Deconstructing Evolutionary Dynamics in Online Social Networks (OSNs)

Nicole Yen Ting Lim
New York University
(email: ude.uyn|032lyn#ude.uyn|032lyn)


Citation information: Lim, Nicole Yen Ting. 2017. Deconstructing Evolutionary Dynamics in Online Social Networks (OSNs). The NYU Student Journal of Metapatterns, volume 1, issue 1. Available at: http://metapatterns.wikidot.com/nyusjm1-1:lim-socialnetworks

Abstract

This paper aimed to investigate if evolutionary dynamics could be universally applied to both living and non-living systems in the world. Evolutionary dynamics has already been successfully applied to living systems, so this paper used a case study on online social networks (OSNs) to determine if evolutionary dynamics could also be applied to non-living systems. The case study solely examined the follower relations in OSNs and sequentially employed a newly coined metapattern of 1) Variation, 2) Selection and 3) Propagation to three OSNs (Facebook, Instagram and Twitter), to verify if evolutionary dynamics could be applied to non-living systems.123 From monitoring the follower relations of six Facebook, Instagram and Twitter users, it was found that evolutionary dynamics could be applied to non-living systems like OSNs because they abided by the metapattern’s three basic conditions: 1) Trait differences, 2) Non-random sieving and 3) Cumulative enhancement.45 The case study showed that variation among users produced a spectrum of trait differences across all three types of OSN accounts. OSN users had control over what accounts they followed and the accounts that followed them because they could sieve through all OSN accounts in the Internet and select non-random users to have follower relations with. From analyzing the algorithms and network topologies of all three OSNs, once a particular follower relation was made, similar follower relations would propagate. Overtime, the accumulation of such follower relations enhanced a user’s OSN account.

Introduction

Evolutionary dynamics has been used throughout history as an analytical tool to study the development of living systems in the context of biology.67 The concept derived from Charles Darwin’s theory of evolution, where natural selection was used to explain the mutational changes and general progression of species and natural systems over time.89 Similar to the Darwinian principles in the theory (Appendix A), evolutionary dynamics adopted the requirement of variation within a system, selection based on a specific set of factors and the propagation of the most successful outcome.10 The concept has since transcended biology to describe the evolution of non-living systems in various social frameworks.1112 Since there are many fundamental and common principles between biology and culture, evolutionary dynamics has recently been used to rationalize phenomena related to cultural evolution (Appendix B).13

For the purpose of this paper, 1) Variation, 2) Selection and 3) Propagation have been fabricated into a metapattern (Appendix C), and it is used as a criterion to determine whether evolutionary dynamics could be applied outside the realm of biology.14 Hence, if a non-living system met the criterion, evolutionary dynamics could be applied to the non-living system. This metapattern mimicked a cycle, where entities within a system must sequentially go through each stage within the metapattern and return to the first stage, similar to how feedback loops function.15 For each stage within the metapattern, there were certain conditions that had to be abided by. Variation required trait differences among the entities of the system.16 Selection mandated that entities in a system had to be sieved non-randomly according to a specific set of factors.17 Propagation necessitated that only the most successful entities could accumulate and continually enhance the system.18 I make the assumption that OSN users make responsible and advantageous follower relation decisions that have an overall positive effect on their lives (see Parameters under Methods). Therefore, the propagation similar positive follower relations would enhance a user’s account in a beneficial manner.

OSNs have been chosen for this study not only to signify non-living systems, but because they are very relevant and important to how today’s society functions. Facebook, Instagram and Twitter are not only the most well-known OSNs, but they have been major influences for cultural evolution because they have drastically changed how individuals interact with each other, how information flows between different media platforms or actors, and how people influence each other. The three specific OSNs were also chosen based on their varying network topologies (Appendix D) to try to emphasize that evolutionary dynamics can be truly applied to non-living systems, even ones that are very complex. Besides using OSNs for business marketing and entrepreneurial endeavors, they are imperative to expanding one’s social sphere, communicating with people and maintaining relationships. Today’s Internet culture is largely based on the number of follower relations in one’s OSN account. The notion of having more follower relations insinuates more popularity and connection to the rest of society, so there is a socially constructed desire to want to have more follower relations in OSNs.

