 Okay, we are now live excellent Hello everybody Welcome to the April 2019. We can media research showcase. I'm your host today Jonathan Morgan of the research team Today we have at least one presentation Our first presenter will be archong who's currently a data scientist at uber and he and arc will be presenting his graduate work at the University of Michigan iSchool on Group membership and contributions to public information goods the case of wiki projects Hopefully will be joined by swati goal after orks done and Hopefully swati will join us and present on her work on studying the thanks feature on wikimedia websites so Let so cross your fingers with me that swati joins us today because I'm really interested in hearing both of these presentations With that arc if you could share your screen with us and I'll let you I'll let you take it from here okay, thanks for the introduction Jonathan and Very excited to have the opportunity to present my graduate work at the wikimedia foundation So the topic sorry Sorry Give it actually give us just a minute. We have Looks like swati will be joining us. Okay, and it's joining us now So let's give her a chance to join up. Sorry for the interruption and Okay, good Also, your slides had gone away for a second. I want to make sure we saw those emerald do we have arc slides? Slides are up just one second awesome perfect um Jonathan, can you see my slides? I? Could tell just a second to go, but now I cannot see them anymore. I Yeah, I'm just to share to the full screen So I think that the full screen might actually be the problem How about this? Yeah Yeah, that's visible to both of us. So if you're able to move forward with the slides like that That sounds like that'll be ideal. Does that work? Yes, I'll pretend it in that way Excellent. All right, and now without further ado take it away arc. Thank you. Okay, cool So today I'm going to present My graduate work group membership and the contributions to pocket information goods the case of wiki project This is the joint work with my advisor yenshen and imam So now there's a lot of information goods online are produced by individual Individual users rather than one single publishers For example, when we go to online shopping We read the reviews on Amazon all these reviews are produced by individual users who either Who either have experienced that have the experience of the goods or we have used that and also they help Yelp online review in the case of health support networks the information and Support you receive are also produced by the peer patients In the photo sharing website the photos uploaded are produced by individual users instead of One single publisher. So The the study we look at is focused on what we call the public information goods In the traditional economics The public goods is characterized by two features one is non-rivers which means that the benefits of one individual from using a publishing of From a public good is not affected by the number of users who who Who are using that so this is by nature for the public information goods where for example when I read a Wikipedia article The benefits that I enjoy from reading it is not affected by how many people who are or who have read that article the second feature of public goods is non-excludable meaning that Nobody is being excluded the opportunity to consume a public goods and this is true especially by choice in many Public information goods cases like in the Wikipedia we We provide the the articles to all to everyone who has access to the internet and do not explicitly exclude exclude them either by by By forcing them to pay a registration fee or having an account so the production of public public information goods and Motivating people to contribute to public public information goods is an interesting problem. And the problems There are many cases where team and groups or organizations of individual users Has been has been observed in the production of public public information goods For example in the software development and community the github. We have a github team where Software developers who are interested in the same topic or in the same or in the same area Come together and collaborate with each other So the question that we want to answer from this study is how does the group membership team membership encourage one's contribution to the public information goods and Secondly, if we identify such such an effect What could be the potential mechanism behind the impact of that? So There were some previous studies looking at how the group membership induced the group membership can drive pro-social behavior Both from the theoretical model laboratory evidence and the feud evidence So in terms of the wiki project and editors contribution behavior our later talk more about What why we define wiki project as a group? So from the previous literature, there were many studies looking at How the wiki wiki project interacts with editors contribution behavior From the perspective of the experience composition of members within the wiki wiki wiki PDM members What is the mechanism through the division of labor or reduce reducing the coordination cost? one common limitation in this study is The causing to protect interpretation in in in these studies are not quite clear meaning that We are not pretty sure whether it is because Joining wiki wiki project It is joining a wiki project that drives an increase in the observed contribution behavior of or it is the other way So in this study, we want to view this gap So now let me give a brief introduction of what a wiki project is so the wiki project is a