Creative Class or Precarious Workers: Animation Recreators in China's Video Platform

About this archive

This page reproduces the dissertation's original English text for the English edition of MAD DOC. The wording and terminology are the author's own; only PDF pagination and Markdown heading levels have been adapted for web reading. The research is based on 69 semi-structured interviews and fieldwork, and its sample, period, and interpretations should not be generalized to the entire MAD·AMV community without qualification. See Research Methods and Survey Design for the site's later research framework.

Original author: CHEN, Huaisong
Master of Arts in China Development Studies dissertation, The University of Hong Kong, July 2024.
Download the locally archived original PDF · Source page

Abstract

“CREATIVE CLASS OR PRECARIOUS WORKERS: ANIMATION RECREATORS IN CHINA'S VIDEO PLATFORM” This is a dissertation studying gig economy phenomenon of animation recreators on Chinese video platform Bilibili, pictured the business model of creative gig workers in video production industry. Research conducted 69 semi-structured interviews and field research, acquired personal information and opinions of animation recreators within ten years. Results showing that these creative workers displaying much more diversified features comparing to traditional gig works, however, there are arguments from both sides of the future development within these creative workers. The supporting economy seems booming with potential risks, while the creator communities maintained sustainable. In general, this research provides an unique insight of nowadays new form of gig economy, related to current status of China's flexible employment and social media platform economy. Submitted by CHEN, HUAISONG for the degree of Master of Arts in China Development Studies at the University of Hong Kong in July 2024 Keywords: Gig Economy; Platform Economy; Creative Industry; Video Platform; Animation Recreators; Bilibili; MAD

Declaration

I declare that the work in this dissertation is entirely original, with the exception of a few instance where proper acknowledgement is given, and that is has not been included in any previous thesis, dissertation, or report that has been submitted for a degree, diploma, or other qualification to this University or any other institution. Signed: (CHEN, Huaisong)

Acknowledgement

This research could not be done without the essential help provided by Dr. He WANG of the University of Hong Kong. She represents the best supervisor that a master student could ever have. And I would like to thank the MAChDS program at the University of Hong Kong, including all the lecturers who ever taught me and Mr. Gee Yeung, for giving me one of the most important and enjoyable educational experiences of my life. I would like to express my gratitude to all the MADers I met on Bilibili. Despite I never meet many of them in person, they have all done their best to help me finish this dissertation,. And I would also like to express my gratitude to all the friends I made in the University of Hong Kong. What I learned from communicating with these classmates was just as much as what I learned in the lectures. I am also very grateful to my parents and my younger sister. Without their support over the past 20 years, I would not be where I am today. And I hereby express my gratitude to all the important friends who have appeared in my life so far also. And to those who are yet to come, those whom I will cherish in the future, please understand that I am doing my best to meet you with the best version I can have. CHEN, Huaisong July 2024 Hong Kong SAR, China

Table of Contents

Declaration … 03 Acknowledgement … 04 Table of Contents … 05 Table of Internet Slang Related to the Topic … 07 I. INTRODUCTION … 10

1. Uncleared Definition: Gig Economy … 10

2. Background: A New Emerging Form of Gig Economy, Particularly in China … 12

3. Research Question … 15 II. LITERATURE REVIEW … 18

4. Current Statistics and Observations on Gig Economy … 18

5. Gig Workers under Platform Economy … 25

6. From Japanese Doujin Economy to Animation Recreation … 32 III. METHODOLOGY AND DATA COLLECTIONS … 39

7. Methodology … 39

8. Data Collections … 41 IV. RESULTS … 43

9. General Description of MADers … 43

10. MAD and AMV and Other Various Recreation Forms … 48

11. Professional Video Production Skills and Learning Path … 62

12. MAD Related Gig Works, Career Perspective and the Supporting Economy … 66

13. Platform, New Technology, the Sustainability in MAD Community … 84 V. DISCUSSION AND CONCLUSION … 90 Appendix: List of Sample Interview Questions … 93 References … 96

Internet Slang Related to the Topic

Not listed in alphabet order. ACG Animation, Comics and Games. Specific refer to commercial or cultural products made by Japan. In most cases, the term can be extended to include physical goods or digital derivatives related to ACG. In a few cases, the term can be extended to include electronic or physical items made in China that have a Japanese style. Most people do not recognize or even oppose the use of “ACG” to refer to European and American animation, comics, and game products.

• Sometimes people use the phrase “ACGN” to include Japanese light novels. Nijigen “二次元(にじげん)” in Japanese, literal translation as two-dimensions. Since ACG culture mainly exists in digital or virtual status, which has only two dimensions (no thickness). This slang has almost the same meaning with ACG, and also commonly used among the ACG community. Hence, ACG will be the main term used in this dissertation. Doujin “同人(どうじん)” in Japanese, with an original meaning of “same type of people”, referring people with same hobbies, interests etc. Now being narrowed down as a term describing recreation of ACG

works, regardless of forms. Recreation The original word is “二次创作” in Simplified Mandarin Chinese, considering there is no cultural elements attached to this term, hence it is replaced by its literal English translation in this dissertation. In the field of ACG, recreation generally means using elements in existing works to create amateur's own work regardless of forms. However, referencing the source is strongly encouraged. MAD Music Anime Douga, is Music with Japanese “アニメ動画” in Romaji, which literally means animation video with music. A recreation using (usually commercial) comics, animations or games to create amateur's own video. People in this field tends to believe MAD should have a high quality. AMV Anime Music Video, western way of calling MAD. Some times only refer to animation recreation. Jing-zhi Both Japanese Kanji and Chinese character written as “静止(static)”, usually called MMV in the West, which is Manga (Japanese Comics) Music Video. Contrary to AMV, Jing-zhi and MMV refer to recreation videos made from ACG comics. Because comics are frozen images, static seems to be a proper term to describe it. GMV Game Music Video, technically, GMV = MAD — AMV — MMV But GMV sometimes could refer to games outside of ACG.

Chun-jian “纯剪” in Mandarin Chinese, meaning “pure editing”, which is a form of AMV but purely use edition without post production. People usually use the phrase “纯碱(sodium carbonate, Na2CO3)” as Chun-jian in this field, because this is usually the first results typing in Chinese pinyin. Hun-he “混合” in Mandarin Chinese, meaning “mix”, a MAD has both AMV and Jing-zhi parts. Only slang with high using frequency in this dissertation are listed above. Please noted that due to the nature of the Internet, the sources of some slang listed above are not able to determine. Hence the interpretations are summarized from observations during the field research of this dissertation, based on interviewees' explanations, public opinions and some widely recognized websites created with long history within the online community. That is to say, meanings of slang listed above are likely to change after a certain periods of time.

I. Introduction

1. Uncleared Definition: Gig Economy

When neoliberalism swept all over the world, the increased mobility in the labor market directly led to an increase in the proportion of short-term contracts between enterprises and workers in the market. Enterprises rely on freelancers rather than full-time long-term employees to fill temporary or even permanent positions. The resulting “Gig Economy” has in fact reduced the labor costs of enterprises and given birth to a variety of new professions and new industrial processes. From the perspective of China's development in the past decade, the gig economy has also appeared in multiple industries and deeply participated in the lives of ordinary people in some areas, reflecting one aspect of China's vigorous economic development. The origin of the gig economy can be traced back to the United States in the 1970s. Due to changes in labor laws, the decline of unions, and the rise in unemployment rates, US companies were able to reduce wages, reduce welfare costs, and reduce the risk of unfair dismissal lawsuits by establishing short-term employment relationships for targeted specific needs (Friedman, 2014). Discussions about the gig economy, the usual focus in the Western World is on Uber, Deliveroo, and Airtasker. This is because these companies with their enormous market coverage have disrupted existing consumption patterns and even expanded the market size, by increasing suppliers and facilitating consumers (Healy, Nicholson, and Pekarek, 2017). Healy, Nicholson, and Pekarek (2017) stated in their research that the rise of

the gig economy has posed new challenges to traditional business models, labor management practices, and regulations. This is because the gig economy represents a new form of service delivery, connecting buyers and sellers directly or indirectly through platform companies. Although the gig economy still accounts for a small proportion in the broader labor market, its growth is extremely rapid (Healy, Nicholson, and Pekarek, 2017). Even if researchers believe that economic, industrial, and political factors may slow down or prevent the growth of the gig economy, making it unlikely to become the mainstream form of the labor market in the future, the article still emphasizes the necessity for the academic community to systematically study the gig economy. Besides a short-term employment relationships and a new form of service delivery, there is also another perspective of defining gig economy. For example, whether gig workers should be considered as employees or self-employed individuals, and whether specially tailored labor laws are needed to protect these gig workers, are issues that need to be considered from the perspective of labor laws (Todolí-Signes, 2017). Or in many cases, scholars regard service providers on the platform as independent contractors, thus, in some sense, it is an expansion of traditional individual operators (Donovan, Bradley and Shimabukuru, 2016). Nevertheless, Todolí-Signes (2017) believed that for the new types of workers that appear in the new environment, there is a need to legislate relevant new laws. The current research framework on the gig economy is chaotic, partly due to the rapid development of information and communication technology (ICT), leading

to an emergence of the platform economy, which largely reshaped the original form of the gig economy, causing scattered discussions and unclear agendas (Malik, Visvizi, and Skrzek-Lubasińska, 2021). Malik, Visvizi, and Skrzek-Lubasińska (2021) studied 378 papers on the gig economy up to February 2020, concluding that scholars need to pay explicit attention to the different scenarios of the gig economy at the local, regional, national and potentially transnational levels. Meanwhile, this research roughly cleared the conceptual boundaries of the platform economy and the gig economy. That is, the platform economy mainly emphasizes the connection between employers and employees through online digital platforms, while the gig economy maintaining the original short-term contracts and freelance features, can generate new economic collaboration models due to the emergence of platforms, which include but are not limited to the possibility of geographically dispersed virtual teams appearing in some new fields, creating the possibility of higher value-added gig work. This possibility has already partially appeared in reality, and this paper will further discuss it in detail in following content.

2. Background: A New Emerging Form of Gig Economy, Particularly in China

Since July 2020, with the issuance of a document by the People's Republic of China State Council supporting “flexible employment (Ling Huo Jiu Ye)”, discussions on the gig economy in China seem to have become more conceptually muddled. Flexible employment, platform economy, and gig economy appear to overlap in describing similar phenomena, yet they each have their distinctions. Regardless, according to the content of Document No. 27 of 2020 by the State Council of China,

“flexible employment” includes individual operations, non-full-time, and new forms of employment channels. Among these, “new forms of employment” particularly emphasize industries such as digital economy and platform economy, including online retail, mobile transportation, online education and training, internet healthcare, and online entertainment. This undoubtedly demonstrates China's current push towards diversification and higher value-added directions in the gig economy. Therefore, new forms of gig economy also emerged. In China, a clear trend is that the emergence of flexible employment has led to innovation in occupations. Many young people born in the internet era prefer jobs with an internet background, such as freelance writers, live streamers, and translators (Lei, Niu, Zhang, and Jiang, 2018). Lei, Niu, Zhang, and Jiang (2018) argue that the bond between individuals and organizations is becoming increasingly weak. Non-full-time jobs offer more flexible working hours and allow for economic expression of interests and professional skills. This is underpinned by the rapid growth of China's human resource outsourcing market over the past 20 years. The internet entertainment industry has already formed a vast market. According to the Digital China Development Report released by the State Internet Information Office in April 2023, the number of online music users in China reached 684 million, with an internet user penetration rate of 64.1%. Sixteen industries with distinct characteristics of new cultural business forms, such as animation and internet cultural entertainment platforms, achieved a total business revenue of 4.386 trillion yuan, a year-on-year increase of 5.3%.

Currently, internet entertainment content platforms are reshuffling the entire entertainment industry. New production and distribution methods, along with new consumption patterns, are forcing traditional media and entertainment groups to contend with new formidable competitors, and gradually losing their once dominant position as the former are more popular with the younger generation (Chalaby, 2024). When users can create videos and upload them to sharing websites, it attracts more new users to the site, ultimately generating more advertising revenue for the platform. This platform also makes it easier for users to find videos they like, creating a positive feedback loop between user engagement and advertising income (Parker, Van Alstyne, and Choudary, 2016). According to Chalaby (2024), the scale and market valuation of these companies during this process owe to their creation of value using assets and workers beyond their boundaries. Clearly, this does not include any long-term stable employment relationships, but the inherent value of the market itself creates conditions for spontaneously formed new forms of trade. In this dissertation, this research focused on one particular platform called Bilibili. Undoubtedly, the highest-quality video sharing platform in mainland China today is a platform called Bilibili. Usually referred to as the “Chinese version of YouTube”, Bilibili consistently receives praise for its superior video content compared to other platforms. Its long-term growth has garnered high expectations from both the public and investors. According to Bilibili Inc.'s 2023 Annual Report, the platform achieved a significant milestone by surpassing 100 million daily active users in the latter half of 2023. Individual active users spent nearly 97 minutes per day on

Bilibili's application, resulting in a 17% increase in total user engagement compared to the previous year. Despite ongoing losses in 2023, Bilibili managed to reduce its net loss by half, and the gross profit margin improved from 17.6% in 2022 to 24.2% in 2023, with positive operating cash flow of 640 million yuan generated in the fourth quarter. A critical factor contributing to this success is the substantial number of individual creators leveraging the Bilibili platform to share content. In 2023, Bilibili received an average of approximately 21.5 million monthly submissions, representing a 46% year-on-year growth. Furthermore, over 3 million creators accessed revenue channels provided by Bilibili, marking a 30% increase from the previous year. Notably, the number of creators earning income through video and live-streamed product endorsements grew by 133%. Clearly, a significant portion of these creators operate within the framework of gig economy or platform economy, maintaining varying contractual relationships with the Bilibili platform in different levels. In this study, we delve into the development of Bilibili's oldest creator community over the past decade from a micro perspective as a new emerging form of gig economy.

