From Qidian Collections to Tieba Indexes: Reconstructing the Platform History of Chinese Transformation Fiction

Abstract

Histories of Chinese bianshen xiaoshuo (变身小说, transformation fiction) often take the form of remembered reading lists or compressed claims that a particular platform “used to be full of” the genre. Surviving web traces support a more precise history. In February 2006, readers on the Jinjiang forum already described Qidian as a major place to find male-to-female stories and referred to a transformation-fiction collection on its front page. A September 2006 Yamibo discussion preserved a passage about a Qidian Girls channel “transformation fiction collection.” The present-day page for the Baidu Tieba community Bianshen Wenxue Ba records March 11, 2009 as its founding date. In 2017, a Tieba moderator published a cross-platform index built from Qidian, SFACG, and Ciweimao/“Shuke” data, complete with an explicit crawling rule. In 2026, Qidian still hosts a curated “transformation fiction” booklist, while SFACG maintains active “变身” and “性转” discovery pages containing recently updated works (JJWXC 2006; Yamibo 2006; Baidu Tieba 2026a; Tieba Index 2017; Qidian 2026; SFACG 2026a).

The central result is therefore a network of platforms, taxonomies, community indexes, and archives; a single-origin narrative becomes one historical claim to test within that network. This article proposes a “three clocks” method. Event time records when the historical occurrence happened; page time records when the page we can inspect was posted or edited; archive time records when a crawler, archive, search system, or researcher captured the evidence. Adding page function, source role, independence, original labels, and reproducible URLs produces claims that can be revised when earlier evidence appears. The method is useful for transformation fiction and for any cultural history built from forums, commercial web-fiction platforms, tag pages, and disappearing websites (Brügger 2018; Nielsen 2016; Hockx 2015).

Keywords: transformation fiction; Chinese web fiction; Qidian; Baidu Tieba; SFACG; platform history; web archives; digital preservation; community memory

1. Research question: which kind of “history” are we trying to prove?

Web-fiction history is a history of versions. Works are renamed, migrated, deleted, reposted, or rewritten. Forums change domains, hide replies, and reorganize boards. Commercial platforms redesign channels, tags, recommendation systems, and ranking pages. Search engines expose one temporal slice of a page, while later readers describe older platforms using newer vocabulary. Michel Hockx’s study of Chinese Internet literature makes changing online texts, technologies, and publishing institutions central to its method. Jin Feng approaches Chinese web romance through production, consumption, gendered readership, and emergent subgenres. Shaohua Guo, in turn, traces Chinese digital culture through successive infrastructures such as BBSs, blogs, microblogs, and WeChat (Hockx 2015; Feng 2013; Guo 2020). A history of transformation fiction therefore begins by specifying what a piece of evidence is expected to establish.

This article separates five claims that are frequently collapsed into one. The first is earliest visible evidence: the earliest public trace currently located by the researcher. The second is platform classification evidence: whether a platform at a particular time grouped “transformation” stories through a collection, tag, list, channel, or search surface. The third is community formation evidence: whether readers sustained discussion, recommendations, indexing, and collective memory around that category. The fourth is cross-platform migration evidence: how works and readers were reorganized across Qidian, SFACG, Shuke/Ciweimao, Tieba, and other infrastructures. The fifth is contemporary continuity evidence: whether a label remains active in 2026 or has been replaced by another discovery mechanism.

The distinction turns a fragile “origin” claim into an updateable historical statement. A February 2006 Jinjiang thread states that Qidian had many male-to-female stories and that its front page had a transformation-fiction collection. That trace supports the claim that Qidian had strong genre visibility by early 2006. It also points researchers toward earlier pages and snapshots (JJWXC 2006). A 2014 Yamibo participant remembers starting web-fiction reading in 2003. That is a memory source: useful for discovering works and search targets, and stronger when paired with pages carrying contemporaneous timestamps (Yamibo 2014).

The same discipline applies to official platform histories. Yuewen’s current corporate description dates the founding of Qidian to May 2002. That is a strong infrastructure anchor; the transformation-fiction collection requires its own page, forum, and archival evidence (Yuewen 2026). Research on Qidian’s production strategy, online-literature institutions, and newer algorithmic platform logic explains how categories, rankings, payment, authors, and reader feedback became a commercial cultural system in which niche genres could become visible (Xie 2013; Shao 2015; Wang & Wang 2025). Genre history becomes more robust when structural research and specific page evidence are connected while each retains its own evidentiary role.

