 Okay, so what is this paper all about? So this paper is about crowd-sourced innovation and We are actually dealing with the other side in a sense that we are not concerned so much about the crowd workers But we are concerned about how the manager the community manager Can actually contribute to a crowd work and what he or she can actually do and Just to give you a brief wrap up what do we mean by online idea crowd sourcing? So what is it actually? It's it's an organization or a tool that helps you to develop new products or services in collaboration with your potential or actual users customers or employees and It could be one of these groups or it could be all of all all of them at the time or at the same time and so it's it's very diverse how this can take place and So a group of individuals whose very diverse skills and abilities is then trying to to solve a problem a solving test that that was developed by an institution or organization and then they are doing that and Before that can take place. There will be a sponsor. Usually this will be a firm or a company that has been identified by this sponsor and That will be posted on a platform and how the platform looks like and also be very diverse There can be platforms that are set up by the firm itself or there can be wide label solutions that are developed by software companies That help you to set up this platform. So any form you can see in practice So but the key idea is that a group of people is developing a new idea Okay, so just to make this clear in terms of what we are talking about There might be multiple categorization of how on an idea crowd sourcing takes place But one form how it takes place is crowd contests So that means that there is a specific problem that will be solved and then the sponsor offers a price to those Workers who have actually solved the problem and most of the other workers will probably don't get any price at all And they just walk away without getting paid for that. Then we have crowd cloud Complementers, which are for instance Platforms where you can develop further applications the most prominent examples probably an app store where you have an iOS system And then in an app store others can actually provide Software add-ons or something like that and they can also complement the original platform Then we have collaborative communities Collaborative communities means that there is a group of people who are jointly working on a problem or solving a Puzzle or they are using the same service. So you could also think about drive now or car to go For instance, there are people just sharing the same car And they are using the same platform for that and then we have this micro tasks where you're working on a very specific small task for instance renaming a file or something like that and The platform I'm looking at or the platforms I'm looking at they are mostly located in this collaborative community domain But they're also partly located in this crowd contest domain Okay, so what would be a typical example for an online idea creation tasks? So for instance Ford came up with this with this Competition where they said welcome to this ideas competition ideas for the vehicle interior Where everything revolves around the future or further development of the car? Use your experience and creativity to submit your own innovation approaches and discuss them with other members The best ideas are determined and awarded by a jury Below you will get a brief overview of the most important information and completions and then they will tell you what you get for the Participation in this competition for instance So then what happens is on the other side you have this people this diverse group and they come up with ideas For instance, they say it's something like that something that my wife is rather small and always has the problem that the Sun wizard Doesn't reach far enough down a downwardly extendable Sun wizard would be very good And then they say lowering the sill so the part of the car where you access the car is certainly useful But from a static point of view very complex So it might not work larger windows on all four sides would be nice So you can actually see better and then you say a dimmable interior lightning it would be nice When we are traveling with children often the lightning inside must be turned on when it's dark outside But this bothers you while you're driving, okay so then they come up with all of these ideas and the question is how can you actually motivate them to come up with these ideas and so our study is about how you are actually from the sponsor side so from the side of the Crowds working market, which could be considered kind of an supervisor or something like that How you could actually motivate the crowd to take part and we are interested in first of all how the sponsor can help to Yeah to to have people participate and The question is how they should behave so how the sponsor should behave that people are taking part in these activities And we are also interested in crowd dynamics because we know from other markets For instance that crowd dynamics are very particular for instance in Crowd investing people get very excited about a project, but at some time they are less excited And then nothing takes place in the internet anymore because attention disappears over time So that's actually also our theoretical Underpinning we have so one thing we are saying is that you know It's very difficult to get people excited in the internet in the beginning They might be very excited, but at some point they are just not excited anymore And they think let's do something else and if you think about the community manager on a crowd sourcing platform Well, this is actually the person who could help to get people excited again, right? But then the question is what should the community manager do to get people excited again and get them back on the