3. Research Question

What many people nowadays are unaware of is that Bilibili, in fact, gradually transformed into a platform for all types of videos after 2018. From its establishment in 2009 until its recent-began commercialization, Bilibili remained a minor community for ACG (Animation, Comics, and Games) enthusiasts for a very long time, in which ACG particularly referred to related works in Japanese “Nijigen (二次 元)” culture (meaning two dimensions, which is not in real life). Among these

enthusiasts, a prevalent form of creative expression involved making videos by remixing existing commercial animations and comics, adding their chosen music — a practice that was mainstream at the time, also originated from Japan's “Doujin (amateur)” culture. These animation “recreation” remix videos were commonly referred to as “MAD” or “AMV”, abbreviations for “Music Anime Douga” (Japanese of “music animation video”) or “Anime Music Video” in English, respectively. The creators of these videos playfully called themselves “MADers”. According to the recollections of the interviewees in this study (Interviewee No. 2, 4, 55, 69, etc), MAD used to account for maybe more than 50% of new uploaded videos at Bilibili till the year 2014. What's more, because this form of artistic expression often utilized various digital techniques, the MADer community on Bilibili has long been regarded as the groups with specialized video production skills. Based on early attention to this group and some preliminary interactions before formal research, it is found that this community has extremely diverse personal backgrounds. Through animation recreations (which will be mentioned as creating MAD videos or MAD creations hereafter), they have acquired a considerable level of professional skills and, using these skills, earn income in various ways during their spare time. Even though MAD has fallen from mainstream popularity to become one of the most marginalized video formats on the Bilibili platform, MADers continue to find their own profit models within Bilibili's content environment, and MAD itself has not disappeared despite its general lack of attention. The gig economy phenomenon surrounding MADers shares some similarities with existing gig economies like food

delivery, online-ordered taxi, or live streaming platforms, but there are significant differences as well. What's more, notably, a considerable proportion of MADers seem to transition from gig work to video-related professional positions in relevant industries. To date, no research has looked in detail at the identity, skill level, learning channels, profit models, and employment potential of these Chinese animation recreators. The economic phenomenon surrounded MADers appears to be a unique but highly relevant case study in the flourishing landscape of internet content economics. This study aims to survey these animation recreators in Bilibili, mapping out the skill sets, incentives, working conditions and career prospects of MADers. It provides a new perspective for researching gig economies and platform economies, shedding light on one facet of China's current “flexible employment” related groups.

II. Literature Review

In this section, the study provides a detailed summary of existing literature on the gig economy, focusing on the valuation of its worldwide volume, the current status of the gig workers, and the role of platform economies. Additionally, considering the research question's relevance to the creative industry, the study also compiles information on the economic of Japan's relevant industry and the impact of existing platform economies on creative content production.

1. Current Statistics and Observations on Gig Economy

According to PwC Legal's “Gig Economy 2022” report, based on statistical analysis across 13 major European countries, both employee and self-employed phenomena are prevalent. Specifically, in Austria and Spain, gig workers are commonly worked as employees, while in Belgium, France, Germany, Greece, Italy, and the Netherlands, self-employment is the dominant form for gig workers. France currently has 3.6 million self-employed workers, with an average annual growth rate of 3.3% since 2009. Among them, gig workers constitute approximately 7% of individual entrepreneurs and 0.8% of the working population. Around 700,000 Italians engage in gig economy work, contributing 0.7-1.3% to Italy's GDP. The gig economy plays a significant role in the UK labor force, with approximately 4.4 million people in England and Wales working for gig economy platforms at least once a week in 2021. While in the Netherlands, gig workers account for 0.9% of the total working population. The report suggests that countries like Belgium and Norway have yet to fully tap into the potential of the gig economy. Overall, there is a lack of specific data

on gig economy scale or its overall economic importance in countries including Austria, Denmark, Greece, Norway, Spain, and Sweden. Ireland even faces serious doubts about the accuracy of its statistics. However, one clear conclusion remains: the gig economy, initially limited to transportation and food delivery services (such as Uber and courier services), is now expanding into more diverse business sectors. Especially during the COVID-19 crisis, individuals increasingly seek online employment opportunities, allowing them to maintain social distancing. According to a survey by the Pew Research Center, as of 2021, 16% of Americans have earned money from online gig platforms. These gig workers engage in various tasks, including driving for online-taxi apps, shopping for or delivering groceries, performing household chores, running errands, making restaurant or store deliveries for delivery apps, and other similar activities with same operating logic. Notably, there are significant age differences in gig work participation: 30% of individuals aged 18 to 29 have earned money through online gig platforms, compared to 18% among those aged 30 to 49, and even lower for those aged 50 and above. Ethnicity also plays a role, with 30% of Hispanic adults participating in the gig labor force, significantly higher compared to 20% of Black adults, 19% of Asian adults, and 12% of White adults. Overall, at the time of the survey conducted, 9% of U.S. adults had earned money through online gig platforms in the past 12 months. Among them, 31% relied on gig work as their primary income source. Notably, 42% of lower-income individuals considered gig work their main job in the past year. Regarding income, 23% viewed gig income as essential, 35% as important, and 39%

as “nice to have but not needed”. Still, many people turned to gig work during the height of the COVID-19 pandemic, for reasons either to save more money or to offset income fluctuations (Anderson, McClain, Faverio and Gelles-Watnick, 2021). Research data specifically focused on China's gig economy remains relatively limited. According to a document from China's National Development and Reform Commission in 2021, the number of shared economy employees in China reached 75 million in 2018. However, as of the document's publication, the official definition of flexible employment remains ambiguous, and the National Bureau of Statistics does not provide specific statistics or a corresponding policy framework. In the same year, the China Information Center released the “China Shared Economy Development Report (2021)”, which estimated approximately 830 million participants in the shared economy, including around 84 million service providers (a year-on-year growth of approximately 7.7%). Additionally, platform company employees numbered approximately 6.31 million (with a year-on-year growth of approximately 1.3%). During a press conference on China's national economic performance in 2021, the head of the National Bureau of Statistics stated that flexible employment in China had reached around 200 million people, with examples such as over 4 million food delivery riders in some platforms and approximately 1.6 million individuals engaged in anchor-related positions on various platforms, nearly tripling from the previous year. According to one of the newest official report by Ministry of Human Resources and Social Security, comparing the first quarter of 2020 to the first quarter

of 2023, the proportion of new flexible employment recruits at the national level increased from 13.8% in Q1 2020 to 28.8% in Q2 2022 but then declined to 19.1% in Q1 2023. The report speculates that the economic recovery in 2023 reduced the demand for flexible employment. However, compared to the period from 2018 to 2020, the demand for new flexible employment recruits actually increased from 2020 to 2023. Among the total job seekers, the proportion of new flexible employment seekers rose from 18.6% in Q1 2020 to 23.2% in Q1 2023. Notably, industries and professions closely associated with the digital economy exhibit higher penetration rates for new flexible employment positions. Sectors such as culture, media, entertainment, and sports have 27.7% of their positions seeking flexible workers. Particularly in internet-related professions like film production, new media operations, and performing arts/management, the penetration rate for new flexible employment positions exceeds 50%, reaching a high of 79.4%. Thus, flexible employment has become a predominant trend in these industries. New flexible employment positions tend to have lower educational and work experience requirements for job seekers, offer higher wages, but provide lower benefits compared to traditional employment positions (Information Center, Ministry of Human Resources and Social Security of the PRC, 2024). In China, informal data sources also provide some insights, as official statistical methods are too complex. Notably, in November 2020, Tsinghua University's Institute of Economics, in collaboration with ByteDance Public Policy Research Institute, published a report on China's gig economy. According to

calculations in the report, China's gig economy reached a lower limit of 900.31 billion yuan in 2018, accounting for 2.16% of the country's GDP. By 2019, the gig economy had expanded to 990.86 billion yuan, representing 2.64% of the GDP for that year. Considering the GDP deflator, the gig economy contributed 10.43% to the total GDP increment from 2018 to 2019, with gig economy's absolute growth accounting for 29.4% of the overall GDP increment. Notably, the report categorizes the gig economy into seven sectors: transportation, shared accommodation, food delivery services, live streaming, professional skills services, knowledge payment, and content creation. Among these, content creation dominated both years. In 2018, content creation reached 742.3 billion yuan, surpassing the second-place food delivery services at 481.64 billion yuan. The following year, content creation further increased to 854.2 billion yuan, exceeding food delivery services at 627.57 billion yuan. Additionally, if we consider gig work related to live streaming, its scale surged from 188 billion yuan in 2018 to 494.2 billion yuan in 2019, which is more than two times, showing the most significant growth among the seven categories. From a macro perspective, the development of China's gig economy has been studied based on panel data from 285 mainland cities between 2013 and 2020. Gao, Zhang, Wen, and Si (2024) discovered significant disparities in gig economy scale between the Pearl River Delta (Hong Kong, Macau and Guang Dong), the Yangtze River midstream and Chengdu-Chongqing regions. These disparities have been increasing over the observation period. However, gig economy indices in China's major strategic regions have shown positive trends, indicating optimism for both

current performance and future prospects. Notably, the internet has played a crucial role in boosting the gig economy in recent years. Nevertheless, the researchers caution against potential talent drain effects in super-large cities and emphasize the need to promptly enhance gig worker employment support systems. Regarding online gig works worldwide, the current global landscape reveals that there are 545 online gig platforms, headquartered in 63 countries. These platforms connect workers and clients across 186 countries (World Bank, 2023). While acknowledging the lack of reliable data sources to estimate the overall scale, a World Bank report on online gig work (2023) attempted a global online survey using experimental random domain intercept techniques (RDIT) across six regions and 17 countries. The total amount of estimated sole-platform registered online gig workers globally range from 154 million to 435 million, almost 4.4% to 12.5% of the total work force, with a caveat that these figures may be underestimated. Researchers estimate that there may be about 132.5 million gig workers worldwide who work as a primary job, 173.7 million who work as a secondary job, and 106.2 million who work as a fringe job. The length of time spent doing gig work and the proportion of their total income that comes from gig work vary widely. Approximately 60% of lower-middle or low-income countries and nearly 50% of upper-middle or high-income countries' surveyed firms confirmed an increasing trend in outsourcing work to gig workers over these years. With the development of the digital economy, the demand for gig workers in low- and middle-income countries has also been steadily increasing, even exceeding

that of industrialized countries. In sub-Saharan Africa, the growth rate of recruitment information on large digital labor platforms is as high as 130%. Unlike upper- and middle-income countries, the main driver of the growth in demand for online gig workers in middle- and lower-middle-income countries is small and medium-sized enterprises. However, as in developed countries, most informal workers in low-income countries are in particularly serious situation, with more than 90% of workers lacking social insurance (Sharif and Qiang, 2023). Also, it's essential to note that while gig work creates new job opportunities, the gig economy still faces risks and inequalities. Many individuals who lack internet access or digital devices such as laptops, smartphones, and tablets are excluded. Additionally, most online gig platforms operate not globally but regionally or locally, connecting employers and workers from several countries within one nation or a region. Although local platforms may not receive as much attention as global ones, they play a crucial role not only in the local labor market but also in the local private sector ecosystem, which is exactly the same focus of this paper. In general, the statistics of the gig economy in various countries around the world are still incomplete, mainly because the definition and statistical caliber are vague. However, the data that can be estimated or calculated show that the gig economy is an important part of the economy of various countries, occupying a proportion that cannot be ignored. Especially under the influence of the epidemic, the importance of the gig economy to some people has further increased. At the same time, companies also have the need to hire gig workers. These phenomena are common all

over the world, both in developed and developing countries. In addition, people have clearly discovered the significance of platforms for the gig economy and emphasized the importance of platforms. Especially in China, the gig economy related to the Internet-related entertainment industry, content creation, etc., even exceeds the food delivery platform in terms of total scale. However, existing literature assessing the overall volume of the gig economy lacks detailed descriptions of business models within specific industries. In this study, we aim to partially address this research gap by exploring one perspective of the creative gig economy, particularly within the ACG style video production industry.

2. Gig Workers under Platform Economy

In a report by the McKinsey Global Institute, researchers identified a four-group taxonomy to analyze the specific situations of gig workers. These groups are: Free agents, who actively choose independent work to derive their primary income source; Casual earners, who use gig work for supplemental income by choice; Reluctant workers, who have to rely on gig work as their primary income but would prefer traditional jobs; and Financially strapped individuals, who engage in gig work out of necessity (Manyika, Lund, Bughin, Robinson, Mischke & Mahajan, 2016). This classification is based on the degree of income need and flexibility among gig workers. Best (2017) affirmed that this taxonomy aligns well with the current gig economy landscape. Meanwhile, Best's paper claimed that it is essential to recognize that gig work should not be solely measured by duration, as different tasks have varying time requirements. Moreover, with the help of platforms and the development of

