00:00:49.980If your bookshelf and your For You page are equally important to your personality, welcome home.
00:00:55.400Pro Society is a weekly podcast that's part book club, part group chat for anyone who thinks Pride and Prejudice and Love Island deserve the same level of discourse.
00:01:04.060Each week, we're connecting the dots between books, the internet, and pop culture with your favorite writers, book talk creators, and plenty of overthought opinions.
00:03:53.780Back when Taylor Swift released her latest album, we did a few episodes talking about some of the online controversies and conversations that that album sparked.0.64
00:04:02.940Namely, the idea that she was sort of pushing trad wife sensibilities.
00:04:07.720that this was her sort of big trad wife pivot.
00:04:09.980And then later, the idea that one of the pieces of merch
00:04:13.760from her launch, this necklace with lightning bolts
00:04:16.460and a star on it, was actually full of coded
00:06:24.480I think, like, there is sort of an unfair kind of ridicule of Swifties when I think it comes to the fact that anybody you meet basically has, like, some sort of fandom thing that they're really, really into.
00:06:40.060I think, like, this is the first time I think she's released an album.
00:06:44.900And, like, a lot of my friends that are Swifties, my sister's a huge, huge Swiftie.
00:06:48.440and she immediately was like yeah it's not good um so i think that was interesting i didn't really
00:06:55.940love uh tortured poet society either like i think there were a couple songs on midnights i like but
00:07:00.700other than like i folklore evermore i was a 1989 fearless or uh not fearless 1989 reputation like
00:07:08.260those are my favorite albums so yeah i think when when this album dropped on the one hand i was sort
00:07:14.140of like you know what I didn't really like it it's not for me I'm gonna move on I know this is
00:07:20.140gonna turn into a huge internet debate and lo and behold it did uh as I think we're gonna get into
00:07:26.620yeah I mean you really summarized it well and this is what I said in the episodes that we did
00:07:33.400about Taylor Swift after the album had just come out I think that there is something that happens
00:07:38.420with Taylor Swift where she hits so many different intersection points gender politics the ethics
00:07:44.140of being a billionaire, class, race, all of these things that we know are these tensions in our
00:07:49.580society. She sits at the intersection point of all of them. And so in one of those episodes,
00:07:55.920I said that I was becoming concerned that I thought the conversation was being, I don't think
00:08:01.960I used the word manipulated. I think I said it was, it felt right for manipulation. I didn't
00:08:06.500have any proof. It was just my sense. And part of that was because, you know, when a conversation,
00:08:11.280You can just tell when a conversation feels volatile, it feels emotional, it feels difficult to participate in in kind of more casual ways.
00:08:26.100Those are the conversations, whether it's Taylor Swift or something else, where I just think, ooh, this could be easily manipulated because there's so many emotions involved in the conversation.
00:08:35.700yeah no totally agreed I think Bridget I have the same feeling as you where I think
00:08:39.460again I listened to this album I sort of was like this isn't for me I'm gonna go back to
00:08:45.040listening to folklore um but it was like I know there's gonna be discourse and immediately it
00:08:52.620was like oh this is going in a bazillion different directions and I think there is a level of like
00:08:57.020hey there is a conversation to be had here I think a lot of the points we're gonna get into
00:09:00.080there is something to be talked about but yeah it is like I agree whenever Taylor Swift comes
00:09:05.680Like, she is just such a popular figure and that, unfortunately, like, a lot of our kind of cultural ideas about women get projected onto her just by being, like, the most famous woman.
00:09:20.580It's very easy to make Taylor Swift kind of the stand-in for, like, white ladies, you know?0.59
00:09:26.480Like, there's something about her that I think is just, it's very easy to project a lot of stuff onto her.
00:09:32.940And so, you know, when I was watching the discourse get quite heated about Taylor Swift and the Nazi necklace stuff, I felt this way then, and I feel it even more strongly today, that I want to be very clear that even though I had a sense that some of the conversation might have been being manipulated in some way, whether by bots and authentic accounts, much of it was real people making genuine critiques of Taylor Swift.
00:10:01.620In the episode that we did about the lightning bolt necklace, we got listeners weighing in in the Spotify comments. Thank you for that. And it's not like those were bots, right? Those were people that I can visually see. Like, I know this listener. This is a listener of the show. They are real people engaged in real, genuine discourse because they think it is important.
00:10:21.120And so I want to make that super clear. And I feel more than ever, that has been, that reality has been confirmed. And so the reason why we're talking about this again, it's because this week, a company called Gudea put out a report about the conversation being driven about Taylor Swift online.