platform? Because the attention of the crowd you could consider as a very scarce resource and the question is how you can tap it actually and So what could the community manager do to get more attention while he could or she could for instance ban specific Contributions so if there is HB hate speech on this platform saying like Ford never worked You know, it's just a shitty car then he might eradicate this this comment from the platform and if there are a lot of positive comments and these positive comments are very similar He might just simply merge the comments So you don't have to read all the comments all the time that you have seen before So that's what a community manager for instance could do So here's our hypotheses and we have three in total So first is if you think about a work environment usually it's like that social exchange theory very far back Would have told you that the experience you have at work also determines your outcomes, right? I mean if you if your work experience is pleasant if you go to work You think it's a great job then you might be actually more excited about it and you might actually contribute more and If you recognize the performance of the workers that might be a good thing for them for two reasons I mean if you recognize what they're doing then that might feel good like for psychological reasons They might simply say well, thank you very much On the other hand, they might also feel well if you recognize their performance Then you might think well, I'm doing a good job and if you do a good job, you might get paid better later on So for that reason that might be also good if you recognize their performance So community managers could also provide valuable resources in a sense that they could actually provide more information So they could actually say well, you didn't really understand the interior what we mean is only the front part It's not the windows or something like that and you could actually tell you what is actually meant by this task They are supposed to do But on the other hand if the community manager is too active and it's just making its own suggestions Then the crowd might not have to engage anymore because if the community manager is doing their job Then the creative process in such a platform might be reduced So for that reasons, we have two Hypotheses the first one is to say a community manager is actually a good tool to engage Crowd participation, but on the other hand, you shouldn't engage that much because if he engages too much Then there might be crowding out and the crowd might simply not participate anymore On the other hand, we are also Arguing that the content that the community manager provides a place of role So it's not only that he's present and says like well, that's a great thing that you're all here So it's a quite a matter of what the community manager actually says so and from psychology literature We know that there is this transformational leadership model for instance, and that means if you try to stimulate your workers intellectually or if you actually Consider them individually then that might be a good thing and they might respond more often But the thing is if you think about a crowd community It's very unlikely if you touch it each an individual Crowd worker individually that is a huge task And if you only approach one individually that might not transfer to all the other guys so if you are very friendly to one person and You consider this person individually and approach that person Others might not start to to work more or come up with more ideas Because they are not targeted by the community manager So what we are saying is that intellectual stimulation might actually crowd out to other workers But if you consider them individually that might only work in small groups because you even have the time to consider them Otherwise if you have 1,000 people a day that might not work in the first place Okay, so here's the data where you're using we have 22 large and medium-sized International companies from 2011 till 2016 who have posted their projects and these are all crowdsourced innovation projects and The data comes from large software vendor So most of the papers we have seen so far looking on innovation on the internet are looking mostly on one particular platform But due to the fact that we got it actually from the people who developed the software for these platforms We got 22 of these different platforms and So but we cannot tell which software whether that is Then we use the data and what we do is we aggregate the individual activities and look only on the daily level So what we are saying is like we're interested in how many activities took place on a specific day on a specific platform by a community manager or for instance by the crowd and so on So how does it work? So most of the platforms or we are observing you in the beginning have a suggestion phase so in a suggestion phase everyone can make suggestions people can make comments and So they can suggest whatever you want like you have seen before For instance with this Ford automotive examples everyone can make a suggestion Thereafter Once you made the suggestions and the community manager might engage with that then you have to vote on these particular suggestions And then either a good idea came up or it didn't and if there came no good idea up Then there will be another suggestion phase and can make again suggestions and comments And you have again a voting phase and then it goes on and on and on and at some point the platform will stop it And we'll say well the sponsor is satisfied Let's just stop it here and then the process ends. So how do these Suggestions look like well, they're all user written well It's not like with Alexa or Siri or something like that where you can actually talk to the system but you you have to type it in and comments are usually as The name already says like comments