In this paper, I will exclusively investigate the follower relations in Facebook, Instagram and Twitter. I will sequentially employ the metapattern to determine if evolutionary dynamics can be applied to the non-living system.

Methods

In order to employ the metapattern in the case study, I monitored my own follower relation activity on Facebook, Instagram and Twitter, as well as surveyed five other users’ follower relation activities. These five other users were my five closest friends who were kind enough to let me pry into their OSNs. Similar to myself, each of these OSN users has had a Facebook, Instagram and Twitter account for at least five years. Moreover, I also analyzed the network topologies and follower recommendation algorithms of each OSN to find any similarities or differences in follower relation patterns across the three OSNs.

Parameters

The case study solely examined the follower relations in OSNs. It only considers active personal Facebook, Instagram and Twitter accounts, and the number and types of follower relations a user has in an OSN account. Different types of followers can be categorized into several groups, but are not limited to, relatives, friends, colleagues, acquaintances, people of specific interest to the users, and celebrities. The case study excludes:

  • Fake accounts
  • Secondary accounts
  • Blocked accounts
  • Advertisements
  • Other websites and related links
  • Newsfeeds
  • Issues, such as spamming, censorship, cyber bullying/attacking, hacking, etc.
  • Other applications for government, business, dating, etc.

Network Topologies

All three OSNs are popular free social networking platforms that are accessible on a variety of different devices. They allow Internet users to register, create profiles, state personal information, upload photos and videos, send direct messages, build relationships, and connect with other users. There are different privacy options on each individual OSN and they are based on a user’s preferences. Users have the option to make all their activity on the OSNs visible to everyone or choose to block specific connections or information from people.19 For the case study, we assume that all OSN accounts are available to the public to be viewed and interacted with.

Facebook allows users to create a detailed profile by having an “About” page where they can choose to write personal background information about themselves, such as a birthdays, workplaces, education, places they have lived, family members, life events, contact numbers, addresses, and much more. There are several key components to the website, like the “Timeline”, which is essentially a virtual bulletin board that chronologically shows a user’s activity. Messages or posts can be left on a user's timeline in the form of text, photo or video. On a user’s timeline, one can update his or her status, a feature similar to Twitter that allows users to broadcast short announcements to their followers. Facebook followers are called “friends”. Another component is the virtual photo album, where an unlimited amount of photos can be uploaded from any device.20 These components are interactive where “friends” of a user can like, tag, react and comment on each other's posts, statuses or photos. All interactions are published on a “Newsfeed”, which is distributed in real-time to a user’s friends. Facebook users can choose whether or not they want to be searchable on the search engine and choose which aspects of their own profiles are public or private. Facebook also allow users to join specialized groups or events that could be either public or private, as well as allow third-party developers to build applications and widgets to be distributed throughout the Facebook community.21

Lim-fig1.jpg
Figure 1. Network topology of Facebook. Showcases the “Timeline”, “About” page, “Friend Request” tab, “Message” tab, virtual photo and video albums, likes, comments, groups, events and “friends” lists on a user’s profile.22

Unlike Facebook that has a more broad and varied platform, Instagram is mainly for photo and video sharing. The website is owned by Facebook and is actually more commonly used as a phone app. It has similar features as Facebook, such as the ability to upload photos, videos, “stories”, and synch with other OSNs, phone contacts or email addresses.23 Instagram is unique because the uploaded photo can include features like filters, borders, and the user can also edit the colors, brightness, contrast, saturation and structure of a photo. The "Explore" tab of the app is like a search engine that allows users to find new or existing followers using photos, people, tags or location services. From the "Home" tab, Instagram users can view, comment or like photos from the people that they have follower relations with. Unlike Facebook, the user can develop a follower relation with another account on Instagram even though the user is not followed back by the account he or she chose to follow.24