00:10:39.460What did this report find? Well, here's their executive summary. Between October 4th and October 18th of 2025, the public conversation surrounding Taylor Swift and her album The Life of a Showgirl evolved into a highly complex narrative ecosystem.
00:10:54.460Gudea analyzed 24,679 posts from 18,213 users across 14 platforms, identifying a polarized online environment shaped by a mixture of organic cultural discourse, symbolic reinterpretation, and targeted inauthentic activity.
00:11:11.660A key finding of this analysis is the role that what they're calling inauthentic narratives played in triggering authentic engagement.
00:11:19.720The false narrative that Taylor Swift was using Nazi symbolism did not remain to fringe conspiratorial spaces.
00:11:26.260It successfully pulled typical users into comparisons between Taylor Swift and Kanye West, who we know is basically like a Nazi now,
00:11:34.900is like designing clothing lines with swastikas on them and releasing songs that are odes to Hitler, right?
00:11:41.040So the report finds this demonstrates how a strategically seeded falsehood can convert into widespread authentic discourse, reshaping public perception, even when most users do not believe the originating claim.
00:11:53.400Additionally, Gudea found a significant overlap between accounts pushing the swift Nazi narrative and those active in a separate AstroTurf campaign attacking Blake Lively.
00:12:03.000This overlap reveals a cross-event amplification network, one that disproportionately influences multiple celebrity-driven controversies and injects misinformation into otherwise organic conversation.
00:12:16.100Not surprised to hear that there's the overlap with the Blake Lively situation, too.
00:12:24.080something that's also kind of is drawing up is like i know we've done a story on tangody before
00:12:30.320about um the whole like lena dunham thing where there was like a misinformation about her having
00:12:35.440sexually assaulted her sister not true yes very much but it was taken from like a grain of fact
00:12:41.840and then turned into something else i feel like a lot of times the taylor swift sort of discourse
00:12:46.700is similar where it's like there's things you can criticize i'm like and we're gonna end it to that
00:12:51.500there's plenty of things you can criticize about her and about her music and about the way that
00:12:54.920she presents herself to the world that doesn't necessarily mean like trad wife nazi you know
00:12:59.860like it's like yeah from step one to step like i don't know 10 000 something this did also make
00:13:06.460me think about was so this album came out i want to say it was the october this year right it was
00:13:12.520it was earlier this year i can't point to when this started but something that i've noticed like
00:13:17.380within the last couple months primarily looking at i don't really use twitter anymore twitter
00:13:22.180is just a nightmare of misinformation instagram reels for me has gotten like i don't really i
00:13:30.560don't know i don't i i will just if i'm using reels i'm just like uh mindless scrolling i've
00:13:35.500been getting so much like rage bait content that is very like specific to topics that i'm interested
00:13:42.000in which has been really interesting i something i keep going back to is i keep getting these
00:13:44.860videos that are like specifically about trans people that feel very like yeah like they're
00:13:50.520trying to create like rage bait content and i'm like i can't tell if this is legit if this is
00:13:55.540somebody just trying to get interactions with their post or if like this is somebody's actual
00:13:59.500beliefs that they're putting out there anyways that's to say i feel like i've seen an increase
00:14:05.620of this type of content that's specifically banking on the fact that that's how you get
00:14:09.240use that's how you you you tap into a yeah current hot topic issue you say something really crazy
00:14:14.620it gets what you use you get money from that i don't know if like the fact that i'm seeing such
00:14:19.420an increase of that lately also is sort of tied to this well you bring up a good point that there's
00:14:25.400you know this this report focuses on accounts that are creating content you know authentic
00:14:30.840human accounts and then inauthentic uh accounts that maybe they're bots maybe not but uh what's
00:14:38.360left out of this report entirely and often the whole discourse is the role of the platforms
00:14:41.800themselves, you know, specifically like Meta, who is really good at targeting to our very
00:19:03.140The data set reveals a multi-layered structure in which cultural commentary, political accusations, symbolic interpretation, and conspiratorial narratives intersect.
00:19:11.580While the majority of users behave typically, so we're not bots, just we're regular people saying things that they actually believed on the internet, 3.77% exhibited non-typical behavior amplifiers and accounted for 28% of the conversation volume suggesting coordinated influence.