2. Materials and method: three clocks and one evidence card

The source base has four layers. The first consists of contemporaneous or near-contemporaneous community traces: the 2006 Jinjiang thread, Yamibo discussions from 2006 through 2014, and the 2017 Tieba index. The second consists of current platform and community pages from Yuewen/Qidian, SFACG, Baidu Tieba, and Japanese TSF sites used for comparison. The third consists of preservation institutions and technical standards, including the National Library of China, the Library of Congress, Perma.cc, Common Crawl, and Webrecorder. The fourth is research on Chinese Internet literature, platformization, fan archives, and web history, which explains why each source layer has a different form of bias (De Kosnik 2016; Helmond 2015; Brügger 2009).

Each source can be described with a compact evidence card:

The “three clocks” respond to a basic problem of web history. A page that remains live in 2026 may describe an event from 2009. A forum post from 2006 may quote an even earlier page that has since vanished. A search engine fetched in 2026 may expose a version last edited in 2025. Brügger treats the archived website as a historical document shaped by both the original site and the archiving process; later methodological work similarly emphasizes selection, temporal incoherence, missing resources, and corpus construction (Brügger 2009; Brügger 2018; Brügger, Nielsen & Laursen 2020). The Memento framework makes the time dimension explicit by modeling different representations of a web resource as time-addressable mementos (Van de Sompel et al. 2009).

The practical rule is page function constrains the historical claim. A corporate profile is excellent evidence for a platform’s stated founding date. A contemporaneous forum post is strong evidence for how readers then found and classified works. A community index shows how fans performed cross-platform organization. A current tag page establishes present-day discoverability. Institutional web archives can preserve a separate trace when the live page disappears. These materials become most informative when they remain distinct and are triangulated (Nielsen 2016; De Kosnik 2016).

3. 2002–2006: infrastructure first, then visible traces of a genre

Yuewen’s current corporate page states that Qidian was founded in May 2002, growing out of the Qidian Original Literature Association, and highlights its role in developing online paid reading (Yuewen 2026). Hockx’s history of Chinese Internet literature likewise places the commercialization and institutionalization of online writing in the broader transformation of literary production during the 2000s (Hockx 2015). By the time “transformation fiction” became visible as a platform category, it was already operating inside an infrastructure capable of using channels, collections, rankings, recommendations, and payment to shape reading paths.

The earliest strong contemporaneous anchor in the present source set comes from the Jinjiang forum on February 8, 2006. A reader looking for male-to-female time-travel fiction was told to “go to Qidian,” where, according to the reply, many such stories could be found and the front page had a transformation-fiction collection. The original poster then named several works they had already read on Qidian, including Candidate Saint, Reborn Girl, and other titles (JJWXC 2006). The thread establishes three useful facts at once: “male-to-female” and “transformation fiction” were already connected in reader vocabulary; Qidian had a strong reputation as a discovery site for the material; and readers moved across communities, asking Jinjiang users for alternatives to works they had encountered on Qidian.

A September 2006 Yamibo discussion adds a second angle. Participants debated whether a Qidian work belonged to yuri and treated “transformation fiction” as a recognizable category. One participant reproduced a passage described as coming from the Qidian Girls channel’s transformation-fiction collection, presenting the genre as a Qidian specialty that could bridge different readerships and inviting authors to contact editors for inclusion (Yamibo 2006). The evidentiary status is especially instructive. What survives in our present workflow is a forum-preserved quotation. Its source role is therefore a contemporaneous circulation record, while the original Qidian collection page remains a separate archival target. It is strong evidence that this classification text circulated among readers in 2006, and it also preserves a URL clue for a future archive search.

Together, the two traces reveal more than a list of old titles. By 2006, “transformation” existed both in a commercial discovery imaginary and in cross-site interpretive debate. The platform helped a set of works become visible together; communities argued about what the grouping meant and how it related to yuri, romance, gendered readership, and other categories. A web genre is therefore partly a narrative pattern and partly an infrastructure for discovery (Feng 2013; Nieborg & Poell 2018).

4. 2009: Tieba turns the topic into durable community infrastructure

The current page for Bianshen Wenxue Ba gives a precise community anchor: March 11, 2009 is listed as its establishment date, and the page describes the forum as a place for discussion of literary works involving gender transformation (Baidu Tieba 2026a). Its present follower and post counts describe the community’s 2026 scale. Those live metrics and the 2009 founding date belong in different evidence-card fields because they refer to different moments.

