 Hey everyone, it's Sarah Whitver and this is our last ADHC talk of this spring 2023 semester. I am so excited about today. We are joined by Carrie Hill who is the digital scholarship librarian at Auburn University Libraries. And Carrie is going to talk about an archive of one's own. I'm super excited to hear about her fan fiction project and I have some things that I was going to say but I can't remember what they are because I always forget when I'm introducing people but Carrie I'm going to turn this over to you and let you take off from here. Awesome, thank you Sarah. So just starting out just thank you for letting me speak today and I just want to preface this by saying, I have not talked about this publicly a lot. So I really welcome feedback from folks, because it'll be helpful in preparing this for future use and publication so we'll start with that. So just as we get started, you know, you might be wondering why study fan fiction, why, what led you to researching this? Well, this really it comes out of my master's thesis and why I did for that. And when we were preparing to pick our topics, develop our research questions, our advisor said to pick something we can live with for a long while without getting fed up with it. And as somebody who has been a part of fan communities since my like, between and teen years, I thought, you know what, this is a topic that I can live with for the duration of my master's program and not get tired of it. I had also found a paper exploring fan information behaviors and I thought, well that fits me really well. I had given me a chance to really contribute to new and different research. I felt like there were a lot of topics I could pick that were sort of safe. You know, there's, there's lots to talk about there lots of things to site but I, I thought it would be cool to do something that was part of just a full emerging field or subfield of LIS research. And practically, you know, I was doing this during the height of the pandemic so we were, I was developing this topic in fall of 2020. So I needed something that I didn't have to get IRB approval for. I wanted to not rely on surveys in a time when everyone's stress levels were immensely high. So doing something that was in a way firmly within my wheelhouse with just programming and looking at fan fiction seemed like the right way to go, trying to get master's work done in a time like that. So, alright, go in into this. Okay, so my original goal was to try and figure out our tagging behavior similar across fandom user groups or do different tagging dialects sort of emerged. So, I wanted to see if it was reasonable to assume that tagging behaviors are similar or not. But, and by analyzing that the way that fan writers describe their works and posting them online for others to find. I thought we could gain insight into the way that non librarian readers and writers conceive of how their work should be categorized. So while designing a study with that goal in mind though I realized we had a need for a standardized system for researchers of different fandoms to categorize tags. So currently studies that code and organize fan tags, commonly build their own coding schema for analyzing characteristics of that tag set. But developing a standard taxonomy would enable us to readily compare tag data across studies from multiple fandoms and platforms and it would potentially reduce the time required to conduct a study by eliminating that need for every researcher to induce their own coding schema. So prior work had been done to develop a fan tag taxonomy so you know I didn't feel like I had to reinvent the wheel I wanted to build on what Ludi price had already done. So that taxonomy had yet to be tested on a fandom outside of the one for which she developed it so for this study I applied her taxonomy to another fandom to explore its application outside of the Marvel comic universe. And within that I originally sought to answer for questions. So to what extent does prices taxonomy translate to use in a new fandom. How does tag expression vary between those two fandoms. At what points does the taxonomy fail to describe tags in a new fandom, if there are any, and what changes might need to be made to develop that taxonomy for pan fandom use. The rationale behind investing all of this effort to study the way that fan taggers organize their work presumes that information organization is highly influenced by culture. So researchers balker and star sort of did the, did a similar work on this. And through their examination of a variety of classification systems. They found that classification systems, all of them inevitably express a point of view on the materials they organize. So by imposing categories of classification. Expresses assumptions that creators of the classification system have made about their world. So for example, in another study, a researcher. Olson points out the Library of Congress subject heading specifically list a category computers and women, but they give no corresponding category for men. They explicitly expresses a view of those classifiers that computing is associated by default with men and not with women. So each choice made in the design of taxonomies and ontologies inherently valorizes one point of view and silence is another. There are alternatives to top down information organization structure structures that seek to solve that problem of imposing the world view of designers by generating taxonomies are by generating organization structures from the bottom up instead of from the top down. So one solution, which is creating a folksonomy through social tagging. It seems to offer some solutions to the problems identified there but it presents some new ones as well. So on sites like our cover our own tags, let users filter this vast amount of content in the fan fiction collection by letting users only see content marked with their selected tag. So that process of organization through user generated tags has been referred to by multiple names like social tagging, collaborative tagging, collective tagging, but all of those referred to the same thing, which is just publicly labeling and categorizing resources in a shared online environment. So some tags result in this body of tags that describe the collection, which has been referred to as a folksonomy since that term was coined back in 2005. And that word plays on folks and taxonomy to embody the bottom up classification system built by users through tagging, and it sort of represents the antithesis of a