00:19:29.580influence this narrative environment unfolded across major platforms including twitter slash x
00:19:34.840reddit blue sky tiktok and fringe ecosystems like 4chan and kiwi farms creating a cross ecosystem
00:19:41.800narrative network capable of rapid escalation so just to like underline that i believe the main
00:19:48.860takeaway here is that while some of this was inauthentic or bot accounts amplifying inauthentic
00:19:55.400narratives, that triggered real people to engage in real conversations and real discourse. So again,
00:20:01.740I'll probably say this 100 more times throughout this episode, because it's a misconception about
00:20:05.720this study that I'm really trying to squash out. The study is not saying everybody who engaged in
00:20:11.940this conversation is a bot that can be dismissed. What they're saying is a small percentage of this
00:20:18.820was bot or what they say as like atypical or non-typical behavior from these accounts, which
00:20:24.320could be bots but the majority of that the conversation was authentic conversation from
00:20:30.360real people and so to joey's point about fandoms some of the way that real people engaged with
00:20:37.960this kind of content that might have been raised by inauthentic users were doing so to defend
00:20:43.980taylor swift to criticize the irrationality of this like nazi necklace conspiracy and to
00:20:49.700contextualize the incident through historical conflicts with kanye west and so yeah i would
00:20:54.420call that like swifty fan behavior of like i like i like taylor swift i have seen what i believe to
00:21:02.840be a untrue conspiracy about her i am going to engage with this conspiracy content to debunk it
00:21:09.760and thus i am actually sort of helping a probably inauthentically generated narrative to spread
00:30:24.700And this idea from the report where you basically, by engaging with this kind of content, real people were boosting it and helping it to spread.
00:30:33.300This is like part and parcel of online manipulation campaigns.
00:30:37.840The pattern of inauthentic provocation leading to authentic user discourse is a hallmark of successful narrative manipulation.
00:30:44.540It demonstrates how small bursts of coordinated activity can reshape cultural perception by forcing mainstream audiences to respond to extremized framing.0.99
00:30:53.800Okay, so if you are a diehard Swifty listening, you might be thinking, okay, so does this report mean that anybody who is critical of Taylor Swift is just a bot and whatever criticisms they raised can just be ignored because it's bot shit?0.97
00:31:10.040And the answer is no, absolutely not, not even a little bit.
00:31:13.760So we do have some criticisms of the report, which we'll get into in a moment, but even taking it at face value, I want to make super clear that the report, as you just heard in that summary, does not suggest that all or even most of the accounts driving this kind of Taylor Swift conversation online were bots.
00:31:31.740And I want to really underline that because one of the responses I've seen online from this report is that the report is saying that anybody who criticized Taylor Swift is a bot and can be disregarded.
00:31:42.280I've seen a lot of Black women saying, I have been critical of Taylor Swift and I am not a bot. I'm a real person with real issues. I understand where that's coming from, which, again, we'll talk about in a moment. But I just want to make clear that the report is not saying this. And so if anybody is saying, oh, well, I told you so, all these criticisms about Taylor Swift are just coming from bots. You were stupid for even engaging with it. That is not correct.
00:32:08.180So in this report, they basically categorized different buckets of Taylor Swift conversation along the lines of the risk of those conversations being manipulated by inauthentic narratives.
00:32:19.560Topics that were in what they call their low risk cluster are topics that were largely human driven and not really driven at all by bots.
00:32:26.560So this was things like people being uncomfortable with some of the racialized messages, I guess, and some of Swift's songs like Eldest Daughter and Opalite.
00:32:35.320um it was humans saying that stuff not bots side note i am actually not listen to any of these
00:32:42.120songs so i i gather from the discourse that a lot of it was about racialized narratives in these
00:32:48.260songs but i'm gonna need joey to confirm that for me or just icky like icky narratives in general
00:32:54.360yeah um i think i right i think because i'll be like of this discussion
00:33:03.620again i didn't listen to the album and think oh this means she's a trod wife now this means she's
00:33:10.120a racist trod wife the song opalite though specifically there were a couple of i think
00:33:17.820it was the situation where i think she just um due to being a white woman was achieved this level of0.97
00:33:26.560wealth and you know probably has mostly been surrounded by other white women from similar0.84
00:33:32.080backgrounds maybe she just didn't like think through it enough like I'm willing to give her1.00
00:33:36.780the benefit of the doubt of that there was a little bit like looking at the lyrics it's sort
00:33:40.720of like this feels icky like there's something about it that I'm like oh um I don't know about
00:33:46.640that I'm crediting this again I I'm gonna give her the benefit of the doubt again this is a0.95
00:33:52.280celebrity I don't know in person I don't know what her fucking beliefs are I don't know anything0.95
00:33:55.740about her we've never met shocking guys i've never personally met taylor swift i'm willing to0.96
