Disseminate: The Computer Science Research Podcast - Marvin Wyrich & Justus Bogner | How Software Engineering Research Is Discussed on LinkedIn | #56

Episode Date: July 8, 2024

In this episode, we delve into the intersection of software engineering (SE) research and professional practice with experts Marvin Wyrich and Justus Bogner. As LinkedIn stands as the largest professional network globally, it serves as a critical platform for bridging the gap between SE researchers and practitioners. Marvin and Justus explore the dynamics of how research findings are shared and discussed on LinkedIn, providing both quantitative and qualitative insights into the effectiveness of these interactions. They reveal that a significant portion of SE research posts on LinkedIn are authored by individuals outside the original research team and that a majority of comments on these posts come from industry professionals, highlighting a vibrant but underutilized avenue for science communication.Our guests shed light on the current state of this metaphorical bridge, emphasizing the potential for LinkedIn to enhance collaboration and knowledge exchange between academia and industry. Despite the promising engagement from practitioners, the discussion reveals that only half of the SE research posts receive any comments, indicating room for improvement in fostering more interactive dialogues. Marvin and Justus offer practical advice for researchers to better engage with practitioners on LinkedIn and suggest strategies for making research dissemination more impactful. This episode provides valuable insights for anyone interested in leveraging social media for advancing software engineering knowledge and practice.Links:ICSE'24 PaperMarvin's HomepageJustus's Homepage Hosted on Acast. See acast.com/privacy for more information.

Hello and welcome to Disseminate, the computer science research podcast. As usual, Jack here. Today is another episode of our Cutting Edge series and we're going to be focusing on science communication today, so very on brand for this podcast. And to talk about science communication, I've got two guests with me today. I've got Marvin Wyrich, who is a software engineering researcher at the University of Saarland in Germany, where he focuses on empirical software engineering and psychology. And I also have with me Justice Bogner, who is a assistant professor in software engineering at the Free University of Amsterdam, also known as The View, where he also focuses on empirical software engineering. So the specific topic of this, or focus of this podcast today,

will be a paper Marvin and Justice published recently at the International Conference of Software Engineering, and actually won the Distinguished Paper Award, so well done for that, guys. And the paper is called Beyond Self-Promotion, How Software Engineering Research is research is discussed on LinkedIn. So welcome to the show, guys. Thanks for having us. Yeah, thanks for having us.

Cool. So it's customary on the podcast, before we get into the fun stuff about science communication, for you both to tell us a little bit more about yourselves and how you ended up where you are. You go first, Justice. Okay, sure. Yeah. Yeah.

So basically, when I was close to finishing high school, I actually wanted to study classical music, but I was just too bad. So there was no chance of this happening. So yeah, I did the next best thing and studied computer science. Yeah, somewhere during my master's, I basically discovered that there is this other route, not going into industry, but doing a PhD. And yeah, a professor encouraged me to do that.

I did it. I did my PhD on software engineering, basically on the evolvability of microservices. It was a software architecture topic. And yeah, towards the end of my PhDd i actually had the intention to go back into industry but i i got an offer for a postdoc position and i didn't think that it would be so easy and uh yeah then i thought let's try it and yeah it turned out that i during these years i really yeah discovered my my passion for research and decided to try an academic career. And yeah, until now it's working out.

So I'm an assistant professor at the FU in Amsterdam and I'm happy on my tenure track, basically. And yeah, decent chances by now, I think, to get a permanent position. Yeah, that's it. Fantastic. Do you still play music in your spare time as the love yeah what instruments are they that you play play so I play piano and um also sang basically uh classical music amazing I've been trying to learn the guitar for years and I can let I can play Wonderwall by Oasis and that's about it. So yeah, I'm very limited in my musical skills.

Anyway, cool. So yeah, over to you, Marvin. What's the story? I would say I have even less talent in classical music than you. I've heard him play the piano. It's actually quite good. So for me, studying software engineering was the primary choice. I liked it.

