icm2re logo. icm2:re (I Changed My Mind Reviewing Everything) is an ongoing web column edited and published by Brunella Longo

This column deals with some aspects of change management processes experienced almost in any industry impacted by the digital revolution: how to select, create, gather, manage, interpret, share data and information either because of internal and usually incremental scope - such learning, educational and re-engineering processes - or because of external forces, like mergers and acquisitions, restructuring goals, new regulations or disruptive technologies.

The title - I Changed My Mind Reviewing Everything - is a tribute to authors and scientists from different disciplinary fields that have illuminated my understanding of intentional change and decision making processes during the last thirty years, explaining how we think - or how we think about the way we think. The logo is a bit of a divertissement, from the latin divertere that means turn in separate ways.

Chronological Index | Subject Index

The benefits of disagreement

How to give your team an antidote to groupthink

How to cite this article?
Longo, Brunella (2016). The benefits of disagreement. How to give your team an antidote to groupthink. icm2re [I Changed my Mind Reviewing Everything ISSN 2059-688X (Print)], 5.11 (November).

How to cite this article?
Longo, Brunella (2016). The benefits of disagreement. How to give your team an antidote to groupthink. icm2re [I Changed my Mind Reviewing Everything ISSN 2059-688X (Online)], 5.11 (November).

London, 4 March 2017 - In my forthcoming book 99 STARS, a collection of case histories on competencies and skills for the digital age, I included the case of a training programme I designed, among other courses, in the late 1990s and kept tuned and slightly reviewed for each edition until 2003-2004. It was aimed at officers working in libraries, enquiries or "ASK-A" and customer services.

The originality of the course consisted in a new approach to what is known as triage process: I managed to transfer to the participants what I had learned applying Herbert Simon's theory of problem representation and administrative behaviour, besides my own over ten years of insights into bayesian logic for human cognition and information retrieval systems (I had acquired such skills - 'hands on' techniques, tactics, databases design, language tips and interactions with and feedback from users about the best ways to extract intelligence from structured and unstructured collections of data - through my former information specialist experience).

More than ten years later, some of the lessons learned from that short course (lasting approx 16 hours distributed in just two weeks time) have become matter of academic experiments and literature on social learning.

Talking about public policies design and governance of relationships in computer mediated environments my mind goes often back to one of the exercises I designed for that course, the title of which was 'The top ten of disagreement'.

It turned out that 'The top ten of disagreement' was and I believe is still is a very interesting type of antidote to groupthink, an extraordinary simple and immediate device for governance of relationships in digital environments.

Teamwork and the online issues desk

The course allowed the participants to acquire practical procedural knowledge and confidence useful to deal with informational problems, particularly the most difficult ones, in a short timeframe: informational problems are managed and solved rapidly and smoothly as long as they are very simple and very routinised. One of the aims of the course was to help the students to acquire a method so they could recognise ill-defined or not very well structured problems.

The 'The top ten of disagreement' exercise showed the participants, with their own words, that how we analyse and categorise information problems is indeed an inextricably subjective experience and yet also the crucial moment of a whole organisational process we want to engineer up to a point in which we can manage it efficiently and in consistent ways, no matter who is at the issues or front desk, and ideally even independently from the context and from individual personalities and attitudes. Later attempts to automate digital interactions (with conversational interfaces) or extract insights from big data repositories (for instance for personalised medicine) have been trying to replicate that ideal setting for more than a decade now.

How we look at and interpret the formulation of a problem determines an entire course of action in a triage process, including of course the final perceived quality of the responses given by the organisation, the way in which we communicate and interact with each other and care for user satisfaction, and the same possibility of providing a solution or a service.

I must say before going on talking about this case that unfortunately, in its original and very effective formulation, the method I taught through my 'Top ten of disagreement' exercise does not scale easily even when it may seem good enough for a variety of applications.

It was exceptionally fit for purpose with two or three dozens of students, but it would not be so manageable and effective with more people working collaboratively in the same virtual office or organisational unit. I tried to redesign it and to imagine its integration within large scale 'big data' governance experiments. It turned out to be not only counterproductive and substantially unreliable but even uneconomic, as well as many other procedures and algorithms model constructed around inference rules that work very well in small groups, where the participants usually share some decisive characteristics, but then turn totally nonsense as soon as the size of the group simply double.

But let's say more of what it consisted of. The exercise was offered in a light format as a quiz: it presented a dozen of real enquiries captured from logs of various library and other commercial services available in the public domain. The students had to assign each enquiry to one and only one of the three typologies of information problems explained in a short lesson on the theory of problems representation adapted to informational needs.

