Interplay between Selection and Social Influence

Summary

The video script explores the complex relationship between social influence and selection using Wikipedia as a case study. It explains how Wikipedia operates as a collaborative platform and introduces the concept of similarity measures to quantify the common interests between editors. The research presented analyzes editor interactions, revealing that people with similar interests are more likely to engage in discussions, a phenomenon attributed to selection. Post-interaction, an increase in similarity suggests social influence, where shared interests grow stronger. The script challenges viewers to consider the balance between these factors in different social contexts, such as friendships formed over good or bad habits.

Takeaways

• 🔍 The script discusses the complex interplay between social influence and selection, particularly in the context of online interactions on Wikipedia.
• 🌐 Wikipedia is described as a collaborative platform where anyone can edit entries, leading to a convergence of ideas and the emergence of 'true knowledge' through discussion and debate.
• 🗣️ The 'user talk page' on Wikipedia serves as a space for editors to discuss and resolve conflicts, which is a visible record of social interaction for research purposes.
• 📊 The concept of a 'similarity measure' is introduced to quantify how alike two individuals are based on shared interests or activities, such as editing the same Wikipedia pages.
• 📈 A research study is highlighted that uses Wikipedia's editing data to analyze the relationship between social influence and selection, showing how these factors change before and after interaction.
• 📚 The script explains that before individuals interact on Wikipedia, there is often a high degree of similarity, which may prompt the interaction (selection).
• 🤝 After interaction, there is an observed increase in similarity, suggesting that social influence plays a role in shaping shared interests and behaviors post-interaction.
• 📉 The research plot presented shows an average trend where people become more similar before they interact and continue to become more similar afterward, indicating both selection and social influence at play.
• 📋 The script points out that the observed patterns may be context-dependent, suggesting that different datasets could reveal varying degrees of social influence versus selection.
• 🤔 It raises the question of which factor, social influence or selection, is more prevalent in society and how this balance might differ across various contexts, such as the adoption of good or bad habits.
• 🔑 The importance of having diverse datasets to explore these questions further is emphasized, opening up new avenues for research into social dynamics.

Q & A

What is the main topic discussed in the video script?

The main topic discussed in the video script is the difference between social influence and selection, and how these concepts can be studied using data from Wikipedia.

How does Wikipedia work according to the script?

Wikipedia works as a collaborative platform where anyone can edit and contribute to its content. It is a blank slate where information can be added, edited, or corrected by anyone, with the idea that through discussion and debate, true knowledge emerges.

What is a 'user talk page' on Wikipedia?

A 'user talk page' on Wikipedia is a space where users can discuss and debate about the content of Wikipedia articles. It is a visible platform for all users to see the background discussions that happen during the editing process.

What is a similarity measure and how is it defined in the script?

A similarity measure is a way to quantify how alike two entities are based on certain characteristics. In the script, it is defined as the ratio of the number of common items liked by two individuals to the total number of unique items liked by both.

How is the similarity measure applied to the Wikipedia dataset in the research?

The similarity measure is applied to the Wikipedia dataset by comparing the editing activities of two authors. It is calculated as the number of Wikipedia pages both authors have edited divided by the total number of unique pages they have edited together.

What does the research aim to understand using the Wikipedia dataset and similarity measure?

The research aims to understand the interplay between social influence and selection. It seeks to determine whether people become similar before they interact (selection) or whether their interaction leads to increased similarity (social influence).

What does the plot in the script represent?

The plot in the script represents the changes in the similarity measure between two individuals over time, specifically before and after they first interacted on the Wikipedia talk page.

What does the steep increase in the similarity measure before the first interaction suggest?

The steep increase in the similarity measure before the first interaction suggests that the individuals were becoming more similar over time, which may have led them to interact with each other (selection).

What does the further increase in similarity measure after the first interaction suggest?

The further increase in the similarity measure after the first interaction suggests that the interaction itself led to even more similarity between the individuals, indicating the effect of social influence.

How does the research differentiate between social influence and selection?

