Collective Intelligence, although not new, is attracting much attention, especially since the Internet became popular. Generally, Collective Intelligence is presented in a simplistic way, but if we look in detail we can find a diversity of processes that will allow us a better comprehension of the phenomenon.
Collective intelligence is an emerging phenomenon in which the interaction of small amounts of information, provided by different individuals, results in knowledge greater than the sum of the whole.
This is an emergent phenomenon because the actions of the different individuals lead to collective phenomena, not explicitly sought by them. Transferring it to the field of Collective Intelligence, Steven Johnson sums it up this way: “Local information leads to global wisdom” (Emergence. The Connected Lives of Ants, Brains, Cities and Software).
However, as shown by Raymond Boudon in The Logic of Social Action, emerging effects can take different forms. Similarly, the processes by which interaction systems generate a Collective Intelligence phenomenon can be diverse.
In this article, I present some examples of interesting interaction systems leading to Collective Intelligence processes.
1. Cumulative processes
Cumulative processes are the most common ones. The best known examples are Wikipedia and open source software. In both cases, the global wisdom emerges from the accumulation of the contributions of different individuals involved in the creation of the online encyclopedia or software.
This does not mean that the contributions of the different participants are equal. In the case of Wikipedia, the proportion of active contributors constitutes only 0.02% – 0.03% of all visitors (Source: Wikipedia). Similarly, among computer programmers of open source software there is a clear stratification of the community based on the degree of contribution to the development of the technological system.
However, differences in reputation or in the degree of contribution does not have any influence on these processes, since small contributions from the broad base of less active collaborators supplement the work of the renowned ones.
2. Neutralization process
Neutralization processes are those in which the global wisdom does not emerge from the sum of the information provided by individuals but from the average. In these cases, people provide contradictory information that neutralizes each other. The resulting knowledge is the result of this neutralization. Let’s consider a couple of examples.
An example of a neutralization process in Collective Intelligence systems is that in which a group of people is asked to estimate the value of a quantity, for example, the weight of an object. None of the individuals gets the right weight. However, it is possible to obtain the exact value by averaging all the estimated values.
Another example is that of public opinion. This type of process is well explained in the work by Elisabeth Noelle-Neumann, The Spiral of Silence. There, the author explains how people’s expectations about which are the majority and minority preferences in an election process are much more accurate than any voting intention poll.
3. Reputation processes
Reputation processes are those in which knowledge is obtained from the intervention of an influence mechanism.
The first example of a reputation process is a situation where I often find myself. When I write articles, I usually have doubts about some expressions. In these cases, the dictionary is not very useful, so I go to Google and I write the expression in different ways. To determine the correct way, I firstly analyze which one is most common. However, many incorrect expressions are very used. So the second step is to analyze who uses each expression. If one of them appears in newspapers or official documents, I can be pretty sure that’s correct.
The second example of reputation process is the indexing system of the Google search engine. Just as in the previous example, when we look for information on Google, the browser crawls all entries in which that information appears. However, not all the results have the same value. What Google does is to sort them according to their relevance, based, in this case, on the incoming links from other pages.
Possibly we may find many more processes, but what I wanted to highlight in this article is that Collective Intelligence is not a monolithic phenomenon, but is based on different interaction systems that result into different processes. Understanding this diversity will allow us to know which process should be applied in each situation.