00:34:02.600throw this to just like she made some metaphors that she thought were really artsy not really
00:34:09.300realizing they can be read that way like this is the same way i like there was a i'm not gonna
00:34:14.000out here for what she actually said but like my straight friend a couple months ago had like
00:34:17.380left a comment on my instagram and the word she used i was like you cannot say that i was like
00:34:22.180delete that right now that does not mean what you think it is yeah i was like
00:34:25.740you just called me something very transphobic uh like I was like I know that was not what you meant
00:34:30.260but yeah um I'm willing to say I think that's probably what happened that being said though
00:34:36.140I I'm not gonna like criticize anybody for being uncomfortable with that um
00:34:40.940that is all right what's there's a piece of art that's in the world people are allowed to
00:34:47.460interpret it however you want something that recently happened that I think made me sort of
00:34:51.500while I was on like the other side of this was going back to this whole like having to take a
00:34:56.280step back and be like is this really an issue when the Sabrina Carpenter album came out it's
00:35:01.260most recent Sabrina Carpenter album I like I'm not a big fan of her and I think like I had a
00:35:06.480realization where I was like she to me is where I was like similar to I think I'll allow a lot of
00:35:10.340people to see Taylor Swift where I was like she's objectively very talented she's not for me like
00:35:14.520her music is not for me it's not right I'm not the audience she has in mind and I think like when
00:35:18.820that album came out and the cover dropped specifically i was sort of like okay i don't
00:35:23.920love that there's a lot of criticisms i have but i can see how this can be in the hovering to some
00:35:28.380people and it was however then i saw the conversation quickly turn there were like
00:35:32.240people online that were like if you're uncomfortable with this you're anti-sex work and anti-women and
00:35:38.420anti and i was like what are we talking about like i was just like oh that was a weird choice
00:35:43.300i wouldn't do it but good for you yeah i mean whatever happened to just being allowed to not
00:35:49.380like something or being like this is not for me i think that's my point where i'm like i wouldn't
00:35:53.060actually have an issue with this like i don't be like any discussion whatever it's not the album
00:35:56.920not for me but i was like well now i feel like i have to come back and be like no guys i'm not
00:36:02.560secretly a misogynist or anti-sex or anti-sex work or anti there was one particular tiktok that i saw
00:36:09.220that was just like listing off all these things.0.55
00:36:10.840And I was like, this is a straight white girl.
00:36:13.200What does this have to do with transphobia
00:50:02.640When I was 14 years old, I was kidnapped and held captive for nine months.
00:50:07.880I survived and I've spent my life exploring how other people survive what should have destroyed
00:50:13.860them. I'm Elizabeth Smart and these are the Survivor Files. I just remember this low taunting
00:50:21.400voice next to my ear saying shut up don't say anything. Every week I'm with survivors who
00:50:29.020live through the unthinkable. I knew if he woke up without a doubt he was going to hurt me. I
00:50:35.820I started feeling that there was someone at the end of my bed, and I just started screaming.
00:50:43.180They are abducted, stalked, controlled, and nearly silenced.
00:50:47.160But these aren't stories about what's taken from them.
00:50:50.600They're stories about what it takes to make it out alive.
00:50:54.240Listen to The Survivor Files with Elizabeth Smart on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.
00:51:02.020If your bookshelf and your For You page are equally important to your personality, welcome home.
00:51:08.920This is Prose Society, the weekly podcast that's part book club, part group chat for thought daughters, pop culture obsessives, and anyone who thinks Pride and Prejudice and Love Island deserve the same level of discourse.
00:51:21.080I'm Eli Rallo, and every week we're connecting the dots between books, the internet, and the conversations everyone can't stop having.
00:51:28.220I'm going to have to look up this story.
00:51:31.380From best-selling authors and your favorite book talk creators to the latest pop culture moments, nothing is off the table.
00:51:38.140It's like if you can hide some real messages inside compelling characters, then that is a Trojan horse.
00:51:45.280Whether you're looking for literary deep dives, smart pop culture conversations, or a community of readers who love to think a little too much, you're in the right place.
00:51:54.820Listen to Pro Society on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.
00:53:30.000So, Bridget, what were some of the high-risk clusters?
00:53:33.720Well, the high-risk clusters, these are the clusters that were heavily seeded by non-typical accounts and drove the downstream effects with authentic users.
00:53:42.500were things like, yep, the Nazi symbolism conversation,
00:53:47.080conspiracy accusations, political reframing and MAGA allegations,
00:53:51.720and then a bucket that they call relationship politicization.
00:54:09.160So have you all seen the response to this study online? Because in my pockets of the internet, the response has been massive. I don't know that I've seen something kind of instantly polarize the internet in such a stark way in a long time. Have you all been paying attention to any of the response?