Information Communication Technology, workers often engage in multiple gig jobs simultaneously, facilitated by current technology levels and resources like massive open online courses (MOOCs), allowing them to acquire specialized skills for specific work. Best (2017) also presented an in-exhaustive list about three economic sectors that has closest link to today's gig economy: Household sector, IT and Media sector, Transportation sector. These three classifications can be further subdivided into several subcategories. From the perspective of work allocation, it is best to distinguish between Crowdwork and Work-on-demand via app (De Stefano, 2016). Crowdwork Involves online platforms distributing a series of tasks to workers. These tasks are typically performed online. As for the “Work-on-demand via app” type of gig work, refers to the allocation of traditional offline tasks (such as transportation, cleaning, and errand services) through mobile applications. For Work-on-demand via app, platforms need to establish minimum quality standards for gig workers and take responsibility for selecting and managing the workforce. However, De Stefano (2016) found that gig workers are often misclassified as independent contractors rather than platform employees. Such misclassification prevents workers from enjoying the labor law and social security protections they should receive. When gig workers are treated as “independent contractors”, they assume risks and responsibilities that should be borne by employers, such as compliance with social security and minimum wage regulations. While platforms benefit from reduced labor costs, this can lead to unfair competition in the market. Over time, it may contribute to the informalization of the

labor market, increasing worker instability and income uncertainty, ultimately worsening labor conditions in the industry. Gig workers, especially those who are already in a disadvantaged position, face far more difficulties than what is mentioned above. According to a report published by SpringerLink in 2018, gig workers face three common vulnerabilities: occupational risks inherent to the work, work instability, and pressures from the platforms they operate on. Work instability encompasses not only uncertainty related to health insurance and social security but also uncertainty regarding access to production resources. Gig workers are influenced by the tools provided to them and the limitations of their own production materials. Additionally, gig work can impact individual career development due to uncertain training opportunities. The platform-caused worker vulnerabilities are equally critical. These include worker misclassification, information asymmetries, and the culture of surveillance. Gig workers often lack bargaining power in pricing decisions due to platform algorithms, which determine rates without their input. Moreover, platforms closely monitor and evaluate workers, creating significant social and psychological stress. Gig workers fear losing income due to uncontrollable factors, striving to maintain positive ratings to avoid algorithmic downgrading, which could lead to an effective dismissal by not getting more jobs from platform system (Bajwa, Gastaldo, Di Ruggiero and Knorr, 2018). Health of gig workers is not a new topic. In the contemporary gig economy, the prevalent practice of determining income based on the amount of work performed

has historical warnings before platform economy emerging. Research based on data from the 1979 National Longitudinal Survey of Youth (NLSY79) in the United States reveals that piece-rate wages, despite potentially increasing productivity, have negative implications for worker health from a public health perspective (Davis & Hoyt, 2020). Compared to fixed salaried work, workers paid by piece rate are more likely to underreport health issues, particularly among low-income, female, and non-white workers who face higher health risks. Consequently, piece-rate wages disproportionately affect vulnerable segments of the U.S. workforce. In some cases, as an intriguing twist, social media platforms have also become involved in the gig economy from a unique and positive perspective. In Thailand, these platforms attempt to cast gig workers, particularly food delivery riders, as “heroes” (Mieruch & McFarlane, 2023). This “hero narrative” portrays delivery riders as independent partners, emphasizing their pride and service-oriented mindset. It praises their resilience and dedication to work despite challenging conditions, aiming to shape a distinct identity. However, Mieruch and McFarlane discovered that this hero narrative serves to replace the worker narrative, encouraging riders to accept their precarious work conditions rather than asserting their rights as employees entitled to labor protections. By framing gig workers as independent contractors, platforms shift risks and responsibilities from employers to workers, such as social security compliance and adherence to minimum wage regulations. This narrative conceals the unequal power dynamics inherent in the gig economy. Meanwhile, interestingly, some Thai riders use social media platforms to resist. They create virtual

spaces for socialization, mutual aid, and collective action among gig workers, leading to the emergence of grassroots civil society organizations advocating for worker rights. Ultimately, this competition of identity narratives evolves into quasi-union-like organizations facilitated by social platforms, collectively striving to improve working conditions (Mieruch & McFarlane, 2023). Indeed, it's essential to recognize that beyond gig workers facing various socioeconomic challenges, there is also a subset of self-driven gig workers. Drawing from the self-determination theory (SDT) in psychology, recent research has found that some gig workers engage in work with full willingness, intention, and a sense of choice, driven by intrinsic interest and satisfaction (Jabagi, Croteau, Audebrand & Marsan, 2019). Consequently, many gig organizations aim to enhance efficiency through external rewards and/or psychological manipulation. Furthermore, Jabagi, Croteau, Audebrand and Marsan discovered that social networks partially fulfill platform workers' need for interpersonal connections and recognition of individual abilities. These networks even contribute to a sense of belonging within online communities. The design of platforms to optimize gig workers' experiences will significantly impact long-term competitiveness and sustainability. If platforms can better meet gig workers' needs, foster positive social environments, and provide meaningful incentives, they can attract more self-driven gig workers, ultimately promoting sustained platform growth. Similarly, a study of 450 online ride-hailing drivers in a university town in the central United States also showed that the emergence of social networking sites (SNS)

has established a collective labor identity for online ride-hailing drivers in the online social space, which has affected drivers' views on the instrumentality of unions and collective organizations. More frequent online interactions ultimately led to a more active interest in joining the driver union (Maffie, 2020). Maffie believes that online worker networks have become a new and institutional participant in the modern industrial system. Sometimes social media platforms provide emotional value for much more wider range. During the COVID-19 pandemic, many businesses and individuals faced near-devastating impacts. In this case, social media played a supportive role in minimizing these impacts (Arora & Sharma, 2022). Arora and Sharma's research reveals that the fusion of gig economy, traditional business strategies, social media platforms, technology, and entrepreneurial spirit led to breakthrough developments. Notably, platforms like Facebook, Twitter, and Instagram experienced a surge in users during the pandemic, becoming accessible channels for learning and earning without significant barriers. Essentially, people could leverage their existing skills to generate income that was previously unavailable, all while experiencing a sense of fulfillment. Amid economic crises, gig work provided a minimally sustainable yet viable source of income for workers, even under relatively low terms. Kenney and Zysman (2019) further elaborate on the role of content platforms in the gig economy. Although it is difficult to precisely quantify the impact of platforms on employment and to accurately measure the value created by the platform economy, digital platforms undeniably facilitate low-cost and convenient connections

between people. Retailers can operate from a bedroom or a garage, and digital work or products can originate from anywhere in the world. These goods and services are available around the clock, and gig workers have the freedom to choose their own working hours. However, Kenney and Zysman also emphasize the need to be cautious about the power wielded by platform owners. From the platform's perspective, companies like Amazon can leverage their control over data to favor certain retailers with more exposure. Simultaneously, platform owners rarely disclose how their algorithms are constructed, and they can modify these algorithms without notifying users. As a result, users often have to rely on guesswork to understand the platform's mechanisms and make decisions about their actions. In China, according to a study conducted by Beijing University of Technology, community-based gig workers have significant advantages over traditional gig workers in terms of job opportunities, flexibility in working hours, and compensation. Traditional gig work suffers from inefficient information matching and high labor time costs (Yang & Lu, 2024). Yang and Lu point out that community-based networks have transformed job-seeking channels from unidirectional to multi-directional, eliminating information gaps and accelerating information flow. As a result, gig workers can proactively book future work based on online information during their leisure time, leading to relative stability in workload. Yang and Lu also found that the economic benefits for platform gig workers depend on accumulating positive credit evaluations and spreading word-of-mouth reputation. Whether spontaneously formed or organized, community-based operations play a crucial role in ensuring effective

labor benefits for gig workers, highlighting the importance of standardized communities. Chinese researchers have also investigated why employees in enterprises engage in side jobs from a unique perspective. Their findings, apart from income-driven factors and emotional value needs, also highlight career development needs (Lin, Liu & He, 2024). Lin, Liu, and He conducted a study involving 27 semi-structured interviews and 7 online interviews. They discovered that, within career value goals, individuals perceive interpersonal relationships or network connections not as a method, but as an achievement target. In this context, side jobs serve as a means to establish connections with upper social networks, reflecting the social nature of Chinese society. Additionally, the work environment significantly influences employees' side job choices. In China's collectivist cultural context, people tend to prioritize their primary works. Existing literature has primarily discussed the livelihoods of workers in the gig economy, having considered the scenarios with and without platform economies. Up to now, literature also briefly touches on the role of content platforms in gig economy. However, there still remains a gap in picturing the situation of typical content creators on content platforms.

3. From Japanese Doujin Economy to Animation Recreation

The term “Doujin”, derived from Japanese kanji “同人”, originally referred to a social circle, group of people or amateur community with shared interests, tastes, or hobbies. However, in contemporary usage, it usually denotes enthusiasts who engage

in re-creation (or some translation as secondary creation) based on existing material or common interests, such as manga (Japanese comics), novels (usually Japanese light novels), games, and music. A recent study involving 2,593 consumers revealed that 85 individuals have created doujin manga, while another 85 have participated in doujin music creation (Ichikohji & Katsumata, 2016). Additionally, over twenty people have earned income through their doujin manga or music creations. The research by Ichikohji and Katsumata indicates that cultural consumers who venture into creative activities often expand their endeavors to other categories and actively seek to monetize their work. Furthermore, those who create across multiple content categories are more likely to generate income by selling their doujin creations. This symbiotic relationship between consumers and creators within the doujin culture contributes to sustaining the broader content industry. What's more, existing literature mentioned that Tokyo's three-day doujin manga market has been held two times a year since 1975 and currently attracts more than 500,000 visitors each year, 35,000 of whom are sellers of self-published media (Hichibe & Tanaka, 2016). Even though most people remain amateur enthusiasts, they can still distribute their self-made works in various doujin activities or exhibitions, and some can even sell their creations. Gradually, half-professionals and professionals have become involved in doujin activities, selling their own works (Hichibe & Tanaka, 2016). Hichibe and Tanaka found that doujin creation nowadays extends beyond Japanese manga, doujin games are also quite popular. More importantly, these endeavors are driven by strong non-economic motivations, rooted in the joy of creation and the

interactions between creators and users on social media platforms. Similar to doujin manga, the emergence of doujin exhibitions and related platforms has significantly facilitated enthusiasts' creative promotion and sales. As a result, some individuals have achieved both economic and non-economic success, even relying on the income of their creative pursuits for their livelihoods. Indeed, the emergence of the platform economy has lowered barriers to entry, allowing more ordinary enthusiasts to become content providers within the economy. This phenomenon is not new and extends beyond Japan's doujin culture. With the development of the internet and improvements in personal electronic devices, the platform economy has significantly reshaped the music industry, ushering in an era where almost everyone can participate in music creation (Arditi, 2016). Traditionally, music production required large recording studios, often concentrated in major cities and controlled by record companies. Due to the high cost of recording equipment, studios were only feasible in capital-intensive locations. These studios employed workers across all roles in the music production process, achieving efficient music output through economies of scale. However, Arditi's research reveals that digital technology upgrades, particularly the advent of digital audio workstations (DAWs), have enabled people to produce music at relatively low costs — even from home. Additionally, the rise of platforms allows recording studios to be rented out to other musicians, similar to how Uber and Airbnb operate, further reducing music production costs. Simultaneously, musicians can directly connect with collaborators and clients on online platforms, leading to the proliferation of personal studios. Musicians now

have the flexibility to choose their work hours and spaces, fundamentally altering the industry's traditional model. Of course, these changes have also led to the closure of many large recording studios, making music production labor more precarious. Professional producers and engineers now increasingly rely on temporary work and part-time income. The commercialization of China's video platforms is also evident. A recent study conducted over two months, tracking a specific studio, focused on platforms including iQiyi, Youku, and Bilibili and so on, through in-depth observations and interviews, the research finally identified common commercialization models on video platforms and highlighted existing power imbalances (Lin, 2021). It's essential to note that in China, digital video platforms serve as both a cultural soft power incubator and a crucial part of government cultural governance. Consequently, platforms impose certain restrictions on creators, who often sacrifice some creative autonomy during the commercialization process. Lin's research also reveals that creators leverage various monetization methods facilitated by platform commercialization. These include direct advertising integration, such as the Revenue Sharing Program (RSP), where video platforms incorporate ads to generate revenue, and creators participate in ad earnings. Another approach is Embedded Product Placement (EPP), where creators feature sponsors' brands or products within their videos, directly earning advertising revenue. Lastly, there's the franchise chain model, allowing creators to authorize other teams to use their brand or business model locally for video production, training, and income generation. However, navigating this

landscape inevitably involves negotiations with both platforms and regulatory authorities. Among all video platforms in China, Bilibili is a specifically unique one, mainly because it originated as a niche content-sharing platform for ACG (Japanese anime, comics, and games) enthusiasts. During its early days, Bilibili intentionally maintained exclusivity by carefully selecting users, resulting in a distinct community culture that resembles a kind of Utopian existence (Chen, 2021). Unlike other video platforms, Bilibili controls user behavior through a set of community rules, delineating boundaries. In fact, users even need to pass a membership exam to unlock certain interact functions on the platform. This combination of outward exclusivity and internal openness has successfully shaped Bilibili's special community culture. Notably, Bilibili's “danmaku (弹幕)” feature, which optimized from Japan's Niconico video platform, disrupts traditional linear video viewing by providing an immediate but multi-time-layered experience, by seeing people's comments from different time at the same video time point. This encourages users to create and share content while engaging with one another. However, this environment has also led old Bilibili users to resist commercialization and societal norms. Chen's research indicates that in Bilibili's early days, the platform relied on user-generated content and even crowdfunding to secure streaming rights for commercial anime. However, such approach was unsustainable. As Bilibili's influence grew, content monetization and commercialization gradually took hold, leading to increased advertising and business activities — despite Bilibili's initial promise to remain non-commercial. Nevertheless,

Bilibili's strong community environment allowed it to stand out rapidly in China's competitive landscape. In the early days of Bilibili, a prevalent form of creative expression of amateurs involved using Japanese commercial animation or comics for re-creation. This process typically included selecting one's own music and blending original materials through editing or other post-production techniques to create personalized videos. However, up to now, it remains uncertain whether this creative form originated in Japan or elsewhere in the world. Two terms, “AMV” (Anime Music Video) and “MAD” (Music Anime Douga), have both become widely used to describe this type of re-creation across different regions globally. In 2010, a book titled DIY Media: Creating, Sharing, and Learning with New Technologies featured an article detailing the production process of AMVs. From material selection to software choices and editing techniques, this article provided one of the earliest AMV tutorials. According to the records of Knobel, Lankshear and Lewis (2010), there were indeed AMVs edited using European and American animation materials in the early days, but people gradually used the term AMV to refer specifically to the re-creation made by Japanese animation, and researchers at the time did not find the specific reason for the change, only suggested that the development process of western AMV might be influenced by MTV at earlier age. Regarding MAD, its history in Japan dates back even further. A summary from Kurashiki University of Science and the Arts in 2008 revealed that MAD was initially referred to as “MAD movie” (マッドムービー in Japanese). Its emergence

coincided with the rise of YouTube and advancements in computer performance, allowing creators to work with relevant tools. According to records by Nakagawa Hirokazu and Nakagawa Koichi (2008), MAD likely appeared within university clubs in the 1970s. However, production and viewing were extremely challenging at the time, limiting its prevalence to very few accessible places. As technology improved and distribution channels became established, MAD naturally spread. The article then outlined four primary forms of MAD: music-based editing, collecting favorite shots like a slideshow, replacing the opening theme visuals of one anime with another, and editing character stories. These practices were most popular before 2008. Existing literature, including Japanese sources, predominantly those focusing on the subculture of recreations within the ACG field, are particularly about Japanese comics and gaming industries. However, when it comes to the specific economic phenomena related to video production and its creators, there is relatively limited literature available. Research specifically addressing the economic status of animation recreators within China remains an unexplored area.

