The size of groups and the role of members

Card's author : Jean-Michel Cornu
Card's type of licence : Creative Commons BY-SA
Description :

Small groups of up to twelve persons

A cognitive limitation of man concerns the size of the group in which he can, without the help of tools, understand what is occuring. The human being is first of all an animal which can enter into alliances, i.e. "an union between people resulting from an agreement or a pact 1". If many animals can live in herds or packs, very few can choose by themselves to enter into an alliance. Great apes and some cetaceans manage to enter into alliance with up to three individuals, but we humans are limited to twelve2.

For more information: the limit of twelve for human groups

The British anthropologist Robin Dunbar became interested in the relationship between the size of the neocortex of 38 species of monkeys and the size of the respective groups in which they
lived 3. Surprisingly, he has found a correlation between these two elements. He then extrapolated this approach in humans to conclude that the natural limit to the size of a human social network was 148, a number that usually rounded to 150, is called the "Dunbar number" . This number corresponds to the size of the breeders-farmers' villages of the Neolithic, and is still found today in the size of social networks.4. this number – considered by Dunbar as rather approximate – determines the number of persons whith whom we can easily socialize without tools (these tools can be for example the list of friends on Facebook, or simply our adress book, which eanbles us to get in touch with much more people than we can even remember...)5.

The confidence which allows to create alliances requires however to have not only an overview of the various members of the group but also the links between them. We talk of holoptic approach6 (from the Greek holos, whole and optikós, related to sight) in opposition to the panoptic approach7 ( from the Greek pan, all) which enable to see all the persons but not necessarily the links between them.

So, even if chimpanzee have a number of Dunbar around 55 which allows them to maintain packs of this size, they can only enter into alliance with three. Mankind, besides having a high number of Dunbar also has a capacity of holoptism which allows her to create alliances with a dozen people. The maximal size of this alliance corresponds to 144 links between people (by taking into account simultaneously the people themselves and the differences in the link between a first person and the second, and the mutual link of the second towards the first one). So, besides his capacity to constitute a social network of about 150 people (what corresponds to the size of the breeders-farmers' first villages of the Neolithic), mankind is also capable of entering into an alliance which allows her complexer collective actions up toapproximately a dozen people8.

We don't know thus how to cooperate normally in groups of more than a dozen people. To go beyond, we had to develop strategies: set up a hierarchy so that the leader manages at the most a dozen second-in-commands who themselves shall manage a dozen people 9 ; or else have representatives (of God or of the Peolpe) which allow us to focus on a few persons according to a centralized star-shaped structure ; or even trust a single mechanism of exchange in the group - money - rather than having to know every person and every interaction between them. But could we exceed the barrier of twelve to benefit directly from the collective intelligence of a largest number without having a hierarchy, representatives or monetary mechanisms as intermediaries?

The astonishing principle of 90-9-1 in groups over twelve

Beyond a dozen participants, we cannot follow any more the entire interactions in the group. It becomes easier thus for a member of the group not to participate unless it is noticed. If in a small group, participation is a standard and non-participation an exception, in a big group on the contrary, only those who decide to participate do it.

But those who participate are not always the same. We to get involved a lot into certain groups and not into others, according to the interest we have in the group. If the number of people who are active seems to us too low, we have a natural tendancy " to supersede". If on the contrary, more people than what seems necessary are already at work, we tend to remain inactive, even become it if we were active. This explains a very counter-intuitive rule: whatever the peolpe in a large group, the percentage of active people stays even, according to the 90-9-1 principle10:
The proactive persons who take initiatives are between one and some percent.
The reactive people who react when asked are between ten and dozens percent.
Others are not all totally inactive. Some are " observers11" who follow the works of the group, use them for them, even if they do not participate. So, there is a whole gradation in the more or less active roles that a participant can play, allowing him to get involved more and more or less and less.

The percentages observed in the existing groups confirm well the principle of 90-9-1. This rule has curious implications. Let's imagine a group of hundred people. We shall have thus naturally at some non-active people, decides to exclude them to concentrate on the small group of about ten active persons. His new group will keep not the same active persons but the same percentage of active persons which plumet to... approximately one. He will well end up alone ! On the contrary, let us take a group of about fifty people. To exceed five or six reactive persons, it is necessary to make quite a lot of efforts. Let's imagine that this time we add about fifty other people even less concerned and thus who may remain rather inactive. We observe that certain people who were inactive, including in the initial group, become more active to keep the same percentage of active persons in the group. Reactive people come up to a dozen...This surprising behavior is well verified on the spot: we have generally a rough understanding of the number of members of the group who allow some to choose to become active or inactive.