00:54:26.580yeah i i've only seen the stuff that you showed me bridget because i am like so far offline these
00:54:33.620days uh i look at reddit which does not talk about any current events it's all like history
00:54:39.360subreddits and uh blue sky which is its own thing like uh screaming about this or that but but not
00:54:47.280taylor swift i guess i haven't just because i uh unrelated to this has sort of been taking a break
00:54:52.500from social media this week however I did bring this up with a couple people in my life over the
00:54:58.000last 24 hours and I was like oh I'm working on a Taylor Swift story and they immediately were like
00:55:02.260oh the Rolling Stone article and I was like uh yep so it was making it like like it I it's making
00:55:09.440its way around yeah yeah and one of the things that I sort of can't not talk about with this
00:55:14.600is the racial implications at play because I think a lot of the women who were not all but a lot of
00:55:21.700the folks who are critiquing Taylor Swift are Black women, people of color. And I think a lot
00:55:27.660of the people who perhaps self-identify as Swifties, who are very vocal about the fact that,
00:55:32.600you know, like, hey, anybody who's criticizing her, this article says they were all bots,
00:55:36.660which again is not true. I think, you know, you can already sort of see how maybe these different
00:55:44.060types of people are talking about Taylor Swift, but they're really talking about other stuff,
00:55:48.140race, gender, class, politics. And then the conversation kind of becomes a stand in for all
00:55:52.980of this. And I specifically saw a lot of Black women saying, you know, I am not a bot. I'm a1.00
00:56:01.560real person. This was my actual critique. And I want to make that clear. One of the things that
00:56:07.560I think this study perhaps did not make super clear is the authentic voices, who those folks
00:56:14.940were and what they were saying. I think that that perhaps gave a bit of credence to people who were
00:56:18.980like, oh, you don't even need to listen to these people. They're bots anyway. And so one of the
00:56:23.700big questions that people have been asking is like, is this study legit? Like, should it be
00:56:28.920taken at face value? What's the deal? So Mike and I have been in conversation with the authors of
00:56:34.800this study. We're going to have a deeper dive episode into the methodologies that they used.
00:56:39.900But I wanted to talk about sort of what people are saying about this study to give you a sense
00:56:44.880before we go into those deeper conversations.
00:56:47.840A lot of the folks who were genuinely critical of Taylor Swift
00:56:50.860were feeling kind of sus about this study.
00:56:53.400So I wanted to sort of get into some of the buckets of critique
00:56:57.580that I have seen floating around online about this study.
00:57:01.260The first is that the methodology that the study put together
00:57:42.220And so a lot of the criticism of their methodology that I've seen is quite vague, where the person posting the criticism isn't really being specific about which of those analytic decisions they object to, and instead are just dismissing the whole thing as biased or flawed, which feels a little lazy to me.
00:58:01.120There are definitely some thoughtful exceptions to that where people are raising very specific and very valid concerns about the study, and we are going to talk about them.
00:58:11.860And certainly the study is not perfect, and interpreting the results of even the best studies, though, requires being honest about acknowledging their limitations.
00:58:21.220So one also needs to consider the practical limitations of what is possible with available resources.
00:58:30.240Analyses are difficult and complicated and time-consuming.
00:58:33.780In this Taylor Swift data set, the researchers examined 18,000 posts.
00:58:38.940It would have been great if they'd had a trained human expert
00:58:43.840and look at the social media history of the account that posted it,
00:58:47.840what sort of things they talk about, who they're connected to,
00:58:50.680and then make a determination about whether it's a real human
00:58:53.300or an inauthentic bot account pretending to be an ordinary human.
00:58:56.560But a study like that would cost thousands of dollars, maybe tens of thousands of dollars. And, you know, I personally believe that the integrity of our online spaces and discourse is critically undervalued.
00:59:09.940And there should be much more funding to do robust studies to examine who is manipulating our discourse.
00:59:19.760But unfortunately, that's just like not the world that we live in right now in 2025 America.
00:59:26.300And funding for that kind of research is not abundant, to say the least.
00:59:31.040Right. And in fact, to my knowledge, this report from Gudea is literally the only published analysis that provides any data on online discourse about Taylor Swift narrative during this time frame at all.
00:59:45.560Right. I haven't done an exhaustive search. Maybe it's out there. I'd love to see it. Please send it if you have it. But to my knowledge, this is the only thing we've got.
00:59:52.740And so given that extreme dearth of available data, I think it would be foolish to completely dismiss this study because it is imperfect.
01:00:02.080I would much rather rely on data, even if that data is impacted by limitations and caveats, than rely 100% on vibes and my own personal experience with the algorithm that I'm looking at.
01:00:15.900Does that make sense? It's not really an answer. I have more, but I did want to pause there.