III. Methodology and Data Collection

1. Methodology

Research on gig worker populations, which encompass diverse backgrounds within a specific type, often combines quantitative and qualitative research methods to obtain relatively accurate results. Almost all the surveys on gig workers in the literature review above adopted interviews, or interviews and questionnaires, or interviews combined with field surveys to obtain the required data. This will also be the main means of conducting research in this dissertation. However, by referencing all mentioned problems and limitations encountered in the field research process of existing literature, this study will be reasonably modified the data collection process in order to obtain more accurate data. For instance, a study focused on ride-hailing drivers in a university town in the U.S. Midwest considered various factors such as work hours, experience, age, education, and ethnicity. The researchers employed daily observations, interviews, and archival data, followed by rudimentary frequency statistics. Ultimately, they explored this group's attitudes toward social interaction and digital community-based unionization (Maffie, 2020). This model of field research combined with interviews to obtain information became one of the reference cases for this study, but this study is cautious and reserved about its sampling method of directly distributing questionnaires. It's worth noting that the Pew Research Center (2023) critically summarized its research on the gig economy in 2016 and 2021, providing valuable insights for

studying the phenomenon in the United States. For instance, they recommend using forced-choice questions that require respondents to answer directly, rather than providing checklists. This approach yields more accurate and valuable information. Another lesson learned is that changes in question phrasing can make it challenging to compare results. Therefore, researchers should anticipate different patterns of responses and focus on variations rather than specific data points during the questioning phase. After comprehensive consideration, due to the greater uncertainty associated with gig work in artistic creation and the difficulty of summarizing meaningful answers to questions using multiple-choice formats, this study will not use survey questionnaires for data collection. Instead, this dissertation will adopt a combination of field research and anonymous semi-structured interviews. Simultaneously, the sample will use a hybrid approach of stratified sampling and snowball sampling to comprehensively explore the gig economy patterns among MAD creators. Regarding field research, this study deeply engaged in the MAD production process, examined the community communication and business models of MAD creators on the Bilibili platform, communicated with a large number of both new and experienced MAD creators, observed some emergent events within the MAD community, and attempted to leverage MAD production techniques to secure some research-worthy data. To further clarify, the goal of this research is to provide an overall depiction of the gig economy phenomenon surrounding MAD creators, understand the business models related to MAD on the Bilibili platform, and present a fresh perspective on

researching "flexible employment" in China. After a period of field engagement, researcher concluded that the sampling process for semi-structured interviews must follow a combined approach of stratified sampling and snowball sampling while ensuring anonymity during the interview process. An important reason for this approach is that MAD creators who specialize in video production often have richer gig work experiences. Random sampling alone would not fully capture their economic patterns. By sampling from different types of MAD creators and subsequently introducing snowball sampling to identify the next interviewee, potential biases can be mitigated. Additionally, due to privacy concerns and the online nature of the interviews with network users, anonymity becomes a necessary protective and encouraging measure to elicit more information from interviewees. During the field research process, the researchers actively participated in MAD creation, explored communication patterns and business channels within the MAD community on the platform, and engaged in extensive conversations with both new and experienced MAD creators. Throughout this process, they also observed MAD-related online events and their impact on the Bilibili platform. Attempts were made to secure business collaborations using MAD production techniques, and valuable data with research potential was documented based on the researchers' firsthand experiences.

2. Data Collections

This study ultimately completed 69 semi-structured interviews, of which two-thirds (46 interviews) were conducted via voice calls and later transcribed by the

researchers. In terms of duration, voice interviews typically ranged from half an hour to an hour to comprehensively gather the necessary information. Due to certain unavoidable circumstances (unmatchable schedule) or strong requests from interviewees, one-third of the results were ultimately obtained through open-ended questionnaires. In all 23 interviews conducted using an open-ended format, interviewees explicitly confirmed their willingness to answer related questions as thoroughly as possible before the interviews. A few interviewees also expressed their willingness to accept further inquiries from the researchers if needed. In the process of stratified sampling, this study found that Bilibili's MAD authors are broadly divided into two categories. One category is MADers who are completely oriented to the general public, and their goal is to produce MADs with high number of views. Another category is creators who organize MAD contests and participate in MAD competitions, and their goal is better artistic expression and better video production techniques. The latter is the focus of this research because the phenomenon of the gig economy about them is more complex and has more potential for development.

IV. Results

1. General Description of MADers

In this study, semi-structured interviews was conducted using a stratified sampling approach. As a result, alongside interviewing some novice MAD creators, this research also interviewed former MAD creators who are no longer active. Considering that high-quality MADs on Bilibili tend to require a two- to three-month work period, and that contests for high-level MADs on Bilibili are only one or two in a year, this study tends to view the submission of at least one manuscript on Bilibili in a year as an indication that a MADer is still active. Excluding four individuals who declined to disclose personal information, the sampled population generally falls within the age range of 20 to 25 years, with the youngest being a high school student and the oldest exceeding thirty years old. During the field research, it is found that the youngest MADer in China are in their early years in junior high school, while the oldest are generally approaching thirty. Analyzing participants from Bilibili's MAD competitions, it is observed that only two prominent MAD active creators is around 32 to 35 years old now. Overall, the age distribution in our interview sample aligns with broader observations. Within 69 interviewees, there are 41 of them still making their next MAD, and 14 of them confirmed that they will no longer making any MAD. The rest, despite claiming that they are no longer making MAD, still maintain willingness to start the next MAD project if there is any chances. MADers generally gave up on MAD creation because of two reasons: shifts in interests, and lack of time. One of the most

frequently recorded reasons in interviews is that MADers have started working. Compared to the relaxed schedule in college, there are too many things to do every day in the working life, which squeezes out the free time that they originally had. And in the remaining free time, MADers are more inclined to use it for other leisure activities, such as video games, etc., rather than continuing to expend brainpower on creation. Over a long period of time, some people even watch less animation and comics, so their interests have changed, so they give up MAD creation. However, even if most interviewed MADers directly answered the question of whether they are still making new MADs, measuring the real situation of whether a MADer is active remained complicated. The main reason is that these answers largely represents people's current understanding of their own status. Some respondents actually have almost stopped making new MADs, but they still think they are conceiving new ideas, so they choose to consider themselves as active MADers. I am still doing it, but I can not maintain a stable update. When I have inspiration, I do it quickly, but most of the time I have no inspiration. As a result, I make MAD in a very large time period, from a few months or half a year. (Interviewee No.49, translated) I basically stopped doing it because I was too busy at work and I am a “Lan Gou (lazy dog)”. (Interviewee No.30, translated) Based on the interview findings, MAD creation on the Bilibili platform experienced a prosperous peak between 2012 and 2014. It maintained or perhaps gradually declined from 2015 to 2017, then followed by Bilibili's significant reforms

toward commercialization in 2018. During this period, certain MAD creators garnered substantial attention on the platform. After that, the impact of the COVID-19 pandemic also affected the MAD community. This study differentiated MAD creators of when their first MAD submissions to Bilibili based on the timing mentioned above. Results revealed that a significant proportion of currently active MAD creators began producing their first MADs between 2018 and 2021. This trend correlates with Bilibili's commercial transformation from ACG platform to a comprehensive video platform, and the influx of ACG enthusiasts who joined the platform to watch Japanese anime. MAD videos remained mainstream on Bilibili during this time, hence the public exposure of MAD increased. Furthermore, the pandemic played a role, as some individuals sought to engage in hobbies during isolation, leading them to explore creative expressions like making their own MAD. Combining their love for ACG culture, with exposure to MADs recommended by the platform algorithm to their screens, these factors contributed to the emergence of the current core group of MADers. This study also found that the average active duration for Chinese MADers is approximately three to five years. The specific timing of when a MADer begins and ends his or her creation is entirely random, cases of starting as high school students, university students and salary man all commonly exist. Eventually, MADers are likely to stop making new MAD due to busier lives and limited spare times. In the interviews conducted for this study, none of the 69 participants explicitly indicated that their personal lives influenced when they create MAD, and most agreed that

recreation is purely a hobby that depends on whether they have free time. The decline in MAD output indeed typically occurs after three to five years, for several reasons. First, enthusiasm wanes over time — some creators become dissatisfied with their skill levels due to improved esthetic abilities. Second, as creators enhance their technical skills, the production cycle for each MAD lengthens, often exceeding their available mental energy, and some MADers are unwilling to compromise on the quality of their work. Third, some MADers have already realized all the ideas they wanted to explore, leading to a loss of motivation to continue producing. In summary, this research identifies a clear trend: Chinese MADers experience a gradual decline in enthusiasm over three to five years, and such decline accelerates over time. Consequently, the number of MADers who sustain long-term creation beyond six years is relatively small, and often contingent on favorable life condition. Another interesting aspect to prove that MADers are primarily driven by their passion and interest is that, within the MAD community, individuals with formal art education background or with prior art-related work experience before venturing into MAD creation are quite rare. Among the 69 individuals interviewed for this study, excluding 8 interviewees who declined to disclose personal information, 50 of which had no artistic background whatsoever. However, surprisingly, when comparing this data with their subsequent responses, a significant number of participants acquired art skills by their MAD recreation. Then, they received all sorts of freelance work opportunities related to video productions. Among these 69 individuals, those who

eventually pursued professional careers or entered relevant academic studies, who experienced a clear transition from amateur to professional status, are actually very high, displaying a significant role of animation recreation in training people's ability. Of the 69 people interviewed, 27 end up working in video production careers today, and of the rest, 32 have also engaged in video production gig work for at least once, including five students who went on to study related majors in universities because of MAD creation. This dissertation will delve into related topics further in subsequent chapters. Whether or not art-related education helps with MAD creation is a matter of opinion. Among the 69 interviewed MADers, their majors ranged from mechanization, civil engineering, chemistry and chemical engineering and so on. Only a few of them majored in digital media in college, only one majored in screenwriting and directing in college, one was an art student, and one studied architecture. Some interviewees claimed: ... I studied architecture. I chose architecture in order to make videos. Then I took professional courses in my freshman year, which included sketching and coloring... It helped a little, but not much. At least sketching and sense of perspective helped me with Jing-zhi MAD... I didn't learn coloring very well... I learned basically nothing during my undergraduate studies. (Interviewee No.43, translated) ... I once studied art in a painting studio. To be honest, it had little influence on me. Studying painting only gave me some esthetic foundation. (Interviewee No.61, translated)

It is common that MADers with art-related backgrounds holding a very conservative attitude about the role their studies play in helping them.

2. MAD and AMV and Other Various Recreation Forms

Compared to the MAD forms documented by Nakagawa Koichi (2008), today's recreation videos based on Japanese ACG culture have become quite diverse and prosperous due to hardware performance advancements and software technology progress. Therefore, in the following sections, this research will delve into the current methods of MAD creation, dissecting the various skills that creators acquire through this process. Understanding the specific technical expertise involved in contemporary animation recreation is essential for explaining why this group of MADers can secure sufficient freelance income and even transition into professional industry roles. 2.1. Anime Music Videos (AMV) 2.1.1. Anime Story Music Videos (ASMV) Originated from Japan, “editing character's story”, which is described as the most complicated form of recreation by Nakagawa Koichi in 2008, is the only form that hasn't completely outdated. This is a form of recreation that is most easy to make in current days, which has become the most common choice for new learners to start with. For most new MADers, they usually start their first thought when they are deeply moved by a commercial animation. In order to pass such feelings to others, and encourage others to also check on this animation, it is easy for new starters to realize, that the most easy way to do so is to personally edit a trailer, or just a short

film that summarize the story of the entire animation series. Such editing usually called the “ASMV (Anime Story Music Video)”, using one or more selected music, remixing characters' lines and scenes from the original works. In most of the time, ASMV will only use one animation, but sometimes MADers might also mixing other animations in order to fulfill or enrich their stories. I remember that ASMV started to become more common around 2018. I think ASMV was also learned from some country else, but with some local changes. The earliest ASMV was just like taking the good parts of anime and putting them together. So I didn't like the ASMV at that time. Now, they have gradually introduced the editing techniques similar to movies... which is better than the previous ones (Interviewee No.4, translated). ASMV is the least difficult in terms of software using technique, because it only involves pure editing and does not involve post-production. However, it is extremely difficult to produce a good ASMV, because under the guidance of some creators, today's ASMV emphasizes the integrity of the story and the ups and downs of emotions. In this process, the creator must have the ability to control the pace of the video, know how to choose the appropriate lines and shots, know how to match songs and sound effects, and know how to render the atmosphere of the video. The creator even needs to partially understand how the camera is dispatched on the real shooting scene, so that he can imagine what kind of lens is suitable to be used, and then make the lens selection, in order to achieve the level of “people can't tell that this is a recreation clip”.