top down taxonomy. And that's what I call a top down taxonomy by subject specialist. Through this talk you might find it interesting that I am talking about imposing a taxonomy on a folksonomy. There I just want to sort of point out mentioned like, while the folksonomy. I recognize is very useful for. Being able to find works and discover them in different ways. For the purpose of research, having one system that everybody uses can make makes it easier for us to all categorize things the same way. But I don't want to say that one researcher should come up with that taxonomy. I think it's a thing that we need to work through together which also, you know, is very much in the east those of pan fan communities right. So yeah, anyway, side note. So folks on a maze provide this way to organize information and collections that are way too large for any one cataloger or even a team of catalogers to index by spreading that work of organization among the users of the information system. So other methods of organization such as like control vocabularies we use to classify works in a library, provide consistency and arguably more precise recall. They also require costly expert design and the requirement to train indexing labor. They also suffer from a slow response to changes in user vocabularies and shifts in the collection, which folks on a maze incorporate immediately. A lot of me enthusiasts may see them as this revolutionary organization systems that allows users to find items in their chosen vocabulary and a system that sort of challenges that traditional meta narrative that respects a single authoritative voice. Folks on a maze still suffer from some inherent flaws that have been documented in a number of studies. Main failings tend to recur throughout those so issues of synonymy polysemy word form variations, different depths of description misspelt tags and single use tags that have little meaning to the rest of the community. Additionally, though it's tempting to view tax on folks on a maze as democratically decided organization systems. They really aren't. The folks on a maze as a result of many individual decisions, not the reaching of a group consensus by users. So the same minority worldviews that are silenced in traditional classification systems can be silenced when they're just overwhelmed by the massive tags created by a majority of users in the system systems. So one of those very flaws that folks on a maze might aim to, you know, fix reinforcing majority, namely reinforcing majority majority worldviews at the cost of differing perspectives. They might be as prevalent in that organization system as in traditional classification systems. Well, there have been numerous computational approaches to try and address those problems. The way that it's handled on our cover of our own. Which one researcher calls a curated folks on a me. The human based solution instead. So in a curated folks on a me the aggregate tags produced by the users are a starting point, and then system experts employed the decision making to identify and remedy problems of synonymy and homographs. So users create tags in a human or a group of humans combined synonymous tags and differentiate homographic tags which should help the system improve recall and precision. So that should allow for the creation of an organization system that balances increasing find ability with the improvement of the experience for even relatively small groups of users so you know thinking about the way that information systems, you know express these systems of power inherently. This can be a way to alleviate that because bullets research on the complex decision that these tag wranglers are making found that the human workers were making decisions regarding tag merging or differentiation, based on factors including respect for historic oppression, and the avoidance of enacting ongoing forms of harm and oppression. So, while some decisions made by those workers are not the best decisions in terms of precision and recall. They improve the inclusivity of minor user views in a way that unregulated folks on a me is or algorithmically produced ones don't. So yeah, um, so this, like I said this looks at the information behaviors of fan authors and archivists and to do that, I looked at an online fan fiction archive created and run by the organization for transformative works, which is a nonprofit organization run by and for fans to produce to provide access to and preserve the history of fan works and fan cultures, and that is called archive of our own. Interrupt you for a moment. I think we're having slide problems. Is everybody. Are you having problems Sarah. Yeah, I noticed that as well like it's like they're, like kind of layered over each other so yeah they're not advancing properly I'm not really sure what's happening. So it like, I really apologize. No, that's okay. Um, hmm. So the animation is not. Wait, okay so what are you all seeing. So there's like the text from your first slide layered over, and the headings are all layered and there's gaps and weird, weird pixels of weird. Yeah, as you move your mouse, I'm getting like some weird pixelation. Oh yeah, now we see. Now I see. Yeah. It's like, it's like holdovers from the old slide we don't know. I've never seen this problem happen before I'm sure you've never. I don't even know what's going on but now we can see the slide that you have. We stuck on a previous slide and I could tell you were advancing them and I was like but they're not changing. But it is now. So you see information behaviors, and then the box that says of our own. Yeah. I'm really sorry to interrupt. That's weird, but I'm glad that you mentioned it. Okay so it's but it's fixed now. Okay. Cool. So yeah, that archive was released as an open beta in 2009 but it is, and it has grown significantly since then. So as of 2022 when I pulled this screen cap. They had more than 4.9 or they had more than 5 million users and nearly a billion works, right is that I'm reading that correctly yeah. Yeah, nearly a billion works in the collection and that's only grown since then again this screen cap was grabbed in like October of last year. So, there, yeah, there's a lot out there. So works are organized according to the tags that their author assigns when it's uploaded. But users can also assign tags to works that they bookmark and that can help the user find work among the collections of work that they've already bookmarked. But it does not appear next to the work for all other users in the archive