Regression to the Mean Girls：Roger Peng と「みんなのための因果推論」――偶然のビジネスマンになった寄り道

This is Regression to the Mean Girls, where we chat with brilliant data people about their silliest side quests. We are mean as in average, but actually very nice. Come sit with us. I'm Lucy D'Agostino-McGowan. And I'm Sarah Lutz-Feech. And today we're really excited to have with us Roger Pang. Roger is a professor of statistics and data sciences at the University of Texas at Austin and also author of the popular art programming for data science book, among 10 other books on data science and statistics. Thanks for being here, Roger. Hello. Thanks for having me. Okay, wait. I see that you are data sciences with an S, and we are statistical sciences with an S. That's right. We have multiple sciences. But, like, it's not common to have an S. Were you there when they switched to having data sciences in the title, or is that for you? I am a member of one of the few departments that, like, started out from the get-go as data sciences.

Oh. Oh. I guess you are, too. I mean. We, yeah, well, we didn't have data. We were statistical sciences. But, yes, this is true. Yes. We did start with that. But it's interesting because, like, Duke only has one science for their statistics. Have you noticed that? Well, you know, I did notice that. And the funny thing is that the sign outside of our main office area says data science. But the official name of the department is statistics and data sciences. Yeah. I like this. All the sciences. All the sciences, not just the one. Yeah. Incredible. So, Roger, to get things going, we do what we call a media minute. So we just chat for a second about whatever kind of media we're really enjoying right now. So it can be books, podcasts, movies, whatever. I will kick us off. I really like the build your own radio feature on Spotify. So when I find a song I'm, like, really into, I'll build playlists. Anyway, it's really gotten to know me. It's been doing really well. And it showed me someone called Cammie Barnes recently. And it's just like folk but pop and casual bagpipes. And I like it so much.

And so now I'm like down the bagpipe side of Spotify. And it's been honestly a really nice time. The algorithm's been serving up the bagpipes. It has. And I am absolutely devouring them. So it's right. You're reinforcing that learning. You learned right. Yeah. How often do you get past the part where you talk about media? This is actually the podcast is just us talking about what we're doing. Okay. Roger, it's pretty easy for us to get past once I start talking about it because I've been on the same piece of media. We'll let you go next. What have I been doing? So have I been doing anything interesting is the real question. So two things. One, I'm watching a show called For All Mankind, which is like an alternate kind of space history type of show. and they're on season four, so things are kind of like... I feel like I watch a lot of shows where season one is totally normal and then by the time you get to season four, everything has kind of gone just in a hundred different directions. The only way to figure it out is just to end the show.

Yes. One of my favorite shows along those lines was Alias, where it's like they just have to kill everyone every four seasons because it was too out of hand. you know um anyway um so that's i feel like that's kind of happening now um but i've also been we've been watching stranger things actually um from like season one so we started i don't know a couple months ago with season one and we are just about to watch the last episode of the last season oh my gosh so are we but we broke it because it's the length of like a full movie i know it's like two hours yeah we watched the first half and then now time has passed and i've kind of forgotten that is one I feel like that one had gotten very out of hand even relative like I acknowledge things were happening in season one but now I've completely given up on at all pretending like I'm gonna understand yeah and I think it's kind of shady when they throw in a couple episodes there like two and a half hours and we're like that's not an episode of the movie so

anyway we talked about Stranger Things with I think it was Christine Yeah, because she, we were talking about, and maybe because you've been watching them now, like, so quickly in succession, maybe you can tell us if you've noticed this. But she, there's, I guess there's been, like, talk about the, like, two-screen phenomena. Roger, my media. What is your continuing media? All I'm doing right now is, like, Lonely Island. Do you, are you familiar? Have I told you about my Lonely Island? I can't remember if I brought that up when we were chatting. No, I've heard of it, but I don't know it. Well, Lonely Island is, like, they were on SNL. They were their own thing, and they were on SNL. They did all the digital shorts, what I would consider the golden era or and golden era of SNL, which would be like 20, like when I was in high school, so like 2005. That's everyone's golden era. Exactly. Everyone just likes it back when they were. So they were on then. But then the reason why, which is, this is especially funny to me because this has been my media every single time. And Sarah, only last episode, I think, realized that this is current podcast. because the Lonely Island and Seth Meyers for the past couple years have had a podcast