So I studied it at university. And during this time, I also spent a bit of time working in industry as a software engineer. But then I decided to stay in academia and do a PhD. And I think the main reason why I became interested in research is, I think, because I noticed that generating knowledge is really beautiful and I don't know it's it's exciting in itself and of course as a bonus you you can have quite an impact and help other people with it and that's I would say that's how I why I stayed in academia and started to do research and now yeah as you said I'm at the university of salem in germany as a postdoctoral researcher trying to figure out where my path continues yeah awesome yeah i totally agree

with it being a very gratifying process of the creation of knowledge is very it's very enjoyable very rewarding and cool yeah so let's get on to the paper and science communication then so before we do that let's set some context for the chat and tell the listeners and science communication then so before we do that let's set some context for the chat and tell the listeners about science communication kind of what it is and how it in theory should play out between practitioners and researchers so yeah i don't know who wants to jump in on that first maybe you marvin yeah all right so in theory i would say as a researcher you publish a paper and then you do something to make your target audience aware of it and you do something to make your research accessible to them. For example, you write a blog post that summarizes your research in very simple language or you write a post about your research findings on social media, for example, on LinkedIn.

And your target audience, for example, software practitioners, they pick it up from there. I would say that's how it should work in theory and what science communication can be. In practice, however, I would say that our software research community is really good at discussing a lot the impact and relevance of our research. So we are very critical about this but at the moment when we could have the most impact right after the publication of a paper yeah we do almost nothing i would say and in the worst case the paper ends up behind the paywall nobody will even get aware that it is there and the researchers move on to the next project and i say this this is the worst case because we have just invested a lot of resources in what is likely to be important research but it is also research that then reaches no one because let's be honest yeah do you know

single software practitioner who scans the online publication libraries and then pays a lot of money for a single pdf of course not right yeah i often wonder how much they actually generate through those individual sales of papers on on like the who's the company is else is Elsevier, Elsevier, a big one. And yeah, I wonder who was actually, I mean,

obviously they make a lot of money for licensing out to, I guess, universities and whatnot. That's probably where, why the primary money source comes from that. But I'm like, has anyone thought, yeah,

I really wanted this paper going to pay like, cause it's expensive as well. It's like 20 quid a go, I think, or like $30 or something like, yeah, I mean, yes,

it is very frustrating when it kind of plays out like that and ends up behind a paywall and you do all this great research and no one gets to read it. Cool. So given that background then, Justice, what is the elevator pitch for your paper or beyond self-promotion? Yeah, so basically we were very interested in how the current state of science communication is for software engineering. And we decided to study this.

So basically we analyzed a lot of posts on LinkedIn about software engineering research papers, found out several interesting things and then yeah synthesized kind of guidelines how we can improve science communication in in the software engineering research community awesome so yeah we're gonna let's get into the details now a little bit about that and find about how you actually went about doing this then so the first thing is so i've got it written down here on my kind of notes is the linkedin study so you identified to start off with like a few research questions right sort of how how how research is shared how people respond to this um research and then trying to look for some sort of positive examples of um of effective scientific communication so we needed a data set to answer these questions and for the first of all like why linkedin and

then how did you go about conducting the study and actually building this data set you could then analyze i'll come to you on this one marvin i think the interesting thing about linkedin is that you find both academics there and software practitioners so of course yeah a lot of people many people have an account. Of course, not everybody is there active on a daily basis, but at least there is a chance now that some interaction can happen between academics and practitioners. And this is why we found LinkedIn to be an interesting place to look for this communication. Regarding the methodology, we just looked for software engineering papers. We made our life a bit easier by selecting only papers that were published at the biggest conference in our field, two biggest conferences, so we didn't have to decide ourselves what is software engineering research and what is not.

And we tried to find all those LinkedIn posts in the first step that were about software engineering research papers. They either mentioned the paper paper linked to the paper or said a few words about the the findings and yeah then in the next step we analyzed the contents of these linkedin posts but also the contents of all the the comments that were made on these these linkedin posts so this is roughly how we went about this and yeah was this a manual process and if you scouring linkedin to look for these things or was there some way to automate a little bit of some api we can kind of query against to scrape things down because it feels like it could be quite laborious i mean how would you begin to search linkedin was it looking through groups was it sort of just looking for

hashtags or how did you approach actually the initial discovery of, yeah, these posts refer to these papers from these conferences? Yeah, that was actually quite tricky because the LinkedIn API is not made for searching content. It's rather made for recruiters finding other people. So, and in addition, LinkedIn doesn't, it's hard to find content on LinkedIn that is older than one year. And so in the end, yes, we tried to use the LinkedIn search, but in the end, we just went to Google and searched on Google for LinkedIn content. And we did so by looking for paper titles, for example, because we noticed that in a lot of posts about SE research, when researchers talk about their research they mentioned the paper title so this