and for my study. I looked at the tags that authors decided to place next to works. Can you now see a no nothing changed. Okay. That's weird. Now, now I'm getting like weird layers of things again. So I see wearing layers. Well, no, no, no, it makes no I'm seeing information behaviors of fan authors and the screenshot from before and I'm seeing over it wearing her clothes by star book lover X files. Yes. Okay, that's what I should be. Okay. Now it actually for a minute and it looked weird but now it looks now now I can see that it's intentional. Okay, good deal. All right. Good lesson. This I need to include less animations more each separate slide, doing its thing, I think might be the problem that's coming. Okay. What does it do. So, so yeah like price, I focused on the tags created by writers but a similar study by a guy named, or by, I think a man named guy Hagan examine tags created by a three readers when bookmarking works and explored those cultural dynamics that are at play in a systems where writers can see those reader bookmarks, and they can provide sort of a hidden feedback system. So that's just sort of an interesting example of how, you know, examining these behaviors from different perspectives can provide cool different insights. So, A03 describes five types of tags that authors can use to describe their works. These are classes that are imposed by A03 media tags, fandom tags, characters, and within characters we have relationships, and then additional tags follow after that so can you see my pointer on the slide now. Okay, so yes, fandom tag. The media tag is hidden in this view because it just is assumed we're in TV. If we were to go to fandoms and try to browse. The fandoms are organized by what type of media it is. Then we have characters so Fox Mulder and Dana Scully would be that here relationships so Fox Mulder and Dana Scully indicates a friendship relationships where Fox Mulder slash Dana Scully indicates a romantic pairing. And then we have some additional tags here so we've got angst with a happy ending angst mid cannon, shaming and office rumors. Oh my. So I just felt like this was a good example of how those tags can be both descriptive of the work and also a dialogic so giving us some communication of what the author is thinking and wanting to show you with that. So yeah, in this system like I said tag wranglers wrangled user generated tags in the canonical tag to reduce issues of synonymy that occur with folks on Amie's so the stated goal of that is not to change how authors of tag their works but to standardize those cannon tags and synonym relationships as much as possible. So, it doesn't affect how a tag appears beside a word, but it rather creates a relationship that direct search and filtering features on the site to treat the authors tag as the canonical or wrangled tag, when preserve while preserving the authors original tag text, but still tag wrangling likely affects writer choices when tagging their works, because canonical tags are suggested via auto complete as the writer types so, you know, as that writer types the capitalized, the capitalized, you know, the properly capitalized capital F Fox capital M Mulder is going to auto complete and suggest that so where the author might have originally just been typing all lowercase it I'd a lot of suggest that which may affect that writer's choice when they're tagging it. So it would be reasonable to assume that a higher degree of variation and tag spelling word order and capitalization would exist among a three tags in the absence of that auto complete function. But going on. So yeah, prices taxonomy, which you can see a wide view of here. And hopefully a more narrowed in view will follow when I do this. There we go. Nope, it didn't know okay. What if we go there and then go back. Will it show you know, it's like sort of but not very. But, okay, is it showing you like, it's, it's just showing like, like I can see that you tried to like pop up the boxes but the old boxes from previous slides are still there and the new ones are there. Yeah, weird. Oh, okay. You know what, then I'm just going to go to this slide where nothing is in the way. And, and try to go from there. Um, so prices taxonomy expanded one that was made back in the early 2000s to describe tags on things like flicker and sites like that. But she expanded it into a lot of different sub types so that you could more accurately apply it to the fandom context, and so that we would, it would tease out those fandom specific behaviors. So, for this study, I sought to test that on a new fandom. She analyzed tags occurring on works tag Romy, which is a portmanteau of the character names rogue and let me Remy LeBeau, aka gambit from the Marvel comic universe, and it's used to refer to the romantic relationship between those characters. And my notes that she decided to analyze works tag Romy because it's a relatively small fandom that's easier to navigate than, you know, say the Harry Potter fandom, which is huge. And because her experience as a longtime fan of the Marvel Universe and more specifically of the Romy ship would reduce the time needed to research tag meanings and would improve coding accuracy. So fandom specific terminology can be exceedingly difficult to decipher for those who are not also members of the fandom. And that's like another reason for wanting this standardized system of coding because it will be very hard for me to go and try to code Romy tags accurately. So comparing our data is really only possible if you have one study where you have like a big group of fandom researchers that are all working together in that study, or if in your separate studies, you all use the same system. So that's, again, why I'm trying to help create this taxonomy. I decided to analyze tags from works on a three that have the relationship tag, Sam Carter slash Jack O'Neill, indicating a romantic relationship between the characters Samantha Carter and Jack O'Neill from the show Stargate SG one. So, when price performed performed her crawl, she would have scraped tags from from approximately 285 works when searching Remy, Remy, Lobo slash rogue. So I was doing my study, a search for works tagged Sam Carter slash Jack O'Neill yielded approximately 4923 results. So because of time constraints and the limitation of performing the study alone. I can narrow that down so I just scraped tags within the first 50 pages of results with crossover fix excluded because I didn't want to run into tags that I could not code because