Coordination platform at Wikipedia, which is a web-based online encyclopedia Established back in 2001 As of last year it has accumulated more than 5.6 million articles with over eight 800 million single revisions It it's been constantly ranking among top five most popular websites all of the world So one feature of wiki PDM is that it provides free and open access to general public So on average, there were over over 500 million unique visitors each month So it falls under the category of public information goods So the wiki project was a is a platform that's introduced soon after the establishment Of wiki PDM. So a wiki project is a group of editors who work collaboratively as a team usually focused on a specific topic To improve the quality of wiki project articles. So for example There is a wiki project economics where editors who are interested in various economics work together to improve articles within that topic As of May the 2018 there were over 2300 wiki projects On on on wiki PDM so wiki projects provides The basis for a project identity or team identity through different features For example, each each wiki project has emission statements Articulating what's the scope of the project? What is the short term or the long term goal it aims to reach? Each project has also a associated project talk page Which essentially functions as a forum for discussion among members About so what's what is the goal which articles that? The the project should focus on For most projects there is also a user box, which is the demonstration of the common identity For example, the wiki project economics has a classic Demand and supply curve as its project Project user box. So if a user who is a member of wiki project economics, he can post that User box on his user page to demonstrate at the demo to to to demonstrate his project identity The fourth feature is each project almost all the projects has its own list of members Showing who are who have joined the wiki project So here is an example of wiki project economics Participants and there was a list of the unique name of the users and at the time When they joined the project So this is a brief description of what a wiki project is So the question that we want to answer is if one editor joins a wiki project What is the causal impact of that? Of joining wiki project on his contribution behavior So the data we look at Is from the the editing behavior of 9,134 registered editors on the english wikipedia So the selection criteria of this is we focus on the top 10,000 most active users The reason we look at this kind of a selective Center of users is that it gives us enough Enough activity so that we can draw the causal identification and it gives us enough power The treatment effect We exclude the bot accounts, which is which is essentially a computer program that usually aims at administrative work Among these around 9,000 registered editors There was 6,000 of them 6,000 of them eventually joined a wiki project. So the participation rate is around two-thirds The data includes for every registered editors here The timestamp for each revision What is the editor id the article or and the associated project of each revision if that article is under A specific project and how large the revision is measured by the bytes and the characters So we aggregate the user behavior on the monthly basis At the end we have a total of 800,821,000 editor a month of the observations in our data set So before I present the causal the code identification results, let's first look at A graphical evidence What's the behavior of a wiki project Member before and after joining a wiki project So these two figures provides a time trend of the edit size measured by the character number of characters In one revision and the number of revisions for each editor before and joining a wiki project So zero is the month relative to joining a wiki project The dashed line in this figure The solid line in this figure are the behavior of users who have joined a at least one wiki project And the dashed line are the Editors who have not joined a wiki project But who present similar behavior to a user who have joined a wiki project So the dashed line is more like a comparison group through which We can we can tell the difference in the behavior after joining a wiki project Here we can see before joining a wiki project the behavior of the two groups are pretty much the same But after joining a wiki project, although there is a decline in their number of revisions and the size of revisions Those who are in the wiki project decreases less compared to those who are not in a wiki project So the difference here demonstrates that Those who join after those who join a wiki project are more active appear to be more active than those who do not join a wiki project However, this this graphical evidence tells us No causal story of that the reason of that is although the average contribution decreases less less For the for those who join a wiki project compared to those who don't join a wiki project It could be that what drives this before and after comparison is the selection bias in the in the treated editors So basically Joining a wiki project is a separation of editors who are sufficiently active Versus editors who are not Active so the difference we see there is not due to the It's not due to the impact joining a wiki project. So what we want to do is tease out the difference and Find out among that gap. What is the causal impact of joining a wiki project? So the way we do is is we use an instrumental variable regression framework um We we we started with a regression equation