was one way of finding those posts but we also looked for example for the conference names if they appear something like this so we really did a keyword search and then filtered manually for those that that fit our criteria yeah that's interesting that google search is better at searching linkedin than linkedin is better at searching linkedin which is kind of a bit of a strange phenomenon i guess but yeah cool so given you kind of went through this process then you collected all this data tell us about some of the characteristics of this data set that you ended up with then justice so like what how many posts were there what were the responses like yeah fill us in some of the data characteristics of what we're working with here yeah so we we basically um had some attributes that we could collect automatically so we had a had scripts basically that you know analyzed in

the browser the post and extracted some metrics from it so for example how long is the post how many characters how many comments does it have how many reactions how many reposts and so on and we also saved the complete text of the post so that we could later on manually analyze it but yeah in the end we also had some attributes that were more difficult to to automatically collect so uh all of the yeah i think close to 100 posts i think 89 or something um had to also be manually analyzed by both marvin and me so we did this independently um we checked out each post and then documented the attributes that could not be automatically synthesized so for example we wanted to study the post intention. So why did the people post this? And we later on synthesized a hierarchy of post intentions based on this.

But we also did this for the comments, for example. And what were the comments about? Were they, for example, asking questions about the research or were they congratulating the poster on getting a paper accepted? Or were they criticizing the results? Or were they confirming, based on their own experience, that this happened to them? So were they sharing experiences about this? So yeah, this was a lot of manual effort, because we had to really go through every post and every comment also. We also had to label if the poster was from industry or academia because

this was also fundamental to our study to find out how these two groups of people interacted with each other so yeah a lot of manual work involved in this data collection yeah i mean if you give us a rough sort of guide how many nights and weekends were spent scrolling through through linkedin and annotating and analyzing clicks and all this oh wow i'm not sure if i could quantify it but it was definitely a process that dragged on for several weeks i would say i'm not sure marvin if you have more concrete estimates but yeah that was probably dozens of hours i I would say. I think the good news is that we never felt the need to count the hours. So that boring collaboration was quite fun. But yeah, it's a tedious process in the end.

Yeah. I mean, at least you put in the kind of doom scrolling to good use and that there's a nice byproduct there because I spend hours scrolling social media anyway, right? So like at least you actually were getting something useful from it as well. So yeah, there is putting that time to good use for sure cool then so given that we've got this nice nice data set that we can then they can analyze tell us about about the results then marvin so what what did you find oh where to start so we have a lot of data i think the

oh yeah maybe two things that pop out to me. First, when it comes to the intention that you just mentioned, the post intention of the poster who creates such a LinkedIn post, we came up with four different categories. So your intention could either be that it's mostly about self-promotion. So you want to make other people aware that you are now there. You do research on this topic or you have a paper accepted so the linkedin post is mostly telling hey look at me i'm i'm i have i have a paper at this nice conference here but we what we found for example was that

this was only like 11 of the the linkedin posts that we that we found in our sample they were about self-promotion then if you climb up this hierarchy, we come to paper awareness, or what we call paper awareness. Those are LinkedIn posts that tell, hey, look, there's this paper. Here's the link. It's about this topic, and that's it. So not too much info on the results or some bullet points on the results. For example, this is when we reach the next level in the hierarchy,

the results for example this is when we reach the next level in the hierarchy the results awareness but yeah most papers where our most linkedin posts were really about this this paper awareness category and not yet reaching results awareness and then on on the very top of these four categories there's results discussion where you then even add some i don't know implications some food for thought so that you provoke some some discussions and only very few linkedin posts do this so far so this was this was one finding and was interesting for us because it has this direct implication that you you can you can basically climb up this this ladder and improve your science communication game i think we can talk about this in a in a second a bit more and the under the other very interesting thing, well, actually two more things.

