Digital Stigmergy
One of the most common observations of stigmergy outside social insects is digital. The phenomenon has inspired several applications in computer algorithms. Some stigmergic behaviour may be found in collaborative projects on the World Wide Web.
Optimisation
One of the first digital applications of stigmergy was in optimisation algorithms. Software developers saw how ants could find the shortest path between a food source and the nest. They were inspired to write optimisation algorithms resembling this behaviour. The essence of the first optimisation algorithms was to find the shortest path between two nodes, similar to how ants do it (Dipple et al., 2014). Like the ants, the algorithm uses an artificial pheromone trail. The trail is reinforced proportionally to how good option the path is (Dorigo et al., 2000). Through positive and negative reinforcements, the algorithm would eventually find the shortest route. The optimisation algorithms were named after the biological phenomenon as ant algorithms.
The first optimisation problem the ant algorithm was given to solve was the well-known Travelling Salesman Problem. The algorithm solves the problem by finding
the shortest route connecting a set of cities (Dorigo et al., 2000). Furthermore, ant algorithms are often used to solve complex logistical problems. Such problems might contain several thousand nodes and edges.
Travelling Salesman Problem solved with ant optimisation
Swarm Intelligence
Swarm intelligence is referring to a swarm-like behaviour in artificial agents, such as robots. The research field includes swarm optimisation, where several model-based reflex agents are used to collectively solve a problem (Garnier et al., 2007). The artificial agents are provided with algorithms that replicate the behaviour of insects. Without an internal memory or cognition, each agent responds to cues in the environment. The state of the environment acts as an external memory for the agents. The agents have an awareness of their internal state, but lack awareness of the state of the overall solution. From sensory inputs, the agent can sense the environment. Based on environmental stimuli, artificial agents can perform tasks to achieve highly complex collaborative behaviour (Dipple et al., 2014).
A swarm of robots indirectly collaborating to form a shape
The Medium
Stigmergy, in nature, takes place in a physical environment. It is observed in social insects as well as human beings. When moving towards the organisation of markets and optimisation, stigmergy occurs in a digital environment. In order to distinguish between the physical and the digital, a new term is introduced, the medium. A medium is a channel of communication or information. In this context, the medium describes a digital environment.
Peer Production
The World Wide Web was the first global communication medium accessible for anyone to read and write. The environment is not a one-way system where you can only retrieve information. People can also contribute their work openly for everyone to see (Gregorio, 2002). By utilising the Internet, information sharing and collaboration across long distances is made more accessible (Heylighen, 2007). Today there are a whole lot of applications, different software, and connected devices that utilise it.
Peer Production is an example of collaboration on software development utilising the Internet. It is sometimes used in the development of open-source software where everyone can contribute. The collaborative work is self-organised, meaning it is neither organised nor planned in any traditional manner. There is no need for any formal communication between the contributors. In peer production, the contributors include individuals of diverse backgrounds and expertise — individuals self-assign to tasks they find interesting or where they have the competence (Ozturk & Nickerson, 2015). By self-assigning to tasks, the developers tend to be more motivated, interested and knowledgeable in their work (Heylighen, 2007). Some of the most well-known outcomes of peer production are the operating system Linux, the media player VLC, the website Wikipedia, and the web browser Mozilla Firefox (Ricaud, 2017).
Logos of Mozilla Firefox, VLC and Linux
Commonly there are a few central figures in the projects developed with peer production. They determine the general direction of a project (Heylighen, 2007). However, there are mainly decentralised decisions that drive the work forward. By developing conditions that contribute to product evolution, processes are optimised (Ozturk & Nickerson, 2015). The structure of the organisation is not by far as strict as in hierarchically organised projects (Heylighen, 2007).
In peer production, there are no deadlines for when work should be submitted. This adds much flexibility and allows developers to freely decide when and where they want to make their contributions (Heylighen, 2007). It is crucial that work is submitted frequently to allow others keeping track of changes and new contributions. Regular project updates of new modules, requested features, and discovered bugs mean other developers can always find updated tasks to work on (Heylighen, 2016a).
Peer Production and Stigmergy
In terms of stigmergy, peer production deals with a digital medium for collaboration. Marks are left as information within a website or software itself (Dipple et al., 2014). Previous work on the website are triggers for new actions from contributors. The stimulating marks are often code-contributions. Such contributions might be bug-fixes, code edits or newly added code extensions. Both peer production and stigmergy in social insects have a high number of collaborating agents. In the anthill, agents are collaborating in the same physical environment. In peer production, people are the agents collaborating through the digital medium. These people are often physically separated in space.
Similar to the pheromone trail in ant foraging, peer production also experiences reinforcement. Project contributions might act like marks that attract the interest of other contributors. Rapidly growing project parts, or parts of high activity, will attract more attention and are likely to stimulate more contributions. The project experiences positive feedback contributing to the growth of only the most promising parts (Heylighen, 2007). Less promising project parts are likely to get less attention, and through negative feedback, they will diminish. The feedback mechanism makes sure that work efforts are not wasted on the less promising projects, but instead focused on the projects that are likely to be realised.