of lack of fandom knowledge. And I sorted by hits, because I wanted, I reasoned that frequently accessed works, maybe those that are the most discoverable, perhaps indicating that they've been well tagged. So, while that could have unintentionally favored works in even the first paid favored older works that had been in the archive longer, even in the first page of results. We had eight of those having last been updated since 2018 eight of 20. And the third most access work at the time was completed only three months before data were scraped so newer works don't seem to have been totally deprivileged by choosing that. So from here we shouldn't have a ton more issues with boxes and things like that. So hopefully you're all seeing just like a table with occurrence counts and percentages. I'm going to use Python to write the script that requested HTML files, and then parsed those files so that I would get the tag, the tags thick title, the author username and their air three assigned tag class. So, while I didn't use title and author username in the analysis they did let me sort and resort tags as I analyzed, so that I could look at each tag both in an aggregate, or in the context of other tags on the same work, because sometimes tags that provide commentary, they might not make sense when separated from their work and viewed in an alphabetized list of tags, but their meaning is clear when viewed in the context of the work it appears on a three. So, you know, thinking back to that example a couple slides ago, you know if those two dialogic tags were separated from one another, they might not make a lot of sense but when you view them together on the work they do. So, first I coded each tag by type and subtype, according to the taxonomy. And the taxonomy originally described the granularity of fan tags she found on tumblr etsy and a three, and she defined 11 sub types of descriptive tags, three of resource tags, two of ownership tags and five types of opinion tags that you can see here. So, I, it's kind of interesting that even though I had 1000 works that I was working with, and price had 285, I only had 2000 more tags, which is sort of interesting. But, but yeah so this just shows the number of times a type of tag occurred and what percentage that represents within its corresponding tag set. So, when we look at that, compared side by side, we can see that descriptive tags are employed most often in both fandoms by an overwhelming margin. When collecting data I chose to exclude crossovers, like I mentioned. So this first photo shows just all of the subtypes included, but I decided to also look at it with fandom excluded, because you know that could muddy the results of trying to compare because of course she's going to have a lot more fandom tags than mine, because there was only one fandom so I just wanted to compare only the categories that we both had used. So, in that sort of comparison, you can see that the Romy tags only changed slightly when we reduced fandoms at least at the type level. Let's see descriptive tag use drops a little bit and corresponding resource opinion task organizing and plan performance increase by just a little bit. So at the tag type level the relative frequencies of tag use remain similar regardless of if we're looking at fandom inclusion or exclusion. So descriptive tags are used slightly more in the Sam Jack tag set, then in Romy and conversely resource opinion and plan performance are used slightly more often among Romy tags than Sam Jack. So yeah, at the subtype level before excluding fandom tags, we can see that Romy tags are coded as fandom ship friendship organization slash team slash group location citation and explanatory more often than in the Sam Jack tag set. And conversely character genre plot and warning tags occur more frequently among Sam Jack tags than among Romy tags. However, when we exclude fandom tags the occurrence of character tags relatively increases pretty notably to surpass the occurrence of that tag subtype in the Sam Jack data set. So at this level the occurrence of tags coded descriptive event person location resource fan work title of fan work ownership creator source. Several others is so infrequent that it's not visible on most visualizations so those are effectively excluded. They just they sort of barely occur among a of three tags in both tag sets character tags occur most often followed by ship tags, which makes sense given that every work in this data contains at least one ship tag Romy or Sam Jack. For Romy tags, including our excluding fandom tags affects the order of the top three most common tag subtypes. And when included, fandom tags appears a third most frequently coded Romy tag subtype but when excluded ship and character subtypes both increased proportionally and plot also increases slightly, but no other ones are notably affected. And this is sort of them together. So, um, yeah with this genre and warning tag current stand out most different, most notably to differentiate the Sam Jack tag set from the Romy tag set subtypes coded much more commonly among Sam Jack tags. While descriptive tags are the most used tag type in both Marvel and Stargate SG one fandoms. The most commonly used subtypes varies a lot. Or it varies. I don't know if we want to say a lot. The character tags are a higher percentage of the Romy tag set. That could be because there are more distinct named characters in the Marvel Universe than the Stargate Universe. So the Stargate franchise has about three major shows, a cartoon, a mini series and some tie in books. But the Marvel franchise has been producing comics and a number of distinct series since 1939 had released 23 movies by 2021 when I did this. I released 11 TV series by 2021 where in contrast each of Stargate's three shows focuses on a team of four to eight main characters, along with a small ensemble of supporting characters and villains. So, while there's only a slightly higher usage of character tags in the Romy data set. Shiptags occur. Much more commonly. So, about twice as much, where ship tags make up about 30% of the Romy tags where they only make up 13% of Sam Jack tags, indicating that there are more romantic relationships tagged among Romy works. So we consider the fact that the Romy tag set represents approximately 285 works and contains 2,172 ship tags, while the Sam Jack data set represents 1000 works and contains 1332 ship tags. Like, each work in the data set occurs at least once. So among the selected Sam Jack fix additional romantic