I know that I remember. First, we saw a lot of people from the industry commenting on these LinkedIn posts. So, yeah, a lot of LinkedIn posts don't receive any comments at all. But for those comments that we had in our sample, a lot were from the industry and a lot were not only congratulating the author of the accepted paper, for example, but really like showing some interest, some confirmation that they made the same experience and they had some questions and stuff like this. And this was really nice to see that there is some interaction going on. But really the most interesting thing to see for me was that the situation was really different from what we

expected. So we thought that software researchers would communicate their research to practitioners. And what we really saw was that often practitioners took on the science communicator role and discussed software research within their practitioner communities. And to us, this was an indicator that I think first software research seems to actually be relevant if practitioners talk about it on their own initiative i think that's a good sign and second not enough researchers likely talk or take the opportunity to to tell people about their research because right now practitioners feel the need to to step into this science communicator role which is traditionally ascribed to scientists so yeah

that's that's fascinating that it's sort of the practitioners taking on the on the role there because you like you say you'd expect it to be the other way around right but i guess it may be a it's a little bit of a case of maybe i've been guilty for this in the past of like i'll write the paper it'll get accepted i'll present it and then i'll never want to think about it again because it's been i'll move on to the next project, right? So I'm not kind of taking the extra step to be in a communicating it effectively to the community. Is there anything you'd like to add on the results there, Justus? Yeah. So maybe what was also interesting for me was that the discussion was very different if the poster on LinkedIn was from industry or from academia. So basically when the poster

was from academia, then 60% of the comments were roughly also from academia and 40% from industry, which was very nice. But still the majority of the comments basically were only congratulating the poster. Okay, nice that you published this paper. But from a science communication perspective, they were really not so interesting. The real interesting discussion mostly happened when the post was from industry. Because then, yeah, the high value comments appeared,

like sharing experiences, confirming the findings, asking questions, maybe even criticizing. And there really was a discussion going on, simply congratulating and that's it. But yeah, what was also interesting, that industry was a bit discussing among themselves. So the vast majority of the comments were from other industry people. And academia was not really involved in in this discussion anymore which was a bit sad but also understandable because the i think the bubbles are just not as intertwined as they could be or should be

yeah that's what i was going to ask because i mean i guess a lot of the way that discussion happens on any given forum depends a lot on the actual structure of the forum how linkedin allows you to engage with i mean i'm not a massive user of linkedin so i'm not too sure how it works here compared to twitter or x whatever it's called these days but do you feel like that a lot of these this is a a problem of like because i'm friends with my academic circle on linkedin therefore i'm therefore they're my direct connection. Therefore I'm more likely to see their sort of stuff and engage with that.

How much of that do you think this is sort of an encoding of the way that LinkedIn is sort of structured, I guess is probably my question. Yeah, I guess definitely that's one of the reasons why this happened. I could also imagine that it's related to how these people share differently. So in many cases, when the posts were created by somebody from industry, then it was often very nice from a science communication perspective. nearly professional science communicators that kind of have it as their side job or even their

main job, I don't know, to share technology or research findings on social media and kind of build a brand like this. So they, of course, spend decent effort on making these posts easily consumable, presented in an engaging way, maybe with images or call to actions or whatever. And then, of course, the chance should also be higher that people share it and reshare it and engage with it. Whereas if an academic simply posts, yeah, I got this paper accepted at ICSI, an A-ranked, a core A-star rank conference, then of course, this is not so motivating and nice to engage with. i'm just gonna click the like button and move on right and i'm not gonna yeah yeah right congratulations and that's it

basically yeah because it gives you the automation link to my yeah it gives you a few things you can say right just click one of those and yeah and move on and i as we were kind of discussing the results so there's a few sort of questions that jumped out about sort of like concrete numbers of things from the data set so for example you mentioned that a lot of put a lot of posts didn't get any interaction at all what was the breakdown there was it is it was it like how many didn't get any interactions at all so if i remember correctly so we have to kind of differentiate between. So comments, there were definitely a lot of posts that didn't get comments. But yeah, there's also this kind of low effort response that you can do on LinkedIn, which they are actually called reactions. But they are these emoji based reactions.

So you can basically, I think this is even called like, then celebrate, support. And they have a couple of more emojis that you can use to to react to a post and these were actually used fairly frequently so i think there are not so many posts in our sample that really have none of these so definitely people reacted to this but yeah actual comments that was much rarer because there's more effort involved, of course. So if I remember correctly, it was something like 40% having no comments or 30%, so yeah. And then others having like one or two. And yeah, I think only like the top 30%, if I remember correctly,

has more than three, four, five comments, yeah. Yeah, cool. It of on the on those ones that sort of climbed the hierarchy and got all the way to the top have been like other positive examples lots of results discussion how big was that sort of thread on those really like kind of that's the word i'm looking for yeah the ones that climb the hierarchy these posts that actually were really good examples of people communicating well effectively on them how big did they get the discussion on them in the comments yeah like kind of how many people are engaging with it as well did you measure that like okay this got like this was just two people talking all this was like