That distinction demonstrates the value of the three clocks. “Established March 11, 2009” is platform metadata about an event; the current description of the genre is a retrospective community statement that may have been edited later. The founding date can serve as a hard infrastructure anchor, while earlier forum discussions continue to document vocabulary and reader practice. Establishing a Tieba community marks a new stage in organization; the genre’s earlier history continues through the older forum and platform evidence.

A separate long-running Tieba for “transformation novels” still frames itself as a community for stories of gender conversion and contains posts in which readers search for removed or forgotten works using partial plots, platforms, approximate years, character traits, and remembered mechanics (Baidu Tieba 2026b). Such “find this old novel” threads are a form of vernacular bibliography. When a title or listing disappears, community memory supplies enough metadata to relocate a text or at least preserve the fact that readers once encountered it. De Kosnik’s account of fans as digital cultural archivists helps explain why such informal practices can become historically significant (De Kosnik 2016).

Tieba thus provides at least three layers of evidence: the continuity of a topic community, contemporaneous perceptions of platform and category change, and user-built indexes or recommendation systems. Together they form a second cataloging infrastructure outside the commercial fiction platforms themselves.

5. 2012–2014: classification disputes preserve memories of platform change

A 2012 Yamibo thread about yuri fan fiction on Qidian shows that “transformation” was already a familiar qualifier in reader discussion. Participants remembered encountering yuri through Qidian transformation fiction and debated how female protagonists, male/female channel divisions, and transformation premises interacted with platform classification (Yamibo 2012). These are reader interpretations, and their source role is audience-side classification: they document how platform rules were understood by readers. Official taxonomy is recorded separately from platform pages.

The 2014 Yamibo discussion is even more valuable for reconstructing changes in discoverability. One participant observed that after a Qidian redesign, transformation-yuri works had become harder to find; others debated transformation, yuri, sexuality, and specific titles. Another user recalled reading online fiction since 2003 (Yamibo 2014). The durable historical signal is that discoverability itself had become a subject of reader memory. A redesign could transform a once-visible collection into a dispersed set that required search, lists, or community recommendations.

Platformization research supplies the conceptual bridge. Helmond describes platformization as the extension of platform infrastructures and logics across the web; Nieborg and Poell show how cultural commodities become continuously reworked through platform data and user feedback; Poell, Nieborg, and van Dijck connect infrastructures, economic processes, governance, and cultural practice within a common framework (Helmond 2015; Nieborg & Poell 2018; Poell, Nieborg & van Dijck 2019). For niche web fiction, a label’s position in a menu, channel, tag system, or recommendation surface directly affects how large and coherent the genre appears to readers.

A historical dataset should therefore store two variables separately: whether a work was present, and whether the platform made it easy to encounter as part of a named set. A user’s 2014 remark about redesign is evidence for the second variable, while work pages, author pages, lists, and archives continue to address the first.

6. 2017: a community index compresses three commercial platforms into one table

A March 2017 Tieba post titled as an index of transformation fiction longer than 200,000 characters is the most technically revealing community source in this corpus. The moderator states that the sources were Qidian, SF, and Shuke, describes improving the Shuke crawler, and gives an explicit inclusion rule: if “transformation” appeared in the description, tags, or title, the work would be treated as transformation fiction. The post then organized results by platform, showing titles, word counts, update times, tags, and URLs (Tieba Index 2017).

The post documents an early and concrete community search-engine logic. Instead of waiting for three commercial platforms to adopt a common taxonomy, a community member defined a recall rule and normalized heterogeneous fields into one forum index. The author also comments that SF entries carried native labels, while the Qidian section was presented mainly through work and recommendation metadata. Individual SF records in the index display categories such as “性转小说,” “百合小说,” and “嫁人小说.” Community indexing therefore translated platform-specific taxonomies into a shared retrieval layer.

The index also makes its selection function visible. A title, description, or tag had to contain the chosen marker, and the 200,000-character threshold added a length filter. That means a researcher can model what the index was capable of seeing instead of treating its output as a census of all 2017 works. Web-archive research repeatedly stresses that a corpus is produced through selection and capture mechanisms; comparative studies of public web archives likewise show different coverage patterns across archives (Brügger, Nielsen & Laursen 2020; Ainsworth et al. 2012).