Lim-fig2.jpg
Figure 2. Network topology of Instagram. Showcases a user’s profile (personal photo collage), “Newsfeed”, “Home tab”, “Explore” tab, “Likes” page, Follower and Following statistics, “Message” tab, photo captions, comments, and photo editing features.25

Twitter is an OSN that is similar to a blog where users can broadcast short posts called tweets. Tweets can consists of various content, like one’s feelings, news, or an announcement of an event. Tweets often contain hyperlinks or hashtags to connect other users to a trending topic or conversation thread.26 Tweets are also limited to 140 characters due to the constraints of Twitter's Short Message Service (SMS) delivery system, and a user can choose to retweet another user’s tweets. Like Instagram, the default settings for Twitter is public, unless a user personally changes his or her account settings, and other users can follow anyone that has a public account. Like Facebook, you can directly message a user and join specialized groups and lists.27

Lim-fig3.jpg
Figure 3. Network topology of Twitter. Showcases the “Newsfeed”, Tweets, “Profile” page, “Message” tab, follower and following statistics, virtual photo and video albums, likes, groups, and lists in a user’s profile.28

Algorithm

Besides using the search engines on the varying network topologies, users of all three OSNs have follower recommendations based on special algorithms that prompt users to propagate their network relations and enhance their accounts. Each algorithm was made by taking into account all the permutations of functions in the network topologies and every trait of each OSN account, which includes all types of user content and activity. The functions and traits were arranged in carefully formulated sequences of actions to generate algorithms for each OSN to create recommendations of user accounts to follow.29 The follower recommendations are tailored to a user because a user’s content and activity is unique. The algorithms slightly influence a user’s selection process, but mostly affect the propagation stage of the metapattern.30

Lim-fig4.jpg
Figure 4. A Venn diagram depicting all the functions and traits that are factored into the follower recommendation algorithm for eachOSN. The diagram shows the functions and traits that 1) all 3 OSNs have in common, 2) an OSN shares with another OSN, and 3) are unique to an individual OSN.

Results

Lim-fig5.jpg
Figure 5. Table depicting how the metapattern 1) Variation, 2) Selection, and 3) Propagation was applied to the three different OSNs (Facebook, Instagram and Twitter). Bullet points in each column refer to the functions, traits and other elements an of each OSNs that collectively explain how trait differences occurred, how selection was prompted and how cumulative enhancement was induced.

After surveying five of my friend’s follower relation activities on the three OSNs and monitoring my own, each stage of the metapattern, 1) Variation, 2) Selection, and 3) Propagation, was evident in each OSN platform.

Variation

All six users had extremely varied accounts for each of the three OSNs. Figure 5 showed that besides the obviously different network topologies, each account in all three OSNs had very diverse content, number of followers and types of followers. The results showed that content can differ in the form of statuses/captions or tweets, theme of uploaded photos or videos, likes, and comments. Moreover, the results showed that no two users in each of the three OSNs had exactly same number or type of followers in his or her account. For example, my Instagram account had few photos and videos that illustrated the progression of my track & field career. My photos and videos had no captions, but I still liked and commented on the photos and videos of users that followed me and that I followed back. I also did not direct message the people whom I had follower relations with. All my other friends that I surveyed had varying photos and videos with different themes. The photos and videos also had captions of different lengths, languages and depths. All of them had different numbers and types of followers, so each other them liked and commented on different kinds of content from an array of Instagram users. Some of my friends have direct messaged users they have follower relations with but some did not. Hence, every user’s account in each of the three OSNs was unique and had a variety of trait differences that identified and distinguished one user’s account for the rest on all the OSN platforms.