three four people arguing about this paper that they didn't like or yeah yeah so we don't have data on the actual number of people involved but but more on the comments. And definitely were posts that had like 20 plus comments. Yeah. In some cases, there were even posts that came close to like 50 comments, but this was more than the exception. So these were really the top posts that also had topics that were really hot

for a lot of people. I think there was one about developer motivation and its relationship to technical debt or something that really went off the rails. Also AI-related posts sometimes where people very much engage with them because they found it interesting and so on. But yeah, there was a lot with barely any or just a few comments. Yeah. Before we get on to the answer how we

can go about improving science communication it's all i always like to ask kind of the about the limitations or scenarios about kind of when a study whatever maybe it's suboptimal so yeah if you review the study you've done and look look at kind of what you did are there any things that you kind of have to find okay this is maybe a limitation of the way we try to approach this problem? Marvin, I think you want to jump in on this one. Definitely what I mentioned before that LinkedIn makes it kind of hard to find a lot of posts. So we can't make any good estimates on the percentage of research that is shared onin in total and in comparison to what could be there so we had like i don't know 98 linkedin posts and in the end in our sample but we can't

tell how many we missed this doesn't mean that then anything of what we found in terms of category shares and so on would change i think we we had some nice information saturation in the end but yeah definitely there could be more than what we have found yeah the other limitation i would say is we limited ourselves to to papers from two conferences maybe there's a difference for other conferences for other communities i mean communities can be very different even within the the bigger software community right so maybe there's there's a sub-community that is more on the social side of things i don't know who like to go out and put

themselves out and themselves out and and communicate a bit more and this could could be different then but yeah for the for the greater field of software engineering research i would say we have a quite accurate picture on linkedin now nice do you feel like linkedin is a good is a good platform for disseminating research and discussion as it or do you think the platform could be improved like if you would have liked defining um building a platform from scratch to encourage communication would you would you would end up looking like linkedin or would you think you'd build it completely different? That's a good question. I think there could maybe be some more mechanisms that encourages communication.

So I think in the end, it all boils down to communication from both sides. So when we talk about relevance, sometimes software research gets criticized for not being relevant enough or something. Yeah, okay, then let's talk or make or let let practitioners talk to academics give them some some platform there some way to to approach us and also the other way around i think linkedin is very generic in its functionality so you can post people can connect with each other and comment but maybe maybe you could implement some mechanisms to foster collaboration a bit more among industry and academia but overall i would say it's it's not the worst platform so you can more or less you don't have to pay to use it yeah of course with

your data but other than that it's free to use so everybody can create an account and you already find a lot of people also a lot of people from your target audience on linkedin so i think it's free to use so everybody can create an account and you already find a lot of people also a lot of people from your target audience on linkedin so i think it's okay yeah yeah it's interesting the link tonight i've been a bit on all the major social media platforms for for a long time and i find people and myself using linkedin more and more i don't know if it's a fact of like as you get older you kind of go through the cycle of like i mean i don't i've not touched snapchat in years i've not not got i've not got linkedin sorry not linkedin sorry tiktok i've not i've not been i've not caught that bug yet so yeah i don't know yeah i can't remember what point i was even making there but anyway just just to add one thing

so i think it was it comes down to preferences so linkedin in my expression is less polluted with personal stuff so it's it's a networking it's a it's a work-related networking website so this could be attractive to some but also to science communicators we don't propose that everybody now has to go to linkedin and communicate their research there so if you don't feel like it if you don't feel like this is your platform then then try something else There are a lot of ways of disseminating your research, right? I don't have to tell you to do the podcast on this. But of course, if you feel like you would prefer to write a blog post and you also feel like your target audience would like to read blog posts, yeah, of course, then write some blog posts.

If you feel like, yeah, maybe I try doing some videos videos there are a lot of platforms for this i think our message is just do something and don't just end your work after the publication of your paper yeah i think that's that's that's a good message for sure i'll come to you on this one as well just as so kind of how we can improve science communication obviously i seeded that last question within like how could we build a new social media platform to improve this but assuming we keep that when that thing that constant what are the other ways we can improve? Or what are the findings from your research that kind of like, okay, this is one way we can definitely improve science communication. Or what are the ways you think that we can do that?