on other people's suggestions So you can comment on what other people said when they say like four windows would be great Others might say well, that's a great idea But let's do five windows or something a bit and you can also upload media files So you can actually make a picture or something like that You can upload a video and you can explain better by this video or by this media file how your idea actually looks like and Yeah, and in the end there's always a voting phase. So what are we looking at? We are interested in two dependent variables. So what are we trying to explain one is the number of user activities? So that is either do they make suggestions? Do they make comments? Do they upload media files and then we look at all of them individually So one is just how much activity is taking place and then what kind of activity? Okay So our explanatory variables or what we think is actually explaining this user activity is obviously the manager activity so how much is the manager doing and How much contributions are they doing in which phase are you currently or you can only make suggestions or can you only vote? during a certain phase and then the user activities in the past So if the community is very active in the last period It's very likely that everyone else will be also very active in the next period if nothing is going on then Apparently also in the next period will nothing nothing will be going on We also have a couple of control variables So we control for instance for the specific project So for instance we would consider if there's a cash price or not that would be considered by our controls We checked the different platforms because there are multiple types of platforms on these From the software vendor you also look at the weekdays So is it Monday Tuesday or Saturday and we also check what period of the project it currently is Okay, so what we do empirically is that we have to specify a specific model and we use account data model Which is a negative binomial model in our setting and so it's a fixed effects model So it's taking care of any unobserved time invariant factors Okay, so let's come to the empirics law How does it look like as I already told you there are 22 projects in total? We observe more than 1,000 days where people were active on these platforms and the breath the platform or the project length Varied between 16 and 122 days. So an average people engaged in such a crowdsourcing activity 72 days and then they found a solution for this particular problem They also varied in terms of how many people participated in this campaign So in some of the campaigns only 16 people participated in other campaigns we had a lot of people participating up to eight hundred and seventy three users per project and In most of the project we observed that there was only two Suggestion phases followed by two voting phases and then they stopped the project and they most likely found a solution to the problem they had Also, if you look at the periods the suggestion phase usually took 19 days and the voting period was shorter So in seven days one week they voted on the suggestions that had been made and The first indicator that managers actually matter and you have to think about these platforms They don't have a professionalized process and how they come up with this manager, right? I mean it's it's pretty interesting in a sense that if you have an employer and then a supervisor they think very carefully about who that should be and For this crowdsourced communities, you know, they might come up with a person or another person or not no person at all And but you can see that on days when the community manager is active Then you have a hundred and nine sixteen user contributions and when he's not or she's not active Then you have only 31 contributions. So apparently it's it's quite useful The first thing we looked at is what happens if you have different periods What you see in the top here is that if the project goes only one month two months or whether it's a very lengthy project So this project might have taken a lot of months seven months or something like that Then you see the contributions by the users and you see the number of contributions by the users Increases in the end. So the most contributions take place in the end and then if it's a long project then it's the other way around and this is actually in line with Collective attention theory saying that you know in the beginning you have a lot of tension and then the attention just disappears But it's surprising to see here's the other way around actually here The attention only decreases really in the end and if you think about it This is a phenomenon you see for instance on eBay where you have phenomenon like sniping You know, you want to be the last person to make a bit, but a thing is it doesn't really make sense here Right, I mean it's it's not that you keep your last idea till the last second And then you you put it on the internet and you say like listen This was my idea and then they implemented because you also want to have comments on your suggestions And if you look at the managers, you see that there's actually no big correlation, right? I mean you see a year it goes all to the end here It's starting in the beginning but the managers what they're doing is pretty erratic no matter how long the The campaign lasts if you look at the time of the day here you see the user contributions on a certain date of the time of the day you see at midnight and then it goes up to 10 o'clock at night and the red bars tell you if the manager was not active and the blue bars tell you if the manager was active What you can see is first of all if the manager was active all the time Then you have also more user activities But you can also see that when these users are active they mostly get active during the regular workouts, right? I mean it shows you kind of that You know at 