where y it is an editor i's behavior um in in month t measured by both the number of provisions and the size of revisions As a function of whether one is in a wiki project or not So membership is a dummy variable that equals to one if editor i is has joined a wiki project in month t And zero otherwise So the parameter of interest here is beta one Which measures the treatment effect of a wiki project membership? so this regression would tells us would uncover the The treatment effect for beta one if membership status is randomly assigned among the users But in reality it is not randomly assigned And it is usually a self-selection process so what we want to do is find out an external shock like and and and an external force that affects What whether one is in a wiki project and or not and at the same time that external force is outside the control of of a user so that it it um So that it gives an exogenous shock to whether one joins a wiki project or not so that external Shock that we look at is basically an instrumental variable approach And the variation we look at is we look at the variation of one user's exposure To wiki project on the on talk page. So let me briefly explain what that is um One channel through which a user joins a wiki project is through the talk page of an article So let's say i'm editing an article at wikipedia, which is public good And as we can see there is a talk page associated with the talk page we keep Associated with the article public good. So on the top of the talk page um There is a list of the wiki project to which that article public good is associated with So when I editor a public good and I go to the talk page I will see the the wiki the three wiki projects in which The the three wiki projects to which public goods belong so I get exposed to The existence of three wiki projects and maybe I I I saw that there's a wiki project economics And I got interested to that and joins that so this join This this channel Through which one joins a wiki project Is the is the variation that we use to construct the instrumental variable So here is how we do it So the rationale is when when one revised the talk page of an article public good He gets exposed to wiki project economics and then he joins the wiki project economics so The instrument measures to what extent one gets exposed to wiki project through the talk page of an article so the instrumental variable is Among all the articles whose talk page for an editor That are revised by an editor and it eventually assigned a project The fraction of those that have been assigned one back them So let me give you an example of the instrumental variable suppose In one month. I joined a wiki project. So for that user we look at all the articles that He has contributed to let's say he has contributed to five articles public goods game theory microeconomics Another two the other two articles are article x and article y Among these five articles three of them are eventually assigned a project public goods game theory and microeconomics Then we look at at the time when that editor joined the wiki project whether uh, how many of these three are Are are are in the wiki project So let's say only public goods and the game theory are assigned to wiki project Economics at that time But microeconomics is not so in this case The intensity the intensity of exposure to wiki project for that user would be two over three because There were three articles that are related to wiki project economics based on the Based on their status as of now But at the time when the editor joins a wiki project only two of them are in a wiki project So the level of exposure Of that user to the exist existence of the wiki project economics is two over three So two over three is like a measure on how To what extent one gets exposed to the existence of wiki project and it drives the The chance that one joins a wiki project or not so In in economics instrumental variable is a very standard approach to Geocosal interpretation in a in in observation of study There were two requirements on the instrumental variable in the in the analysis one is what we call inclusion restriction So basically inclusion restriction requires that the instrumental variable in this case the intensity Of exposure to wiki project should be highly correlated With whether one joins a wiki project or not So we run a test and see that the first stage of statistics statistics is larger than 10 which is a Usually root of thumb Criteria so the not hypothesis of wiki instruments is rejected Basically, it says the intensity of exposure to wiki project that we construct serves as a strong instruments the second requirement exclusion restriction requires that The the the instruments that we construct is not direct directly affected By the outcome variable, which is the level of activity that we in in in this case so The exclusion restriction cannot be tested. It could only be argued. So the reason that we believe The exclusion restriction is satisfied here is the construction of the instrumental variable Which is whether an article article is assigned a project or not Exogenous to the editor because the the status of an article Is determined by the other users it is out of control of one editor. So that gives the rationale why we Why the exclusion restriction is satisfied here Okay, so here let me present the evidence from the instrumental variable regression here as a comparison I I provide You with both the ordinary least square regression and the instrumental variable regression The left panel of this table gives Fair regression using number of revisions each month