Yeah, I think one finding that I find particularly relevant is that it's sometimes not even so much more effort to to improve your science communication because especially if you are in in these low levels of the post intention hierarchy where you post something anyway and already spend the effort on creating a linkedin post then why not spend like a couple of more minutes to enrich the post so that it becomes engaging as the results in it, the key takeaways, maybe even some opinion from you about the implications of this. Yes, it will take more time, but I think it's also worth it, especially if you have a study where you think the implications for industry are are really important then then yes please spend a few extra minutes and and do this i know academics

are notoriously overworked and and have tons of stuff to do and yeah then people come and tell you yeah and now you also have to do science communication on top of this and i understand that this is is not ideal and yeah some people you know like this more than others and that's perfectly fine but i think also that the universities should should value this more in in their evaluation criteria so for example at the few there's also a criteria a criterion called valorization and outreach. And this plays a role in your evaluations for tenure and also for promotion to the next level, basically. And I think this is nice because some people might want to really focus on this because they are passionate about this and then want to spend time on this that they don't have,

I don't know, for grants or more papers. And we should still value this. If the research is important to share, then people should be incentivized to also spend their effort on sharing it. Yeah, I agree completely. It's a key step in the life cycle of research, right? And the only way you can kind of get that to happen is incentivize people, right? And if you can put the incentives in the right place then yeah hopefully it will it will it will happen

cool so yeah i guess coming back to you on this one marvin where do you go next with this this this research and have you got another study plan for a different platform or yeah what's what's next so i think what jesus just said about the overworked academics is very important. And a lot of people may not feel right now that they have the energy to spend on science communication activities. But since this topic is really close to my heart, and I believe that research needs to be accessible to way more people than it is right now, my suggestion for the way forward is to demonstrate to academics that science communication can be a very rewarding activity. And this is not only about numbers like an increase in citation counts or something like this.

So this is really about this feeling that after you spend like two years on a project where you planned and conducted a study, you maybe developed a tool and you got your paper published. After all this work, when somebody tells you, hey, thanks for this interesting research. I made exactly the same experience. So nice to have finally some data on this. And this confirmation from just a single person of your target audience, it can really make your day. And it can actually fill up your energy level a little bit, right? It can really be rewarding. But we need some data on this, actually, because I get this asked a lot. Is it worth it, right? We need some studies that show the effects of science communication activities on awareness and impact of our software research.

And maybe we also need some studies that discuss the individual challenges that researchers face in communicating their research and how we can overcome these challenges. That would be my proposal for some future work, how we can build on what we already did. Not so much studying other platforms. Of course, this could also add, but I think what we found is more like people are not so aware that this is a thing to do or this is something that can be rewarding. So maybe we should start with this and then everyone will find their way to disseminate their research. That's what I hope for. Yeah, like kind of going beyond the citation camera and kind of the thing that they mentioned as well about kind of you have that one interaction with somebody and it being so rewarding to say, oh, yeah, I used your tool or I read your paper and I realized and then I can have a discussion with them I find similar experience when I see someone like retweet one of my podcast posts and say oh I really enjoyed this podcast I mean that is in itself is like I always say if one person listens perfect brilliant that's

fantastic so yeah I think a way of some way of measuring that would be really nice to sort of be able to quantify it because then you can go to your i don't know the 10 you can make a look at this sort of thing and it can can help you progress in your in your career as well cool so yeah my my next my next question is is about impact and it's what impact do you think or would you like this work to actually spark in a sense? Like, how can people in their day-to-day life take this research in and use it to leverage the findings of the work that you've done just as? Yeah, I mean, we have these two groups that we kind of studied, right? So the practitioners and the researchers. I think for the practitioners, yeah, we don't have so many clear takeaways or actionable

points for them because they are already doing pretty well, I would say. We were even surprised to see how well they were doing. So good job, industry. Keep it up. But yeah, for researchers, my hope would be that the paper first, of course, raises awareness to make people think about, make people consider doing this more and more and we even got already some some positive responses from people that they were inspired by this that they now uh kind of pledged to post something uh about their research periodically or in a in a certain interval and that was nice and yeah we of course also hope that that the guidelines or the advice how to make a good LinkedIn post with software engineering research might also help some people to improve their science communication.