where they, like, are going through all their digital shorts. And so I've been listening to this podcast and I've listened to it many times. I get in loops where I'm, like, stuck on something. So I've just listened to them over and over again talk about their digital shorts, and then I've watched their digital shorts, and I don't know. So that's been most of what I painted a picture of them. That's been behind me. This is, like, I'm just deep in Lonely Island lore. But this week, once again, I think this was my media last time, I also watched the current SNL. So I watched Will. the finale, right? Yeah, Will Ferrell was on this and he was very funny. So, yeah, it was good. I think the last one I was not as into the episode. This one was like, yeah, high marks all around. Well, the last one was Olivia Rodrigo. Yeah, which she was fine, but it just wasn't, I don't know. I felt like the level of show was not as good as Will Ferrell. I mean, Will Ferrell, yeah. I mean, she did a fine job considering she's not like an actor. Right. And actually, like, I was kind of impressed.

I didn't know anything about her beforehand. She made some comment about how everyone says that their favorite SNL cohort is the one when they're in high school, which is the joke I just made, except then she starts naming the current cast members because she's like their age. She is young. She is young. I'm sure she's allowed to be young. You're right. I mean, just, yeah. I mean. Yeah. So that's my medium in it. I'm still watching SNL. And you're never going to change. I don't know. Maybe I will, but it hasn't – nothing. It's not – I've not gotten out of this one yet. I tried – I think one time I said I listened to a biography, which I did finish that, but then I came back. Yeah, it was Leonardo da Vinci, I think, because we were – yeah. Yeah, it's kind of a thing where, like, podcasters talk about how they don't listen to podcasts, but I guess you do listen to podcasts. I do listen to podcasts. I do. And I listen to a couple of podcasts, but the Lonely Island one is my, I listen to that one on repeat. Yeah. And then I find myself telling people some little facts about, you know, the not Andy Samberg. They'll be like, what?

Who? Yeah. Yep. Maybe that's the whole reason I started this podcast. I have a second place I can talk about podcasts that I'm consuming. Okay. Excellent. Media Minute. Well, okay. Now we're going to move on to the context. What was your side quest and what data did you use, Roger? So I think there's a couple ways that I can present this side quest. One is more of a generalized career side quest to, like, become a business fan where I was selling things. I love this. So that's kind of like the high-level way to talk about it. and the kind of, like, detailed, on-the-ground level way to talk about it is kind of my exploration of Twitter and Facebook ads. And so, and, like, and just, well, yeah, that kind of stuff, and kind of trying to do your kind of, like, every man's causal inference, essentially.

I love this. Would you still consider yourself a businessman, or is this like you were, it was like a? I, well, I think one of the things I learned from this experience is that I don't really like being a businessman. So I do have things that I sell, but it's not like something that I focus on very much anymore. Because it involves skills that I don't have, I think. It's what it comes down to. Well, maybe I have them, but I don't know. I'm just not good at it. You don't want to foster them. It could be that you have them, but you don't. Like, you know, I don't know. I don't know exactly what the skills are. I don't think academics generally are, like, thought of as people who do a good job of selling. Well, there are a lot of parallels, but the environment is very different, I think, is what it comes down to. And so, anyway, do you want me to try to give you a little background? Yes, give us the background. So, about, oh, it's weird to say this, but, like, a little over 10 years ago, we, I and two colleagues, Jeff Leak and Brian Caffo, developed this, like, massive open online course on data science.

So it was one of the first ones and it was like hugely popular right at least to our surprise that is to say and so at some point i think it was brian who said you know what we should write a textbook for each of these classes because you know classes have textbooks and that way people can like you know have a set of notes and whatever so we kind of did like a crash kind of program of writing all these texts writing textbooks um and uh for each of the courses and um and so then it came down to well, okay, well, how do we distribute the books, right? So we found some publisher and like a self-publishing platform called LeanPub, and we put it on there. And one of the interesting things about it was that it allowed people to set their own price, right? So you could set it between zero and I think the maximum was, I don't know, $999 or something like that. Anyway, but we had a suggested price, and most people, unsurprisingly, set it to zero. But some people, you know, paid real money, actually. And so it came down to, okay, now we have to sell these books. And I think we had like this in-person class model where, you know, you teach a class and then like you sell the book to the students in your class.