Wikipedia
Wikipedia is an excellent example of digital stigmergy, where peer production is the primary way of writing. Today Wikipedia is the largest and most popular general reference site on the World Wide Web. Wikipedia is openly editable for everyone to contribute their knowledge in developing articles and content (Heylighen, 2007). Contributors choose freely when and to which articles they contribute their knowledge. Commonly contributors work on topics of their interest such as hobbies, beliefs or profession. The contributors in Wikipedia are classified in one out of three categories;
-
Anonymous user, recognised by IP- address
-
Registered user
-
Administrator
All of the contributors are allowed to make edits and develop the site content. Administrators, however, have a few privileges and special permissions.
Stigmergy is recognised in some of the ways Wikipedia coordinate the work on the website (Besten et al., 2010). Contributors are spread all over the world, and they do not commonly communicate directly (Rezgui & Crowston, 2018). Wikipedia includes many mechanisms that facilitate collaboration and coordination of the work (Stvilia et al., 2008). Stigmergy is one of the major mechanisms in developing Wikipedia, but not the only driver of peer production processes observed on the website (Besten et al., 2010).
Wikipedia has several ways of coordinating work through stigmergy
Marks in Wikipedia
The Wikipedia website is a medium that records and carries stigmergic marks. There are several different kinds of marks found on the website, but all of them stimulate subsequent actions. The main content of Wikipedia is articles. Commonly the contents of these articles stimulate people to make contributions (Rezgui & Crowston, 2018). Since all articles are openly editable, anyone can fix small mistakes or add other changes. Some of the other stigmergic components on Wikipedia are tags, a Wikimedia platform and ranked lists.
Tags
Tags are one of the central stigmergic marks in Wikipedia. Tags are cues that direct the work that needs to be done (Besten et al., 2010). The tags resemble the popular hashtags used in, for instance, Instagram. Wikipedia tags are labels describing an issue and can be attached to articles. Tags help in categorising similar issues within the website. The tags are a type of metadata used in Wikipedia. Tags might point out that an article needs rewriting or lack a neutral point of view. Tags tend to draw attention from editors, and it signals that an article is unsatisfactory. By drawing attention and stimulating actions from people, without direct communication, a stigmergic mechanism is obtained through the tags.
MediaWiki Platform
MediaWiki is a platform offering software for editing Wikipedia articles. A talk page for each article provides editors with a forum for social interactions and explicit coordination through communication (Besten et al., 2010). The platform has other features that offer ways of collaborating indirectly. MediaWiki has several editing tools and records all edits in revision histories. Edit histories provide editors with a simple overview of the article evolution. An edit history includes metadata of who performed it and a timestamp for when it was done. Further, the platform has watchlists where contributors can subscribe to articles. Logged-in users will get a list of recent activity in the articles they have subscribed to (Rezgui & Crowston, 2018). Watchlists make it easy to keep track of changes in articles editors find interesting, allowing them to react to changes immediately.
Edit histories and instances in a watchlist serve as marks that stimulate actions from Wikipedia contributors. If editors themselves do not solve an issue, one option is to classify and assign a tag to it. Such action will stimulate other editors to take further action. The edits that most commonly stimulate responses are those that are easy to handle or does not require involvement from other editors (Rezgui & Crowston, 2018). Such edits might be minor fixes or repairing vandalism on a page.
Edit histories of Wikipedia articles
Prioritising new Articles
Even though Wikipedia has abundant articles on all sorts of subjects, there are still articles of topics that have not yet been established. Existing articles might refer to a concept that has not been defined. Hyperlinks to these articles direct the user to an empty topic, which has a template that explicitly invites people to fill in its contents (Heylighen, 2007). Another way of finding empty articles is from a list of most-wanted articles. This list is prioritised based on the number of hyperlinks that are pointing to the article. The more links referring to it, the higher up on the list it is found. This ranking is similar to how Google ranks its page prioritisation in Google Search (Parunak, 2005). In Wikipedia, the entries in the list of empty articles are stigmergic marks. The prioritised list is subject to positive reinforcement. The higher on the list an article is, the more likely it is to be written.
Linux
Linux is the far most used operating system in the world when counting units and offsprings like Android (The Linux Foundation, 2012). It is a prime example of non-profit peer production and human stigmergy. Everyone can contribute to the development of new applications and plug-ins. Sometimes a person comes across a wanted feature, like a driver for a printer. In Linux, they are free to solve their problem and distribute the feature for others’ access. The only criterion for using and modifying the code is that all source code is shared openly. Linux is a collection of system software and application software packages, which are available for download. In this way, users can easily customise their system to their needs.
Individuals contributing to Linux is somewhat rare; mostly companies are participating in the development (Anthony, 2014). Regardless of working for a company or not, Linux developers collaborate. Collaboration often happens through mailing lists, where many people have signed up for updates (Foster, 2017). Whenever someone asks a question in a list, whoever has an answer might submit it to the mail thread. Mail contributions to such lists are often about code commits or patches. Patches are recorded code changes, describing what is missing, must be added or has to be changed in the source code (The Linux Foundation, 2012). Contributors mostly self-assign to a task and address the issue outlined in a patch. Patches are stigmergic marks directing actions of work tasks. Another way of collaboration is directly in the source code (Foster, 2017). Code commits and code maintenance might trigger people to make their contributions.