as the outcome variable And the right panel gives the size of revisions as the outcome variable So both these outcome variables are the log transformation So the the coefficient here should be interpreted interpreted as the percentage change The Sorry the change in the log transformation So from here, we can see that joining a wiki project has a statistical and economically significant impact on the number of revisions both Both for the number and size Um from a backup envelope calculation The instrumental variable regression shows that Compared to those who were not in a wiki project joining a wiki project has a 25 times more Increase on the number of revisions in the month after joining a wiki project for a for a user And the size of revision increases by 10 times more We So then we are also interested in looking at not just revisions because revisions includes both additions and And the deletions so if we restrict our analysis to only revisions that are additions are the same results The same results holds so Having established that there is a large causal effect of joining a wiki project Then we want to see what could be the potential mechanism for the wiki project membership What is the reason that drives the The higher activity of users who join in a wiki project? So one of the plausible reason is that wiki project provides um So a large A large set of devices that gives recommendations to new to users through a very in various format For example, many wiki project has this collaborative collaboration of week Which has started in the previous literature Many wiki project maintains a list of open tasks Um from which users users can select the immediate tasks they can work on Many projects have the monthly news letter which gives a summary of tasks have done that are That are finished and to be done in the future So our conjecture here is the more recommendations project provides the larger the increase in in editors Contribution because the recommendations serves as a focal point Serves as a recommendation So that it decreases the search cost for a user and directs them to a smaller set of articles that they can work on So to verify this conjecture We need to come up with the measure for the number of recommendations and the measure we use is the size of home page So how large the home page of a project is um, the reason we look at the home page is So the recommendation devices that i've introduced previously suggest open tasks monthly news letters are usually listed in the In in the home page So the the larger the home page is the more likely there are more recommendations there So to the extent that The size of home page serves as a good measure for the number of recommendations we provide a regression results that With with number of revisions and additions at the dependent variable and the log transformation of project size as an independent variable So our result shows that um In response to a unit standard deviation change in the log of project size Among all the wiki among all the wiki projects at wikipedia There was a 8% increase in the number of revisions and 13 increase in the size of revisions And similar results if we restrict our analysis to only additions So the the basic start the the short story here is the larger the project size is The more increase in the average in the more increase in the activity of an average wikipedia member So this is here the revisions and additions are restricted to the articles inside A wiki project. So if I join wiki project and economics this table tells There is an eight percent The larger the project is there is an eight percent increase We are also interested in whether there is a spillover effect In terms of the articles outside the wiki project and the results shows that the Sorry to interrupt you. Um, we only have about a minute or two left So just want to let you know. Okay. Sure. Thank you. Um, so So in short what um, this evidence provides the support that um, the Recommendation through wiki projects could be one reason that drives the behavior of wiki project so I'm going to skip the The next result on the similarity of wiki project, which also speaks to our conjecture. So in conclusion We measured the impact of wiki project membership on individual contribution And the result of an instrumental variable regression shows that there was a large effect of wiki project membership both the number of revisions and And the size of revisions And uh through through observation analysis, we We provide support that article recommendation could be one of the reason One of the underlying reason of the estimated impact So this is all for my presentation and thank you very much For everyone who joined and uh, I would be happy to answer any question Excellent. Thank you very much arc. Um So for time, we're actually going to hold questions until after the end of the next presentation Um, when I think we'll have plenty of time for question I have I have one big question that uh, because as you know arc, this is something I've been interested in a long time So, uh, I look forward to having a conversation about it. Yeah, sure. Uh, so I stopped sharing my screen Yeah So, uh, so now we have uh, swati joined us stealthily Just uh, just after the beginning of arcs presentation. And so I will introduce now swati goal. I hope I'm pronouncing that right swati Swati is going to present her research, which she conducted In collaboration with layla zia And with ashton anderson of the university of toronto last summer When she was working as a volunteer with the wikipedia foundation research team Uh swati's presentation is