The quality of quantity

file000401942226As a scientist, Francis Galton had two main interests. Mental abilities, and breeding. He was convinced that both were needed to maintain strong societies. His studies made him realise: ‘Stupidity and wrong-headedness of many men and women being so great as to be scarcely credible’. The solution was, according to Sir Galton, to give control to a selected few, but brilliant individuals. The hope for society was to keep the degenerate stock at bay.

One day in 1906 Sir Galton visited a country fair in Plymouth. There, his attention was drawn to a weight-judging competition. An ox was on display. Bypassers could place wagers on what the total weight of the ox would be (after being slaughtered and dressed). 800 people joined the competition. Among the participants were some that would be more knowledgeable about oxen, experts if you will, for example butchers and farmers. But the crowd mostly consisted of lay-people who had no or very little knowledge of what an ox normally weighed. These would have to be informed by anything they might have picked up in a newspaper, in discussions with friends, or whims.

The analogy to democracy was instantly clear to Galton. Each man one vote. In a paper about his experience (published in the scientific journal Nature) he wrote: ‘The average competitor was probably as well fitted for making a just estimate of the dressed weight of the ox, as an average voter is of judging the merits of most political issues on which he votes..’ Sir Galton probably saw a wonderful opportunity to study (and show) how little the average ‘voter’ was capable of.

After the competition, Sir Galton borrowed the tickets to calculate the average guess of the group (total guessed weight divided with the number of participant). This way he could conclude what the collective wisdom of the group was, as if it had acted as one person.

A few well-educated and informed people, and a lot of ignorant and misinformed people. Galton was convinced that the result would show that experts were better at assessing the weight than the total group. But no.

The average guess was 1,197 pounds. The slaughtered and dressed ox had weighed 1,198 pounds. The group’s collective guess was as good as perfect.

Perhaps expertise was overrated. Perhaps keeping the whimsical masses at bay was not the solution. Galton later wrote “This result is, I think, more creditable to the trust-worthiness of a democratic judgment than might have been expected.” In this case, no single individual guessed more precisely than what the group as a collective did.

The book “The Wisdom of Crowds: Why many is smarter than the few” by James Surowiecki starts out with the above story. In a world where individual talent is put on display, experts consulted for advice, and leaders looked to for direction, suggestions about trusting the masses are almost heretical.  But clearly and convincingly, Surowiecki points out the often underrated potential of groups to make well calibrated assessments in situations of both high and low levels of certainty. Under certain conditions of course (independence, diversity, and decentralisation). Which is the point of the book: to explain why some groups sometimes go awry, and why sometimes groups will beat an expert ten times out of ten. I highly recommend it.

In safety, a select few traditionally decide what is the right way forward. It has been for the crowds to follow such prescriptions. Should anyone deviate, an elite group often gets to specify what the consequences will be. Unless you have been put in a safety managers role, organisations traditionally have little interest in how your mental abilities can contribute to organisational success other than following directives.

However, as Shane Durdin pointed out in a previous post: People are not the problem, they are the solution. The role of safety (professionals) need not be to keep the ‘degenerate stock at bay’, about reminding and controlling, or removing discretion for local decision-making. The role could be about creating the conditions that unleash the human potential to be the best they can be.

Several posts on this website illustrate the potential of such a different way of organising safety. Read for example about Newmont Mining’s approach to improving safety, or the Shared space traffic solution in Holland. And my friend Eder Henriqson told me about a Brazilian company who regularly asked their employees: “What is our next accident going to be?” Not only did this increase the employees’ hazard awareness, but the company also gained insights to issues they did not know existed.

Do you know of other ways to engage the collective intelligence in organising safety, or organisational life in general?


  1. Rob Long Reply

    Daniel, you have hit the mark with Galton but I more admire him for his contributions to mathematics. His discovery of ‘regression to the mean’ is perhaps the most neglected of all concepts in the safety industry. Kahneman re discovered it in his excellent work Thinking Fast and Slow. This constant distortion of numbers in the industry is crazy, so much is made of so little and so much is neglected in the conditions and social arrangements that affect risk. We see all this definitive speculation about cause attributed to data and absurd cultural causality connected to injury data, it is astounding. The nonsense of Bird and Heinrich continues as if injury data is a cultural indicator. In the ACT where I live, the most amazing causality has been connected to injury and fatality data recently in the Getting Them Home Safely Report yet no-one questions that the data as mis-attributed, the lemmings just all assume that such data is a cultural indicator. Without Regression to the Mean this mis-attribution of data makes sense but if you understand Galton it is absolute nonsense.