Selection

The results in Figure 5 showed that all six users of all the three OSNs were highly selective when not only choosing what content would be displayed on their individual accounts, but also the number and types of users they followed and allowed to followed back by them. For example, I only followed people in my current social sphere, who are people that I know well and have previously interacted with in-person, onto my Facebook account. My method of follower selection was similar to the other OSNs users who I surveyed. None of us went onto our search engines and followed random people for privacy and security reasons. All users always had a choice of the users he or she wanted to follow and have follow back. We also had similar methods of follower selection methods for Instagram and Twitter. However, instead of limiting ourselves to following and being followed by other users that we had previous interaction with, we could select specific groups or people of interest to follow. Since all of us had unique interests, the choices of follower relations on Instagram and Twitter were more customized to each user. For instance, I would follow accounts on Instagram that had content on runners, versus one of my friends who would follow accounts that had content on the culinary arts. However, these groups or people of interest often never followed us back (see Non-random Sieving on page 10). Nevertheless, varying OSN content, existing connections, present affiliations, behavior, interest, environment and lifestyle impacted a user’s decision to follow or be followed back by another user.

Propagation

The results in Figure 5 showed that all six users’ follower relations multiplied over time. Besides all six users continually following and accepting follower requests from other users that they already have existing connections, affiliations or interest in, each OSN created follower recommendations to prompt a user to add similar user accounts and expand their social sphere or network. For example, I often went to my search engine on Twitter to follow new users that I have most recently met in-person. I also regularly followed new accounts of groups and people that I have recently been interest in. Most of the new users that I followed often had similar connections and affiliations with my existing followers, and majority of the new groups and people of interest that I followed had common traits with the groups and people of interest that I currently follow on each of my OSN accounts. I frequently form follower relations with new athletes that I meet at the New York University gym because we train and interact in the same vicinity, and I like to follow more hip-hop music artists because I already follow so many on all my OSN accounts. The other users that I surveyed also propagate their follower relations in a similar way.

Discussion

From monitoring the follower relation activities of the six Facebook, Instagram and Twitter users, it was apparent that the metapattern could be applied to non-living systems like OSNs because they abided by the metapattern’s three basic conditions.

Trait Differences

Variation occurred because no two users of any OSN account are exactly the same. Each user’s account had unique traits that another account did not posses. Trait differences in individual active OSN accounts included varying content, different number and types of follower relations and how a user interacts with his or her followers. For example, the way I maneuvered the network topology of Instagram was different compared to the other users I surveyed. I could customize my own unique profile that was different from the rest. Trait differences were evident between my Instagram account and my friends’ accounts because we all had different usernames, captions, photos, photo editing methods, biography blurbs, and number and types of follower relations.31

Trait differences stemmed from, but were not limited to, a user’s preexisting behavior, environment and lifestyle. Different content and activity across all OSN accounts derived from different behaviors, environments and lifestyles of a variety of users.32 The three elements were responsible for trait differences in OSN accounts because they affected a user’s connections, affiliations and interests, which influenced the number and types of follower relations a user makes, as well as interactions that could occur between the various people involved. Hence, Instagram and Twitter accounts often have more trait differences because the OSNs give users more freedom and opportunities to follow the personal accounts of groups or people of interests. These groups or people of interests are usually famous organizations, celebrities and brands that have verified personal accounts. The chances of a user being followed back by users with verified personal accounts are rare, unless a user knows a verified personal account user personally. Unlike Instagram and Twitter, groups or people of interests do not share their personal accounts with the public on Facebook. Oftentimes, they just create generic pages on the OSN and upload content to notify their activity to whoever likes their pages. The groups or people of interests usually never manage their own pages by themselves (public relations teams typically manage these pages) and their follower relations are calculated to portray a certain image.33 Hence, trait differences, particularly the number and types of follower relations on Facebook are more limited to a user’s specialized social sphere of friends (Appendix E) as compared to Instagram and Twitter. The follower relations are more limited on Facebook because a user’s “friends” are almost always people who the user has already had personally interactions with. Instagram and Twitter accounts are less limited and allow more trait differences in the OSN accounts because a user can follow anybody who is public and does not need to be added back to build a follower relation.3435

Non-random Sieving

Selection occurred in the case study because each user had complete control over which account they chose to follow and which account they chose to have follow them back. Moreover, each user also had complete control of his or her activity and the content that was uploaded on all three OSN accounts. Each decision was a user’s personal choice and it was not random. The decisions seemed to be collectively based on a user’s behavior, environment and lifestyle, connections, affiliations, interests, ethnography, ideology, and culture.36 Together, these elements indirectly give a user a set of factors to based decisions on and sieve through the large number of accounts in each OSN to select the ones they specifically want to make follower relations with. Hence, the number and types of follower relations a user makes, as well as the interactions that could occur between the various people involved are never random.