Right. And so at the time, you know, the class would start once a month. It was like a month long class and every month it would start again. And so we would send an email to the class saying, here's the book and then feel free to get it if you want. Right. It's not like required for the class or anything. And so the publishing platform, like, gives you data on all of your sales, right? And so you can click a button and you get this CSV file of every sale that occurs. And so as it would happen, because we had so many people enrolled in this program, and it was, like, a synchronous program at the time. So, like, everyone enrolled the first of the month, basically. And so we sent this email out, you know, on the second of the month or whatever. And if you look at the date, it was like the sales just basically just skyrocketed on like the third day of the month, basically. And we're like, oh, that's it. Yeah, because like it did occur to us that because there's like 100,000 people enrolled in this class, that even if 10% of them decided to buy the book, that it would be massive. It was like the most number of sales that we ever had in our entire lives, basically.

Now, okay, most of them were, you know, for zero dollars. It wasn't like everyone was paying money. but anyway so that was like lesson number one in causal inference in terms of the I guess well it could have been confounded I suppose but it was a pre-post but anyway so then we started we can sell it to the class but then what about like other people people who are not in the class right so the second lesson we learned is that if you don't control the audience then you're not going to sell it to them it's like If you have a captive audience, then that's the best kind of audience to have. They generate. Yeah. Yeah. If you just wander out into the world, it turns out that doesn't work. But I didn't know that at the time. I was like, okay, we're going to buy Facebook ads and Twitter ads. I just did this myself because this was mostly just my interest. And we're going to sell to the masses, or at least to the technical masses. So Facebook and Twitter were kind of like the two platforms at the time in terms of social media.

and it was kind of interesting but Twitter was kind of like at the time I mean I haven't been on X recently or ever so I don't know what the audience was like but at the time it was much more kind of like a technical nerdy platform whereas Facebook was a much bigger but much broader platform and so one of the things I noticed is that on Twitter the ads went to fewer people so I think I guess those are called impressions right but the rate of clicking was a lot higher, was very high. Whereas on Facebook, the ads went to like so many more people. I think it was like 10 times more people or something like that. But the rate of clicking was a lot lower. And even with what we tried to like target certain kind of demographics, the conversion rate was extremely low on Facebook. Now, clearly, I did not know what I was doing. You know, like in terms of how do you advertise something on any type of platform? I didn't really know what I was doing. So it's possible I was doing something wrong. But I think When you're selling a book that's like an art programming book, it stands to reason that it might do a little bit better on Twitter than Facebook.

Yeah, this is fascinating. Okay, I have so many follow-up questions. I don't know. Maybe this is going to take us in the wrong direction. But I'm curious, like, what did you use as keyword? Did you – back then, was it the same as now? Like, now on Facebook, when I'm building an ad, I can specify, like, what types of interests. I can get smaller and smaller audiences where you must be interested in this and this and this. And if you have all three or you could do the union of the, you know, I don't know. I'm curious even how you started because I don't know how it was. Facebook, they probably were collecting things like, did they know our user or data scientist or someone that's technical? Yeah, it was. I think they had like, you know, so this is OK. The decision making process was a little tricky because it was like, you know, you could really narrow it down. But I guess I was worried at the time that the audience might get a little too small. right and you know a book on if you think about like a book on our programming might appeal to someone who's like a data scientist but it also might appeal to someone who's just like a software person right you know who maybe wants to learn r so i think i i tried to balance like okay people who might be interested in data but also maybe people who are interested in like computing and

programming and things like that and so maybe that was a mistake that was probably a mistake but anyway so i think i might have kept the audience a little bit more broad than i should have well i'm curious too because like i know i've only ever done anything on instagram and Facebook, which now is like the same company, but they have different things that happen. Like, the way it works is different. Instagram, at least for what I've done, is a little bit closer to what you're describing, Twitter, where it's like the audience, at least that I've been able to find, is sort of more narrow. Whereas Facebook, you can get like a million impressions, but really only make like 10 sales or something because the growth theory is like not very good. But it's way cheaper. Facebook ads, I mean, I don't know. I'm just curious. Like, was it also, was there, did you find that Twitter charged more because they were, they like had more value or was it the same? I think that's a good question. I think I spent more money on Twitter. Whether that's because they charged more or not, I can't quite remember in terms of the rates. But because I think because you could just set a budget and then that would be it. Right. But I do think I spent more on Twitter just because I think at the end of the day, it was a little bit more interesting.