6 in the morning the first contribution start and then it goes on until 10 and at 10 it decreases again And there will be less activities So here are our first results our main results So the first one and I walk you briefly through this So it shows you that if a manager is one time more active And this is so-called incident rate ratio So it tells you that if a manager is active once more then users are active 1.7 percent more So actually users are also getting more active if the manager gets active and Then you could actually ask well does it matter how often he gets or she gets active and you could simply say well Is he active or is she active? Yes or no and you can see on a particular day if the manager is active then user activities Increase by around about 50 percent So if the manager is active at all then users also get active more often Okay, so that is the most important finding from that slide and you can see So it doesn't matter whether you're in the voting phase or whether you're on the in the regular suggestion phase that this effect remains What you can also see is that if a manager was active the last day So if he was not active today, but yesterday then this also still has an effect on the user activities today So user activities still increase but not as much as if the manager was active on a specific day Okay, so what happens if the manager gets active more often on a specific day So what you see here is if a manager gets active once twice three times four times five times a day How does this affect user activities and here you see again? This is again so-called multiplicative effect. So two means they get active two times as often three times It often and so on so what you see is if the manager is active once or twice a day then not much happens So it's not statistically significant a different from from one But what you see is if the manager gets active three times a day, then you have a significant effect and actually the manager Induces the crowd also to become active But what you can also see is that you know getting three times active on a specific day is not better than getting four times Active on that day only if you are active eight or nine times, then you actually increase crowd participation more So what it tells you is that the more Active you are the better it is for the attention of the crowd But you should get that active because the more active you get Also the less the attention of the crowd disappears which makes sense. I think which is perfectly Makes sense. So what else do we find? So the next thing we are looking at is I mean, how does manager activity? affect the attention of the crowd and what you can see is that it matters whether a manager makes Suggestions comments on the crowd activities or does it matter whether he uploads media files and you can see for the number of Contributions the crowd makes the only thing that matters is The number of contributions in terms of comments the manager makes on other suggestions So the manager shouldn't make its own suggestions It shouldn't upload media files so that the number of user suggestion increases means the manager must comment more on what the users are doing And the next thing we are looking at is we are interested in how does the manager activity Influence the specific content of the users So do users make more comments? Suggestions or do they upload media and what you observe is that if the manager makes suggestions him or herself That doesn't affect the crowd at all. So what the manager should actually do is he should comment or she should comment on other user suggestions or if he or she uploads media files explaining the things more better or better Then this has actually positive effect on other on other community users Okay, so what else did we do? We also looked at the type of content The manager actually provides so the manager could for instance Appreciate user comments and saying like, you know, that's very valuable valuable what you're doing and managers did this quite frequency So in four or three out of four times he appreciated or she appreciated the comments of others You also clarified and gave feedback. So that was very rare. So in 13% they provided feedback They provided new information quite frequently and they motivated the crowd to incur or encourage them to participate more and Finally, they also did this intellectual stimulation where they said well, this is a very intriguing idea Could you do this in some other form or could you do this better or what is the idea behind that? Okay, so what did we find well We actually found that the most important thing is that there was intellectual stimulation If you intellectually stimulated the crowd then they engaged much more than if you did something else What was also important if is to provide information or to motivate the crowd But we didn't find that providing specific feedback or to appreciate the work of a specific crowd worker was important And I think this makes also theoretically sense because if you appreciate an individual person that doesn't trigger a lot of people to Participate more because they might not even realize that you appreciated One individual person. So what's the the core of our results? So the core of our results is that First of all, the community manager has a positive impact So and this is interesting because most of the time if you look at these communities It's not that they have a Specific rule how they select community managers or they often come a talk with a suggestion Who should do it? Maybe some person out of the crowd could do it They know as a lead innovator for instance and then they make this very erratic in terms of who should do it But we know it's important actually We also see that the community manager should be active, but very stably so meaning that the community manager should be active At least once or three times a day But