have the chance to enrage thousands.
  • According to a recent article in The Guardian, every three seconds a Wikipedia page is rendered inaccurate–or more inaccurate than it was to begin with–by a hoaxer, ignoramus, or malcontent.

By the time I finished reading through the article, I wasn’t intrigued any more. And not incensed either. More like apathetic. So why do I post this? The usual reason. This kind of thinking will gain currency, especially when amplified by traditional media, unless people like us push back.

Freedman manages to take a swipe at collaboration and teamwork, at social software, at democratised innovation, the whole nine yards. Where he does give praise, it has the feeling of being faint and damning.

So here’s my take on Freedman’s piece:

1. Deciding isn’t doing

There was a book written many years called Five Frogs on a Log which first described this analogy to me. Five frogs, sitting on a log. Four decide to jump off. How many are left? Five, because deciding isn’t doing :-)

We should not confuse group decision making with groups working. Collaboration and consensus are two entirely different things.

2. Consensus comes in many forms

  • Simple Voting, with an agreement that the vote is binding on the group, no chads anywhere
  • Weighted Voting, same as Simple except that the boss is always right
  • Soft consensus, agreement amongst all those who cared or bothered to turn up
  • Filibuster consensus, agreement from all present in response to being bored out of their wits
  • Real consensus, where group members are given the opportunity to voice their concerns and misgivings in private, prior to a group decision that all will support in public

3. Real consensus is made far more possible by the use of social software
Patrick Lencioni, in his “fable” The Five Dysfunctions of a Team, makes the point far better than I could; all I am doing is placing the argument in the context of social software:

  • Unless there is trust within a team people are unwilling to be open
  • Unless people are open they won’t express their doubts and misgivings
  • Unless they express their doubts and misgivings they won’t feel they were party to the decision
  • Unless they feel party to the decision they won’t commit to it
  • Unless they commit to it execution is all but impossible

Read the original; my summary doesn’t do it justice. Lencioni’s arguments are very useful in the context of social software. Here’s an offbeat reason why:

In the past, firms had relatively low attrition. Job mobility was low, and gold watches and carriage clocks abounded. In such circumstances, consensus really worked. People voiced their misgivings openly, then went with the majority view after debate. Why? Because they knew that their peers and bosses would remember their sacrifice and their commitment to the team. Institutional memory.

Now, as we live with far higher attrition and job mobility, the institution has no memory any more. Sacrifices are not worth anything, people move, bosses move, there’s an I’m All Right Jack Every Man For Himself culture. Consensus is therefore hard to achieve, and infinite loops of decision-making and second-guessing and third-guessing abound.

In addition to the institutional memory problem (which is primarily about sacrifice) there is also a time-and-space problem. The infinite loop mentioned above is now able to move to new levels of infinity a la Cantor. People are often travelling and in different timezones.

Social software can be used to solve all this. It captures the context, helping absentees “get up to speed”. It captures the conversation, allowing concerns and issues to be aired and recorded. Institutional memory is therefore preserved regardless of attrition. It solves the space and time problems, along with more modern “unproductivity tool” problems such as Which Version Are You Looking At? and I Can’t See That On My BlackBerry and “Sorry, my signal is fading and I have to sign off”.

So Mr Freedman:

  • Deciding isn’t doing.
  • Consensus comes in many forms.
  • Social software aids consensus.

21 thoughts on “On social software and consensus”

  1. I’m not sure I’m quite with you on this one JP. Your Five Dysfunctions of a Team makes the argument for me…5 unlesses.

    I’ve come across many people who would love to blog but are genuinely concerned about the response from their bosses. Several I know do so annonymously precisely for this reason. They’re not anarchists or extremists but they hold views that would not be concordant with those of the firms for which they work.

    Remember the silly case of La Petite Anglaise? How stupid was that yet it happened.

    There are very real barriers in the workplace. The fact we should maybe focus on that is a matter of organisational focus. As we tech focused buseinss types say – change comes from the top empowering the rest. The trouble is it is so hard for the top to retain its focus on doing that.

  2. You’re right David, I didn’t make my point well enough. I felt that real consensus was rare and valuable. Everything else flowed from that.

    Also, I left the “prediction markets” aspects alone, they have been critiqued well in other microconversations.

    So thanks for your comments, they help balance mine.

  3. I must take task with Surowieki’s: “The effectiveness of groups, teamwork, collaboration, and consensus is largely a myth. In many cases, individuals do much better on their own.”

    Take the problem of solving a crossword. Although I find them rather tedious my mother loves them and as a child my innate competitiveness meant that I became good at the mechanics of solving them. When I visit my parents I usually end up solving the cryptic (although rather easy) crosswords in Times and Telegraph with her. We could both solve them eventually in half an hour or so (perhaps with 2 or 3 clues left) but because there are certain types of clues that I prefer, and ones that she is better at, when we work together we can often finish them in 5-10 minutes.