Yeah. For this kind of sale. But I think if you looked at the data, I mean, if you just looked at the data in terms of sales, like just the raw sales data, it was almost, I don't know what your experience is, but it was always impossible to tell, like, you know, when the ad campaigns were running. Yeah. So I don't know. Now, did LeanPub have a way to link where people came from? Could you tell if they were coming from a referred campaign or if they were from? Yes. Wait, so let me see about that. Tell me about the analysis. Maybe we could do that. Well, okay. Hold on. So, yeah, you're asking a good question. And unfortunately, you're like digging up old memories of mine that I should have probably prepared a little bit better. No, this is good. We're doing it live. But, you know, in terms of the raw sales data, there wasn't a way, or I still don't think there is a way, to determine where someone came from.

Okay. Right. So you had to kind of piecemeal it with Google Analytics. and so you can embed like a Google Analytics identifier in your LeanPub page and then you could see okay well what percentage of people are coming from various places yes and then you'd have to make a leap in terms of like well because you don't know which particular sales are coming from which particular place but you can know roughly you can kind of know on average like 20 percent are coming from Twitter or something like that that's kind of how it worked right and I guess I'm guessing the Google tag would just tell you that they went to the site, but not whether it's not that particular one. Not that it resulted in a sale. See, now it's like things like, so I use Shopify, and it does all that for you. It tells you exactly where every sale comes from, and yeah. Yeah, that, I'm trying to remember if Shopify existed at the time. It probably, I don't know, I don't remember. I think it did, but I've never used it, yeah.

Yeah. I've never tried it for, like, a textbook, though, so I don't know. Yeah. This is, like, okay. So, all right. I've learned a couple things. One, that you maybe are a businessman. You sold things successfully. That is business, right? I don't know. I guess technically, yes. Yeah. And you had it harder than what people have now because you had to, like, try to do it. So you were doing causal inference, trying to decide, did this, did this, You do it like pre-post, so like regression discontinuities. Pre-post, exactly right. This is awesome. Did you find there was like, it was hard, like if you were running two things at the same time, running one on like Twitter and one on Facebook, like could you, because you wouldn't be able to disentangle which was which, right? No, it would be hard. I don't think I did that actually. Okay. Because it would have been hard to disentangle the two, I think, at least using the data that we had at the time. Yeah And do you think this drove people to take the class like people who saw the book then went and learned that you got to I seriously doubt it Like I think it only worked in one direction

Like I think if you took the class, then maybe you bought the book, but it didn't work the other way, I don't think. Okay. Because at that time, like random, well, no, that's true. Random people did take the class, but I think it. I wonder that people took the class. That's a lot of people. Yeah, that's true. But one of the things that's, like, difficult to understand is, like, how big the world really is, you know? And so, like, you think that, like, that's a lot of people, but there's also a lot of other people out there who are just not interested at all. So it didn't work the other way, I don't think. But it was interesting to see, like, who paid and who didn't. Yes. Because – so you have a book. Let's say it's a suggested price of $15, right? And you can set that slider to any value between zero and whatever, right? What do you think was the distribution of the sales price, like the final sales price for the book? What do you think that looked like? I think it was definitely bimodal. Yeah. But I think it's interesting of do people who pay pay what you recommended or a different amount.

Okay. Obviously, it's going to be bimodal, right? So the question is, what are the details? What are the modes? I had this, like, Gigi Plot 2 in two. It was like a – I taught a class on Gigi Plot 2 in two hours, and so then I, like, released the course. Anyways, it was on MePub. And you could pay zero, and I had a suggested price. And I found most people pay zero. The next highest, most common was the suggested price, and that there were a very small number of people that paid more. No one paid between – now, maybe your expression is different. No one really paid between zero and the suggested price, but there were a couple people that would pay a little more than the suggested price, presumably for some kind of altruistic, like, they feel bad that some people are paying you zero dollars. I don't know if any of these people are. That's called ineffective altruism, right? That's right. Yeah. That's right. Ineffective. Yeah. So, but what percentage paid? For me? So, I don't know because I don't know that I got – the only way I know that anyone paid is, like, I get emails and probably only, like, 50 people total ever have paid.