pairings occur about 332 times. While among the Romy fix additional romantic pairings occur 1,887 times. So one existence of more total characters could account for that, but it could also be an indication of different choices made by the authors in just how much they decide to include or decide to focus on. So it would be really interesting to see a comparison of the number of relationship tags that occur on each work. So, you know, if 100% of them have one, what percentage of works in each set have two ship tags or three ship tags, etc. But that couldn't be disused from the current study because it wasn't a goal of our original collection. So yeah, we see those peaks and genre and plot in the Stargate SG-1 fan. And with that, we have to wonder, you know, it seems really unlikely that there would be fewer genres or story elements represented among like all of those Marvel works than among the Stargate SG-1 fandoms. In fact, prices scraped produce twice as many distinct tag names, despite the scrape in this study, collecting about 2000 more total tags. In the in my coding, the genre type includes expected tags like humor angst, or in her in her coding. The genre type includes expected tags like humor angst and romance, but it also includes tags like voyeurism. And as previously mentioned, I tried to tag as I tried to code as closely as possible to her coding decisions based on examples I could see in her work. And voyeurism is clearly marked as a genre tag in one of her co-occurrence graphs. I've coded voyeurism and tags like it as a story element being a plot item, because I see that as descriptive more of a plot point in a work than an entire genre. So even when trying to model coding after her examples, it's inherently subjective tags may fit in multiple categories and one coders judgment call can vary from another's. But I believe that a more robust dictionary for training would improve agreement greatly. Both fandoms express more opinions that explain pop plot points than convey feelings, but in bookmarking tags that could be different. Guy Hagen found that a three bookmarks contain affective communication more prominently than anything else. So a study comparing bookmark tags and other tags could be really interesting, especially if we're all using the same system to code. And can really compare those results well. But yeah, so that's just sort of the raw of the findings. The last interesting sort of finding I want to point out is that there's only a slightly higher usage of plain performance tags and Romy works. And those tags indicate events and competitions like Sam Jack ship miss 2020, where authors celebrate Christmas by writing works for each other that revolve around Sam and Jack celebrating the holiday together. So they're there for a tag type where we can compare the degree of community building activity present in each fandom. That would make sense for there to be a lot more events and competitions generated among members of a fandom consisting of millions of comic book readers and moviegoers. Then the fandom of a cult hit sci fi show that last produced new show content. 20 years ago. And that tag Sam Jack ship miss 2020 indicates that there was a degree of community engagement happening in December 2020. So although the source content of Stargate SG one is a lot older with no recent releases to spark increase fan activity. Fan activity may have seen a spike during the pandemic, as people were increasingly seeking ways to connect with one another through online events where we would be safe from disease transmission. Yeah, so conclusions, what did all of those findings. What what were the main things, both fandoms most often use tags to describe their works for readers. The tagging practices seem broadly similar between the two fandoms, but more work needs to be done to prepare a taxonomy for pan fandom use so like I was mentioning with genre and story element being really difficult to differentiate between, but it's seeming unlikely that there would actually be more genres present in her data than mine. I think that indicates that we need a little bit more prep work done. Additionally, you know, there were, while I could fit most of my tags into the categories established by price. There were some that really did not work. So one type of character relationship members of the same family didn't have a category that fit it well. Those relationships wouldn't be a ship, which describes romantic relationships between character, but they're also different from friendship. You know. So there may need, while there may be no need for a category of tags describing family relationships in works filtered by Romy. There are several relationships in the top stargate fandom that need that distinction. So a family subtype of descriptive tag. I believe should be added to make the taxonomy more useful across fandoms. Regardless of our questions of intercoder reliability, both studies clearly indicate that fan writers are more deeply descriptive of their works than current library discovery systems allow. Significant overlap of genre story element tags appearing in both studies may indicate a high prevalence of new ways that writers think of categorizing their works for readers to find. While those categories may be new, their existence in literature is almost certainly not. Today's readers might be able to find more literature that appeals to them. If those new categories were applied to the traditionally published works that a user would find in libraries and bookstore. Since popular tags like angst, fluff, family, comfort, relationship and mental health issues can describe works of classic literature as easily as they do fan fiction. So why does it all matter. Like I mentioned, the best way to organize information changes as cultures change and culture among readers and writers is changing with both of platforms like what pad and a03 even good reads that make that relationship more of a two way street of communication than the traditional author providing to the readers without seeing feedback in real time. So it would be interesting to do more study on these things because it would be great from a librarian standpoint to see what the insights we find can tell us about how to make organized materials more discoverable for readers. So that is what I've prepared. Thank you for letting me speak, and we have a bit more time left for questions so I would be really happy to hear questions or feedback to. Yeah. To make this better. Thanks for this carry this has