2.1.2. Zong-man (综漫) AMV, Duo-su-cai (多素材) AMV The Zong-man, which could be understand as “Comprehensive Remixed Animations”, unlike the ASMV, using multiple animation materials to finished the AMV. That is why sometimes MADers referred it as Duo-su-cai in Mandarin Chinese. There is no one anime series that dominates Zong-man AMV, or in other words, people cannot clearly feel that there is any main character in a Zong-man AMV, which is the most significant feature of Zong-man. MADer might plot a vague storyline within their Zong-man AMV, but the main purpose of Zong-man is always using multiple animation materials to create an atmosphere of certain emotions, or focus on giving audience a fancy visual experience by MADers' editing. There are two basic types of Zong-man AMV. One is to directly select well-known shots and lines with similar emotions from well-known anime, and then simply remix them with the music of your choice. Another way is to select music with a strong sense of beats, and then select all the shots with fighting actions or strong dynamic shots from well-known anime, and edit them to the beats of the music accordingly. These two production methods require almost no thinking in edition. In order to edit advanced Zong-man AMV, creators also need to understand the beats of the music and select appropriate shots from dozens or even hundreds of animation series. These clips must be strategically placed to create a visually pleasing experience for the audience. Simultaneously, MADers require color-adjustment skills to reduce a sense of “not-matching” between clips sourced from different anime. Only

harmonization ensures a seamless viewing experience for the audience. Zong-man AMV shares similarities with pop music videos nowadays. Both require editors to have a profound understanding of camera movement, character motion, and pacing to infuse tension and dynamism into the visuals. What's more, Zong-man even required editor have basic ability in voice mixing because the use of sound effects is inevitable. However, compared to pop music MV, which often contained abundant footage of singers performing or dancing in the same place, Zong-man AMV demands editors to manage a larger amount of source materials. Integrating such extensive content into a cohesive final product is absolutely challenging. 2.1.3. Other Types of AMV Traditionally, when MADers mentioned AMV excluding ASMV or Zong-man, they are talking about an MV style video, using a single anime series as editing material, setting to a specific song. Sometimes, single-material AMV may incorporate a storyline, but typically without any dialogue. As some MADers' understandings, the distinction between ASMV and single-material AMV lies solely in the presence of character's dialogue. Some MADers believe that AMV focus more on editing to a specific song, while ASMV prefer selecting songs to create their stories. As a result, using more than one song in an ASMV is not rare. Within the AMV category in China, there is an important branch that has long been popular abroad, especially as mainstream in France, Spain, and Russia. This

category is directly referred to AMV on YouTube, but in China, it is called “composited AMV (合成 AMV)”. The key difference between composited AMV and ASMV or Zong-man lies in what the MADers did to their AMV. In composited AMV, creators are not just editors, but also post-production and visual effects artists. MADers who create composited AMV go beyond simple editing, they need to meticulously extract characters frame by frame from the original anime footage. Such painstaking process allows them to achieve effects like background replacement, animated text moving behind the characters, motion graphics animation, various visual effects, and even integrate characters from different anime into a single scene. Some even realized Pop Art in their MADs. These advanced techniques often require specialized software, creating a “composition” in the project file, usually using Adobe After Effects. Hence, in China, this category of AMV is referred to as composited AMV. Many Chinese MADers who specialize in AMV no matter ASMV or Zong-man will pursue the creation of composited AMV, in order to create fancier MADs or just improve their post-production skills. Furthermore, composited AMV can be further categorized in to five directions. Such division is also originated from the West, in order to classify and score AMV works in MAD competitions organized by western MADers. The five directions are: Actions, Dance and/or Fun, Drama and/or Romance, Psyche, Horror. A portion of Chinese MADers have borrowed this categorization when running Chinese MAD

tournaments. It is easy for audience to get what this AMV is about from these tags. However, some exceptional works defy easy categorization, seamlessly blending multiple aspects. For instance, a single AMV might excel in both Action and Psyche or simultaneously evoke Drama and Horror. The fun part is, difficulties in classification often occur with good works. If talented creators accidentally submit their works to the same category in the competition, the competition for ranking in that category will be extremely fierce. In summary, creating all types of Anime Music Videos requires excellent esthetic sensibility in videos. Creators must understand the rhythm and beats of the music and how to synchronize it with the animation clips you selected. In terms of difficulty in using software, ASMV are generally considered the simplest, followed by Zong-man AMV, and finally, composited AMV. Traditional single-material pure-editing AMV are now rare, most AMVs incorporate at least some extent of composition. Making ASMV leans toward skills akin to screenwriting and directing, while composited AMV demands an understanding of storyboard, animation, graphic design, and visual effects. Additionally, since sound effects often come into play, experienced AMV editors often possess some audio mixing skills. 2.2. Jing-zhi MAD, the Manga Music Video (MMV) Jing-zhi MAD, originated in Japan, has gradually transitioned from a relatively niche MAD production method in China to a dominant mainstream approach over the past decade. The term “Jing-zhi” translates to “静止(static)” in both Japanese Kanji and Chinese Characters. More than half of the widely recognized

high-quality MADs created these years are Jing-zhi MAD. Compared to AMV mentioned above, which utilize existing commercial animations, some MADers encountered a scenario, which is they found a Japanese comics (manga hereafter) they really enjoyed and wanted to do re-creation, but it didn't have an animation version. Because in the Japanese ACG industry, most of the anime is not original, but adapted from manga (many manga are adapted from Japanese light novels). Usually only the more commercially successful manga are adapted into anime for further promotion. This model of commercialization for secondary use of intellectual property is quite common in Japan (Hernández, 2018). Rather than waiting for those manga to get commercial recognition and subsequent animationalized production, some Japanese MADers began to explore the possibility of turning manga into videos over a decade ago. Thus, the earliest Jing-zhi MADs were born. Jing-zhi MADs directly use characters from manga, animating them by moving them across the video time frame. These videos are set to music and often include explanatory subtitles with text animation. Because the characters remain static compared to AMV, this approach is aptly named “static” MAD, signifying a MAD in static-style or with static pictures. Jing-zhi MAD represents the highest technical level of MAD creation on Bilibili today, and also worldwide. This form originated in Japan and has a variation known as Promotion Video (PV), which also widely practice in South Korea. After mutual influence between Japan and South Korea, Jing-zhi and its various new

practices were introduced to China. For a long time, and even until today, the production methods and esthetic direction of Chinese Jing-zhi MAD have followed those of Japanese Jing-zhi MAD creators: My first contact of MAD was from overseas... Some of the best work they did was deeply shocking to me when I first saw it. A Japanese MAD creator named totori... His style is really beautiful. I prefer that kind of refreshing style, hence I learn to do his. (Interviewee No.16, translated) An obvious phenomenon related to Jing-zhi is that, similar to the history of world painting, all significant revolutionary progress made or new directions shifted in the history of Jing-zhi MAD over the past decade, contributed by only few creators, or even one creator who discovered new techniques. For example, during this research, almost all interviewed Jing-zhi MADers emphasized the impact of Nanatsuki from 2020 to 2022, who is a talented Japanese MADer. Nanatsuki's work completely revolutionized the production style of Jing-zhi MAD worldwide. As a result, all new MAD creators today are studying his style, or use some of the presets from his projects, since it became common practice ever since. 2.2.1. Techniques Required of Making Jing-zhi MAD To create a good Jing-zhi MAD, the first step is to extract characters from manga. During this process, MADers typically use Adobe Photoshop software for image cutouts. Although this type of MAD is still referred to as “static”, modern MADers had tried their best to avoid overly static scenes. In current Jing-zhi MADs, when cutting out characters, MADers extract not only facial features but also hair,

clothing, and joints. All the compositional elements extracted into separate layers can be animated. As a result, Jing-zhi MAD characters today have flowing hair, blinking eyes, and even clothing that moves with arm gestures — all achieved based on common manga materials. However, this process is not as straightforward as it may seem. When MADers cut out characters' hair, the original character's head will inevitably end up with a transparent area on the forehead, because that is what would definitely happened when a part of an 2D illustration was removed. This means that most Jing-zhi MADers have the ability to observe the character's structure and illustrate simple drawings to fill in the missing parts. Only then MADers would be able to sway the hairs on a character with complete forehead, in order to animating comics. They are not professional in drawing, but they have to do so. This technique is known as “puppetry”. It's similar to creating a jointed wooden puppet, or traditional Chinese “Shadow Puppets Plays (皮影戏)”, but MADers achieve the same effect electronically using manga on their computers. However, a scene cannot consist of just characters in motion, MADers need to design a background for those characters. For a long time, Jing-zhi MADs would choose relatively simple characters with straight perspective angle, and then composite them with real-world background photos. Especially since the work of Japanese MADer totori, Jing-zhi MADs have increasingly pursued visually appealing lighting and shadow effects. MADers use various lighting and blur effects, combined with color adjusting, to make viewers feel as though the character naturally belongs in

that photo, rather than standing in front of a big background board with no interactions. But MADers are not satisfied; they aim for more exaggerated camera movements. As a result, many MADers today, especially Jing-zhi MADers (some composited AMV also use 3D model for their backgrounds), are learning 3D modeling. Compared to 2D photos, self-designed 3D models allow for more flexibility in matching backgrounds for Jing-zhi MADs. If we compare these techniques to those used in the Japanese animation industry, Jing-zhi MADers who achieve this level of work essentially perform the roles of two professions. One of these roles is conventional CG (Computer Generate) artist specialized in 3D modeling, which involves creating usable model assets. The other role is “animation photography/cinematography ([日]撮影/[中]动画摄影)”. In this position, the task is to simulate realistic camera effects using computer software or create specific emotional atmospheres. In essence, this process of combining character imagery and background through lighting and shadow synthesis in MAD production is almost the same to what happens in animation photography, except Jing-zhi MAD usually use more dramatic lights, shadows and coloring to show MADers' artistic expresses. Another characteristic of Jing-zhi MAD is that it compels MAD creators MADers to acquire an additional skill, which video editors rarely need. Since manga materials lack spoken dialogue, similar to the manga itself, advancing the storyline in Jing-zhi MAD requires textual information. However, directly placing text within the

visuals often disrupts the video scenes established earlier. Therefore, Jing-zhi MADers commonly need to master text animation and graphic design skills. If you closely examine high-quality Jing-zhi MADs today, you'll find that their cover posters exhibit excellent typography or graphic designs. Of course, similar to ASMV, the ultimate goal of Jing-zhi MAD is to convey a story. As a result, Jing-zhi MAD creators also need to focus on plot design, ensuring that the story's ups and downs align with the music's. Additionally, because Jing-zhi MAD involves selecting characters and creating backgrounds, it offers greater creative freedom compared to AMV recreation. Creators can choose the storyboard they desire. From this perspective, Jing-zhi MAD ultimately demands a certain level of directorial skill from creators. At the very least, compared to AMVs, Jing-zhi requires MADers to design their own shots. Interestingly, this has led to an unexpected outcome. Many creators who initially made ASMVs found themselves constrained by the original works as their skills improved. Because pure editing can only use anime footage, which are finished products that have already been done by the animation company, there are very limited adjustments that the recreator can make. Some of the stories they wanted to realize could not be realized because of the lack of relevant footage. Consequently, they shifted directly to producing Jing-zhi MADs. Jing-zhi MAD allows creators to create the storyboard they want, because all they need to do is find a suitable picture of the character, and then the background is completely up to them. As a result, some commercial IP even have animated versions, yet these creators prefer utilizing manga

for their recreations. 2.2.2. Promotion Video (PV) and Shou-shu With the understanding of skills that Jing-zhi MADer processed, this study can further describe two types of video creations similar to Jing-zhi MAD, the PV and Shou-shu. Both forms of video generally have greater commercial value. Most of the production orders that MADers are able to get come from these two video formats as well. This will be elaborated on later in the article. The first type, known as “PV”, stands for “Promotion Video”. However, unlike the conventional English understanding of a promotional video, here, a PV should not be interpreted as an advertisement for a specific organization or brand. Instead, in this context, a PV are describing something more like Music Video. The main difference is that these PVs are not created using real-world footage. Sometimes, an independent musician, possibly on platforms like Bilibili, YouTube, or Niconico (Bilibili started as an imitation of Japanese website Niconico, which is also a site focused on ACG culture), wants to release his or her new covering or original song, hoping to have his or her own music video. These music videos often rely on artists to illustrate characters and combine them with post-production animation effects. In some cases, PV can also serve as promotional videos for manga or ACG style games, especially when referring to PV made by ACG gaming companies. The PV made by ACG gaming companies are animated videos for introducing their games, which is similar to a short original animation trailer. Considering the description above, it's evident to see that the video production

skills required for Jing-zhi MAD is transferable to PV production. In the field research, researcher found that there is a considerable intersection between the groups of PV creators and MAD creators. Most PV creators with basic skills do not venture into Jing-zhi MAD production, as it is relatively challenging for them. However, conversely, a significant proportion of talented PV creators are also MADers. There's another creative video form that has rapidly become popular in Bilibili's animation sections in recent years. This form is known as “Shou-shu (手书)”, which literally means “handwritten” in Mandarin Chinese. It falls between dynamic comics and animation, sometimes serving as fan-made recreation content related to ACG culture, while other times featuring original characters and stories. Unlike traditional animation, the Shou-shu doesn't meticulously animate every frame of character movement, instead, each shot focuses on one or two crucial actions. Essentially, Shou-shu can be thought of as a form of Jing-zhi MAD, but instead of extracting characters from manga, illustrators directly draw the characters used for post-production. Therefore, when it comes to Shou-shu, the best and most readily available post-production gig workers are often MADers. 2.3. Other types of MAD and AMV In addition to AMV and Jing-zhi MAD (Manga Music Video, MMV), there is also GMV, which is Game Music Video. GMV is the only form of creation that may not be related to ACG culture, because it depends on where the original materials used for recreation come from. If it is a Galgame of Japanese ACG culture, MADer often uses the Jing-zhi method to produce it, and the final MAD will look like a Jing-zhi

MAD. If it is a remix of some ACG games' promotion video, it belongs to GMV, which is very similar to AMV. But if it is a screen recording clip of some realistic First-person Shooter (FPS) game, it does not belong to the category of ACG, but it still belongs to GMV. If an MADer combines AMV style production with Jing-zhi style production in his or her recreation, this type of MAD is referred to as “Hun-he (混合)”, which means “mixed” in Chinese. Additionally, if an MAD incorporates both manga material and animation material, it would also fall under the category of “Hun-he”. Sometimes MADers use “Hun-he” to create a video that mixed multiple animations into one story, changing what happened in the original plot, such MADs can be called “misunderstanding MAD (误解 MAD)”. In addition to the MAD styles described earlier, there are some styles that are challenging to categorize. Some unlisted categories include subtle variations related to specific atmospheres delivered to the audience, which has no relevant for this particular study. There is also one style known as “IG style”, which refers to content created by individuals on the Instagram platform. In IG style, creators upload videos of motion graphics (such as shape-shifting square or circles), occasionally incorporating ACG material. These videos rarely longer than 1 minute with no narratives. Trying to define what IG style is, is impossible, even though a large number of creators active on Instagram and Tik-tok make IG Style MAD. MADers on Bilibili dislike IG styles, even dislike those IG style creators: First of all, I hereby claim that I like to cut off all connections with the

MADers who make IG style. It's disgusting when they get into all sorts of cliques, and they go into chat groups to attack other people's work and belittle them. They are always like, “I'm doing awesome effects, I'm greater than you” (Interviewee No.5, translated). Since IG style does not profit in the same way as MAD, this style will be discussed separately in the following section discussing gig economy. When it comes to special creation styles, there's a unique type that deserves mention, even though it doesn't really represent an artistic style. This type of MAD is referred to as “dssq”, which stands for the Chinese phrase “Da Shi Suo Qu (大势所 趋)”, meaning “the prevailing trend”. Specifically, on the Bilibili platform, “dssq” refers to MAD videos created with the sole purpose of achieving high view counts. These “dssq” MADs often fall into the category of Zong-man AMVs, where creators combine scenes from currently popular or classic anime in the simplest way possible. Their primary goal is not making a good video, but to attract audiences to click into their videos by playing tricks on covers and titles. Consequently, this approach has had a serious detrimental impact on the MAD creative community. From the perspective of the gig economy researched in this study, the economy model associated with “dssq” differs significantly from that of highly skilled MAD creators, and this distinction will be discussed further in subsequent sections.