The research differentiates between social influence and selection by observing the changes in similarity measure before and after interactions. If similarity increases before interaction, it suggests selection; if it increases after, it suggests social influence.

What is the conclusion drawn from the research about the interplay of social influence and selection?

The conclusion drawn from the research is that both social influence and selection are at play, with selection seemingly having a stronger effect initially, followed by social influence after the first interaction.

How does the script suggest that the findings may be context-dependent?

The script suggests that the findings may be context-dependent by comparing the Wikipedia dataset with other potential datasets, such as those related to smoking or obesity, where the balance between social influence and selection might be different.

Outlines

00:00

🔍 Exploring Social Influence and Selection Through Wikipedia

This paragraph introduces the central research question: distinguishing between social influence and selection. It highlights the challenge of identifying these phenomena in real-world data sets of friendships. The speaker proposes using Wikipedia as a unique data set to study these concepts due to its collaborative nature, where multiple users edit and debate content. The paragraph also emphasizes the importance of understanding Wikipedia's operation and the concept of a similarity measure as prerequisites for the research.

05:05

📝 Understanding Wikipedia's Collaborative Editing Process

The speaker elaborates on how Wikipedia functions as a collaborative platform where anyone can edit content, leading to the emergence of true knowledge through discussion and debate. Despite the potential for misinformation, the collective effort of the community is trusted to converge towards accurate information. The paragraph also introduces the concept of 'user talk pages' where editors discuss content, providing a rich data set for research on social influence and selection.

10:06

📊 Measuring Similarity in Wikipedia Editing Behavior

The paragraph explains the concept of a similarity measure using an analogy of shared food preferences. It then applies this concept to Wikipedia editing, where the similarity between two editors is determined by the number of pages they have both edited, relative to the total number of unique pages they have edited. This measure provides a quantitative way to assess the alignment of interests between editors, which is crucial for the study of social influence and selection.

15:11

📈 Analyzing the Dynamics of Social Interaction on Wikipedia

The speaker describes a research method that involves tracking the interactions between two Wikipedia editors over time. By plotting the similarity measure before and after their first conversation on the talk page, the research aims to observe changes in their interests. The paragraph discusses the potential for a steep increase in similarity both before and after interaction, suggesting a complex interplay between social influence and selection in the formation and development of online relationships.

🤔 The Interplay of Selection and Social Influence in Online Communities

This paragraph delves into the implications of the research findings, suggesting that there is a significant degree of self-selection in social interactions on Wikipedia, where similar interests lead to communication. After interaction, there is a further increase in similarity, indicating social influence. The speaker also points out the importance of considering the context of different data sets, as the balance between selection and social influence may vary. The paragraph concludes by posing broader questions about the prevalence of these phenomena in society and the potential for further research.

Keywords

💡Social Influence

Social influence refers to the process by which individuals adjust their behavior, attitudes, or opinions based on the interactions or communications with others. In the context of the video, it is observed when individuals, after becoming friends, start to share more common interests and behaviors, as shown by the increase in their similarity measure after they have communicated.

💡Selection

Selection is the process by which individuals choose to interact with others based on pre-existing similarities. In the video, this is seen when two Wikipedia editors first interact, often because they already share similar interests, which is reflected in their similarity measure before they speak to each other.

💡Similarity Measure

A similarity measure quantifies the degree of commonality between two entities, such as people or their actions. In the video, it is used to assess how similar two Wikipedia editors are based on the articles they edit. For example, two friends who both like certain types of food would have a higher similarity measure.

💡Wikipedia

Wikipedia is an open-source online encyclopedia where users can edit content collaboratively. The video uses Wikipedia as a case study to analyze social influence and selection by examining the editing patterns and discussions among users on the platform.

💡User Talk Pages

User talk pages on Wikipedia are discussion areas where editors communicate about content edits. The video highlights these pages as a valuable data source to study the interaction between social influence and selection, as they provide insights into when and why users start communicating.