This historical example suggests a field that future Agent-built archives should preserve for every record: selection_rule. A work might enter through a platform tag, title match, full-text query, community recommendation, or search-engine result. Once the selection rule is explicit, the index boundary becomes evidence about retrieval conditions; historical existence is evaluated through other evidence chains.

7. 2026: the old genre remains active, but discovery interfaces have changed

Qidian currently exposes a curated list titled “专题书单之‘变身文’,” updated in February 2026, demonstrating that bianshen wen remains usable as a discovery term on the platform (Qidian 2026). The page creates an intriguing continuity with the 2006 forum traces: scattered works can still be assembled into a recognizable discovery surface. The visible form, however, is now a curated booklist, whereas the older traces describe a collection or channel-like entry point.

SFACG offers a different kind of continuity. Its current “变身” tag page includes works updated in July 2026. Its “性转” impression/tag surface places long-running and newer works into another discoverable set. A 2025 author’s listing statement even refers to the work itself as “变身文学” (SFACG 2026a; SFACG 2026b; SFACG 2025). The platform taxonomy and creator self-description can therefore be observed side by side, with each retaining its own vocabulary.

Japanese TSF web fiction provides a useful comparative case. A 2018 Shōsetsuka ni Narō creator guide distinguishes transformation, body swap, possession, reincarnation, and surgery or technological redesign. TSFes describes its festival format as a continuation of projects previously run by TS Kaitai Shinsho and identifies the 2021 second body-swap festival as the last one held there before the new site took up the practice. In 2026, Kakuyomu still returns recently updated works through TSF searches, while creators on note continue to organize series through TSF and body-swap labels (Narō Guide 2018; TSFes 2026; Kakuyomu 2026; Uzuki 2026). The comparison shows that adjacent genres can develop different degrees of taxonomy density and archival continuity across language platforms.

Chinese evidence should therefore preserve its own historical vocabulary—bianshen xiaoshuo, bianshen wenxue, bianwen, xingzhuan—while a research layer maps relationships to Japanese TSF terminology. The source layer retains what historical actors actually wrote.

This article also complements the preceding five-layer transformation-mechanism guide. That guide asks what changes inside a story: body substrate, action viewpoint, memory and skill, social identity, and reversibility or control. The present article asks how readers in different periods found stories and grouped them together. One studies narrative events; the other studies discovery infrastructure.

8. After a page vanishes: building a reproducible preservation chain

The fragility of web references predates current social media. McCown and colleagues examined URLs cited by D-Lib Magazine and found substantial link failure even in the mid-2000s. AlNoamany and colleagues later studied traffic to the Internet Archive and found that many requested archived pages had disappeared from the live web (McCown et al. 2005; AlNoamany et al. 2013). Link loss is therefore a structural condition of web history.

China also has an institutional history of web preservation. Research from the National Digital Library dates experimental collection and preservation work to 2003. Later work describes the National Library of China’s web-resource collection and preservation platform. In 2019, the National Library launched an Internet information strategic preservation project that involved cooperation with large Internet platforms (Zhang, Dun & Zhou 2017; Zhao 2020; Ministry of Culture and Tourism 2019). The National Library’s current public statement explicitly identifies long-term preservation of online information and digital resources as part of its responsibility (NLC 2026).

International web-archive practice provides complementary tools. The Library of Congress advises researchers to combine descriptive records, URL search, capture timelines, and replay, while noting that scripts, interactive components, and third-party resources can complicate capture (Library of Congress 2026). Perma.cc focuses on citation durability and links a live URL, creation date, and preserved representation. Common Crawl stores large-scale WARC collections suited to corpus discovery and text analysis. WACZ packages archived content, indexes, and contextual metadata into a portable object (Perma.cc 2026; Common Crawl 2026; Webrecorder 2025).

For GenderLibs, a practical preservation chain can be written as:

original URL → evidence card → independent snapshot → summary + content hash → research citation → later recheck

The independent snapshot and the research article perform different jobs. The archive preserves the state of a page; the article explains why that page matters in a chronology. If the live page migrates, the evidence card retains the original URL. If the taxonomy changes, the captured version retains the historical label. Provenance becomes a research asset that serves both preservation and interpretation.