The analysis of the problem formulation was considered as the first necessary step needed in order to define an optimal and standardisable (consistent) way to find a solution (what we called problem solving strategy).

The simpleness of the exercise format was in itself disorientating - for good - because people that work in enquiries, ASK-A and customer services (even without adding the speeding factor of the digital media) are usually working within efficiency targets (such as not more than 10 or 15 minutes per enquiry and so on) and tend to immediately rush into a mode of mental shortcuts towards the solutions, to think about the tools they master or the best sources to be used or to speed up the communication with colleagues and users and reduce any brainstorming or reflection time to a minimum.

During their virtual classroom experience, the participants did not need to provide any answer nor they are expected to guess what the correct sources should be. Instead, they just had to concentrate and decide how much the formulation of the problem was defined and structured following the instructions given in the lesson.

What I cared about with the design of the exercise was not to influence the students' analysis and decision making process. I did not want to expose them to mine or others' judgements of the problems formulations before they could go through their own analysis in their own time: whatever intuition, prior belief, notion, value, behaviour or practice should possibly stay out of their mind, so they could open up their attention to my short lesson on Herbert Simon's guidance on problem solving.

Everybody would assume mine were the "correct" answers and of course these were given as the correct ones. Therefore these were disclosed to each participant only after the exercise was done. When they would have instinctively looked around and asked their colleagues "what you think of this", they were reminded they had to read the lesson before taking the quiz: not knowing other students' choices nor my own recommended answers was part of their learning goals.

So there was, all in all, an implicit pressure to social compliance as in any educational setting and at the same time a quite decisive demand for cognitive dissonance.

Policies and criteria they would have applied sifting enquiries in their own usual workplace were not very pertinent to the task given by the exercise. The usual mental tricks and shortcuts or biases they would have applied did not easily filter into the quiz experience or at least the interference of the participants' context would be very limited (this would be usually disclosed and become explicit during the follow up discussion in the virtual classroom as part of a final de-briefing stage for each exercise).

In sum, in this more than in any other of my online courses, to concentrate on the wanted learning outcome - strategies for information problems formulation - we had to prevent groupthink and the other typical nasty teamwork poor performance problems I mentioned in previous articles of this short series on governance of relationships (see icm2re 5.6 - 5.10, cascade effects, polarisation, social compliance and so on).

Immediate to understand by everybody, the measurement of the level of disagreement on the way we categorised the enquiries did not mean at all that this was an absolutely scientific or objective indicator of the group cohesiveness. Neither that group members were completely immune to external influence, or not responsive when exposed to the first partial results of the Top ten, providing evidence of the probabilities they could indeed agree with some others participants and with the tutor. For this reason I was used to disclose and comment the data about the disagreement within the class only after one or two days, when the majority of the participants had familiarised with the whole experience, had already taken the quiz or an unusual low level of participation called for encouragement.

The awareness that there was disagreement and this could be particularly significative for some types of problems' formulations and not for others lowered the risk that people adapted their beliefs to conform with the opinions of others. There was always the possibility that somebody overlooked their own usual ways to proceed and experiences and get stuck into their first "gut" interpretation of some of the problems' definitions but the majority was keen on seeing the exercise for what actually was: an opportunity to analyse the many facets of the ways in which an information problem can be formulated and dealt with. Such opportunity, in turn, would speed up the possibility of an alignment towards an efficient way to treat a problem, to ensure consensus on how to handle the process and to find a solution in consistent, replicable ways.

Lessons learned

In their book Information Rules (1999), Hal Varian and Carl Shapiro warned digital entrepreneurs and researchers about the dramatic changes caused by the attraction of information policies within the public sphere, a phenomenon worsened by public engagement with governance processes through digital technologies.

Nobody really knows how to prevent and to manage social compliance and political polarisation in computer mediated environments.

We know something: for instance that some basic social norms (affecting privacy and other human rights) are pretty much the same in both physical and digital social environments, that in the physical world crowd dispersion is a very effective measure to control social risks whereas in the online world the opposite seems true (large numbers of accounts are needed to reach break even for efficient functioning of platforms and processes - for crowdsourcing, streaming, social networking, cyber security and so on).

But overall there is still no general understanding on how to prevent and avoid the dysfunctions and unintended consequences of groupthink in the online environments. My hypothesis, thanks to the successful experiment of this exercise, was that size matters as well as management of people interactions in promoting and fostering meta-cognition and self-reflection as the very ultimate antidote to groupthink.