So, you never analyze the data? No, I haven't. So I want to hear. I have two pieces of information because all I know is when they do pay, I get an email. So I could see what that about is. And that tended to be what I asked for or sometimes weirdly a little more. Yeah. So definitely the mode is at zero and then at the suggested price and then a couple of blips like a little bit higher. Yeah. Like rarely would someone put in like $8. Like it doesn't seem the psychology doesn't seem to work that way. No. But it was like, I would say about 10% paid something. Paid, okay. Yeah. And it seemed across the board, like of all the books that we had available, 10% seemed pretty universal, actually. Really? Which I think is interesting, yes. Yes. Yeah. I mean, granted, across the courses, the people are the same, for the most part. They're drawing from the same population, at the very least, and sometimes literally the same. But so, OK, so it's not super surprising that it happened across the books.

But I have had it like other things that have kind of like, for example, like the when we had our remember when we had a podcast or not you and I, but I had a podcast. Yeah. And we had a Patreon. Yes. And and based on the kind of rough numbers that we got from the podcast provider, you know, it's a little hard to tell how many people there are. But it also was kind of like 10 percent of people pay for the Patreon. I was among those 10% Roger I don't know if you remember that but then my credit card like it expired and I never fixed it but I like stayed in the like in the like they still sort of like somehow Patreon I don't know anyways I remember this because after like the first year I was not actually being charged but I was somehow they were like oh it's still in the same way and then on the other hand there were people who I know stopped listening and like continued to pay for like five years they were just getting you a cup of coffee they were being sweet I guess so yeah So anyway, 10%. That's the universal law. This is fascinating. So do we think this is transportable? I mean, do we think this is a data science, like, or is it our user?

Like, did you do any for Python? No. We didn't do anything involving Python, unfortunately. Yeah. Okay. If I had to hypothesize, I would imagine the Python community would be – well, what are we hypothesizing here in terms of who would they pay? Would they pay? Yeah. I think there would be a lower percentage paying if I had the iPod. Yeah. Only, of course, it's like a much more heterogeneous mix of people, I think, in Python. Our community is, like, altruistic. Oh, my gosh, I'm looking at this ggplot 2 and 2. Let me tell you something. Two things. I lied before. Okay, I'm remembering why I lied. I didn't mean to lie, but I have a code that I would give, like, my students that would make it free. It's possible to have it free. Mistake number one. You're not a business person. No, no, but maybe I am. I have a minimum price of not zero. I have a minimum price of $7.99. Okay. And so this is actually probably why, because probably what happened, you know, I was just saying that people would weirdly pay more. It's probably because the suggested price is $14.99 and the minimum is $7.99. So a lot of people probably pay $7.99 and then other people pay $14.99.

And because I did this so long ago and forgot about it, I'm just seeing these $14.99s like, why are these people paying $14.99? But they weren't actually paying more. They're just paying what I asked for. I think Leadpub changed their policy so that you can't set the slider to zero. You can't set the minimum to zero. I think the only reason I can still do that is because we basically were grandfathered in. You were? Okay. Yeah. And so, but now I think if you're like a new author and start like a new book, I don't think you can set the minimum to be zero anymore. Yeah. They're like, that's a dumb idea. Don't do that. We're not going to let you do that. We're not going to let you just give away your work for free. Exactly. Yeah. Well, I'd say I walked away wondering, like, what is it? What do people who sell things for real see in their, like, ad dashboards? You know what I mean? Because I've, like, I kind of walked away from the whole thing thinking that, like, this seems like mostly a waste of time. But only because I think of the structure of, like, of, like, the audience that I'm trying to sell to and the thing that I'm selling is probably not the ideal thing.

Not that I know what the ideal thing is. But what I concluded was that I already have, like, the prime audience at my fingertips, right? I'm just wasting time going on Twitter and Facebook trying to find other people. Like the next marginal person is going to be super expensive to capture, I think. Right. Like those 100,000 people. To have an email list of 100,000 people. I mean, that's like people sell that kind of stuff all the time. Well, yeah. And to be clear, we did not. But the funny thing is that I didn't even realize that. Like, I had no idea how valuable that was, right? And so, anyway. Or, yeah. I mean, so, anyway. It was just, I just learned the most basic, you know, kind of business things in terms of, like, you know, products and markets and product market fit and whatever. That's how I feel about social media. There's, like, definitely a marketing way to do social media effectively.

and then there's like things that I just like tried and then saw what worked and I was like, what if I do it again? And it's like, I, yeah, sometimes that is a way to learn. I don't know. I have never attempted a business endeavor. I have no data like these to afford. Yeah. I, I don't know. It's hard to know. It's, this is one of those situations where like, everyone assumes that you're either going to succeed or fail, but like the modal experience It was really just like, well, I'm not really sure what happened. Like, that's actually what real life is like, I think. Like, everyone assumes you're going to have perfect information at the end of the day. But I walked away with more uncertainty than less. Yeah, this is, I, hmm. Wait, so, wait, Lucy, can you say what you're selling? I sell mahjong tiles. Have I told you that before? Oh, yes. Okay. All right. So you're selling like a physical product. I am. But that's different, too. I also think that's different from a digital product. Yeah. Yes, it is different.