titled thanks for stopping by a study of thanks usage on wikipedia And swati, um, I will uh, I will let you take it from here if you can unmute yourself Um And then see if we can get screen sharing set up If you can hear me swati So swati according to us you are still muted. Although I see you in the hangout Layla and I are working hard behind the scenes right now trying to debug this with swati um Well, uh, while we're doing that perhaps We can check in and see if there are any um Maybe we have time for a question then Isaac, do you have any questions from irc or youtube for ark? Uh, there's none on youtube on irc. I think there was some interest and just a little more Explanation around this confounding effect of choosing to join a wiki project and how that was dealt with so any kind of Re-vamping of that and talking about that again would be appreciated Uh, okay, so so So the confounding That we want to We want to clarify through the instrumental variable is that when we compare users who join wiki project versus who do not join Right there was a selection effect on that It's my so joining a wiki project could just be a sign that separate users who are sufficiently Motivated for those who are not so we want to find an external shock that that is Exogenous to these users. So these these are out of control for them So so think of it as think of the exposure as more like a nudge to these users So if I if I'm interested in an article and I added it and I saw the wiki project economics there It serves as a nudge telling me that okay, there was this um this thing called the wiki project economics versus another user whose As motivated as I am but just because some factors that he cannot control He did not see that sign maybe because that article was has not yet been assigned a project yet So that kind of a difference which is out of control for an editor um serves as a A shock a a local shock to them So that helps us to tease out the the the confound here So that is basically the the idea of how we how we how we use that Intensity of exposure to wiki project to users I had it so yeah, so that my question is actually kind of dovetails with that Layla is currently working with swati to to debug some fun technical issues But in the meantime, I'll start I'll start my question and then we can revisit it later if we if uh when when swanti gets her gets her technology issues result but So I my question is so you you frame this around group identity and group theory And group membership And I wonder if that's necessary And I wanted to know if what your thoughts on that are so so you're really the mechanism's exposure to the existence of these wiki project pages essentially That's not necessarily in my mind the same thing being aware that a page exists that contains resources Written by people who have similar interests to you Um, isn't the same thing as being a member of a group in and of itself um But I could definitely see that being aware of that could be helpful So maybe uh now now i'm good to say, uh, welcome swati and we can we can have this conversation after her presentation But um swati, uh, are you ready for us? I saw your video for a second there, but I do not hear you Okay, I can see your video. I think she can't unmute Yeah, that's really unfortunate um Do you have another computer Jonathan the other thing we may want to try is for swati just call in It's we have never done that but that could work So you can advance your slides while you're logged in here, but then your voice will come through the audio Yeah In that case, let us try The layla are you sending her the phone information? I I did Excellent okay All right, swati we're gonna make this happen um Thank you all of all of everybody out there and research research showcase land for your patients today Uh, Jonathan another option I have is just to call and so what is mobile number if she wants from the hangout Yeah, yeah, I see those options works. I see a thumbs up. Okay. Let me give it a try Excellent. Thank you layla All right, swati I think we can hear you Can you say something? Can you hear me? Yes, we can Excellent. All right, so let's and We can see your screen So, uh, all right great job swati. Uh, I'll let you take it from here I'm Sophie. I am a high school student and this past summer I worked with layla from the wikimedia foundation And um professor anderson from the university of toronto look into the thanks feature on wikimedia How it's used and how it might be impacted There are 14 edits per second made on A lot and until recently showing gratitude for these edits used to be hard And the thanks feature was created to address that problem it was rolled out on By 2013 in the aimless wikipedia and it has just been introduced across wikimedia And the idea behind the thanks feature is that as an editor I can go to the edit history Of any page and I can click on the thanks button next to any revision Whoever made that revision will receive notes Thanks them for that specific revision And then the fact that I think that they're not the revision that I think number four will be painful But we studied thanks. Um, this is the research station page at the bottom and Here's what we work Uh, I'm going to review my first look at Can we read these things? So that's a fact that friends of editors have sent or received things As an editor so those who are editing more frequently Are also using thanks more frequently And then less experienced editors think more experienced editors more often Which basically means that it's it's more typical for things to be sent upwards And now we're going to look at