9. Competing evidence: one history can leave several valid versions

The most productive moments in platform history often appear where sources disagree in form. The February 2006 Jinjiang thread refers to a Qidian front-page transformation-fiction “column,” while the September Yamibo thread preserves language about a transformation-fiction “collection.” They may refer to the same editorial surface, to related pages, or to successive configurations (JJWXC 2006; Yamibo 2006). The evidence graph can preserve both observations while leaving exact page identity open for future snapshot discovery.

The 2009 Tieba establishment date comes from current platform metadata and can be treated as a strong event anchor. The current Tieba description of the broader genre is a later community narrative and belongs in a different evidentiary field (Baidu Tieba 2026a). A 2014 reader’s memory of early-2000s reading is valuable as a search lead and memory source, while a timestamped 2006 post has a different relationship to the period being described (Yamibo 2014).

The 2017 index exposes another type of divergence through its own rule. It actively searched for the word “transformation” and imposed a 200,000-character threshold, so works using other labels, shorter works, and pages inaccessible to its crawler fall outside its designed visibility (Tieba Index 2017). That boundary is useful historical information: it tells us how one community operationalized bianwen for retrieval.

Web archives introduce further version differences. Brügger emphasizes that an archived web document is shaped jointly by the historical website and the archiving process. The Library of Congress similarly explains that dynamic scripts and complex embedded resources can be difficult to capture completely (Brügger 2009; Library of Congress 2026). The disciplined historical statement is therefore about capture status: an archive’s absence records what that archive holds, while other contemporaneous pages, indexes, bibliographies, and creator records can continue the search.

10. An executable reconstruction protocol: turn “the first” into an updateable evidence graph

The source pattern above supports a seven-step protocol that an editor or Agent can execute repeatedly.

First, define the claim. “By February 2006, Jinjiang readers treated Qidian as a major discovery site for transformation fiction” is a testable statement. It yields a clear search plan and can later be refined by earlier evidence.

Second, collect contemporaneous pages. Timestamped forum posts, platform announcements, tags, creator pages, and indexes are high-value targets. Preserve only the excerpt needed to support a claim in the article, while storing full provenance separately.

Third, separate the three clocks. Event time, page time, and capture time receive independent fields. A current Tieba page can carry event_date=2009-03-11 while its research retrieval occurs in 2026.

Fourth, record the selection function. Booklist inclusion criteria, search queries, tag-generation rules, and community index rules belong in provenance. The 2017 “title/description/tag contains 变身 + at least 200,000 characters” rule is an unusually clear historical example (Tieba Index 2017).

Fifth, construct at least two evidence chains where the claim warrants it. An official platform page can pair with a contemporaneous reader page; a forum-preserved quotation can pair with an archived snapshot; a creator page can pair with platform taxonomy. Multiple pages quoting one another remain related sources and should be marked accordingly.

Sixth, preserve original labels. 变身小说, 变身文学, 变文, 性转, and TSF remain separate source terms. A research ontology can map them at a higher layer. This supports cross-platform search while keeping semantic change observable.

Seventh, publish an updateable evidence graph whose chronology can expand with newly recovered sources. Each temporal anchor connects six entity types: platform, page, label, work, community, and archive. When an earlier snapshot appears, the graph gains a new node and the “earliest currently observed evidence” field changes. Previously verified nodes retain their value.

The method replaces a race to declare a single origin with an accumulating evidence boundary. It is also well suited to automation. Agents can search old URLs, check live-page status, create independent captures where permitted, compare label changes, and add each discovery as a provenance-rich record. Human readers can then inspect the same graph and understand why every historical statement exists.

Conclusion

The surviving public evidence already supports a platform history of Chinese transformation fiction with explicit evidence grades. Qidian’s infrastructure dates to 2002. By 2006, contemporaneous Jinjiang and Yamibo discussions connected Qidian with concentrated discovery of “transformation fiction” and preserved clues to collection-style editorial organization. In 2009, Bianshen Wenxue Ba established a durable community infrastructure. Discussions from 2012–2014 preserve reader perceptions of channel divisions, yuri/transformation relationships, and discoverability after platform redesign. In 2017, a Tieba moderator used an explicit crawler rule to build a normalized index across Qidian, SFACG, and Shuke. In 2026, Qidian lists and SFACG transformation/sex-change tags continue to make the field discoverable (Yuewen 2026; JJWXC 2006; Baidu Tieba 2026a; Tieba Index 2017; Qidian 2026; SFACG 2026a).