It's interesting, too, because I found, and I don't know if, like, there's something wrong with how we, not wrong, but the way we, like, sort of condition academics, I think we sort of get conditioned a little bit to devalue our time. Or I don't know. I mean, I guess. Oh, that's for sure. Yeah. We devalue our time at essentially zero. Yes. And then it's almost seen as like a character flaw if you want to value your time. Like I have found selling a physical product much easier because if you're selling a digital book, what exactly are you selling? You're sort of like back paying your time that you spent on it in theory. And then if that's how you think that like products work. I mean, the other thing you could say is you're selling someone something that's worth something regardless of how much time you spent on it. And that's why they're paying for it because it's giving them value in that way. But for a physical thing I find it a little easier because someone is going to get this thing into their house and it like art that I created and you know well you know we verging into one of my favorite areas which is accounting um oh and uh

one of your favorite areas it absolutely is yes oh man I love accounting why do you love it I don't know I think it probably has something to do with my foray to the business world and I'm like I'm not good at actually this business stuff but like the numbers aren't really interesting but you know What's interesting to me, I'm in a partnership with someone else, and she got the accounting. She went to NC State and got some kind of accounting award. I can't remember what it was called. So she does a lot of, like, that, even though, like, in terms of her actual expertise is, like, in marketing and communications. But she handles our accounting more than I do because I don't think my brain works like that. Even though I'm numbers related, but I like to estimate. I don't like to be, like, you know, going down the line. like on so tell me there are some like there are some like cool almost philosophical questions in accounting you know one of the things it's like with selling a physical product there's a marginal cost right and so like yeah you cannot produce the thing without spending some money so at a minimum you want to recoup you know that cost right but when you sell a digital product it's like the marginal cost is basically zero right and so right this is the issue that you're talking

about right i mean it's like exactly yeah and like and it's true in other fields like i've heard doctors say that like you know you work in a field where there's lots of procedures then you make more money than if you work in a field where you're just kind of giving out advice basically right so right and so like because that kind of feels like it has a marginal cost of zero right uh but with digital products you accrue all this expense when you develop the products right right and then you depreciate and then you amortize it after you start selling it right and so like that's kind of how the pricing gets worked out but anyways let's but that's all just accounting right i mean that's true that's true that kind of account and i can see the look you don't find this interesting at all no i no i it is interesting i just think it's like too hard to like like you have first you have to assign value to your i guess there's like opportunity cost which you could have been doing otherwise instead of you know i don't know it feels hard to like assign that value and then be able to like what is it is it like would it depreciate i mean like for a physical object you could let it like depreciate but what happens to like this digital thing

Well, it amortizes, right? Yeah. What does that mean? That's the same thing as depreciation, but it's for non-physical objects. But, like, why? Because it's less relevant? Is that why it would amortize? Yeah. Yeah. Or it becomes, yeah, so if you have, like, a patent, for example, like, over time it becomes less valuable, right? And so. Makes sense to me. Yeah, or a copyright or something like that. I think I said everything that needs to be said. if there are lessons to be learned I'm not sure what they are but if you're maybe if you're an academic out there who is interested in selling stuff don't talk to me talk to Lucy if you want to sell some mahjong tiles I can give you the skinny on that not if you're trying to sell any kind of like you know something that comes from your brain So a lot of junk tiles come from your brain, you know? That's true, yes. I do think your point about a physical product is an interesting one and the distinction there.

And I'm really glad to now know the equivalent of depreciate for non-physical items, Roger. I don't think I would have even thought a word for that existed. Well, I've made my contribution, I can tell. Plenty. If you were to rank your analysis on a scale of one to five fetches, how many would you give it? Oh, okay. Sorry. I just got the reference. It took me 10 seconds. When Sarah and I first started this, we were initially saying we were going to rank it, and then we were giving everybody five fetches. We're going to have to make them rank themselves, or else there's going to be no distribution on these fetches. I think the analysis, three. Three fetches. Three fetches. Yeah. That's pretty fetch. Yeah. I mean, it is at least some fetch. Yeah. It's happening. We're making it happen. Yeah, exactly. Yeah. Do you still do ads or no? No. I don't run any ads anymore.