all four of these general bullet points specifically So then five percent figure has worked well over a couple of years ago more like More than four percent, but the message of space features has been increasing over time And what's interesting to note about this is that this upward trend exists even for projects in which the total number of editors has decreased The greater percentage of editors are using the same feature even in projects where there are Yeah, you know products where fewer editors in total And so then we looked at which projects you think and a ranked list of all projects is provided Among language with opinions you were the five languages in which there was the greatest percentage of the editor population Involved in certain things and something interesting to note about this is Especially at the lower end that the higher ends as well You have a couple of projects in which there are only one or two people thinking So there are definitely projects with things the things features used a lot But also projects in which it is almost Not used at all And so now we talk about the disparity in things Used within projects More experienced editors and those who are editing more frequently are as you might imagine things more often And this can be a huge disparity because sometimes the top 20 percent of the editors in a project So we're seeing 260 times more things than the bottom 20 percent And the main behavior with this is that things are typically set upwards This is not saying just that The more experienced editors are receiving more things But that's what experienced editors are typically sending things up to more experienced editors So here you see Editors range by percentile so the bottom you have the percentiles of editors Based on the number of edits they have made throughout their Wikipedia career And then the one of this is the things they have given So almost all the things are being given by the top 5 percent Which is to be expected because a lot of these lower percentiles are people who have only undergone the one that lets you try And the day of success received is very similar to the only real difference. Is that the top 5 percent? That's the top 5 percentile The average number of things received is 14 So it's interesting however is when you make this a ratio of the Things given to edits you'll see that the portion to the edits the number of things given is actually decreasing as We get higher than the other counts. This is showing the smallest subset of languages like the languages and so it's possible it's being more smooth If we were to do it across all projects, but both trying to have an every project exam Especially the get for the editing in 90 percentile, 95 percentile And then to look at disparity again the idea of things being sent upwards We have a sample set of a lot of languages and you can see that Even among all the editors so the bottom 20 percentile editors It can look at all the things given by these editors The average edit count of the people who gave the things Was lower than the average edit count of the people who received the things and the same holds true for tenure And the most noble example in which this is broken Is norwegian after the tenure graph And also norwegian for the edit count graph Which is interesting because it is such a clear trend in the other languages It could be used simply to A different structure of the project or Some quirk of the project relating to Wikipedia, but it could also have some larger cultural implications That might be interesting versus your work to consider And then this holds true even for more experienced editors So if we look at the top 20 percent of editors We still see that the average person giving a thank is less Is less experienced than the average person receiving a thank So the takeaways from this first part of the study are first that The usage rate of the things which are higher increaseable, but there is still a lot of room to grow Even among editors who give thanks There's still a small group that gives thanks at a much higher rate We've shown that there's a strong correlation between receiving thanks and having a higher edit count although there hasn't been any Part of the length thus far The most experienced editors send or receive the most thanks in absolute terms But if you look at the number of things they send and receive in comparison to their edit counts It's actually The least the lowest ratio or think ways or takeaways If we actually look at the number of things Sorry to interrupt you in all projects as opposed to the number of thankers and we control for population size We see a lot more variation in the number of thanks than the number of thankers The number of thankers is actually pretty constant, but the number of thanks given with regards to population size has a lot more variance We see that things are on average sent to more experienced editors. They're sent upwards And things are typically received in groups as opposed to being spread throughout the year So people tend to receive thanks in clubs and so to see More on some of these takeaways that word is representing this presentation. You can go to the documentation And now to go to the second part Of the study which was looking at the impact of thanks on Editor activities the thanks editors consistently have higher edit counts the day after receiving a thank When compared to their un-thanks peers and this holds for Novice editors and it also holds for experienced editors So what we did is we matched editors who received a thank on some day with those who did not Controlling for features such