A typology of crowds

Over the last few days, I’ve been involved in an email discussion on “The Crowd,” which will be excerpted on PBS’s Digital Nation site. One thing that has long bothered me about discussions of online crowds is that they tend to yoke lots of different sorts of groups together under a single rubric. Important differences end up being glossed over.

With that in mind, I’ve been trying to think through the various forms that online crowds take. As a rough starting point, I came up with four:

“Social production crowd”: consists of a large group of individuals who lend their distinct talents to the creation of some product like Wikipedia or Linux.

“Averaging crowd”: acts essentially as a survey group, providing an average judgment about some complex matter that, in some cases, is more accurate than the judgment of any one individual (the crowd behind prediction markets like the Iowa Electronic Markets, not to mention the stock market and other financial exchanges).

“Data mine crowd”: a large group that, through its actions but usually without the explicit knowledge of its members, produces a set of behavioral data that can be collected and analyzed in order to gain insight into behavioral or market patterns (the crowd that, for instance, feeds Google’s search algorithm and Amazon’s recommendation system).

“Networking crowd”: a group that trades information through a shared communication system such as the phone network or Facebook or Twitter.

Clay Shirky, who is also participating in the discussion, suggested a fifth crowd type for this list:

“Transactional crowd”: a group used to instigate and coordinate what are mainly or solely point-to-point transactions, such as the type of crowd gathered by Match.com, eBay, Innocentive, LinkedIn and similar services. (I would think that contests like the Netflix Prize also fall into this category.)

Each of these “crowds” (and there are surely others) has its own unique characteristics and its own unique strengths and weaknesses. Some crowds, for instance, gain their usefulness from the individual talents of their members. Others (notably the “averaging” sort) gain their usefulness by essentially filtering out those individual talents. Some crowds might be called “hives,” which implies some degree of individual unconsciousness about how one’s work or behavior fits into the larger whole, while others aren’t anything like mindless hives. Some crowds become more useful as they get bigger; others work best when kept to a small scale. “Crowdsourcing” and its cousin “digital sharecropping” may draw on any or all of the different types of crowds, to various effects and with various ethical implications.

As this nascent typology indicates, there’s not really any such thing as “The Crowd.”

UPDATE: Tom Lord, in a comment, suggests a sixth category:

“Event crowd”: A group organized through online communication for a particular event, which can take place either online or in the real world and may have a political, social, aesthetic, or other purpose.


Filed under Uncategorized

12 Responses to A typology of crowds

  1. Please consider adding some taxonomic category that would better encompass “flash mobs”. These are not *quite* transactional, I think. C.f., for example, the recent political dust-ups over the annual pillow fight in san cisco, and he one off recent “riots” in Berkeley (that started in large measure as a flash-mob dance party).

    Not every on-line crowd exists as a web site.

  2. Martin Lessard

    Tom Lord has a good point about “flash mob”.

    However I would tend to tag it under Shirky’s “transactional crowd”.

    Each member of mob is having as a matter of fact an unique “point-to-point transaction”: in exchange of her/his own presence, s/he get a temporary membership for the event. No member (conceptually speaking) is meant to meet again…

  3. BrianSJ

    From a user-centred design perspective, this typology is interesting. How designers view ‘users’ collectively determines the nature of the interaction, the approach to service etc. I think the typology could be worked up as a set of perceptions that are important for service design, innvoation etc.

  4. Etienne Pollard

    Should that be “data exhaust” or “data footprints” rather than “data mine”? Data mining is what non-crowd actors do to the (mostly unconscious) output of the people whose behaviour and choices they monitor, but as far as the crowd are concerned this is a side effect – they just want to buy something from Amazon, search for something on Google, listen to good music on last.fm, and so on.

  5. I think back to the usenet group days and I am pretty sure I’ve seen all these crowd kinds back then also. Other than the user-interface of the now blogging/online tools, has anything changed from those usenet group days? Not really.

    And what about a category for the good old-fashioned personal blogging crowd, you know, the personal journal, web-log writing type…? Did they disappear by the wayside?


  6. A very interesting post, Nick. I would like to translate it to Spanish and publish it in my blog. Let me know if I can.

  7. Resource Crowd: This could potentially be wedged uner the Transactional or Social Production crowds, but what about crowdfunding platforms such as Kickstarter or Kiva? These aggregate small amounts of money into a large enough whole to fund a particular goal. It seems to be a distinct category of The Crowd from the ones you have listed.

  8. Abbagliati.blogspot.com

    I guess that your silence regarding my offer to translate to Spanish your post means… No, I can’t do it :(

  9. Geordie Adams

    Hi. Enjoyed your post. Two comments. One, the crowds that you suggest are not mutually exclusive and often you can have several types of crowds in any one engagement, we regularly see and analyze data that supports that. Two, our company (PubliVate)focuses on crowdsourcing in the public sector only and clearly, for them (and I suspect for many other business verticals and individuals), there is a “Learning Crowd”. Our data show a consistent, exorbitant number of non-attributed idea views (sometimes as much as 3:1 ratio…meaning non-registered people are viewing up to 3 times the ideas as registered contributors) underlining one on one feedback that people are still learning and becoming comfortable with being empowered, particularly in government.

    Perhaps a third point and this is what we concentrate on is not necessarily the types of crowds, although I see benefit in that, but measuring the crowd (particularly – if you can – those crowds/organizations that are doing multiple crowdsourcing exercises over time). we measure them in a variety of areas including through a collaboration index we have created with the goal of, of course, improving the overall collaboration…if we don’t have that as an objective than we will still get value from the 1-2% who are making 40% of the contributions but it would be tragic not to teach and progress others who also have a lot to offer.


  10. Abbagliati: Feel free to translate it. Nick

  11. Abbagliati.blogspot.com

    Nick, here is the Spanish translation of your post:


    Thanks for letting me do it! Enzo.

  12. aimee whitcroft

    Hi Nick

    Excellent point! It reminds me of my ol’ market research days, when of course the type of crowd/groups to be dealt with was pivotal (there’s no such thing, if it’s to be useful, as the generic term ‘crowd’).

    It seems, however, like you need a more layered classification system, with some groups and sub-groups. A tree, if you will, along Darwin’s lines. Happy to help, of course :)