just been so fun. I think one of the things that I'm constantly have in the back of my mind as I'm hearing you talk is the way that I see our current generation of undergraduates try to search. And this is very indicative of the ways that they think about search terms and findability and discoverability of information. And so I think you are right about the implications for sort of understanding this from a library's perspective and figuring out how to leverage it. And also, and thinking that and this is just a sort of an observation and my own, like, desire for findability but I think about the fact that my Kindle is really hard to navigate. Right, like my kid will wake up in the middle of the night and come and take my Kindle and read like he'll open up like 10 books. And then the book that I'm reading disappears somewhere into it and there's no way to like, other than adding it to a collection. There's, there's like no way to, it would be really nice to have a tagging system there. I was just thinking about like the whole bottom up tagging and bookmarking system. It, the way that we do it in Zotero I think about the way that we do it. You know, I was a huge delicious user. Yes, health remembers delicious but oh my God I miss it every day. So yeah that actually, when the term folksonomy was coined in 2005. That is a site that Vanderwaal was studying, you know, particularly and delicious is also where Smith developed the original taxonomy that price based hers off of. It was a big one that first had a lot of social tagging and people, you know, doing collections like to be read, you know, right. So one of our two things. I don't know if I maybe missed it because I was talking to a student so I had to mute this for a minute. So I miss like the middle section. But have you looked at book talk already. So not as a part of like this study. Yeah, this was just on a three. Because also, I was doing this in 2021. I wasn't as much on tick tock at the time. So, like, on book talk, you know, I'm doing some cool applications for that, in terms of like reader recommendation and categorizing things in libraries, just at my own public library I've seen like a, the big hits on book talk. You know somebody had this like notion of playlists. I think has really been imported from, like, how much of a regular part, I think, especially the upcoming generation that like Spotify plays a role in, like these user sorted playlists as a way of content sharing. And I think I see that less on a three less on fanfic.net and more on sites that are that host like image based content. So like, but manga sites regularly use these sort of like many library playlists that are user created and tagged in that way. And I think it's a cycle through content. And I don't know. I don't know if that's useful. But I like, I don't know, it's very intuitive. The way it's built, and I think that book talk leans into that formatting a lot of the time because it enables you to create playlists so like, especially like larger book talk influencers will have playlists like dedicated to genre or dedicated to tags, right like hurt and comfort or certain ships and things like that. And I don't know I feel like it makes it really easy. It makes a lot easier to access what you're looking for those those playlists formats I don't know why I haven't I haven't done enough research on it but it is something. Yeah, like, just, it's like not grouping by genre so it's real weird but it's interesting and might be helpful for you. I don't know. I don't know that is that's really interesting. Because some of the, some of the research that I looked into a lot for this was fine birds research on like personalized collections and how in those collections you really see the point of view of the collector being expressed by how they categorize things. So, even just the idea of, you know, this collector sees this work as this thing, primarily. So like, you know them saying, Oh, primarily hurt comfort is what's important. It doesn't matter if it's an adventure genre, a romance genre, a family thick, the important thing that you want to pull out is hurt comfort. I think it's cool how the playlist format allows you to connect with other, you know, collectors who are interested in the same primary thing and, you know, and in that playlist format you almost get their ranking to have like oh this is what I see is the top. You know, this is the first one you should read and you know, going down through that list so that is that's really interesting. It's a really complex social network, almost in its own weird way. But it's, I don't know, there's like, it's like selling, not selling for monetary value but like presenting your curation has become really commonplace in a really unique way and that's like what is a library but like, you know, it's a lot of curating. Right. And it's like, it's no longer gate kept in a way like beholden by universities or infrastructure because it's all, it's like this is all digital curation. Yeah. And I don't know, I think, like, you know, people become influencers off of their curation which is insane. That's none of it's like some, like, if they're a writer or not, you know, but yeah, sorry. Yeah, no, I follow accounts. Yeah, I'm trying to think of but yeah, the guys and influencer based just off of, you know, the books that he pulls together and that, you know, is having me thinking wow, you know, library work does not make a lot of money but maybe maybe I should go be a book influencer instead. You know, apply curation that way. I think like that. Oh, go ahead Sarah. Oh well I was gonna say like I think, you know, kind of in reference to that like I think that in on a three, like you mentioned like kind of like separating out user bookmarks. But also right like you can have, I'm pretty sure it's been a while since I, since I like play it on a three but I'm pretty sure that you can search things by bookmarks you can filter by bookmarks. You can look at like, and there are like different ways to do collaborative collections I'm not sure how like prevalent they are but there is some like, you know, other kinds of curation that are happening. And definitely like there are sort of, you know, like groups of people who like will form a, you know, form a group to put their collections together like and you couldn't write you can like put another writer's work in your collection because you are, you know, curious Okay, so like there's like, I'm wondering about how, and I was like, because like as a user, like