All users have the ability to sieve through the Internet (a holon) to select OSNs (clonons) of their choice to pick unique individual OSN accounts to follow (sub-clonons).37 Facebook is probably the most selective OSN out of the three in the case study. Follower relations on Facebook are mainly relationship-driven, unlike Instagram and Twitter that are is not only relationship driven but respectively visually and ideology-drive.3839 Fundamentally, I tend to “friend” most of the people in my social sphere on Facebook. On Instagram, I tend to follow groups and people of interest who intrigue me because their accounts represent something that I personally think is aesthetically pleasing, such as fashion brands and flower displays. On Twitter, I tend to follow groups and people of interest that share the same ideology and thoughts as me, like inspirational speakers and comedians. When I first created my Twitter account, I started to add users and accept follower request from users that I already knew in my own social sphere. I have had previous personalized interactions with the users either in-person or online, and they were usually family, friends, colleagues, or acquaintances. After I added these highly selective categories of users, I then started to follow groups and people of my own interests.4041 The selection process was similar to Instagram.4243 On Facebook, I just started liking events and pages of groups or people of my own interests but the pages were not personal accounts. Although the other five users had different social spheres and interest as me, their selection process was similar across the three OSNs.

Since follower relations are reciprocal, I realized that my unique content and activity also influenced how people chose to follow my OSN accounts.44 For example, I centered my Instagram account on all things related to track & field. My photos, likes, comments, hashtags, etc. all had something to do with running. Over time, I noticed that athletes and running clubs started following me because they had looked at the content of my account and were interested in me. Hence, trait differences play a major roll in determining what kind of selection process takes place for all OSN users.

Cumulative Enhancement

Propagation of follower relations occurred in all six users’ OSN accounts over time. Users either utilized the search engine on each OSN to follow similar existing followers or follow more accounts based on follower recommendations. The propagation of follower relations increased the size and diversity of all users’ social networks. It is Internet culture and arguably human desire to have more follower relations in an OSN (see Introduction). Since it was assumed that all preexisting follower relation were positive, propagating similar relationships was believed to enhance the accounts in a beneficial manner, whether it was expanding one’s social network, maintaining present connections, or updating news, interests and other information that is relevant to the user.

The follower recommendations were results of the algorithms of each OSN. All permutations of functions in a network topology and every trait of an OSN account, which includes all types of user content and activity, are arranged into a sequence of actions for an OSN to generate an algorithm for creating follower recommendations.45 These follower recommendations are unique to a specific user because all user content and activity varies (see Figure 4 for the all the functions and traits that are factored into the follower recommendation algorithms for each OSN).

Facebook is a great example of an OSN that constantly gives users follower recommendations. Facebook’s follower recommendation algorithm is a robust link prediction system that not only takes into account numerous functions and traits all at once, it produces different kinds of scenarios, or this case “friends” who a user should follow.4647 Besides, using my search engine to type the names of users I want to follow on Facebook, I look at the “People You May Know” section on “Newsfeed” to see if there are more people that I could follow to propagate and enhance my account. Usually, Facebook bases “friend” recommendations on mutual “friends”, tags in similar photos and comment on similar posts. The first time I joined the NYU track & field team and became Facebook “friends” with one teammate, all other users who mentioned they were NYU athletes in their profiles popped up in my “People You May Know” section on my “Newsfeed”. Other OSN users that I surveyed had a similar experiences but with different social groups.