Nice. Yeah, that's great. You've already been collecting pledges. That's great. Cool. Yeah, my next question is about surprises. And whilst working on this project, Marvin, what was the most interesting thing you kind of learned? The most interesting thing was, I think… It could be surprising as well, the most surprising thing. I think it was this one thing that we really, against our expectations, saw these discussions among practitioners without much involvement of academics. And because I still believe that research papers

are really hard to access for people who are not trained to read research papers. So seeing somebody from the industry, maybe without an academic background, telling others about it was really a really nice thing to see and really not what we expected to see. And then even people reacting to it and say, hey, this is interesting

or criticizing the study or whatever. That was so nice to see because as I said earlier, we in our community, we sometimes are really critical of our own work and of the relevance of our work. And I think those stories and this data that we have,

it really tells a story that the work we do or some of the work we do is really so interesting that practitioners take it up on their own without us even communicating it. And yeah, talking about it, I think this is both surprising to me, but also very, very nice to see. Yeah, how about you, Justus the is that the same story for you or something different jumped out while you

were working on it uh no i definitely agree on this one point that marvin made that i was very surprised and also encouraged to see how many non-paper authors actually shared the the research of others that was really nice to see and surprising to me because I would have expected that basically, I don't know, 90 plus percent of people that posts these papers on LinkedIn, that those are the authors of the papers. But yeah, that wasn't the case, which was nice.

Another surprising thing for me was how difficult it actually is to search LinkedIn, but also to crawl the content of of linkedin so basically i i wrote a python script for that um based on selenium webdriver and it was extremely difficult because it also was different when you were logged in or not logged in but when you were not logged in then there was a limit on things you could do so linkedin makes it really difficult to to analyze their content we were actually close to thinking about asking our network if they know somebody at linkedin that would be willing to

provide a data drop for us but in the end the script somehow worked yeah so that was also surprising to me and maybe the last thing was that i was also surprised to find yeah a fairly small percentage only of posts about papers from these conferences because we had like the two main flagship conferences in software engineering also several tracks from these conferences so not only the main research track but also also other tracks and like i think the last five years or something, Marvin, is that correct? Or even more? Yeah.

And we found basically only 100 posts that talked about a bit more papers because some mentioned several papers. And this is really, really a fraction. So, of course, part may be due to LinkedIn being very difficult to search. So I definitely think we missed parts, but also via Google, we didn't find many more posts. So this was a bit sad.

Yeah. On the plus side, we, while searching, we had some false positives of other science communication posts that were not the two conferences that we found. So we had to exclude those. But yeah, it was still nice to see that there were other posts, for example, about journals

that had a similar name. Yeah, that's nice. Cool. Yeah, I guess it's always nice as well to ask about how this paper actually came about. What was the origin story between how did you two guys end up? Let's say, okay, we're going to go and analyze LinkedIn. How did the paper come about? What was the origin story between how did you two guys ended up let's say okay we're going to go and analyze linkedin like how did the paper come about what's the story there uh yeah i guess marvin can explain this better i mean i go there from my perspective but science communication is actually his his topic so i think it's better

if he answers that topic yeah yeah i think it's it's really really an important topic and we can do a lot there. I think that's the main motivation to really do something about the state of science communication in our field, to have more of an impact, to make more people aware of the really good work that we do and help more people by simply making our research accessible. I think this was really the main motivation. Now, more unofficial motivation that didn't make it to the paper is that I also really like collaborating with users. So, yeah, let's say we just wanted to have another collaboration. So, one part of the motivation to start something. And I was happy that users quickly agreed to this topic and said, yeah, this is really relevant. let's do something about this and so this collaboration started. So this is there have

you got are you guys collaborating on anything else at the moment that you're working on together? Not yet we have we actually shared an office for two years at the same university and there we started to collaborate when I was really new into my phd i had no experience nothing i started on my first paper and i was like oh this is overwhelming i have no idea what i do and this was the first time i approached just this because i i thought yeah he's a nice guy and i i thought i told him i know this is not your topic but i need somebody to look over this and i think you could be interested in it. And so our first collaboration started. Yeah, from there on, we just kept doing projects together.