01:00:19.740Yeah, it does make sense. And I should mention this, too, that, Mike, you and I were able to speak to an outside expert, Molly Dwyer, who works for a kind of similar company that does similar kinds of reports.
01:00:31.460And we'll hear from her. We have a fascinating conversation. Also, Joey, Tumblr fan. Tumblr is what got her into this work in the first place, which is very interesting.
01:00:39.380What's funny is that when I was trying to figure out what expert voices I wanted to turn to to help put together our conversation about what's going on in this report, you were like, oh, I found this person, Molly Dwyer.
01:00:51.900They put together a similar report about bot networks impacting the conversation around the Cracker Barrel logo redesign.
01:01:34.360Yeah, I mean, we as scientists, we are trained to be skeptical.
01:01:38.700We are trained to find the reasons why your data does not support your conclusion, right?
01:01:46.160Because by doing that, we eliminate the bad data and then we're left with like the data that does actually support the conclusion.
01:01:54.760And that's how we know things about the world, right?
01:01:58.280And so I think, you know, rather than a binary question of whether the methodology for this particular report about Taylor Swift discourse is sound or unsound, I think the better question is, you know, is the methodology sufficiently robust that we have confidence in its conclusions or that the conclusions are at least in the neighborhood of correct?
01:02:17.880And in my opinion, I think the answer is yes.
01:02:21.440And, you know, like you said, we spoke with Molly Dwyer.
01:02:24.280Listeners will get to hear from her on Tuesday.
01:02:26.840And I really encourage them to listen because she really had a lot of good stuff to say.
01:02:33.080She was like, yeah, their approach is like pretty fine.
01:02:35.720You know, there are places to quibble, certainly like, oh, maybe I would have done this a little bit differently.
01:02:41.120Maybe the exact percentages are a little rough.
01:02:44.000But like the overall pattern of the results seem pretty well supported and also importantly in line with other studies that have looked at similar questions in different contexts, which is an important way for us to, you know, have confidence in analyses when, you know, it's not possible or practical to do a perfect design of the study that you might want to.
01:03:14.000And so just briefly, there are two aspects of the approach that they use here that I do want to emphasize.
01:03:20.860One is the algorithm that they use for classifying accounts as either typical or atypical.
01:03:26.940Atypical is sort of synonymous with like a bot account or an automated account.
01:03:31.220And then the other piece of their analysis that I wanted to highlight was how it was longitudinal,
01:03:35.020that they didn't just look at the mix of typical and atypical accounts posting about Swift,
01:03:39.960but they looked at how that mixture changed over time.
01:03:42.320And they found an early surge of posts from atypical accounts right after her album launched, which kind of helped set the tone for the narrative.
01:03:52.400Some of those high risk topics that you talked about earlier.
01:03:57.900Those atypical accounts were then later eclipsed in volume by more typical users as they joined the conversation and it became like a viral trending topic.
01:04:07.800uh and then later uh those typical humans kind of lost interest in this story because you can
01:04:15.380only talk about a lightning bolt being a piece of nazi iconography for so long before it becomes
01:04:20.980boring uh those humans like stopped posting and so the proportion of the inauthentic bots uh
01:04:28.360the inauthentic accounts increased at that point as they were like trying to keep the conversation
01:04:33.380going. And I just think that that kind of temporal analysis adds a lot of nuance to help us
01:04:39.480understand the role of bots and inauthentic accounts in shaping these conversations.
01:04:44.540Nuance that really gets lost in black and white social media posts that either want to claim that
01:04:50.260this report either totally vindicates Taylor Swift or is completely flawed and tells us nothing.
01:07:20.000and anyone who thinks Pride and Prejudice
01:07:21.880and Love Island deserve the same level of discourse.
01:07:24.920I'm Eli Rallo, and every week we're connecting the dots between books, the internet, and the conversations everyone can't stop having.
01:07:32.020I'm going to have to look up this story. I'm obsessed. I'm obsessed.
01:07:35.620From best-selling authors and your favorite book talk creators to the latest pop culture moments, nothing is off the table.
01:07:41.560It's like if you can hide some real messages inside compelling characters, then that is a Trojan horse.
01:07:49.020Whether you're looking for literary deep dives, smart pop culture conversations, or a community of readers who love to think a little too much, you're in the right place.
01:07:58.640Listen to Pro Society on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.
01:13:13.500And so the idea that they would just like make things up doesn't really hold a lot of water.
01:13:22.040And especially in this context that you described where companies like Reddit and X and Meta have shut off researcher access to their APIs, which we used to be able to use to do this kind of research.
01:13:35.020It used to be a very low barrier for a grad student somewhere or even just a normal person sitting at their desk at home. If they had this sort of question, they could do this research pretty easily using Twitter's API.