then doing more hardly has any effect on crowd participation So if you're there like every day and you do three times a day some commenting on other users then this will be actually a good thing and The question is what you should you actually do right? I mean should you provide more information or what should you do and what we see is that if you can Intellectually stimulate these people which makes sense in a crowd-sourced innovation context, right? I mean you should innovate something so in if you intellectually stimulate them. That's a great thing And you can increase Community participation But if you individually consider someone as for instance in a regular workplace where you might do this to to encourage people to To work better or to to come up with better ideas But that actually doesn't help in a crowd-sourced innovation context and finally What we also observe is if you look at the people when they are working What you see is that they are working during regular work hours and most of the time when they are doing it you see that This work hours You know, you don't know what they are doing actually So you don't know whether they do this during their work time or whether they do it as their work or whether they Do it on top and we think that is actually worthwhile to do further research. Yeah, that's pretty much of the results So thank you very much and I have to apologize that the technique let me down and so thanks very much Thank you. I was just wondering because you were proposing a inverted u-shape Relationship and I didn't see that in the regression table. So did you find a? like squared term squared interaction or so Because it would we didn't put it in the nonlinear estimator because the problem is if you interact Variable in a nonlinear estimator. You don't know what you're really getting out, right because you're Interacting two nonlinear variables or like curves and you don't know where you actually are so you shouldn't do that but what we did was we looked at the Number of contributions individually so that was a slide which still popped up before shut down And what you could see is that you know one contribution doesn't play a role So two didn't play a role three played a role and then the effect size increased But if you increase it even more and we didn't show that because they were very little Days where you had so many contributions But if you if you put this together and you say like, you know, they made 50 to 60 contributions You could see that the effect size again disappeared and became smaller So yeah, so we consider this as evidence for an inverted u-shape actually Um Thank you for sharing that with us. I was wondering more on a on an analogy type of thinking The results that you found for the role of community managers Are they? This is just gut feeling right. Are they similar to what you would see in other types of communities? Because I mean innovation contests are very specific a decade ago. We would have called it open innovation type of things and Back in those days there was always the the strong assumption that what you see there in this Social process is stable across types of communities. Is that something that you would assume to be true here as well? well, I would guess for a couple of the results so saying for instance this thing with the intellectual stimulation I would guess this The effect size I would say is probably larger for crowd-sourced innovation communities, right? I mean if you do a micro task, you know, how can you stimulate the people intellectually? I think it's probably very difficult to do it. I guess you could do it No, I was more referring to there are for instance communities around people with shared interests that Like patients for instance, right where it's not about generating something new or solving a problem where it's about sharing a Mutual topic or interest where you see self-help Activities that type of things right and that's that's a different type of community setting. So I was thinking is that similar? well, I mean you could think about things like Wikipedia for instance where you say well you make this entry into this online dictionary and I would guess Similarity is for instance in the fact that first of all if you do the work for the for the crowd, right? I mean they don't participate anymore. That's a almost trivial result, right? Because but in this context I think it's very interesting because you have to show what should the community manager do We all don't really know what they should do, but it tells you don't have them make Do too much and I would say in the Wikipedia context, you know if you would create too much context or content yourself that might actually result in a situation where people are not contributing anymore because they are saying it's no longer their thing and This we got to the kind of contribution the community manager makes like I would say for this individual consideration the same holds right and I had to think a Long time about this why we actually see this result, but I mean it's pretty clear, right? I mean in a Traditional work environment Well, if you if you talk to someone like specifically like you know you might for she might feel appreciated the same Yeah holds in a classical work environment, but in this online Communities like Wikipedia or whatever type of community would be must not be idea creation If you're tortured someone individually, there might be some crowding out because they might say well That's a nice guy having this individual consideration, but you know you only You only tortured one person, right? And so this this doesn't work in in this crowdsourced Communities I would say And I would say it generally doesn't work, okay, I would say Since we run really over time And the coffee is waiting outside if there are further questions do not hesitate to quest to ask To ask me and we have a break right now