I book talk doesn't totally do it for like I find it actually hard, I heard, I find it hard to find things again. Like oh yeah that sounds really cool and then I'm like but where is that playlist and what was that thing and like I can't like, you know, so like, like I think that there are ways that that's also, you know, like I was just thinking what I said about like what you were saying it's like it is such a it is such a like, but I was like I was like oh I'm showing my, like, I interact with this stuff and a three rate is such like a, it's such a millennial sort of thing. And it's in the way that it's like, people are thinking about putting it together this is clearly an offshoot of live journal and you know, fanfiction.net. Yeah, Tumblr, you know, it's like those spaces where like, you know, and so the thing about book talk, I think is probably at least as much about just me being less comfortable with that sort of format and, you know, I mean you're totally right. I like the difference there is. I'm going to lean on comp pedagogy terms, but they probably outside of my field here is that book talk is almost like an inquiry based research style or methodology, like seeking something you have to like, I don't know. Does that make sense? Yeah, yeah, it does. It does. Yeah. And also like, you know, like what I think I like the world of what I'm usually if I'm on TikTok, I'm, you know, probably looking for interested in something other than like that's not where I'm going for my book recommendations right and so. And there are there are, in case you want to try it more. There's there's definitely some genres of book talk, like, like book talk is just the archway right and then right yeah like if you have a thing like if you like it it's it's like if you I'm not going to say I can't think of anything off the top of my head right this second but like if you want if they were doing just fan fiction curation on book talk right. Yeah, would probably be able to seek out her comfort playlists of good for just that genre and then dial it down into the fandom as well by her tailing your algorithm by the content that you follow. But that's the weird thing there's not I don't feel like there's I think that the algorithm on and maybe that's what I was trying to get to the algorithm on on Tik Tok is far more predatory, I guess, then like the algorithm you would find on a three and one is specifically seeking a capitalist venture that is other thing whereas the other one like they're for two different completely different. Yeah, but so that issue that we were getting into at the beginning of like organization systems you know serving power structures. Like those majority views are going to be really privileged on book talk because people, you know, more people are going to be interested so they're going to be shown to people where you get a little bit of a leveling of that on a three where when you search for a tag. You can sort it by its kudos, the data appeared, you know, things like that, but you're going to see everybody's you're not going to have, you know, accounts that and I'll stop sharing so that we're all just talking heads. You're not going to have accounts that are like privileged over others where you have like an influence or where there's are always going to show up. It's going to be more. A little bit more equalized that way. Yeah, additionally, I mean this, this sort of, you know, the sort of like feedback I'm hearing where like, you know we've got even in this call two of the four of us find different systems much more easy to use. The way that we categorize information is highly dependent on culture and this just kind of drives them and cultures are changing so rapidly among generations based on the online places we frequent right so you know, like Sarah was mentioning a three feels like it's coming from a very millennial place in life style of thinking, but Gen Z is not that much younger than us yet. And the curation is so much different. So it'll be really interesting to see how we adapt to that in the future. As you have these different groups that are all trying to be served by the same, you know, public libraries, who are expecting different ways of, you know, finding their information. They're coming at it from, you know, often completely differing perspectives on, you know, like Sarah, both Sarah's you mentioned earlier, like the way that undergrads now are searching for things in, you know, in your academic library collection is a lot different than the way that we were searching for them when we were students, and then the way that the system is designed for them to be able to find things. So, yeah, I think that fan categorization can be an interesting little microcosm for exploring just the broader changes that are going on between even close generational cohorts in the in the age of like online searching. I think that really just points kind of platform wise to that exigency though, right, like archival of our own really is meant for a holding container of things that people want to be able to go back to, right, whereas something like, and I up front I stay very far away from tick tock for very personal reasons. I have clustered myself in a research hole, and I'm not allowed to look outside of it because I'm a magpie. I can't get distracted until I finish the current project that I'm in. But what I understand of tick tock in general as a platform is that it's constantly pushing new content in front of the most as possible, whereas archive of our own is much, much more of what it calls itself and an archive where, you know, the stability and findability of specific items is privileged in the way that it's organized. Right. So that algorithmic exigency is is radically different for the two things so discoverability on tick tock is very important, but find ability to different things. Right. Discoverability is constantly seeing things that you may not have expected to see, right, whereas findability is like a known item search. Right. So, I think that's a huge difference I know I, I spend a lot of time on Instagram and unless I bookmark something into my own account, I can never go back and find it. Right. Even if I know the account that it came from half the time it's really hard because of the way that people push their content and because of the way that the Instagram algorithm populates my feed. It's, it's very hard to find something if you saw it while you were scrolling and then two days later you want to go and find that