The larger contribution is methodological. A genre history needs the history of works, platforms, community indexing, and web preservation at the same time. Three clocks separate event, page, and capture times. Evidence cards separate page function from source role. Selection rules expose what an index was designed to see. Those structures make it possible to distinguish the earliest evidence currently found, formal platform classification, later community memory, and present-day discoverability.

For GenderLibs as a long-term research asset, this kind of provenance-rich archive is more durable than a pile of saved pages. It preserves the pages and their position in community culture. When a booklist disappears, a forum migrates, or a tag is renamed, the evidence chain still records what existed, when it was observed, how it was classified, and which claim it supports.

References

1. Hockx, Michel. Internet Literature in China. Columbia University Press, 2015. https://cup.columbia.edu/book/internet-literature-in-china/9780231160827/
2. Feng, Jin. Romancing the Internet: Producing and Consuming Chinese Web Romance. Brill, 2013. https://doi.org/10.1163/9789004259720
3. Guo, Shaohua. The Evolution of the Chinese Internet: Creative Visibility in the Digital Public. Stanford University Press, 2020. https://www.sup.org/books/asian-studies/evolution-chinese-internet
4. Xie, Qiren. “中國大陸文學網站的創作生產策略與困境：以起點中文網為例.” 2013. http://www.tjctaiwan.org/word/4834172014.pdf
5. Shao, Yanjun. “媒介融合时代的‘孵化器’.” 当代作家评论, 2015. https://fh.pku.edu.cn/docs/2018-11/20181108190721423955.pdf
6. Zhu, Li. “电子媒介时代网络文学的生产机制.” 2021. https://wkxb.ldu.edu.cn/__local/D/72/62/B6CFFA953E9279CA8EB9936CE1F_E5E88F19_DC9E8.pdf
7. Wang, Xiaoying. “超长与微短：互联网时代‘文学’的两副面孔.” 2015. http://www.semiotics.net.cn/userfiles/images/cd6ed6a56cea2e7d000eb84884cb1b80.pdf
8. Wang, Lu & Wang, Xuan. “Algorithmic drive, aesthetic transformation and the production logic of Chinese online literature within the context of the platform economy: The case of Qidian.” 2025. https://doi.org/10.22599/wcj.74
9. Helmond, Anne. “The Platformization of the Web: Making Web Data Platform Ready.” Social Media + Society, 2015. https://doi.org/10.1177/2056305115603080
10. Nieborg, David B. & Poell, Thomas. “The platformization of cultural production: Theorizing the contingent cultural commodity.” New Media & Society, 2018. https://doi.org/10.1177/1461444818769694
11. Poell, Thomas; Nieborg, David; van Dijck, José. “Platformisation.” Internet Policy Review, 2019. https://doi.org/10.14763/2019.4.1425
12. De Kosnik, Abigail. Rogue Archives: Digital Cultural Memory and Media Fandom. MIT Press, 2016. https://mitpress.mit.edu/9780262336772/rogue-archives/
13. Brügger, Niels. The Archived Web: Doing History in the Digital Age. MIT Press, 2018. https://mitpress.mit.edu/9780262039024/the-archived-web/
14. Brügger, Niels. “Website history and the website as an object of study.” New Media & Society, 2009. https://doi.org/10.1177/1461444808099574
15. Brügger, Niels; Nielsen, Janne; Laursen, Ditte. “Big data experiments with the archived Web.” First Monday, 2020. https://firstmonday.org/ojs/index.php/fm/article/view/10384
16. Brügger, Niels & Schroeder, Ralph, eds. The Web as History: Using Web Archives to Understand the Past and the Present. UCL Press, 2017. https://discovery.ucl.ac.uk/1542998/1/The-Web-as-History.pdf
17. Nielsen, Janne. Using Web Archives in Research. NetLab, 2016. https://ranke2.uni.lu/assets/pdf/Nielsen_Using_Web_Archives_in_Research.pdf