Part by mastery of critical core skills (information retrieval, data literacy and critical thinking) and part by genuine accident (inventing new means of communication and sharing data among virtual teams), my instructional design experience with "The top ten of disagreement" in the field of information problem solving was an extraordinarily effective way for me not only to teach how to handle enquiries services with remote users in efficient ways but also to get a grip of the limitations and opportunities of computer mediated environments for teamwork.

I saw the risk of involution of a way of working that tends to produce very easily the evidence needed for social compliance, by way of replicas that inhibit the selection and use of the right data at the right time by all the intended users. The potential for developing dysfunctional and unethical influence and censorship and self-censorship mechanisms is enormous.

The online environment seen as a reliable ecosystem with self organising capabilities for problem structuring and problem solving at scale that just need to be mined (through big data applications) is an extraordinary illusion that pleases marketing and trade practitioners, it may create entertainment and an infinite sequel of commercial communications but it risks to complicate public life, to ignite radicalisation, to feed trolls and fraudsters, slander and scams, pillories and disinformation campaigns, idealising the idea that the relevant evidence is produced - mostly anonymously and with altruism - with a revolutionary bottom up approach opposed to the traditional top down dissemination of information used by mainstream publishers. These simple and polarised ways to look at social interactions and communications do not help online governance of relationships and are terribly misleading.

We can teach everybody, not only people working at enquiries and issues desks, an antidote to groupthink and that is metacognition and critical thinking.

Unfortunately, the pressure of shareholders and public stakeholders, the tensions and frictions between polarised and conflictual views among teams working for regulators, governments, political parties and non governmental organisations has determined in recent years the success of what has been called a tribalist style of team management that points towards the outcome not the process: cohesive groups of people play games in restricted online cultural environments.

Communications, behaviours, reactions may be magically aligned and extremely efficient but for leading groups and audiences towards precipices of stupidity and uselessness.

The online working platforms - from intranets to all sorts of collaborative applications and even social media - are socio-economic computer mediated environments in which efficiency and alignment are generally sought with little space left to individualism, creativity or experimentalism.

Promoting awareness, understanding and acceptance of disagreement is an extraordinary powerful way to counterbalance such "native" flaws in the design of collaborative platforms and processes so far.

Learning how to represent, restructure, prioritise and ultimately to solve problems within the computer mediated environment gives an antidote to confirmation bias and groupthink, particularly if we see there are different ways to approach the formulation of a problem. It also opens up spaces for people to come to term with their own emotional overreactions, cognitive failures or poor performances that are absolutely normal.

However, a positive appraisal of the benefits of disagreement does require design of minutia, detailed attention to microscopic interactions, incredible amount of time and energy that seem at present incompatible - as mentioned - with the expectations created by machine learning and algorithmic framework at scale.

For a number of reasons many corporate leaders and consultants have exploited tribalism tendencies in recent years as they are easy to spread through remote teams and social media: ethnic and cultural identities create the immediate glue needed to forge an expectation of unconditioned agreement and cohesion. Trust builds up quickly where there is only the appearance of understanding and collaboration, leveraging on a constructive attitude and the willing for an immediate return in terms of engagement and sense of belonging or friendships and recognition.

The rhetoric of the "one team" can be an extraordinary trove for propaganda, for efficiency targets and for delivery but also the worst possible narrative for the organisational culture. It can easily transform the interactions of the most creative ensemble of people in a coffin for innovation and creativity, leading straight towards groupthink.


I do not know in how many contexts and projects 'the Top Ten of disagreement' exercise can be easily translated and adopted but I would not be afraid to try it anytime there is a suspect of groupthink attitude taking over.

In recent years there has been an increase in the number of laboratory experiments by behavioural economists willing to study and understand more about social learning dynamics, contagion and political polarisation, not only within computer mediated environments and not only for information sharing, advertising or for policy making.

The entirety of the digital world - from education to health services - increasingly relies on word of mouth, viral campaigns and informational cascades and positive externalities: these dynamics are in turn also the sources used for inferences about people profiles and personalities and recommendation arranged by "deep learning" algorithms with very arguable benefits.

Looking at the disagreement about the way we read, we listen and we understand the formulation of a problem and measuring such differences in objective terms makes the fixity element of group thinking dynamics (as I defined it in 5.8) explicit, the automatic connections and biases more likely to emerge, the distorted ways to deal with privacy or confidentiality matters less marginal or instrumental.

I do not have any convincing way to prove its efficacy, but it believe the "Top ten of disagreement" is still a very good device I can think of for an antidote to groupthink in a controlled, organised environment.