exact. Right. Yeah. I think it's really important to just to note that like a oh three like I don't know I don't know if it's actually correlated or not like it's very possible it's just like more of a coincidental thing. But fan fiction is such a respite for like marginalized voices. It's a very queer space and a lot of slash going on over there and that I feel speaks to like the notion of find of find ability I think also equates far more to visibility than discoverability does. Right. You talked about pushing new content right up front I think right. I think that's what we're getting. First, and you're not on. I don't know if it's a white talk but like one of the big issues with it just in a general sense is that you know you'll get like a black content creator and they'll do something and then a white content creator will utilize what they've done do the same thing without crediting them and then they will go viral for it. It becomes the discoverability without a way to like move backwards to the origin origin origination I can't talk whereas like there is no room for that given on a three. So to some extent it's just a much more inclusive environment in terms of time authorship and writership and all that. Right. Yeah, personalities are really far removed on a three. I feel like it's very work centric. Things like book talk that is very creator centric very curator centric if we're talking about people putting together playlist right where on a three that work is going to be forwarded. The author of it is going to be the one that you see most not somebody who has put together a list of it. Yeah, and yeah it's actually it's well researched that fanfic is, you know, like you said it's a respite it's a place for people in the queer community to be able to clear the fandoms that they love because in, you know, in a capitalist structure where these shows and these books are having to be pushed toward the broadest audience you know sometimes those voices can be silenced but in fanfic, when we're all just writing for each other, you can, you can write flash right, thick that you know puts pairings together that seem like they could be happening in the show, especially in shows where like queer baiting is a big issue where it's like, you know, they hint at it so much but they're never going to show it to you. Fanfic can be that place where fans finally express it fans finally give each other a place to see it and read it. So enjoy that together. So yeah. Yeah, when I think it is like, I was just saying Sarah I'm thinking about like what you were saying about discoverability because I think it's a different kind of because like, there's findability. I guess I don't know like, look at I think about like if I go on. Like, in a new fandom, you know, and I'm like, I want I really want to go like read. I want to go find like the heart comfort slash fix between these characters in the right in this fandom like I can get pretty specific. Right. And like that makes like the the tagging schema and like the way that a three is organized this stuff like that's like would you call that I guess would you call that something. Would you call that discoverability or would you call that findability. Or really, really, really misused. Yeah, but I'm thinking about discovery as much more like, like just things passing in front of you that you didn't know were there whereas. Okay, yeah. Findability is like following a path to an item. Yeah, actually seeking like I'm looking for something specific even if it's not like that thing. Yeah. Yeah. I'm thinking about like a Venn diagram right now, and like discover the findability and visibility, like an overlapping space and like where the platforms land on it. And it just, I keep thinking, I keep thinking like where I would put scout on that. I'm sorry, I mentioned that because of the tagging system. And in our search for our library specifically but like, they're sometimes the tagging is like so out of hand and sometimes it's just like so non existent and it just doesn't. You know, it's based what system. It's our EDS. Okay. And I think, Vanessa, that it's based on the subject headings that are existing in different records that come from different places. It's the same as like what imports into Zotero if you notice that like, sometimes they are just, you know, you'll import a short article and it'll have like 15 tags. Why does that even have that is because of the metadata that is attached to that item in the record that you accessed. And it's very, it's very messy. And you can actually you can find that happening on a three as well, especially with works or, you know, really what turned into like collections of work where each chapter is maybe a different fandom, where they just a writer might spam it so many times that it appears in nearly everybody searches but it. It's very annoying. You know, that's because I like her comfort tag be there for a chapter that's about totally different characters than you want to read about right and so I'm actually getting a hurt comfort Fox Mulder Dana Scully you're actually getting like hurt comfort. Remy Lobo rogue. But you were searching for, you know, one thing and but it appears because of how you know the the user has taken advantage of that tag system. So that issue can come up with a three as well. Yeah, I also wanted to know what happens carry if you run it through that word cluster model that I use for my Instagram hashtags. And Vivo like it's an analysis. Yeah, it's the it's a linguistics. Phenonyms modeling. It's, it's, I can send you some information on it but it's finding those hyponemic and hyponemic. Yeah, that would be interested in seeing that too. Yeah, because because um, then you could sort of figure out what what things are actually can contextually connected to each other as far as your tagging system goes, and you could filter out the ones that are irrelevant, you know, the ones that are just lightly connected. Yeah, you should go back I'll send you prices articles because she actually did social network analysis, which is part of why her data doesn't match up completely to what I would want to look at. And yeah, I'll send you her articles because I'll send you my stuff. Yes, that would be good. Yeah. All right. I think we've gone over time a little bit. It's been so long, but this was, it was so great to get to discuss this with you and to really have my people at this talk. No apology necessary. Thank you so much, Carrie. I'm going to stop recording and conclude the presentation.