18. Van de Sompel, Herbert et al. “Memento: Time Travel for the Web.” 2009. https://arxiv.org/abs/0911.1112
19. Ainsworth, Scott G. et al. “How Much of the Web Is Archived?” 2012. https://arxiv.org/abs/1212.6177
20. McCown, Frank et al. “The Availability and Persistence of Web References in D-Lib Magazine.” 2005. https://arxiv.org/abs/cs/0511077
21. AlNoamany, Yasmin et al. “Who and What Links to the Internet Archive.” 2013. https://arxiv.org/abs/1309.4016
22. Zhao, Danyang. “国家图书馆网络资源采集与保存平台的技术实现.” 数字图书馆论坛, 2020. https://dlf.istic.ac.cn/dlf/ch/reader/view_abstract.aspx?file_no=202009007&flag=1
23. Zhang, Wei; Dun, Wenjie; Zhou, Xiaoying. “国家数字图书馆网络资源保存的实践与探索.” 数字图书馆论坛, 2017. https://journals.istic.ac.cn/dlf/ch/reader/view_abstract.aspx?file_no=201706007&flag=1
24. Liu, Qing & Kong, Fanlian. “中国网络信息存档及其与国外的比较——基于国家图书馆WICP项目的研究.” 2013. https://www.lis.ac.cn/CN/10.7536/j.issn.0252-3116.2013.18.014
25. Yuewen. Qidian platform profile. Accessed 2026-08-16. https://www.yuewen.com/
26. Jinjiang Literature City forum. “求男变女，穿越时空的文。（不要玄幻的）.” 2006-02-08. https://bbs.jjwxc.net/showmsg.php?board=17&boardpagemsg=64399&id=13441
27. Yamibo. “这类型小说属于真百合还是伪百合？” 2006-09-22. https://bbs.yamibo.com/thread-33332-1-1.html
28. Yamibo. “有没有人看起点的百合同人小说呢.” 2012. https://bbs.yamibo.com/forum.php?mod=viewthread&tid=181249
29. Yamibo. “变身小说，你是否了解？（吐槽向）.” 2014. https://bbs.yamibo.com/thread-212393-1-1.html
30. Baidu Tieba. Bianshen Wenxue community profile. Accessed 2026-08-16. https://tieba.baidu.com/p/9491164236?see_lz=1
31. Baidu Tieba. Bianshen Xiaoshuo community index page. Accessed 2026-08-16. https://tieba.baidu.com/f?kw=%25E5%258F%2598%25E8%25BA%25AB%25E5%25B0%258F%25E8%25AF%25B4
32. Baidu Tieba. “三月二十万字以上的变文索引.” 2017. https://waptieba.baidu.com/p/5018703708?is_jingpost=1&lp=5028&mo_device=1
33. Qidian. “专题书单之‘变身文’.” Updated 2026-02-20. https://book.qidian.com/booklist/detail/646421/
34. SFACG. “变身” tag page. Accessed 2026-08-16. https://book.sfacg.com/stag/679/1/
35. SFACG. “性转” tag/impression page. Accessed 2026-08-16. https://book.sfacg.com/tag/%E6%80%A7%E8%BD%AC/1
36. SFACG. Author listing note using the term “变身文学.” 2025-05-02. https://book.sfacg.com/Novel/737447/985243/9087942/
37. Shōsetsuka ni Narō. “TSF文化はよくわからないけどTSF描写を書いてみたい方へのアドバイス.” 2018. https://ncode.syosetu.com/n2405es/
38. TSFes. Site description and project history. Accessed 2026-08-16. https://tsfes.com/about/
39. Kakuyomu. TSF search page. Accessed 2026-08-16. https://kakuyomu.jp/search?q=%EF%BC%B4%EF%BC%B3%EF%BC%A6
40. Uzuki Raina. “TS音楽魔法世界シリーズ　男と女が体を入れ替わりつつ音楽で青春魔法バトルします.” note, 2026-04-18. https://note.com/april_liner/n/nc4cabdf6810b
41. National Library of China. Statement on preservation of online information and digital resources. Accessed 2026-08-16. https://www.nlc.cn/web/dsb_footer/dsb_zcwm/dsb_cbwjs/dsb_sm/index.shtml
42. Ministry of Culture and Tourism of the PRC. “国家图书馆启动互联网信息战略保存项目.” 2019-04-23. https://www.mct.gov.cn/whzx/zsdw/zggjtsg/201904/t20190423_843029.html
43. Library of Congress. “Tips on Searching the Web Archive.” Accessed 2026-08-16. https://www.loc.gov/programs/web-archiving/for-researchers/searching-the-web-archive/
44. Perma.cc. “About Perma.cc.” Accessed 2026-08-16. https://perma.cc/about
45. Common Crawl. “Overview.” Accessed 2026-08-16. https://commoncrawl.org/overview
46. Webrecorder. Web Archive Collection Zipped (WACZ) 1.2.0. 2025. https://specs.webrecorder.net/wacz/1.2.0/