💡Data Set

A data set is a collection of related data points used for analysis. In the video, the data set refers to the information gathered from Wikipedia, particularly the editing and communication patterns of users, which is analyzed to understand the dynamics of social influence and selection.

💡Plot

A plot is a graphical representation of data. In the video, a plot is used to illustrate the relationship between the time two Wikipedia editors first interact and their similarity measure before and after this interaction. The shape of the plot helps in understanding the balance between social influence and selection.

💡Context-dependent

Context-dependent refers to how certain observations or outcomes can vary based on the specific circumstances or environment. The video suggests that the balance between social influence and selection can differ depending on the type of data set used, such as Wikipedia editing patterns versus smoking habits.

💡Obesity Research

Obesity research is mentioned as a contrast to the Wikipedia study, indicating that in some contexts, social influence plays a more significant role than selection. The video compares findings from obesity studies where social influence was more dominant than selection, highlighting the variability in different contexts.

💡Friendship Formation

Friendship formation is the process by which individuals develop social bonds. In the video, it is discussed in the context of whether friendships are more likely to form based on pre-existing similarities (selection) or if these similarities develop after the friendship has formed (social influence).

Highlights

Introduction to the core question: the difference between social influence and selection.

Explanation of the challenge in distinguishing between social influence and selection in real-world datasets.

Overview of Wikipedia as a collaborative platform where anyone can edit, leading to discussions and debates that improve content accuracy.

Definition of similarity measure using an intuitive example involving food preferences.

Application of the similarity measure concept to Wikipedia editors by comparing the number of articles two people edit together.

Research method: tracking the interactions between Wikipedia editors to analyze how similarity influences their conversations.

Discovery that people tend to become more similar before they talk, and continue becoming similar after interaction.

Key finding: people select each other to talk when they are already similar, and this similarity increases further through social influence.

The relationship between the steepness of similarity curves and the phases of selection and social influence.

Conclusion that selection plays a larger role initially, while social influence takes over after interaction.

Importance of the observed trend in Wikipedia data and its applicability to understanding broader social dynamics.

Critical reflection on the dataset: this finding may be specific to Wikipedia and not universally applicable.

Comparison to other contexts like smoking habits, where social influence may outweigh selection.

Obesity research is mentioned as an example where social influence has a stronger role than selection.

Final observation: the interplay between selection and social influence is context-dependent and varies across different societal behaviors.

Transcripts

So, we saw the big question what is the difference  between social influence and selection. Which one  

of the two is happening are both happening now  it’s very difficult to for us to go head and  

look at the real world data set of friendships  and ask this question if we ask this question  

of can we have some very concrete data set where  this experiment can actually be conducted and the  

facts be unravelled there was one such data set  which gave some very promising observations. So,  

our motive for thisvideo chunk is to understand  this piece of research on Wikipedia ok.  

Let me now motivate theconcept by by explaininghow  Wikipedia works. Number one and number two by  

explaining what one means by similarity measure  these two are important for us to goany further  

ok. So, how does Wikipedia work? Wikipedia  basically is a blank slate you can basically  

anyone can go write anything on Wikipedia for  example you can open the Wikipedia entry of  

India edit it if you think some information is  incorrect there delete it and the right content  

that you think is right and if someone else things  what you have entered is wrong they will come and  

edit it and change it to what they think is right  this goes on and on one frequently asked question  

is can we really trusts such a database where  anyone can come and enter anythingyeah very  

intriguing but it is known to be true that whenlot  of people get together and start discussing and  

debating when they have conflicts of their  ideas true knowledge is known to emerge .  

Initially of course there will be some false  information here and there but a lot of people  

watching it will converge to the right answer.  So, Wikipedia is actually believed to be very  

trustworthy these days that’s about Wikipedia but  what happens in the background of Wikipedia there  

are many people who will be coming and editing  it if there is conflicts on theirbelief of what  

should be put and what should not be put they  talk in the background with each other and the  

background talking is also visible for all  of us there are calleduser talk page one can  

click on this tab and then take a look at what  all discussions they have about the content.  

