Crowd-driven socialization

When a large crowd come together due to a certain common interest, we call it Crowd-driven Socialization, or Crowd socialization. I have been working on this for many years, initially on an online platform called the sMesh Central (or SC), which is for a casual type of online crowd socialization (e.g., ball games, presidential election, major media event, etc.).

Most of this blog is about a so-called planet-scale terraAI (or TAI) project which allows participation of the general public. As such in this post I will try to push the concept further and attempt to work out some details.


Crowd-driven socialization is different from person-centric socialization (such as how facebook is typically used with friends and familiy) in the following ways:

  1. It can be at an extremely large scale. For example, a national election can easily involve hundreds of millions of people.
  2. There could be people holding opposing interests in the crowd. For example, when watching a NBA basketball game in a stadium there could be fans for the other team present.
  3. The crowd are mostly strangers to each other, and as such the style and focus of social
  4. The topic is highly focused. For example, it is generally considered inappropriate to discuss religion on a socialization site for NBA games.
  5. Unless the target event is owned by a proprietary party (e.g., the Olympic games, the Super Bowl, etc.), then large-scale crowd socialization tend to be highly scattered, i.e., there is no central hub where users tend to go first.

The current state is that many online social platforms (such as facebook, YouTube, Twitter, etc.) are all aiming to become the hub for live or hot events. There are also small startup such as the sMesh Central (which I developed) in this space.

There are several types of crowd-driven socialization, discussed below.

Casual Crowd Socialization

This type of crowd socialization is casual in the sense that the crowd come together unorganized, typically for a relatively short period of time, and are not aiming to achieve anything together.

For this type of socialization the following are of utmost importance:

  1. Summarizing and guiding the crowd sentiment becomes important. For large-scale crowds (consider the hundreds of millions of people interested in a presidential election) socialization tools such as discussion threads or chats become ineffective. Rather, it becomes important to give people an easy outlet for expression, then have a quick way to show the summarized crowd sentiment.
  2. Guiding opposing interests. For events where there are multiple conflicting interests (e.g., the fans for the two teams in a ball game),
  3. Crowd-driven censor
  4. Aggregation
  5. Influencer mechanism
  6. spams

aggregation is a must.

Working Crowd Socialization

This type is a working crowd in the sense that there is a relatively clear goal regarding what is to be achieved together. This crowd is different from the typical sense of a team because the size of this crowd could be extremely large, and the manner of collaboration can be somewhat ad hoc.


  1. Wikipedia
    The can be considered to be a leading example of such Working Crowd Socialization.
    The working style in Wikipedia involves mainly writing textual articles, so the workflow is relatively straightforward.
    Ref: Wikipedia:Editorial oversight and control
  2. Amazon's Mechanical Turk (AMK)
    The AMK is a paid service offered by Amazon. It is designed to split a large amount of work into small pieces, then manage the workflow of spreading them among a large ad hoc workforce. The types of work involved tends to be relatively simple, and the result collected are processed separately.
    An usage example of the AMK is in image recognition. In order to train a Machine Learning system to recognize a certain types of objects, it generally needs have a good set of training examples. In this case a training example is acquired by having an image categorized by a human worker, and through AMK we are then able to prepare a large set of training examples.
  3. Web search engines
(drum roll) TAI's crowd-driven socialization

TAI's crowd-driven socialization is much more complex than the above types, since we aim to involve a very large number of people in building a very complex and advanced system.

The TAI needs to go beyond the above examples in the following areas: * Manage more complex workflow * Now what would it take for a very large crowd to work together on something much more complex? * TO BE FILLED

As such, this will be discussed in a separate post Crowd-driven Socialization in TAI (upcoming).

comments powered by Disqus