01:13:48.440But, you know, Elon Musk shut that off
01:13:51.460because he didn't want people poking around
01:13:53.700to publish data about how completely filled with bots
01:14:09.700for the kind of research that looks into this stuff
01:14:12.660that we as just like members of the public
01:14:17.720Like, we really have no choice at this particular moment, but to make peace with the fact that a lot of the data that we are able to use to answer questions like this are going to come from private companies.
01:14:30.380That's just the state of things until we change it.
01:14:32.920Yeah. And, you know, one of the other kind of buckets of criticism I've seen floating around the Internet is, oh, Gudea, they're an AI company.
01:14:42.260They used AI to put this report together.
01:14:44.100To me, this is kind of a weird claim to debunk the whole study, which I just, I don't think it's fair. The company's full name is Gudea AI and its website says that it offers AI services. But it is important to note that just like other social listing tools like Hootsuite or other tools that maybe you've used, this is machine learning. It is not what we call generative AI.
01:15:06.100So like a quick and dirty summary of the difference is that machine learning has been a staple of technology for decades.
01:15:13.560It analyzes existing data for predictions and classifications and decisions.
01:15:17.600Kind of like when you go on to Netflix, the recommended shows, that's machine learning.
01:15:23.880The way that your Gmail knows what emails are spam and filters them out, that's machine learning.
01:15:31.560While generative AI creates entirely new content through learning complex patterns from data sets.
01:15:38.580So think of that as things like Sora or ChatGPT, which is why it takes a lot more processing power than just a general machine learning.
01:15:47.460So the real difference here is creation versus analysis and prediction.
01:15:52.480So this company having AI as one of their work streams doesn't really mean anything, really.
01:15:58.920It's not a very, I don't think it's even a real criticism. This is kind of a tangent, but Mike, you and I got into a whole thing about how companies will put AI in their name, even if they aren't even really AI companies, as kind of a marketing term, even though they're not using AI the way we think of it, like with ChatGPT.
01:16:17.880and I mean I don't think it takes a genius to guess why a company would do that like why I
01:16:22.580would be like oh my company is podcast.ai instead of just podcast because it's marketing it makes
01:16:28.720people look at them twice it probably helps them fundraise and get more money I totally get it
01:16:34.460yeah I mean just ask friend of the show Ed Zitrone right like a huge portion of our whole
01:16:40.820economy is propped up by AI marketing hype depending on what sector you work in you would0.94
01:16:47.220be a fool not to put AI in your company name. So I would probably say I mentioned this earlier,
01:16:53.000but probably my biggest criticism of the study is that I just don't think it does a great job
01:16:58.740of really breaking down the authentic conversations about Taylor Swift, even though they make clear
01:17:05.900that the majority of the conversations that were happening were authentic people and not bots.
01:17:10.240I just think that a lot of the nuance was easily lost there. Like, who are these authentic people?
01:17:15.680what were the narratives that they actually cared about? What were the narratives? How were those
01:17:18.840narratives moving alongside these inauthentic bot-driven narratives? Like, I can kind of see
01:17:23.720how people who are critical of this study are now saying, oh, you know, the study is saying that
01:17:30.960everybody who hates Taylor Swift is a bot. Even though the study didn't actually say that, I can
01:17:35.500sort of get why that's the narrative that is sort of spreading because the study does not spend a
01:17:40.540lot of time fleshing out that nuance. Again, the study itself is in no way saying that everybody
01:17:46.500that was critical of Taylor Swift is a bot, but I get how this seems like a very limited view
01:17:51.340into these authentic voices to the point where people are saying they feel like their voice is
01:17:55.620being erased by this report. And I guess I would be remiss to not mention the Rolling Stone of it
01:18:03.860all because the findings are reported in an exclusive piece for Rolling Stone. Now, I don't
01:18:09.540of how this lands for the both of you. The person who wrote the piece is somebody who's been on the
01:18:13.700show before, Miles Cleef. You know, Rolling Stone is a music publication, right? I have spent my
01:18:21.960fair share of amount of time in like celebrity media. And a lot of times publications like this
01:18:27.420are working hand in glove with celebrity PR. And so it's hard for me to say that an outlet like
01:18:32.900Rolling Stone is going to be impartial when they're reporting on a huge musician like Taylor
01:18:37.740Swift, right? People have pointed out that Rolling Stone did a Taylor Swift takeover during one of
01:18:43.260her album's launches. They rated her album five out of five stars. I don't really have the answer
01:18:48.060here, but it's a question that I've seen come up kind of a lot. And the way that the article is
01:18:53.800framed, for folks who don't know, the people who write articles don't also write the headlines.