as total edit count and tenure and also a short term edit count So if I received a thank on some day in my sense or not And we had had similar edit histories based in the long term and the short term We could potentially be matched and then we would compare the subsequent edit count over the next few days And the next few weeks As we mentioned and see if there were any differences Uh, originally we tried propensity score matching But then we switched to ground-trace matching because We showed that there was a correlation between higher values for each of the features we use and a higher future edit count So the way we did the matching is we looked at two editors and we tried to match only when All the features for the two editors were very similar favoring The un-thanks editors and so I mean by that can you see in this example result? This is an example result from Polish Wikipedia Where we took a group of editors and we attempted to match Thinked editors with un-thanks editors And what you can see is that if you look at the average values for the thanks editors This is the un-thanks editors for tenure edits Previous thanks short term for this edit short term previous thanks These numbers are really similar But the average for the editors who was think is Rightly lower than the average for the editors who are un-thanks Which would indicate an expectation that the un-thanks editors would have Would be more engaged in the future But if you look at the next day edits of those who were think It's much higher than that for those who were un-thanks And the reason we did it this way is it shows that It is likely to be receiving a think on that day that caused The discrepancy in the future edit count that we saw And we also show that this isn't just because of a couple of results Screw the data because overall Within the matches and the majority of matches The spanked editor was the one who had a higher Future edit count compared to their un-thanks match And we looked at these results in portuguese with the pedia and metawiki as well So what do these results mean? What it means is that we should consider possibly making the future more easily discoverable by editors Or just giving thanks as it seems that they can increase short term edit activity and can act as motivators But it also raises a lot of interesting questions to pursue further One of which could be the cultural implications of how the thanks feature is used Another is trying to see or trying to show that Thanks features could come down to have long term effects So we could expand first expand the study conducted here to more projects And then try to show not just that receiving a think Is strongly correlated strongly linked to having a higher edit count the next day But that receiving multiple thanks possibly leads to more engagement and more edits over the long term course of all with media in this career Thank you. Thank you too Thank you very much Wati I'm so glad we were able to make that happen. That was a wonderful presentation Uh, so first we will go to We'll go to Isaac uh to check in Isaac, do you have any questions? Isaac Johnson research scientist with committee foundation Do you have any questions from irc or youtube for swati? No, no questions from either Excellent Um, then I will open it up. Does anybody uh in the hangout right now have uh questions for swati And you know, I've got tons of questions. So get them in now if you have them All right, you asked for it Uh swati, I thought this was really really interesting. So I have I have a couple questions first. I think um So these are these are not necessarily questions where I think that Um, I guess I'm more interested at this point in in learning What you think after you've spent so much time studying the usage of this feature um This is my first question is Why do you think that there's a such a pronounced disparity between the people who give thanks And the people who receive thanks in terms of both their tenure and their edit count So I think part of it is probably due to the fact that if I am editing more I am more likely to receive a thank um I think it's possible that that could be the entire explanation Except for on Norwegian right That struck me too. It looked like the trend was reversed on Norwegian if I was reading your chart correctly Was I wrong? So so that's interesting, right? So I I think I I think I agree with you that it makes sense the more you edit The more likely you are to get thanked for an edit. There's more things to thank you for um, but then How would we account for uh, a situation where the where the the The dynamic was reversed, but it looked like a basically the same strength but in the other direction What's broken about norway? as well, um, I was thinking that it could potentially be because um In some languages, there are very simple people who are thinking who are making up a majority of The things that are sent it could be just a couple of Or actively editing very actively but also very actively And what I was thinking that it could also cause Sort of broader cultural impact especially because for some of the other studies So we lost we lost your audio there um I think I got the first part of your comment It sounded like your first first comment was well This may be that that the overall You know number of thankers and thank instances in norwegian wikipedia was so small that You know, you see a lot of skew in the data compared to some of the other wikis And then the second part it sounded like you were saying that there were there could potentially be kind of cultural Um culturally mediated reasons why