    This is far quicker than a simple doubling of manpower. The reason is obvious of course – everyone finds some clues easy but others are a stumbling block. With those stumbling blocks removed by someone else we can both concentrate on clues that are easier for us.

    Now instead of two people make it N people. For any clue there will be someone for whom the solution is obvious. Any crossword could be solved within a few seconds with enough people simultaneously looking at it. This of course is analogous with Linus’s comment about all bugs being shallow given enough eyes.

    Surowieki is probably American and probably has an MBA and has been brainwashed by his country’s culture that business is about “making decisions”. But let’s lay to rest this bizarre idea that teams are counterproductive when it comes to solving a task.

  4. Hey Matt, the comments you take exception to aren’t Surowiecki’s, they’re Freedman’s. Maybe I didn’t make that clear enough.

  5. “Any crossword could be solved within a few seconds with enough people simultaneousy looking at it” – well no it couldn’t actually, the solution would exist within the mass but the practicality the grid being filled by such a large group would militate against the doing of it.

    The crucial distinction, I think, is between actual groups or teams working on something and individual discrete contributions adding to the collective wisdom. These are different types of collaboration. The former can be very impractical as Freedman suggests, the latter is incredibly powerful in many situations.

    The key question is how to focus the power of the latter example on as many problems as possible and Freedman may have a point in suggesting that in current structures this isn’t always feasible.

  6. I’m tempted to say that groups working together are actually lots of individuals working together, and like typewriter keys, we have to find ways of avoiding jams. Crosswords are something I know a bit about, having had reasonable results in Cutty Sark and Langs Supreme competitions in the early 80s. I used to do the Times in less than five when I was good, but for some reason declined to around 11 when I was in competition.

    You can have many people work on one large crossword by having a moderator who fills a large visible to everyone grid and a fifo queue. Just an example.

    Emergence and wisdom-of-crowds techniques are not suitable for everything, no one has suggested that. But that is different from saying that collaboration doesn’t work. Or that crowds have no wisdom.

    Until we have thought police in operation, no one can force a person to respond a particular way to something that’s said or written. One of the biggest values of a brainstorming approach, which can be replicated in software by exposing output “little and often” is the serendipitous way other people make other connections that help solve the bigger problem.

    The failure of current structures has a lot to do with corruptions in hierarchy caused by unclear or unshared objectves, power craziness, silo thinking and a risk aversion. These were not caused by social software but by painful readjustments in the concept of the “firm” or “organisation”, coupled with less-than-optimal compensation structures. We have evolved into selfish people….

  7. Crosswords are a nice example because if someone solves a clue that I’m stuck on then not only does that mean less work for me, but the solution may give me first letters of clues which then enable me to solve them more easily. This cross-pollination effect of small tasks within a complex structure is more realistic model of how a team should operate so that it is genuinely more than the sum of its parts.

  8. I agree. It also has a “commit early and commit often” feel to it which is useful in untangling complex dependencies, genetic algorithms using wetware

  9. Sometimes there can be real value in going back to the old songs. In this case the singer was Joyce Rothschild-Whitt; and the song was her paper on “The Collectivist Organization,” which was published back inb 1979 in the AMERICAN SOCIOLOGICAL REVIEW. (Since I got to this through JSTOR, it is, unfortunately, not readily available for public consumption.) The paper examines several organizations that deliberately opposed the “rational” hierarchies of Weber’s model of bureaucracy. Regardless of what the abstract says, I do not think I shall be accused of being too reductive if I render the punch line as follows: “Sometimes it works; sometimes it doesn’t.” (Those happen to be the words that Merce Cunningham used to describe his impressions of what it was like to use chance operations in creating his choreography!)

    The REAL sin of reduction in this discussion seems to be that of assuming that “what works” works for ALL groups and ALL tasks. Now I am sure that none of this have intended to be quite so extreme; but, just to belabor the obvious, making strategic decisions for a business on the ropes is just not the same as solving a crossword puzzle (even one from the LONDON TIMES)! So I think the major contribution of Rothschild-Whitt’s study is the way in which she tries to home in on those settings in which “real consensus” is most effective; and my guess is that her conclusions in 1979 are still valid today.

    Reading this study reminded me of Andy Grove’s book, ONLY THE PARANOID SURVIVE, in which he talks about the value of “broad and intensive debate” when painful strategic decisions need to be made. Here are a few tidbits from this part of the book:

    “The more complex the issues are, the more levels of management should be involved because people from different levels of management bring completely different points of view and expertise to the table, as well as different genetic makeups.”

    “The debate should involve people outside the company, customers and partners who not only have different areas of expertise but also have different interests.”