3. Professional Video Production Skills and Learning Path

From a software usage perspective, nearly all MADers are proficient in professional-grade software. Based on data collected from the interviews, out of 69

respondents (excluding one who refused to answer), only two individuals did not explicitly mention using Adobe software. The remaining 66 MADers have mastered either Adobe Premiere or Adobe After Effects. Most of the MADers are skilled in both, because Premiere has obvious advantages in editing, while After Effects are almost the only post production software used in professional video-formed media creation, including movies, TV series and animations. Those who started making AMV before 2022 also familiar with the Adobe Audition. Because before the widespread adoption of artificial intelligence, MADers relied on Adobe Audition to manually extract clear dialogues from anime's background musics. Similarly, for those creating Jing-zhi MADs, Adobe Photoshop is essential, as it is practically the only method for extracting images from anime scenes. These are demand-driven learning outcomes. In fact, even creators who don't use Adobe software for MAD production, still need to rely on other professional tools. This is because only specialized software allows adjustments on every single parameters to meet individual artistic expression. When it comes to post-production, aside from Adobe After Effects, there are currently no direct substitutes throughout the globe. However, for video editing, some MADers opt for DaVinci Resolve. These creators typically focus solely on editing and don't require extensive post-production features, allowing them to overlook the convenient integration between Adobe Premiere and After Effects. Additionally, some users prefer Final Cut Pro, especially if they work with Apple products, as Final Cut Pro is specifically designed for video production by Apple. It's worth noting that using

DaVinci Resolve and Final Cut Pro for color grading and editing is also common practices in current media corporations, simply rarer in ACG related industry. MADers' software learning experiences generally follow a progression from simple to complex. The key difference lies in the starting point: some MADers begin with mobile devices or low-performance computers. Consequently, they initially encounter user-friendly editing software, such as “Hui Sheng Hui Ying (会声会影)” or “iJianJi (爱剪辑)” Still, over half of MADers directly start their journey with Adobe software, with many considering Adobe Premiere as their first serious editing tool. In earlier years, around 2014, some creators used Sony's Vegas, which was considered a highly professional editing software at the time. Building on this foundation, MADers — especially those making Jing-zhi MADs — gradually began exploring 3D software. Within the 69 interviewees, 30 of them clearly confirmed that they have learned at least one 3D modelling software. An interesting detail is that 3D modeling gained traction around 2017. Between 2018 and 2020, the majority of creators learned CINEMA 4D. However, after 2020, even those who previously used CINEMA 4D shifted their focus to a software called Blender. Up to now, within the 30 interviewees who learned at least one 3D modelling software, 28 of them learned Blender, and 15 of them only learned Blender. There are likely two reasons for this shift: first, Blender is better suited for achieving ACG style textures compared to CINEMA 4D; second, Blender's open-source nature has led to the development of user-friendly plugins. Some MADers, aiming for advanced effects, also learned Houdini and other 3D modeling software. A very small minority delved

into more intricate “puppetry” animation, studying tools like Live2D Cubism and Spine. How did MADers acquire such professional video production skills?This research found three primary pathways that contributes most: i) Study by oneself, ii) Check out online tutorials, and iii) Ask MADer friends. According to the data of 69 interviews, excluding the six people who were unwilling to answer, 26 people completely rely on online tutorials, 12 people completely rely on self-study, and 1 person completely relies on asking others to learn. Among the remaining people, 15 people rely on tutorials combined with self-study, 5 people rely on tutorials combined with asking others, 2 people rely on self-study and asking others without reading tutorials, and the last 2 people use all three methods. Some noteworthy details emerged from interviews with MADers, thanks to those MADers from Chinese Taiwan and those who frequently check MAD-related content on YouTube and Twitter. They generally find that Bilibili's search mechanism is accurate and language-friendly, making it easy to find Adobe After Effects tutorials. However, for Blender tutorials, YouTube offers more comprehensive resources. Additionally, Bilibili tutorials often cover comprehensive basics of software operations, so many MADers start their video production journey there, gradually understanding the logic before seeking specific effects. Notably, learning 3D software differs significantly from learning After Effects. Each 3D software has its unique logic, and due to relatively fewer users, both Bilibili and YouTube lack comprehensive foundational tutorials on these software. Consequently, some MADers turn to online

courses to learn 3D software effectively: Initially, there were no teaching materials on Bilibili, so I learned the basics of CNIMA 4D and After Effects from “51 self-study network”. Then I had the opportunity to access some sorts of integrated material websites. I find foreign paid tutorials on these websites very useful, including courses on software system and various theoretical lectures. About 80% of the video tutorials I've learned come from this source. It's more challenging to find this kind of tutorials on Bilibili. And tutorials on Bilibili are usually shorter. (Interviewee No.7, translated) However, it is worth noting that the vast majority of MADers use the Adobe series software in their cracked version. Due to the high cost of licensed professional software, it is common for creators to use cracked versions for video production. Whether pirated software is a problem in the long-run is something that there is no way to conclude at this time.

4. MAD Related Gig Works, Career Perspective and the Supporting Economy

The complexity of gig economy phenomenon related to MAD is the most significant finding from this field research and interviews. This issue extends beyond how MADers secure freelance projects or whether they can pursue a professional career. MAD, as a unique art form, coupled with its rich expressive capabilities, has surprisingly created commercial value from various unexpected perspectives, even if some aspects still appear nascent. Notably, the gig economy in the MAD context differs significantly from other gig work forms, such as ride-sharing or food delivery. Despite the industry's inherent risks, this study believes that the vast potential lies in

the imaginative space behind diversification. Based on 69 interviews and the summary of the field study, the researcher found that there are five perspectives that need to be considered in the way a MADer makes money: A) Profits purely from MAD: i) Incentive money for video creation of Bilibili platform ii) MAD competition prize money iii) Video production commissions for specific MAD B) Profits from video production commissions C) Profits from other MADers D) Profits from starting a business E) Earnings from entering video production occupations Since most respondents indicated that there are varying degrees of non-disclosure agreements in place regarding the topic of relevant business collaborations, what follows will anonymize the names of companies and people that appear and describe the overall level of revenues only to the minimum extent permitted. First, the incentive money. As recalled by the MADers interviewed, the Bilibili platform launched the “Video Creation Incentive” function around 2017 to encourage video uploaders to submit as many high-quality videos as possible. A video uploader had to upload enough videos, and those videos had to perform well enough (in terms of data such as viewer watch time, likes and comments, etc.) to accumulate a specific score. This score is calculated by an undisclosed algorithm, and only when this score

reaches a fixed value can the “Video Creative Incentive” function be activated. From the moment that incentive is activated, the amount of video views will be converted into money according to a percentage, and transferred to the creator's Bilibili account, which can be withdrawn to a bank card. Around 2018, the most craziest number gained by a MAD... I think it was over 50 million views on Bilibili, a MAD made by my friend. At that time, Bilibili gave him more than 100,000 yuan as creative incentives. This is actually a very exaggerated income for re-creators. (Interviewee No. 16, translated) When Bilibili's creation incentive money was at its peak, a video with roughly 10,000 views could have a reward of 30 yuan. However, in 2022, this reward to creators was drastically downgraded without a clear official explanation of why from Bilibili. MADers interviewed generally said that they have stopped paying attention to the creation incentive, especially after 2022. Since the vast majority of the 69 interviewees have an average video view count, of 2,000 to 100,000, or even less, for a single MAD, that aspect of the income has been lowered to less than 100 yuan per MAD, or even less than 10 yuan. Considering that most people's MAD production cycles are around two to three months, this income looks even less. It is also affecting those MADers who have massive subscribers: I have communicated with them when doing business with big Up-loaders, those having hundreds thousands followers. Although this type of MADer has so many subscribers, the income that MAD can bring them may be less than 30000 per year now. It is impossible to consider MAD as a major source of income. Secondly, the

commercialization of MAD has been done very, very poorly. (Interviewee No.27, translated) In fact, in most MADers' thoughts, the advent of creative incentive payments has had a serious negative impact on the development of MAD. This is because the system of creative incentive money, when practiced, encourages the development of inferior works instead. Because the total population of viewers watching MAD is basically the same, they are the general public who are already interested in ACG culture themselves. The algorithm of the Bilibili platform mainly recommends MAD to the homepage of these people as well. For a good MAD, it generally contains emotional ups and downs, and the story needs to be told from the beginning. Therefore, good MADs generally require the viewer to stay long enough to feel the video push up step by step towards the final climax. One social media, most viewers will not have the patience to wait, especially since most MAD are likely produced by materials that is not familiar to the audience. But if a MAD just piles the most recognizable characters from the most famous anime at the beginning of the video, it's much more likely to draw people in and keep people stay. Ultimately, inferior and homogenized MADs are more likely to get higher views and make higher monetary gains by incentive money, encouraging more people to make such MADs. This trend, as previously mentioned, is called the “dssq”: Since Bilibili has been offering creative incentives, the so-called “dssq” trend emerged, it had a negative impact. More people started looking for ways to earn the first bucket of gold at Bilibili. I understand it, I can't say it's a mistake... Then things

took a turn. After this period, people began focusing on learning techniques again, improving their skills and thinking about their future career prospects. (Interviewee No.2, translated) In addition to the Bilibili platform's incentives, learned from foreign countries since Bilibili's founding, Chinese MADers have also organized MAD competitions on the platform. These contests often set a time period in the near future, and contestants are required to submit a video made during that time. The organizer of the competition will also try to raise some money, or the organizing MADers will pay for the prize money for the final winner (the latter is the most common phenomenon). Most of the people who will sponsor the contest are also generous MADers; it's not a lot of money, but since it's a MAD contest and it's a way to show off their work for other creators to see, the majority of MADers are happy to participate in the contest and try to earn the prize money along the way. And such contest prizes, in turn, encourage MADers to try to produce some new MADs: ... Anyway, in that competition, its prize money was very high. Then I came in seventh place with 1000 yuan. My participation is just for money, to be honest. ...Eventually, the MAD quality of that entire competition was very high. (Interviewee No.56, translated) Behind MAD's limited revenue is the copyright issue that its commercialization have encountered. This issue is not only appeared in China, but also in Japan. The Japanese ACG websites Niconico and YouTube have helped promote anime and manga and their merchandise to a large extent, but those

recreations can essentially be considered copyright infringement of the original work, and thus there is considerable debate in the Japanese animation industry about whether or not to crack down on MAD (Hernández, 2018). It's a symbiotic but conflicting state with legal issues involved, while MAD as a consumer-created product that economically promotes the original product, making the issue incredibly problematic. But staying in the gray area means that MAD won't be outlawed anytime soon. And MADers with high subscription numbers can still ensure that their videos have a group of regular audience. This leaves space for some products to exploit the loopholes. Some games, both ACG or non ACG style, are looking for MADers to use their games to make GMV. These commissions mostly go to the dssq MADers, who tend to have more views number, more deserving of advertisement placement. Meanwhile, Multi-Channel Network (MCN) companies, which are brokers for bloggers, also seek out a portion of MADers with a high number of followers: With about 20 or 30 thousand subscribers, MCNs will start paying attention to us, then we'll get jobs. About 80% of them are music promotions, the rest are games, usually the most popular games at that time. They want you to use their songs to edit an AMV, then post it on your own account. With my current number of followers on Bilibili, people usually approach me with an offer around two thousand yuan for editing and promotion. (Interviewee No.10) Although sometimes certain games would search for Jing-zhi MADers, for their excellent production technique. In this case, they paid more than the dssq:

Some smart phone game makers, at some important time points, will give some funds to recreators to do the recreation of their products. If it's a MAD, it's usually posted on my own account. Those posted on my own account, is usually free, that I can create whatever I want to express, and I'm unlikely to be interfered with. Actually, I've done a recreation of that game before, and it was completely my own idea, so it's free, and it ended up getting about 100,000 views. The company made that game saw it, I think that's why they approached me. And a MAD like that ended up charging them a couple thousand yuan or more per minute. (Interviewee No.46, translated) The reverse is also true, MADers need to show a good views data to the outside world in order to negotiate higher price from advertisers: In fact, I often have to delete my AMVs and re-upload them. As soon as an AMV doesn't get high enough views, I have to delete it. Either I'll post it again after a while, or I'll simply stop posting it. Because of advertisers, whether they're advertising game or music, they don't just look at how many people following your account, but more critically, how many views you've gained on your previous videos, and that determines their offer. Only the ones with good data performance will make more money. (Interviewee No.68, translated) The status now is that the vast majority of MADers, including those have talented or high-level skills, are struggling to get more than 100,000 views on their videos, even though good MAD work is often forwarded to each other within the MAD community. Fortunately, because MADers have very good technical skills, the two types of videos mentioned when introducin Jing-zhi above, the PV and Shou-shu,

often have MADers involved in their post-production. This has become the main income source in gig works for the vast majority of MADers. But the problem is that, these jobs that involve post-production cannot be done by a MADer who only makes AMVs, especially those who made ASMV and Zong-man. PV commissions generally come from individuals who want to create music videos for their songs. These songs can be original compositions or cover versions. Based on the descriptions from interviewees and observations by researchers, most music video creators on Bilibili appear in their videos themselves. For those who do not appear on camera, while having sufficient financial resources, they usually want a good PV. The group which satisfies these two standards, is commonly known as “Vtuber”. The term “Vtuber” originates from YouTube and refers to people who choose to present themselves online using a virtual looking. The “Vtuber” is “Virtual YouTube Blogger”. Bilibili also hosts a large number of Vtubers, many of whom work as live streamers. By receiving donations from fans during live streaming, Vtubers can generate considerable income and are willing to invest that money in a high quality PV. Because releasing a music video with a beautifully crafted PV can help them attract a larger audience over the long term. As for MADers, it also generates a great money: Normally, when taking commissions, I feel like... For a well produced outcome, I can earn around five to six hundred yuan for one minute PV. (Interviewee No.32, translated) The commission for Shou-shu videos involved a complex value chain.