Now this big data set is available to our disposal  and one can conduct any kind of research on this  

data set I am now going to explain one such  research which actually helped us understand the  

inter play between social influence and selection  as I told you just now I helped you understand how  

Wikipedia works that was the first prerequisite  the second prerequisite is understanding what  

one means by a similarity measure. So, let me  motivate what is the similarity measure now. So,  

let me define what is the similarity  measure it is actually quite intuitive  

let me motivate it with a good example. Assume I like the followingdishes ok I like  

upmalet’s say idlidosa, pizza and let’s say  fruit punch these are the five things that  

I like ok upma idli all those things are south  Indian dishes and let’s say my friend Priya she  

likespizza burgerlet’s say pasta and idli ok these  four things as you can see we don’t have a lot of  

commonalities but we have we like pizza and idli  that is common between both of us. So, what I do  

is in the denominator I put the total number of  items that we like whatever I like and whatever  

Priya likes both put together as you can see is  seven items out of which we both like two items.  

So, I say the similarity measure between me and  Priya is two by seven this is what we mean by  

similarity measures now I am going to use this  on my Wikipedia data set that I was telling you  

right. So, what do I do there are two people  two authors who are editing Wikipedia pages  

randomly here and there let’s say I am editingthe  Wikipedia entry of India andlet’s say Chennai  

floodsand Wikipedia entry on iPhone eight and  things like that different Wikipediapages I am  

editing and a friend of mine let’s say peter is  editing Wikipedia pages again some some other  

Wikipedia pages some ten other Wikipedia pages. The similarity measure between me and Peter is the  

total number of Wikipedia pages that we both are  editing, total number of unique Wikipedia pages  

that we both are editing. In the denominator and  in the numerator we write those Wikipedia pages we  

both have edited just like the similarity measure  that I defined between me and Priya the similarity  

measure between me and peter on Wikipedia is  total number of Wikipedia articles that we  

both have edited in the numerator divided by the  total possible unique Wikipedia pages that we have  

edited both put together that’s in the denominator  this gives a good measure of howsimilar we  

both are on ourWikipedia editingtransactions. So now that you understand what ishow Wikipedia  

works and you also understand the definition of  similarity measure I am going to use these two  

things in answering the big question of social  influence versus selection ok. So, here it goes  

I will do the following I will look at two people  who have spoken to each other on Wikipedia I told  

you people do talk in the talk pages in the  background if I observe that two people have  

spoken to each other please note I have a data set  here in that data set I observe if two people have  

spoken to each other if let’s say two peopleI will  say Rama and prem have spoken to each other on the  

wiki talk page then what I do is I plot this  what is the plot the plot is the following.  

This is my y axis this is my x axis. I will see  when exactly they spoke let me call that time  

t equal zero and I will observe what were they  doing before and what were they doing what did  

they do after this time equal zero. What do I  mean by that? Let me define what is the x and  

y axis properly on the y axis I will putI will I  will put thesimilar their similarity measure ok  

their similarity measure is given by let’s say  point zero zero five point zero one zero. So,  

on and. So, four point zero one five. So,  on up to let’s say point zero zerozero  

four zero ok I am doing this on the y axis This is my similarity and this is my time I  

repeat rama and prem have started editing some  Wikipedia articles and their similarity measure I  

am seeing when they first spoke this line denotes  the time when they first spoke by spoke let’s be  

clear I mean first time they spoke first time  they spoke on the talk page on the talk page and  

I observe what was their similarity measure before  and what is their similarity measure after it was  

observed that the plot looks something like this  it looks like this let me correct this it just  

goes steep it just goes like this that’s it. Point to note is the is the steep increase and  

further increase and then it it it becomes sort of  constant what is happening here let’s just pause  

and observe lets revise time similarly measure  rama and prem there similarity measure was. So,  

this is time when they started let’s say some  few days before and the first time, they met  

this probably some minus something minushundred  minus two hundred or whatever is the scale ok.  