I don't have an X account. I don't really have a Facebook account. Yeah, it's dark times in the social media world. Yeah. I, too, have felt similarly. I wonder if you would have any luck on, like, LinkedIn, though. Yeah, that's a good, yeah. And there are more and more ads on LinkedIn now. Yeah. It's funny to think about. It didn't even occur to me, but I think because ads on LinkedIn came later. Yeah. LinkedIn has had such a renaissance in, like, the last year or two. It's crazy. I mean, yes, but. Yeah. But, you know, the same thing happened on LinkedIn that happened on Twitter, which is that you have these, like, growth hackers, you know, that are just, like, doing their thing on LinkedIn, and it's like, oh, you've got to, like, filter through all this stuff. which I find annoying. But it's worse than it was on X because I feel like at least on Twitter, there were growth hackers, but at least it like had to come from their own brain. Like now it's all, it all sound exactly the same because they all use the same like generative AI. Yeah. So it's like, maybe it's easier to filter through, I guess,

because it looks the same, but it's like kind of exhausting. It's exhausting. Yeah. Yes. Yeah. I like even, I feel like it's not just, I mean, I don't know how they do it because I'm a Luddite and don't understand like large language models or whatever anymore. But they, but it seems like people are not only posting using them, but they're also commenting using them. Yeah. Like literal full interactions that are like no human. Why would you have this interaction? Right. Like what exactly is it doing for you? I think I mostly meant that people are actually on it now. And the fact that there's like video and stuff like that is kind of interesting. I mean, the video can be the same kind of thing where it's like, you know, the like kind of textbook inspirational stuff. But I did once get a Hot Ones video on LinkedIn where it's like, do I need to watch Matt Damon eat hot wings in my professional network? I don't think so. That's not really the forum for that, in my opinion. No disrespect to Matt Damon. I get enough of that on other forums. I was like, this doesn't feel right here.

sorry that you haven't found out that my next venture that i'm proposing for us is to make a hot one just like our version of it you know i mean like i mean would we eat hot wings yeah and then we would invite we would it'd be like a podcast or we would like i love hot ones please don't get me wrong it's just that i don't think it belongs on my linkedin i think sarah says she has boundaries yeah that is what it is let's be on the other hand it's like let's cut let's break those boundaries i'm breaking i gave a talk last week about professional social media and was joking that like until like two years ago i wasn't even connected to my husband on linkedin it was like you can't even like you can tell i think very little about my personal life on linkedin which i didn't super do on purpose but i do think it's funny because sometimes when he reacts to things He just looks like a stranger. Who is this man? It's like, wow, this guy's, like, obsessed with me. Weird.

All right. Do you have any final thoughts or takeaways, Roger, for us? Any advice? Advice for – no, I have no advice except that, you know, I think side projects in general, especially in academia, is good just to keep yourself busy. And I also think that, like, okay, now I'm just getting serious, but, like, I feel like in academia, people often talk about, like, being innovative, you know? There's a lot of talk of innovation. But to me, it's, like, that's the nature of what all these side projects are. Because some of them, like, turn out into something, and it's like, oh, that person's so innovative. It's like, well, yeah, but up until, like, a minute ago, they were just wasting time, right? Yes! Right? And so that's why side projects are important. Because it's, like, you're wasting time up until the point where it's, like, ingenious, right? Exactly. There you go. You get the concept. I love it, too. Yeah. Okay. Well, thank you. Thank you for sharing with us. You go, Glenn Coco. You go, Glenn Coco. We're mean as an average, but actually really nice.

Tune in next time. Would you like to say that with your whole heart? Are we still on the podcast or are we done? I think we could be. I want a compilation of you every time you say that, because every time you say it, it's just like 5% different. Never once has it sounded the same. We hit it real hard and we're very high. What do you think we're not actually mean, but we, what? We're married as an athlete, but we're actually married. Which I realized at this stage is actually self-deprecating. That's like... For me, this is... I don't think we actually are average. This is average. Two average girls trying our best here in Akateria. Big world.