The number of participants structures the groups

A group working normally will thus have approximately 1 % of proactive people and 10 % of reactive people. It will be necessary to make efforts to increase the percentage of reactives to 20 %, see up to 40 % in very exceptional cases. In order to have a big group producing as much as a small group of a dozen people without requiring too much efforts, the group will have need at least a hundred participants to have at least about ten or dozen reactives.
Between twelve and hundred participants, we are thus in the case of an intermediate group: too big to allow us to keep the pace up with all that's happening and hence to manage it in a constraint way , and too small to do as well as groupe of a dozen persons without requiring major efforts of animation. Beyond a hundred contributors, it is possible with a reasonable investment, scale up and then have a group with over twelve active people Au-delà de cent participants, nous pouvons avec un investissement raisonnable, "passer à l'échelle" et avoir un groupe dont le nombre d'actifs dépasse la barrière des douze, on conditions that we take into account the reactive behaviors (at least 10%) and not only pro-active ones (at least 1%). There is also a high limit: beyond a thousand people, organizers and other proactive persons which undertake some tasks of management, are themselves a group of over twelve, jeopardizing the coherence of the group12. A group of several thousands people seems then more complex to manage. The French-speaking network of botanists Tela Botanica implemented intermediary members to identify better the proactive persons and started to constitute a group so they could exchange between them. Beyond, in groups of several dozen thousand members, the number of proactive persons soars, exceeding a hundred and allowing other forms of regulation and a centralized and collaborative governance. Some very large groups exist where management is not done by constraint but by opportunity. It happens for example with wide online projects such as the various linguistic versions of Wikipédia encyclopedia or else the collaborative mindmap. The understanding of what eases the implementation and the development of such big groups is still unclear.

To know more about the subject: for proactive persons, Internet is divided in three

When you wish to work with a large group which stays limited to few hundreds, even one or two thousands, it is imperative to seek to work with the reactive people (ten to up to some dozen %) and not limit yourself to the proactive ones (one and some %). on the internet, the difference.On the Internet, the difference is reflected in the notion of push and pull tools.

A pull tool is a tool which obliges to "pull" information from where it is. This is the case for conventional websites but also forums and major web 2.0 tools for which we must be proactive to get their information. On the contrary a push tool is aiming bring in (to "push") information to us, or more precisely to the tool we consult everyday. In our everyday life, its the answering machine (with a pull tool, we would have to question each of our friends' or boss' voicemail to see if there is not a message left for us or for the group... It's also like that with our letterbox that we check regularly and where our letters are sent. We then just have to "react" to what we received.

In the case of Internet, the push type app is mail. To work with several other people, est le courrier électronique. Pour travailler à plusieurs par exemple, mailing lists enable to exchange directly in each others' mailbox, without forcing people to go proactivelyy on the group's website. But there are now several sites which we consult regularly, Facebook, Twitter or other social networks. One of the major difficulties of working together with a great number of people is that we can't check everything sytematically: letterbox on the way back home, answering machines and voicemails, private and pro mailboxes, Facebook and Twitter professional or private accounts. More and more people only check regularly their mails, Facebook or Twitter, sometimes two of them. In terms of push tools, and so in a reactive approach, the internet is then divided in three, even if it is still possible to seek proactively information through channels we use less regularly.

In companies, there is often a privileged channel. For example the use of mail is compulsory and it is then possible to push information directly to the different employees. In this case, and to prevent proactive people from being frustrated - being the more motivated even if ten times fewer than the reactive people – it can be interesting to allow push as well as pull methods. It is possoble to associate a forum and a mail to get the advantages of a mailing list and the pull tools: when a new subject is posted on the forum, most contributors get it by mail. Then they just have to email back to and their answer will be on the forum. Those who wish to adopt a proactive approach but avoiding bllocking their mailbox can go straight on the forum to read the topics, other people's contributions and then contribute. According to the number of participants, and to avoid drowning those who receive the information by mail in too many messages, it is possible to adopt a reactive approach on all posts for the majority of the group (for groups limited to several hundred people)13 ; or to send emails only on former questions, or to receive by mail only the initial questions, a selection of contributions prepared by the managers and summaries of discussions for larger groups. Those who wish to get the details of all contributions must then get the information proactively on the forum14. Ideally, the choice of receiving all contributions or only important mails by push in his mailbox (questions, summaries, invitations ...) should remain the choice of participants, regardless of the group's size.15.

In the case of large groups bringing together people from different backgrounds (individuals, various organizations ...), when selecting a tool push, e.g emails or on the contrary Facebook, part of potential contributors are excluded. To avoid this, one must be able to get information and contribute through the channel he regularly uses. This tool which will enable to receive all exchanges or only initial questions and summaries through a chosen channel (Facebook, Twitter) and to answer simply directly with this tool, is still to be developped. This tool must also enable those who wish it, to get proactively contributions on a forum type tool and even contribute from it.

The importance of large groups of 100 to 1,000, based on reactive people

Although very large groups now represent a new horizon showing that it is possible to work with thousands of people and maybe more, large groups of a hundred thousand people are of particular interest for two reasons.

Before beeing very large groups of several thousand or even tens or hundreds of thousands people, groups start with only several hundreds members. It is therefore important to understand the way large groups operate to allow the emergence of very large groups. Besides, many subjects have not for vocation to gather thousands of people. Even if it is necessary to increase - sometimes a little artificially – groups of several dozens persons to exceed a hundred, it's not always always possible to increase all groups beyond several hunders or thousands of people. The work groups of the Internet Engineering Task Force (IETF) which develop each standrads for the internet are typically of some hundred people. Same for the different groups to which the project Imagination for People provides support as a partner and which are interested in identifying and supporting projects for a particular side of social innovation (Fab labs, third places, currencies, innovation in the South, energy, group management ...).
However, these large groups require to take into account the particular reactive and not just proactive persons who in this case are not enough.