    1. Daniel Hummerdal Post author Reply

      Andrew Townsend (another contributor on this site) has pointed out the widespread statistical naivete of the safety community (have a look at his earlier posts here).

      Perhaps there is opportunity to write some critical posts providing questions and alternative perspective on the usual stats?

  2. John Culvenor Reply

    Go Francis Galton. I think the crowd just got lucky. Anyway…

    I have studied in controlled designs safety solution creativity (thinking them up) and solutions discrimination (which ones are best) with variables of skill (creativity skill v safety specific) and also group size (team v individual).


    Creativity expertise is an advantage in solution development but safety expertise is not.

    However safety expertise is an advantage in safety solution discrimination. That is, recognizing the good ones and therefore knowing which ones to use or develop further.

    With regard to teams, consistently research has shown teams are poor at creativity. The sum of individual efforts are consistently superior. But rarely had teams been assessed on discrimination. So I tested teams v individuals on safety solution discrimination. What I found with team judgements to safety solutions is that the team judgement was better than the mean of the individuals in the team.

  3. John Culvenor Reply

    Each time we need to make a decision, can we get 787 people to contribute? Sometimes, but not often.

    If Galton had compared the average estimate of the small number of novices and a small number of experts (e.g. butchers) I dare say the experts would win.

    So the novice theory presents as being usable if you can muster (nice pun eh) enough people each time you need to make a decision. Otherwise using expertise (more knowledge in one head) is probably going to be more economical.

  4. Eder H Reply

    Hello Daniel! Interesting post. In fact, this “collective intelligence” needs to be thought beyond the traditional “stock” of procedures designed by a few planners. Organizations need to learn about real working constrains and how to improve individual and collective intelligence to cope with risks intrinsic of normal work. “What is going to be the next accident?” may be a way for both understanding the present and prospecting the future. Thanks for the post.

  5. Les Henley Reply

    Hi Daniel,
    I’m prompted to ask – hasn’t our (harmonised) WHS legislation taken this whole issue into account by requiring employers to consult with those affected workers?

    If wefail to do this, as safety professionals, maybe in the belief that we know better, then we are immediatley in breach of the legisaltion we are attempting to comply with.

    I have long held the persepctive that those who do the job have a better chance at finding the safe way than I who have never done the job.

    As a safety professional I rarely present myself as a technical expert in finding the ‘safe way’. Rather my technical expertise is in demonstrating the various tools of risk management and leading those affected employees to work it out for themselves using those tools.

    But, due to my wide exposure to a range of industries, products and processes, I do have a knack for thinking outside the box and generating creative, yet simple, solutions that can be adapted by those who need to do the job.

    1. Daniel Hummerdal Post author Reply

      Hi Les,
      Consultation is likely a step to improving decisions by engaging more diversity (or at least i suspect that was the intention).

      In your opinion, does this consultative requirement work in practice/do any good?

      1. Les Henley Reply

        Hi Daniel,

        Like any new ‘skill’ it takes time and a good facilitiator to enable inidividuals and groups to establish effective participation.
        The first barrier that needs to be overcome is existing/past culutres where workers were not engaged – the ‘boss’ tells them what to do.
        This can be broken down by encouraging participative questions suchs ‘what do you think we should do?’

        The second barrier is limited understanding of risk management principles. I break this one down by engaging a whol eteam – supervisor and subordinates – in a workshop looking the difference between hazard and risk (hazard identification), understanding likelihood and consequence (risk assessment), worst case and most likely case risk outcomes, the hierarchy of controls – I use this to help generate a variety of control options, ‘can you think of an option for each level in the hierarchy?’
        Option selections via cost benefit analysis – does the option change the level of risk, what are the phsycial and finanacial costs of each option and what would be the benefits.
        I have found that this approach has resulted in a significant downturn in workers compensation injuries – reduction in numbers and reduction in severity.
        I the last 2 years, despite $6M increase in wageroll, workers comp premiums have reduced by almost $300K.
        So yes I believe it works and does good.
        We haven’t reached (probably won’t) end game of nil incidents yet but I’ve recently had to ask my boss to broaden my role beacause I’m no longer fully utilised looking after just WHS & IM.

Leave a Reply

Your email address will not be published. Required fields are marked *