However, tracing back to the ultimate source, researcher discovered that nearly all Shou-shu projects are requested by a well-known ACG gaming company in China. Other gaming companies and industry giants like Tencent and NetEase also invest in similar mode on Bilibili, but usually smaller in scale. The game company's marketing department sets promotional goals for their game and allocates funds accordingly. Then a significant portion of their budget goes to Bilibili, asking the platform to encourage users to upload videos related to the game. While videos with high views may even receive additional cash rewards. Meanwhile, another portion of the budget is allocated to content creators with massive subscribers who specialize in creating ACG content on Bilibili. These targeted uploaders, which usually has massive subscribers, are also required to create Shou-shu videos related to the game. However, during peak demand seasons, even these uploaders with their own teams struggle to handle the immense workload. Consequently, some of the production tasks are outsourced to smaller video studios. If the workload still becomes overwhelming to smaller studios, members may contact their friends to collaborate on fulfilling the commissions. Throughout this entire process, whether it's creating high-quality game recreation videos on the platform, or the team of massive uploader, or the video studio, or the studio members' friends, the majority of them are MADers, perhaps accounting for over 90% of the workforce (Interviewee No.2, 5, 8, 11, 16, 27, 31, 35, 37, 41, 46, 53, 57, etc). Whether it's PV or Shou-shu commissions, they have become essential ways for MADers to participate in the gig economy and earn considerable income. Some

perceive this income as higher than what they would earn in full-time employment, while others believe it's not quite at that level. Overall, MADers typically price their video commissions based on the cost per minute. As video quality requirements increase, so does the price. Additionally, shorter deadlines command higher prices: Sometimes, in about ten hours of works, I can earn... 1,000 to 2,000 yuan, so I'm making a pretty good profit (Interviewee No.3, translated). Comparing work full-time or take commissions full-time... Taking commission is relatively flexible in terms of your personal schedule, with less intensity and earning less money. Meanwhile, the pressure of video-designing occupations in Internet companies is pretty massive, which will squeeze out your last trace of physical strength. As a result, taking commissions full-time is equivalent to spending 80% of one's work energy and earning 60% to 70% of the full-time occupation money. (Interviewee No.56, translated) The income from taking personal commissions are definitely higher than when working full-time normally. But it cannot last long. As for how much, it's not convenient for me to disclose (Interviewee No.54. translated). Most MADers could earn around 2,000 to 10,000 yuan per month from these commissions. Some intentionally limit their workload, resulting in lower income. While earning 10,000 yuan per month is almost working on video production every second in that month. What's more, to explore the gig economy potential for AMV creators, the researcher, trying to work as a role of part-time video editors, had messaged several

Bilibili content creators who upload knowledge-sharing videos, inquiring whether they need part-time video editors. The selected Bilibili accounts are all with over one million subscribers. Surprisingly, some creators didn't even have their own video production teams. Among those interested in hiring part-time editors, most had no concept of the market rate for editing services. After negotiation, these knowledge-sharing content creators typically willing to pay ranging from 500 to 1000 yuan for an average 20 minutes video. Compared to AMVs, the content they needed was relatively simple, just cutting out any mistakes from the original recording and adding subtitles. With skills learned from making AMV, the researcher successfully accomplished this gig job, proving these is also potential opportunities for AMV MADers. MADers can also earn income from other MADers. In fact, the MAD community has long maintained a culture of sharing without financial compensation. However, in certain cases, some Jing-zhi MADers choose to sell their project files. This decision is driven by the fact that high-quality Jing-zhi MADs often involve complex software operations and reflect the creator's unique style. Analyzing project files allows other MADers to understand how skilled creators produce their videos, making it a valuable resource for learning and improvement. According to the Interviewee No.15 and 69, they both earned over 10 thousand by selling project files or relevant tutorials. According to statements from interviewees No. 48 and No. 15, IG-style MAD creators on Chinese Tik-tok have another way to earn income: by taking on

apprentices. Some MADers charge a fee,which ranging from a few dozen to several hundred yuan, to mentor a follower in video production. Similarly, some Tik-tok creators establish paid group chats for video production discussions. Up to now, this phenomenon has not yet become widespread on the Bilibili platform. Another interesting way of getting income from MADers is related to Zong-man AMVs, especially the dssq AMVs. As mentioned in previous sections, this type of AMV required using diversified scenes from many different animations, collecting all the materials by oneself is always considered a massive workload. Therefore, some MADers are selling their own clip collections (镜头提纯) online for those who are unwilling to do these works all by themselves. Especially many dssq MADers are making MADs mainly for advertising income or platform incentive money, using the simplest or least time-consuming way to create MAD is reasonable. However, such businesses also worsening the overall quality of dssq MAD because many MADers purchased similar clip collections. As MADers receive an increasing number of video production requests, some creators begin to consider entrepreneurship. Interestingly, as early as around 2014, a Bilibili editor quit the job and formed video studios with groups of MADers, establishing companies. Since then, new video production companies have emerged almost every year. The value chain mentioned in Shou-shu production, those small video studios are all established by MADers. These companies often have names ending in “Ying-hua (映画, which means “movie” in Japanese)”. Even in 2023, many MADers continue to establish new video studios, with some forming companies and

others remaining independent. Among the interviewees, several are founders of video studios, each with an average of over 50 MADers. However, during the initial stages, income remains relatively limited: The studio was founded in August of the year 23, and every month... Cash flow is probably around tens of thousands yuan per months, some months may not be very good due to fewer advertisers seeking services during those times. And then... It's been quite volatile couple of months (Interviewee No.8, translated). People in this business... Maybe at the beginning, getting small requests? Maybe a few hundred yuan of small PV, but the more you do the more skilled, you earn more. You can also choose to start your own independent business, then pick up some four-digit or five-digit value jobs. (Interviewee No.11, translated) Take one step further, it is actually possible that an amateur MAD creator, transition into a gig worker in PV and Shou-shu, then transition into a professional video designer in ACG industry. The experiences of the interviewees themselves, as well as stories they shared about their friends, consistently demonstrate the feasibility of this transformation. Most MADers, especially those with higher skill levels and greater work expectations, prefer opportunities within the ACG gaming companies. These companies often require high-quality promotional content, necessitating a team of capable creators proficient in 2D and 3D animation. Their task is to produce compelling promotional videos for critical milestones in game development. It is still called “PV”, but these promotion videos are genuinely short original animation films. Even if direct employment with ACG game companies isn't feasible, there are

numerous mature “Ying-hua” companies in the market that handle outsourced work from these gaming companies. Many of these outsourced companies were founded by former MADers, or led by Lead Artists and video production team leaders being MAD creators themselves once. These companies actively employ high skilled MADers to contribute to their projects. One reason why these positions need to recruit MADers is that the current art colleges in China do not train talents in this field: Around 2020, many MAD creators actually didn't have a clear views about what future might be. Those who transitioned into professional promotion video production, is those being active in the MAD community during that period. They coincided with the rise of short video on social media and the expansion of the ACG gaming market. The overall economic environment was characterized by significant growth. Meanwhile, in China, there was a significant shortage of skilled video production professionals in ACG style. Unlike countries such as Japan and South Korea, which they have majors like “video design”, Chinese universities lacked a comprehensive programs specifically designed to train the artists that ACG gaming needs. Consequently, video post-production occupations in the field of ACG actively competed for talent like MAD creators, offering improved compensation and benefits. (Interviewee No.16) According to the interviews, transitioning from an amateur MAD creator to a professional video designer isn't typically a consideration for those with strong educational backgrounds. For example, the Interviewee No.3 works as a dentist, making it clear that he wouldn't trade his risk-free job to make video full-time.

However, for most MADers with limited education (such as junior college) or degrees in fields with challenging job prospects (such as civil engineering), video production roles are highly appealing. The majority of MADers currently working in video production report earning significantly higher incomes than their undergraduate classmates or their original career paths: Compared to my classmates in the major I studied before... If you have to compete about income, my current profession may be more than twice as high as theirs. (Interviewee No.31, translated) As for video-making occupations in ACG gaming companies, some remote cities have an average monthly salary more than 10,000, and the better cities are about 20,000. This income is much higher than my classmates, because they are traditional industries. My undergraduate classmates graduated about earning 5000 per month, I was able to get at least 10,000 by then. (Interviewee No.11, translated) However, there are also warnings about entering gaming industries: If a MADer possesses a skill level not high enough to be employed as video designer, but still wants to enter the ACG gaming industry, it is to see if he is willing to choose someplace lower. He can either start with advertising or marketing, or just join an outsourcing companies. (Interviewee No.18, translated) ... But in terms of working years, it will definitely be much less than my old classmates, because video post production cannot reach the age of forty or fifty. I currently expect to work till the age of 35. (Interviewee No.31, translated) The gaming industry has brought dividends to this position, but as the

threshold becomes lower and lower, more and more people will inevitably be eliminated in this industry. Personally, I think it is not suitable for long-term development. (Interviewee No.54, translated) The best time has indeed passed. If I had made the decision four years ago, around 2020, to skip undergraduate and directly join that game company, my accumulated income would be very considerable by now... Back then, even pure editing roles had opportunities within that company... No one could foresee its remarkable growth in recent years. Now, entering such companies is significantly more challenging, and the required skill set has expanded considerably. (Interviewee No.69, translated) Some interviewees discussed other alternative career paths. Interviewee No. 63 mentioned working for a company that produces promotional videos for various state-owned enterprises. He emphasized that beyond the ACG industry, there are abundant video production opportunities across society. Many industries are embracing new media, leading to a high demand for video-related roles. While salaries may not be exceptionally high, job opportunities exist for internships, part-time positions, and full-time roles. Even for average level MADers, these positions are well within their capabilities. And an Interviewee (identity number remains confidential as he required) worked for a company that produces the most short-form video advertisements globally. Although the quality of their ads isn't top-tier, they focus on high volume, churning out numerous short, standardized ad videos for comprehensive client promotion. This interviewee highlighted that

MADers have a technical advantage, allowing them to earn approximately 10% more than recent college graduates in similar roles. However, transitioning toward traditional film and television industries poses greater challenges for MADers. Interviewee No. 55 described the TVC (Television commercial) industry's attitude toward MADers: The most obvious problem is that when you want to go to a regular advertising company and find an interview opportunity to submit your showreel, other people's showreel are all edited from real-life footage, while you are an anime material. Although the thinking framework of editing may be similar, the HR may worry that you may not have experience in editing real-life footage, which can be considered a more risky choice than others. (Interviewee No.55, translated) Of course, there are exceptions. Interviewee No. 59 shared his journey: after graduating from university, he learned video editing due to the AMVs. Over six to seven years, he successfully transformed from a complete novice to a commercial film editor in China. According to this interviewee, with sufficiently high editing skills and the right opportunities, becoming a professional editor is indeed possible. An important point to note is that a certain interviewee (identity number remains confidential as he required) mentioned that, due to the saturation of MADers in the related industry, certain aspects within the MAD community can impact employment opportunities. Since many top-tier MADers are acquainted with each other, conflicts between one MADer and a well-known counterpart can lead to social exclusion from the entire MAD community. One MADer experienced such a situation.

When he attempting to find a video production job in the relevant industry, the recruitment team at one company discovered that the applicant was a MADer. They directly inquired with their friend, who is also an active MADer, about the opinion of this applicant MADer. Given the predominantly negative attitude toward the applicant from other MADers, the company decided to reject them at the resume stage. Interestingly, among the interviewees, including two MADers from Chinese Taiwan, it was noted that many people reach out to them via Bilibili, hoping they would accept certain requests. To them, this phenomenon is quite uncommon on platforms like YouTube or Twitter, even though a significant number of MADers from countries such as Japan, South Korea, France, and Russia gather on those platforms. One of the MADers declined all business collaboration requests because they couldn't receive Chinese RMB, and transferring New Taiwan Dollars from mainland to Chinese Taiwan was also challenging. However, another MADer borrowed his relative's Alipay account, and successfully received the money. As a simple summary, the gig economy around MAD has an obvious diversified feature compared to typical gig work models like ride-hailing and food delivery. The core of this lies in the rich artistic expression of MAD. However, from an economic perspective, there are risks associated with the MAD gig economy. Firstly, legal issues related to derivative works could potentially lead to the entire MAD community being banned. Secondly, the majority of MAD creators' income is closely tied to the ACG industry, even relying entirely on few major ACG gaming company. Currently, the ACG gaming market is still growing, but whether MAD

creators and their economic livelihood can survive if the industry faces a downturn remains an unanswered question.

5. Platform, New Technology, the Sustainability in MAD Community

5.1. The Role of Bilibili to the Development of MAD Among the 69 interviewees, excluding 20 who remained neutral or declined to answer, 40 expressed a positive attitude toward Bilibili's role. Only 9 individuals held opposing views. The most common perspective is that Bilibili is the sole platform in China that brings together all MADers. It is here, over an extended period, that MADers can create, discuss and engage in artistic expression. Existing literature believed that the emergence of video platforms has also produced a collateral effect: the creation of creative sharing communities. In addition to creators providing content and appreciating each other's work, there are two different types of audiences: commentators who browse content for entertainments and new potential creators who learn from such creations (Qiyang and Jung, 2019). The research by Qiyang and Jung (2019) found that short video and social media platforms have tremendous potential for collaborative learning of creative skills, and it was observed that some newcomers indeed achieved technical improvement. This coincide with the scenario of MAD on Bilibili. What's more, comparing the MAD creative environments on two different Chinese platforms — Bilibili and Tik-tok, which is even larger in scale: According to descriptions from interviewees No. 15 and No. 48, on Douyin (Chinese version of Tik-tok), over half of the MADers have experienced malicious verbal attacks online.