01:18:58.500That's the editor. And so Miles Klee, who wrote up the piece, is not the person who wrote this
01:19:03.720headline, probably. But the headline is, Taylor Swift's last album sparked bizarre accusations of
01:19:10.280Nazism. It was a coordinated attack. Now, if when you get into the nitty gritty of what the report
01:19:16.400is saying, that's kind of a bold claim that the report doesn't actually make. That title is saying
01:19:22.580like, case close, smoking gun, coordinated attack, in a way that the study, when you actually read
01:19:29.260it, like we've all done, doesn't actually make clear. And so I just, you know, in our interview
01:19:36.380with Molly, who runs Peak Metrics, a similar kind of social listening company as the one that did
01:19:40.780this study, she told us that, like, at baseline, some percentage of internet discourse is generated
01:19:47.440by bots. It's just the reality. It is, unfortunately, in this day and age, somewhat commonplace. So
01:19:52.280framing a report that suggests unusual activity around Taylor Swift discourse as a coordinated
01:19:57.340attack isn't really right because like sadly bots is just part of the internet now that's just like
01:20:03.640part of how online discourse happens and so again you can sort of see how it sets up this narrative
01:20:09.140that taylor swift is a victim of something who needs to be defended and she kind of is in a kind
01:20:14.620of way but also what they're describing is at this point somewhat commonplace does that make sense
01:20:19.820yeah i i think also to to that headline specifically saying like bizarre accusations
01:20:26.400of Nazism two things can be true where it's like I would agree with that that like yeah
01:20:32.540the high risk like straight up accusations of Nazism like she made this necklace which again
01:20:40.380she personally didn't make the necklace but like whatever people were mad about the lightning
01:20:44.880bolts like yeah on the one hand that is kind of insane that is sort of a a far-fetched claim
01:20:50.880on the other hand like saying this was like saying that the album sparked accusations of
01:21:00.260nazism is like ignoring some of the real criticism of it which again i don't i don't know like
01:21:08.020i don't know how newsworthy some of it is i don't know how like some of it being
01:21:13.280again taylor's writing a song that i was like oh i i like eldest daughter the lyrics were kind of
01:21:20.160corny i didn't really like it i don't think we need a whole article about that like so it's
01:21:26.040really yeah like two things can be true it can on the one hand like there were bizarre accusations
01:21:29.580there was a lot of misinformation but saying that's all it was like
01:21:34.800yeah and again that is just sort of the media atmosphere now i mean like i we've done a bazillion
01:21:40.960episodes about like review bombing and every form of media like this just sort of is the reality now
01:24:07.440that it's been it sparked such backlash and I guess that's kind of my so what here is that
01:24:16.560you know everybody people who listen to this podcast Swifties Taylor Swift critics people
01:24:22.100who don't care about Taylor Swift at all just want to have a robust internet landscape all of
01:24:26.720those people coming together really remind us that the internet is not just bots in chaos it's also
01:24:33.220real people trying to make real sense of culture in real time. And so I just think that we got to
01:24:39.480remember that. I don't think that this report serves to invalidate anybody's critiques or
01:24:45.160discomfort that they might have felt or any of that. I think it just shows that some conversations
01:24:49.780are more vulnerable to outside manipulation than others. It's just not a judgment. It's on anyone.
01:24:56.120It is just a function and a reality of this current media climate that we are all kind of navigating together.
01:25:04.960This feels like the pop culture equivalent of, like, whenever leftists online disagree with each other and they immediately start calling each other, like, CIA assets or something.
01:28:55.820If you want to help us grow, rate and review us on Apple Podcasts.
01:28:59.400For more podcasts from iHeartRadio, check out the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.
01:29:11.220I survived nine months in captivity and I've spent my life exploring how other people survive
01:29:24.220what should have destroyed them. I'm Elizabeth Smart and these are the Survivor Files. Every
01:29:30.240week I'm with survivors who live through the unthinkable, abducted, stalked, controlled,
01:29:35.640and nearly silenced. These are stories about what it takes to make it out alive.
01:29:41.220Listen to The Survivor Files with Elizabeth Smart on the iHeartRadio app, Apple Podcasts, or wherever you get your podcasts.
01:29:48.900If your bookshelf and your For You page are equally important to your personality, welcome home.
01:29:54.980Pro Society is a weekly podcast that's part book club, part group chat for anyone who thinks Pride and Prejudice and Love Island deserve the same level of discourse.
01:30:03.280Each week, we're connecting the dots between books, the internet, and pop culture
01:30:06.740with your favorite writers, book talk creators, and plenty of overthought opinions.