The results in Figure 5 also showed that the propagation of follower relations is limited by a user’s behavior, environment and lifestyle because those factors influence how many people a user knows and interacts with, as well as what his or her interests are. For example, I am a more extroverted person that lives in a big city and has a hectic lifestyle. Thus, my social network is bigger than my friend who is an introverted person, lives in a suburb and has a slow paced lifestyle. I am also a very open person unlike my friend who is more private and not as willing to expand his social circle to build new relationship or maintain old connections. Hence, I have more follower relations in all three OSN accounts than he does. Since I live in an environment that is cosmopolitan and developed, I seem to be also exposed to more groups and people of interests, such as celebrities and famous brands, than my friend who lives in a suburb that is more residential and has less activities to offer.

Conclusion

Since each stage of the metapattern was evident in each OSN, and all three of them abided by the metapattern’s three basic conditions, the criterion was fulfilled and it is indeed possible for evolutionary dynamics to be applied to the non-living system for analysis and insight. However, just because the metapattern was successfully employed in this case study on OSNs does not mean that evolutionary dynamics is in fact applicable to all non-living system or universally applied outside the realm of biology. There are bound to be non-living systems in the world that disrupt the cyclical nature of the metapattern and do not abide by its three basic conditions. More research has to go into the field of study relating evolutionary dynamics, cultural evolution, and metapatterns together in order for the universal applicability of evolutionary dynamics in all systems to be conclusive. Therefore, I propose that more case studies have to be conducted on various types of non-living systems to find the true relationship between evolutionary dynamics, cultural evolution and metapatterns. Nonetheless, we now know that evolutionary dynamics can be applied to OSNs and the concept certainly helps us understand more about the current state of our society and help us better plan for the future.

Appendix

A) Darwinian Principles:

Darwinian Principles derive from Charles Darwin’s theory of biological evolution in his book On the Origin of Species published in 1859. Darwin’s theory states that all species arise and develop through natural selection where the species with the most desired and advantageous inherited traits in an environment at a give point in time increases the species’ ability to compete, survive, and reproduce. Darwin’s basic principles of evolution by natural selection are:

  1. More individuals are produced each generation that can survive.
  2. Phenotypic variation exists among individuals and the variation is heritable.
  3. Those individuals with heritable traits better suited to the environment will survive.
  4. When reproductive isolation occurs new species will form.”48

Darwinian principles have since been applied to other disciplines, such as cosmic and cultural evolution.

B) Cultural Evolution:

Cultural evolution stems from Darwinian theories and comprises of similar principles. It is a theory is used to explain how cultures evolve, adapt and become increasingly complex over time. There is the unilinear and mutilinear theories of cultural evolution, but the concept is that there are several factors that cause human behavior and culture to evolve over time and explain the state of our current society.495051

One of the most prominent factors that has driven cultural evolution and has influenced the conditions of today’s society are OSNs, and that is why OSNs have been chosen as the case study for this paper.

C) Metapatterns:

Metapatterns are essentially patterns of patterns. We use the term to describe various concepts and patterns in our world. It can be used to describe:

  1. Shape, dimension, material, and methods of construction
  2. Functionality
  3. Experiences

In other words, metapatterns “are attractors – functional universals for forms in space, process in time, and concepts in mind.”52

D) Network Topology:

Network topology or network geometry refers to the arrangement of a network. This includes the physical topology, which is the layout and design of the network, and the logical (or signal) topology, which are the nodes and connecting lines of the network.53

E) Specialized Social Sphere of Friends:

My semantic = People who have had some sort of personal interaction with a user prior to being added by the user or having sent a follower request to a user to reject or accept. These people are specialized because each individual user has different unique groups of people they interact and connect with. All the unique groups of people form a larger social sphere encompassing all the groups, and no two social spheres are the same. These people have had developed a relationship with the user before following each other on the respective OSNs. These people are usually categorized into, but not limited to, family, friends, colleagues, acquaintances, etc. In Figure 5, I used “Friends” instead of followers to honor Facebook’s terminology.

F) Limitation of a user’s connections with people, number of interests and extent/knowledge of interests:

My semantic = Since no two people are the same, each user of the three OSNs have varying behaviors, environments, and lifestyles which not only influence the content they post on their OSN accounts but the number and types of follower relations they make.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License