I think we also did one that went over two and a half years. It was a mapping study. Really crazy work, but it was a lot of fun. But right now, we need a two-month break or something. I mean, it's only two months now since the conference where we presented the paper and then we saw each other again. And yeah, really nice. Cool, yeah, because my next question is all about the creative process. And I guess maybe part of that is finding something

that you work really well with and then kind of the rest of them follow us from that. But yeah, I'll put the question to you both and then you can choose who wants to to go first on it and it's like how do you go about generating ideas and then once you've done that how do you decide what to work on for two years maybe for if it's a big project yeah do you want to who wants to go first with this one yeah i can try because it's a difficult question so one part of it is that i i very rarely try to systematically generate ideas so mostly when i find interesting stuff or read an interesting paper then it sometimes just happens that i think ah okay based on this i could adapt this in the

following way and then perform a study and then i i have a basically a simple text document in which i document such ideas so it's more it's usually very short it's like a sentence that mentions i don't know the goal and maybe the research method but yeah sometimes this also happens when i am in the process of writing something and then i i stumble across something which i don't know and then i search and i can't find a good paper and i think okay maybe this should also be a study and then i also write it down in this in this document of mine but yeah i am doing this really systematically is something that i probably should do more now i have also been criticized of doing too many different things

because there are definitely more senior researchers in the community that think or that think it's beneficial to be known for one thing or at least for a few things. And this is, of course, easier than to build a brand

if you are really well known for one topic or I don't know, one research method or yeah but yeah so far it has worked out for me and i so basically all studies that i do with marvin are actually usually not very close to my research topic they are kind of yeah they are always interesting and i always love doing them and especially love collaborating with Marvin because we mesh really well together on this. But they are never beneficial for my own research line. And I don't think it's bad because, yeah, I mean, the work is good.

This paper won an award. In the end, more people also get to know us, which is nice. So, yeah, maybe you don't have to be so strategic about it after all yeah i like that sort of it must be it must be good in a way to sort of get exposed to an area that's not yours because that's i think sometimes you can end up just very single trap single-minded but being exposed to other ideas allows you to kind of think okay well maybe i can take that and apply that here and vice versa you kind of that exposure to to many different things you need breadth and depth right i think kind of get that exposure to many different things.

You need breadth and depth, right, I think is what I'm trying to say, I think, to create good days. And I hope you've got that goldmine of a document backed up somewhere, Justus, as well. Yes. Cool. So, yeah, same to you, Marvin. How do you approach this process?

I think it's very similar to what Gustav just said. I just want to add that maybe it changes over time. So doing a PhD, you are the one who pushes most of the project. So you get really deep into the data, really deep into one specific topic. And there you need to be motivated over a long time to work on a specific topic. So it's really important to pick something that is interesting content wise and after your phd you begin to yeah do more of a mentoring and supervision task so you scale the amount of research you do you collaborate with more people and those people usually drive these projects those are are the PhD students then. And there I find it more important to work with the right people, basically.

So to work with people who are really interested, who are rigor in what they do, who don't want to take shortcuts in their methodological rigor and who do good quality stuff and are also fun to work with, of course, because in the end you spend a lot of time with them so i would say right now it's it's like 50 50 so i need to be interested in a topic of course to say yes to a project but i also need to have a good feeling that this person that i i'm about to work with is a good person yeah no that's a nice answer finding the right people is is half the battle right if you get good people then yeah naturally normally flows from there and yeah cool so yeah two two great answers to that question that's my favorite question i love seeing how people work and everyone has a different answer like it's crazy i have interviewed what well over 60 people now and everyone's answer is different so it's yeah yeah we're all different cool so yeah

we've arrived at the time for the last word now so and what's the one takeaway you want the listener to get from this podcast episode today? Then who wants to go with this one? You can vote for the last word. There we go. So for me, I think it would be that I would really like to encourage researchers to spend a bit more of their time on science communication. And the more experience you get, the more efficient you will be with this. And I think it can really pay off and it's it's a

very nice feeling to get reactions to yeah signs of yours that you communicated so uh this is probably the one thing that i would like to to share yeah yeah damn that that's what i also wanted to say but uh let me then add that maybe for the practitioners out there researchers in the in the field do a lot of useful work, right? So listeners of this podcast likely know this, but maybe I want to encourage you to follow along two or three researchers whose stuff you really like and follow them online.

And if they post something, feel free to comment on their stuff. That really, as I said, that makes a day. Ask them questions, don't be afraid. Maybe criticize their assumptions or whatever. If researchers receive some comments on their work, that's really what keeps them going and doing better things. And yeah, so it's a two-way communication problem, I would say,

and both sides need to step up the communication game a bit. Yeah, I think that's fantastic. Worthy message to end the podcast on. So thank you so much, guys it's been a pleasure today to chat with you both and i'm sure the listener will have loved it as well and yeah we'll see you all next time for some more awesome computer science research Thank you.