In brief

Once a group exceeds a dozen members, each person takes a proactive reactive, observer or inactive posture, and can switch from one to another according to various criteria. We observe in a rather counterintuitive way that the percentage of active persons remains outstandingly even (90-9-1principle): proactive people are between one and some per cent and reactive people between ten and dozens per cent.

We can deduct from it that groups can be identified by the number of members:
  • Small groups up to twelve persons who can be managed in a constraint way (While waiting for an action of each of the various members) ;
  • Intermediate groups between a dozen and a hundred people who require more efforts in management to obtain reactions ;
  • Large groups between a hundred and one or several thousand people who enable to produce collaboratively... under conditions to focus on reactive persons ;
  • Intermediate very large groups of several thousand people among where the proactive members's group is hard to keep coherent ;
  • Very large groups over dozens of thousands people where proactiove persons are numerous enough to make management less constrained ;

Large groups between a hundred and one or several thousand people are of particular interest: they are a must for groups who are likely to become very large, and mostly they are a size that corresponds to the number of people which can be gathered around many specific topics. But they need to take a particular care to members who act proactively (they can be approached in online systems with push tools such a email, Facebook or Twitter rather than pull tools as the web or forums) and not only to proactive people who are not numerous enough.

Mot clé: #taille28

1 alliance. Wiktionnaire [online]. [Accessed 30 January 2014]. Available from:
2 CORNU, Jean-Michel. Donner : une capacité naturelle mais limitée. In : Tirer bénéfice du don: pour soi, pour la société, pour l’économie [online]. Limoges, France : FYP, 2013. Stimulo, ISSN 2265-7754. ISBN 978-2-916571-87-4. Available from:
3 DUNBAR, Robin. Theory of mind and the evolution of language. Approaches to the Evolution of Language. 1998. P. 92–110.
4 GONCALVES, Bruno, PERRA, Nicola and VESPIGNANI, Alessandro. Validation of Dunbar’s number in Twitter conversations. arXiv preprint arXiv:1105.5170 [online]. 2011. [Accessed 30 January 2014]. Available from:
5 CORNU, Jean-Michel. Donner : une capacité naturelle mais limitée. In : Tirer bénéfice du don: pour soi, pour la société, pour l’économie [online]. Limoges, France : FYP, 2013. Stimulo, ISSN 2265-7754. ISBN 978-2-916571-87-4. Available from:
6 NOUBEL, Jean-François. Intelligence collective, la révolution invisible.  TheTransitioner [online]. 2007. [Accessed 30 January 2014]. Available from:
7 BENTHAM, Jeremy. Panopticon; or, The inspection-house: containing the idea of a new principle of construction applicable to any sort of establishment, in which persons of any description are to be kept under inspection: and in particular to penitentiary-houses, prisons, houses of industry ... and schools: with a plan of management adapted to the principle: in a series of letters, written in the year 1787, from Crecheff in white Russia. To a friend in England. Gloucester, Royaume-Uni : Dodo Press, 2008. ISBN 978-1-4099-5202-2.
8This corresponds approximately to the maximum size of a human family, the size of human groups before the establishment of villages in the Neolithic or the maximum size of the small jazz bands that do not have a conductor to ensure direction, unlike "big bands"...
9In constrained environment such as fire brigades in action, a hierarchical level is added as soon as the level n-2 exceeds 12 people (and not the level n-1 immediately dbelow as in the other cases). During a forest fire for example, the trucks of 4 people have a leader each. When it is necessary to mobilize 4 trucks (16 people 4 leaders) a leader of higher grade is named.
10 Règle du 1 %. Wikipédia [online]. [Accessed 30 January 2014]. Available from:
11 Les observateurs dans les groupes. Fing : groupe intelligence collective [online]. [Accessed 30 January 2014]. Available from:
12This does not happen with reactive people that react to proposals from managers or other reactive people but interact less with each other and therefore do not constitute a sub-group but only a part of the main group.
13 In 2012 the Internet Nouvelle Génération Foundation has developed a tool enabling to contribute by email (push approach) on forums, to set on one's social network (pull approach) when the question is about about collective works such as Digital Question or Digiworks gathering between one and three participants : Réseau social de la Fing. Réseau FING [online]. [Accessed 30 January 2014]. Available from:
14The Adeo group (13 DIY trademarks across the world: Leroy Merlin, Weldom...) tested in 2013 the combination of email and forum in order to send only questions, selections of contributions and summaries to the 1,500 members of the group who were exchanging on the definition of the group's strategy. In that case, everyone would receive by mail the same (limited) information and only the proactive members would search, if they wanted, the details on the forum (pull tool).
15The group on digital prospective from Franche Comté uses a discussion list to exchange, but some members have chosen not to receive mails from the list (eventhough they belong to it in order to contribute). But they receive carbon copies - for the moment in a manual way – of important emails: summaries and invitations.