you would see this in this wiki um I didn't hear any more than that though. Did you could you have more to say on that topic? That was basically what I was saying. Yeah Um, I was also saying that there were a couple of other instances in which Norwegian was a little bit different than the languages we studied. Um, for instance in I believe it was you know ratio of fingers to editors especially in the The people who were the more experienced editors was higher than Interesting so norways an outlier and other in other dimensions as well. Thank you Um, Isaac, uh, it sounds as though we have another question from IRC. Would you like to relate it? Yeah, actually two one. I think is a clarifying question, which is did the research look at any non wikipedia, uh, We could wikis just you know essentially what were the The universe of wikipedia is that we're studied. Um, so I'll let you do that one because I think that's a short one And then there's a call So, yes, um We did a couple of the studies across all projects and then for some of those For instance the looking at the A, um A disparity between senders and receivers those we only did in language with a couple of a set of language wikipedia but then for Looking at the impact of receiving the things on editors Patrons we expanded that to Okay, thanks. And then the second one was whether you noticed whether the thanks clustered on similar edits And whether multiple thanks maybe had a a larger effect On a given user to the second part of the question About multiple edits may be having a larger effect that did not study, but it's something that I think somebody that should See if that would have a larger impact But yes, we did notice that uh edit clustered or think clustered around us Awesome. Thank you once again swati. Um, so we have a few more minutes that we can go and I'd like to open it up um for Uh, if we have another question or two for either arc or swati today um now is the time And if nobody else has questions I can always just continue with the question I was asking Anything from irc or the room for swati or arc? Nope Okay arc So basically my question was do you really need to consider wiki projects to be groups in order to Perform the analysis you are performing or under or interpret the results that you were uh, That you discovered Yeah, so, uh, I think let me answer the question to that In in two parts the first part of our result is more like um Considering wiki project as a as a platform right where people have similar interests into that And it happens that it shares many feature common features we see in the groups They have a common interest that they have a common go So the value of the first part of our research is to give a causal um estimation, um if I join such a platform, which Which which is pretty like if the usual teams and the groups we see in online offline And then communities what's the value of that? them By saying that joining a group there could be many ways Through which a group motivate people right so previous evidence have shown that By joining a group you will see emotional support from in-group members Like what you did you You received things from them and there were many wiki project please wiki projects where a new a new user is is automatically Is automatically received automatically receive some written message or a thumbs up For the contribution so all of these could be driven. So what we have shown is um, the resources provided by a wiki project and um gets Um gets users who restrict it to a small set of articles that they can start working on So reduced by by coordinating their effort. It is one of the channel. So in that sense, I think This speaks to one mechanism One plausible mechanisms through which um And what can can can motivate users can motivate users Yeah, that sounds reasonable to me um Okay, well before we have a couple more minutes before we get shut down. So I have one more question and it's for swati uh Swati, have you received or given thanks on on wiki before and uh When you gave it, what did you give it for and when you received it? How did you interpret that? So the real editing that i've done on wikimedia is um for writing the research page um, and receiving it thank the only thing I received was from raila, but um I actually Even knowing the person who it was um and having seen communicating. They were on a daily basis It's so so really good to receive the thing. Um, and so for me it was very believable when we got the results and we saw the Large impact of large where appears to be a large impact of receiving a thanks on having a higher edit activity the next day Excellent. Well, I'm glad to hear you say that because I just thank you for your latest edit to the understanding thanks research page on meta So I hope that will lead you to make many more research edits to wikipedia projects in the future Also for everybody out there listening I linked to that research report in the showcase uh in the showcase abstract on media wiki.org which is available through the rc channel And also below the video on youtube. Um, so if you'd like to find out more about that paper Arc if you have any public documentation around your paper I'm happy to link that from the media wiki page as well so that people watching this in the future Can follow up with that you can just send me anything over email Sounds good. We'll do that And with that thank everybody who joined us today This has been the mark the april wikipedia research showcase and we will see you all not next month We're taking next month off. Um, so we will see you all again in june. Thank you very much