And the time when they first met and thentime t  equals hundred and then two hundred and then three  

hundred. So, on whatever you want two hundred  minutes or two hundred seconds or you can you can  

have any scale here but these are time when they  first spoke to each other did you see something  

happened here something really fascinating what  is the curve tell you the curve tells you that  

before they first spoke to each other there was  a sudden surge of similarity between them these  

two people were very similar because of which  they spoke maybe because of which they had a  

talk on the spoke on the talk page maybe do you  see after they spoke there was increase in their  

similarity measurewhat does this translate  this translates the following because that  

their interest were common that’s what similarity  measure means because that their similarity their  

similarity was high their interest were common  they actually spoke to each other after speaking  

there was further increase in their similarity. This translates the following two people become  

friends if they agree on a lot of let’s saydishes  in my previous example of me and Priya maybe if we  

have two by seven similarity we may not become  friends if we agree on a lot of food interest  

maybe will become friends. I don’t know whether if  this is true or not but assume that people become  

friends because they have a lot of common food  interest what will happen after we become friends  

once we become friends we tend to have all the  more common food interest you probably will hang  

out together and eat more often the same kinds of  dishes give each other at our testimonials of this  

kind of food in this kind of place and we may want  to try these things in each other’s presence.  

So the fact that we are similar makes us talk the  fact that we have spoken right now makes us become  

all the more similar fine this plot was given  to you forreasons ofbetter explaining things to  

you but but the fact here is if you look at the  text book this plot is the plot of the average  

by average I mean they have looked at all possible  people who have actually spoken and they observed  

what was their similarity measure measure before  and what was their similarity measure after this  

is the average plot that you seeah in the text  book which says that on an average people tend  

to become more similar and then talk and then  become all the more similar not just this.  

This this what’s coming next is what makes  the research all the more interesting they  

observed. I repeat on an average people become  similar people because this is average I told  

you this is average taken over all let me write  that down average taken over all possible pairs  

of wiki editors who spoke to each other sorry who  spoke to each other and we observed that that when  

they speak at this time there is sudden increase  in their similarity measure before speaking and  

after speaking this is the average curve ok. mega inference that one can make here is that  

this is steeper than this now that’s that’s  that’s the climax point in the piece of  

research this is steeper than this let me  write that this is steeper as compared to  

this this is more steeper than this what does  that mean that means that you become very very  

similar and then you talk and then you continue  to become better and better similar right. So,  

what is this and what is the take a minutes  pause and observe. What is this and what is  

this in our language that we are motivating  from the past few minutes which one is social  

influence and which one is selection when do we  select we select when people are similar. So,  

I am sure you would have figured out the  answer right now two people select each  

other to talk when they are similar this is  selection here happening on this part of the  

vertical t equals zero line selection and  on this part once they spoken to each other  

maybe they will share more common interest and  they will started eating pages together that’s  

what is happened that’s what this means there  is steep increase in the curve here as well  

and this is going to be our social influence . So we just now sawthe plot and we observed the  

interplay between selection andsocial influence  on a particular type of data set now let’s get  

a little critical and ask this question fine  the plot is very clear it says thatinitially  

there is a lot of selection happening here and  then social influences also happening selection  

seems to be more in play than social influence  while both actually happening that is with this  

dataset maybe this kind of an observation is  very context dependent dataset dependent.  

Now instead of taking Wikipedia dataset  if I were to take some other data set,  

let’s say the way smokers become friends with each  other maybe there is a lot of social influence  

factor there the unselection factor correct and  in fact obesity research says otherwise whatever  

we saw here it’s the reverse of that here there  is more selection than social influence while in  

the obesity research what we inferred was what  we inferred was there is more social influence  

than selection this is a this is a very context  dependent observation the people have made in fact  

thisthis opens up brand new questions namely what  is which is more in play in the society right. So,  

be it withfor bad habit’s is itmore of social  influence than selection for good habit’s is it  

selection and less of social influence one can  ask ask many such questions and thankfully we  

do have some datasets on which we can experiment  this questionsand and then find our answers.