    “This kind of debate is daunting because it takes a lot of time and a lot of intellectual ergs. It also takes a lot of guts; it takes courage to enter into a debate you may lose, in which weaknesses in your knowledge may be exposed and in which you may draw the disapproval of your coworkers for taking an unpopular viewpoint. Nevertheless, this comes with the territory and when it comes to identifying a strategic inflection point, unfortunately, there are no shortcuts.”

    “Don’t justify holding back by saying that you don’t know the answers; at times like this, nobody does. Give your most considered opinion and give it clearly and forcefully; your criterion for involvement should be that you’re heard and understood. Clearly, all sides cannot prevail in the debate but all opinions have value in shaping the right answer.”

    “It is important to realize what the purpose of these debates is and what it isn’t. Don’t think for a moment that at the end of such debates all participants will arrive at a unanimous point of view. That’s naive. However, through the process of presenting their own opinions, the participants will refine their own arguments and facts so that they are in much clearer focus. Gradually all parties can cut through the murkiness that surrounds their arguments, clearly understand the issues and EACH OTHER’S point of view. Debates are like the process through which a photographer sharpens the contrast when developing a print. The clearer images that result permit management to make a more informed—and more likely correct—call.”

    As you might guess, Grove’s book was intended to reflect on his personal experiences at Intel. Grove is no longer in charge over there, and things do not seem to be going very well these days! This does not been that Grove’s book is a valid or reusable reflection on his own experiences, but it should at least provide grounds for our own reflection!

    Finally, I need to voice a bit of concern about this whole idea of “institutional memory,” particularly since I was involved with a HICSS “mini-conference” this this topic for several years. I believe that it extremely important to recognize that an institution’s “memory” (whatever that may mean) does not reside in the records in collects and archives; it resides in the ACTS of the people working in that institution. (I believe that the best way to read Grove is as a reflection on ACTS, rather than on the RESULTS of those acts.) Thus, I fully agree that any failure of institutional memory can be attributed to the volatility of that institution’s work force. However, I do NOT believe that social software can confront this problem, because a RECORD of a conversation can not, nor will ever be, a satisfactory substitute for ACTS OF CONVERSATION.

    I have said this before, but I suspect it cannot be repeated enough. Life is a matter of both nouns and verbs. We have been very good at coming up with technologies that facilitate the management of the nouns in our lives. Unfortunately, we still do not have as good a handle on managing the verbs; but, with a deliberate pun intended, the verbs are where the action is. Whatever our technology does, what WE DO will always be more important. I am frustrated that we have not come to grips with this simple principle; but I am hopefuly that, as long as we have conversations, we shall eventually become more comfortable with it!

  10. I was not saying that collaboration doesn’t work or that crowds don’t have wisdom – quite the opposite in fact. I was seeking to highlight that humans get in the way of it just as much as they enable it.

    As for the crossword example I highlighted – I was merely questioning the “in seconds” element of the claim – even using a moderator it would take much longer than that even if the solution was potentially available within seconds. But your admisison of expertise in this field now allows me to cheekily suggest that you acting alone could complete a puzzle far quicker than any group of ten of my friends working together. In certain cases the wisdom of the individual exceeds that of the crowd.

  11. Humans can and do get in the way. But I feel that with social software, we can find ways of optimising this process.

    If I take the crossword example, and extend it a bit, see what you think of it.

    Today I can complete the Times crossword faster than an average ten-person group. Probably.
    Tomorrow I start explaining to people how to do the crossword, taking a leaf out of Anthony Grey’s wonderful Crosswords from Peking.

    Over time, many heads will be better than mine. I will stop being a bottleneck. And as crosswords scale up, and grow from 15×15 grids to global multilingual 225×225 grids, there will come a time when I don’t scale.

    But my group will scale. Especially with the context and learning and sharing that comes with social software. [So that’s how you do it. I don’t get that, can you run that by me again? hey guys, this sounds silly, but do you think 13 down is syzygy? etc etc]

  12. No argument from me when it comes to the potential of social software to solve “crosswords” because they are characterised by having a definitively correct solution. Thinking off the top of my head, however, I wonder if other problems that require original creativity (for want of a better expression) may not be so susceptible to large-group solutions. I’m sure you’ll have a riposte to that.

    Maybe I’m just a little jaundiced by the frequent propagation of mythical opinions and “expertise” across the blogosphere where ranking hierarchies clearly lend excessive authority at this stage of the blogosphere’s development to some who blogged early. Groupthink exists within social software too.

  13. I’ve always considered the “wisdom of crowds” to be more related to aggregated knowledge, and less to the ability of crowds to create something meaningful like a Wikipedia article. For example, on the show “Who Wants to be a Millionaire”, I believe the stat is that the crowd was correct about 75% of the time, which makes it by far the best way to make a decision. However, try to get that same crowd to write an article and it would be all over the map.

Let me know what you think