Some MADers would form a group chat and add their disliked MADer into that group chat (usually on QQ), where they could collectively insult them. Such online bullying is prevalent within the Douyin MAD community. Additionally, the “Kai-he” (Open the Box, 开盒) phenomenon is extremely serious too. This behavior involves certain MADers revealing the victim's real name, ID number, home address, and even parents' personal information and workplace in the comments section of the victim's video. Of course, this personal information is obtained illegally by the attackers. As a result, almost all friendly and talented MADers on Tik-tok are now on Bilibili. According to the description provided by the Interviewee No.32, there is a significant difference between YouTube and Bilibili platforms. YouTube lacks effective communication channels among creators and has limited social features. As a result, MADers often need to rely on other chat platforms for chatting. In contrast, connecting with new MADers on Bilibili can be quite simple. Additionally, YouTube's recommendation algorithm for MAD content may negatively impact creators' motivation: The recommendation mechanism of YouTube is quite “abstract”, relying entirely on its AI review. It actually has little to do with whoever subscribed you. If the AI determines that your video quality is high, it will recommend it to others... the number of views on YouTube is quite unpredictable, it is all depends on luck. (Interviewee No.32) In fact, it is the social attributes of the Bilibili platform that ensure low communication costs for MADers engaged in the gig economy. Because MADers can

easily connect with unfamiliar MADers through Bilibili, the cost of getting to know each other has been minimized, allowing all MADers to truly form a community. Even within the case of technical-focused MADers who dislike the dssq style MADers, there are still many of them who know each other. It is quite easy to find another unfamiliar MADer by asking the MADers you know. Additionally, whether it's game company staff, other types of creators on Bilibili, or all sorts of agencies, they can all directly contact MADers through Bilibili. These factors ensure that MADers can participate more deeply in the emerging gig economy. Bilibili itself also has an integrated commercial section for users to order and accept requests. 5.2. MAD Contest and Teams: Core Sustainability of MAD From the current perspective, the existence and long-term development status of the MAD community are not primarily determined by commercialization but rather by the robust vitality naturally formed within it. The MAD community, up to now, remains being fundamentally an enthusiast community, and continues to attract new creators as long as there are ACG culture enthusiasts in China. This influx of creators has been historically verified. Despite the sentiments expressed by veteran MADers about the golden era of MAD on Bilibili from 2012 to 2014, they also acknowledge that over the past decade, the total number of MAD creators has increased significantly, and their technical skills have noticeably improved. Behind this growth lies the role played by MAD competitions and MAD organizations, ensuring the continuity and development of the entire community. Over the past decade, a large number of MAD organizations have emerged

within the MAD community, of which there are three main types: Team, Club, and Studio. By convention, there is an unwritten rule that a MADer can only be a member of one team max. As a result, the exclusivity of teams leads to an outcome that MADers are generally proud if the overall level of the MAD team they are in is higher. At the same time, the vast majority of MAD contests are dependent on teams for their existence. Typically, each team hosts a MAD contests as an organizer at a fixed frequency. And, these contests tend to represent the level of membership of the team, as team members tend to prioritize submitting MAD works to their own contests, and new MADers participating in the contests usually want to join a particular Team through that contests, thus creating a virtuous circle between the Team and the MAD contests. While there are equally Club and Studio-sponsored contests today, they are generally not large, according to an experience organizer: In fact, many organizations don't know how to organize competitions... The specific manifestation of the situation, is that the vast majority of the competitions didn't have a second one next year, or the first one didn't receive any MAD submissions... However, for big competitions, such as the Initial MAD Team and the “Feng hua Hui Zhan (Contest hosted by the Initial MAD Team)”, members will provide a guaranteed number of high-quality MADs for the competition, so that the quality of the contest can be maintained at a high level every year, and high-level competitions can allow more talented newcomers to enter IMT. (Interviewee No.17, translated) When it came to the question about the current state of the MAD community,

41 out of 69 interviewees were neutral or choose not to answer, but still 16 out of the remainder expressed positive attitudes, and only 12 were explicitly dissatisfied with the current state of the MAD community. When it comes to the future of MAD in China, 18 out of 69 interviewees are optimistic, while 11 feel that the situation is getting worse. The difference, however, is that 13 people believe that there are two sides to the issue, with some MADers getting better, while the whole may continue to be in the doldrums. Most of the negative attitudes are about expression, that today's MAD creations are becoming homogenized, and that nowadays MADers value video presents and visual effects more than artistic content. And, of course, there is the issue of failed commercialization. As for the positive group, on the other hand, believes that Chinese MADers have come out of their own way and have caught up with the world's best in some aspects. The attitudes on both sides have their validity, and again, this is a question that will have to wait for the future to answer. As the interviewee No.17 claimed: Honestly, the MAD community is the biggest reason why MAD is still alive. The original MAD community is divided into differences levels... pure interest, and with some professional, the threshold of entering different levels of community is basically linked to the quality of the work... which gradually becoming an elite group. The future of MAD is thriving, I really mean it, because I know some adjustments on Bilibili's development directions. This is an economic downturn stage, the gray area of copyright is becoming loose, and the operation model of Bilibili has become more

perfect. (Interviewee No.17, translated) 5.3. New Technologies This study also explored the attitudes of MAD creators toward artificial intelligence (AI). Over time, MAD creation has always progressed in tandem with advancements in video production technology. However, AI poses potentially disruptive implications for creative work. Among the 69 interviewees, 28 believe that AI will be beneficial in the long term, and 26 see positive implications for creativity. In this year, 2024, AI has already improved MAD creators' efficiency by assisting with image and sound processing, resulting in better effects than before. Meanwhile, in the short term, AI still can't fully replace human involvement in specific video production tasks. But, AI has indeed prompted many MAD creators to consider additional creative possibilities in recreations. Overall, the attitude remains positive. Interestingly, four individuals firmly believe that AI is not a positive development, while seven hold a considerably negative view of AI's impact on creativity. Their reasoning lies in the simplification of production processes, enabling more creators to achieve previously challenging effects through user-friendly tools. These creators are those who lack the ability to delve deeply into the creative process, yet suddenly gain access to advanced video production capabilities, ultimately exacerbating issues of homogeneity.

V. Discussion and Conclusion

This study provides a detailed account of the origins and active periods of the MADer community. Compared to other gig workers, MADers exhibit a decline in creativity and energy as their years of creation increase. Consequently, MADers have relatively shorter engagement in gig work compared to typical gig laborers in traditional industries. Additionally, the study discusses the software and technology skills mastered by MADers, emphasizing the channels through which they acquire relevant expertise. Notably, there exists a close connection between MADers and the platforms. Bilibili, in particular, facilitates the progress of MADers by offering convenient conditions and low learning costs, enabling MADers to stay up-to-date with the latest video production technologies. In a context where China's universities do not specifically cultivate video creators in the ACG style, MADers occupy a unique niche not only as gig workers, but the only capable workers in this field, granting them advantages not available to other laborers. This study thoroughly investigates various but most commonly seen profit models related to MAD, including earnings based on MAD content itself, earnings based on skills associated with MAD, earnings based on creative production process, entrepreneurial ventures, and even the potential transition of MADers toward professional video production occupations. The diversity of MAD as an art form allows for significant expansion possibilities. Unlike traditional gig workers in fields like ride-hailing or food delivery having only one way to profit, MADers have the advantage of tailoring their income models to align with their specific skill sets. These

are the good sides being a creative class. However, the current commercialization of MAD carries two unavoidable risks. Firstly, recreation operates in a gray area of copyright law, even the software used by MADers are generally cracked version. Secondly, these gig workers' main income sources are often trace back to only a few giant ACG gaming companies. Consequently, the overall future of the MAD community remains highly uncertain. Its sustainability depends not on its internal dynamics but rather on how original content creators address intellectual property concerns, or whether the gaming industry remains the growth. Thus, MADers can also be seen as precarious workers. However, due to a healthy mechanism of continuous improvement within the MAD community, existing MADers still maintain a relatively positive outlook for external prospects. Therefore, this study suggests that it is reasonable to hold an optimistic expectation for the long-term development of MAD and its economic potential. However, this research does have certain limitations. On one hand, sample selection still lacks unbiasedness to an extent, as many MADers or former MADers with large subscriber numbers on Bilibili tend to decline interview requests. Moreover, among the interviewees, most are restricted by confidentiality agreements, preventing them from providing precise information. Additionally, some MADers are not solely animation recreators; they often possess other skills. The situation of creators who simultaneously excel in multiple areas remains beyond the scope of this study. Overall, MAD and its online creator community does reflect the diversity of China's current development of gig economy and flexible employment, also

demonstrated the current situation of some young people in China who are earning extra income on the internet platform. While this dissertation provides the results of one in-depth study of a particular group, it is important to note that the MADers group is still extremely unique and its long-term development may show new trends that is beyond what is described by the conclusions of this study.

Appendix List of Sample Interview Questions Please noted that questions might be altered based on interviewees' answers to the previous questions during the interview, in order to get more detailed information on related topics. Questions were asked in Mandarin-Chinese translation to the interviewees.

1. Time and Categories of Participation in MAD·AMV Creation. 1.1. When did you start making MAD·AMV? 1.2. Why did you start making MAD·AMV? 1.3. Are you still making MAD·AMV? 1.4. If you no longer continue to produce MAD·AMV, when and why did you stop? 1.5. What type of MAD·AMV do you produce? 1.6. Why did you want to make this type of MAD·AMV? 1.7. Have you ever changed the categories of MAD·AMV you choose to produce? 1.8. If you changed the categories, Why?
2. Techniques Learned from Participation in MAD·AMV Creation. 2.1. What software did you master through the process of making

MAD·AMV? 2.2. What kind of technique have you mastered by MAD·AMV Creation? 2.3. How did you learn these techniques? 2.4. Why did you choose to learn these techniques? 2.5. Why did you choose the way you mentioned to learn these techniques? 3. Personal Information Related to MAD·AMV Creation. 3.1. Could you please introduce yourself? 3.2. Do you have any art-related background? 3.3. Is there any impact of your personal status on your MAD·AMV creation? 4. Income Related to MAD·AMV Creation. 4.1. Have you made any income purely from producing MAD·AMV? 4.2. Have you gained higher income, or somehow increased your potential income level, because of the skills you learned from making MAD·AMV? 4.3. If you ever earned income from MAD·AMV or the skills learned from MAD·AMV creation, could you tell me the nature of your income? Is it a request? Is it part-time? Are you starting a business? Or any kind of employment? 4.4. What is the approximate range of income? 4.5. What kind of videos did you make for earning income? 4.6. What is your average time consumption for finishing one video production job? 4.7. Do you adjust your direction in personal creation out of the need for

income? 5. Related Economy and the Value Chain. 5.1. If you ever got a request from others, how did they reached to you? 5.2. If you ever got a request from others, who is paying you? 5.3. If you are working full-time in relevant industry, do you know how many of your colleagues are MADers or former MADers? 5.4. If you are working full-time in relevant industry, do you think it is easy for a MADer to get a job in your occupation? 6. Personal Opinions. 6.1. What role do you think the Bilibili platform has played in the creation and development of MAD·AMV? 6.2. What role do you think the Bilibili platform has played in your self-improvement and income growth based on MAD·AMV? 6.3. What do you think of the creator community atmosphere of MAD·AMV? What do you think of the future of the MAD·AMV creator community? 6.4. What do you think the emergence of new technologies means to the creation of MAD·AMV? 6.5. How do you view the impact of artificial intelligence?

References

Anderson, M., McClain, C., Faverio, M., & Gelles-Watnick, R. (2021). The state of gig work in 2021. Pew Research Center, 8. Arditi, D. (2016). Informal Labor in the Sharing Economy: Everyone Can Be a Record Producer. Fast Capitalism, 13(1), 97-104. Arora, M., & Sharma, R. L. (2022). Integrating gig economy and social media platforms as a business strategy in the era of digitalization. In Integrated business models in the digital age: Principles and practices of technology empowered strategies (pp. 67-86). Cham: Springer International Publishing. Bajwa, U., Gastaldo, D., Di Ruggiero, E., & Knorr, L. (2018). The health of workers in the global gig economy. Globalization and health, 14, 1-4. Best, S. (2017). The US Gig economy: economic imposition or economic choice?. The Business & Management Review, 8(4), 60. Bilibili Inc. (2024). Bi li bi li gu fen you xian gong si 2023 nian du bao gao [Bilibili Inc. 2023 Annual Report]. Retrieved June 18, 2024 from: https://ir.bilibili. com/media/2xvhmd00/bilibili-inc-2023-annual-report-c.pdf Chalaby, J. K. (2024). The streaming industry and the platform economy: An analysis. Media, Culture & Society, 46(3), 552-571. Chen, Z. T. (2021). Poetic prosumption of animation, comic, game and novel in a post-socialist China: A case of a popular video-sharing social media Bilibili as heterotopia. Journal of Consumer Culture, 21(2), 257-277. Davis, M. E., & Hoyt, E. (2020). A longitudinal study of piece rate and health:

evidence and implications for workers in the US gig economy. Public Health, 180, 1-9. De Stefano, Valerio. (2016). The rise of the just-in-time workforce: on-demand work, crowdwork, and labor protection in the gig-economy. Comparative Labor Law & Policy Journal, 37(3), 471-504. Desilver, D. (2021). Q&A: How pew research center studied gig workers in the U.S. Pew Research Center. Retrieved June 18, 2024 from: https://www.pew research.org/short-reads/2021/12/08/qa-how-pew-research-center-studied- gig-workers-in-the-u-s/ Donovan, S. A., Bradley, D. H., & Shimabukuru, J. O. (2016). What does the gig economy mean for workers?. Friedman, G. (2014). Workers without employers: shadow corporations and the rise of the gig economy. Review of keynesian economics, 2(2), 171-188. Gao, J., Zhang, L., Wen, Z., & Si, Q. (2024). Zhong guo ling gong jing ji qu yu cha yi, dong tai yan jin ji shou lian xing yan jiu [Regional disparities, dynamic evolution, and convergence of China's gig economy]. Zhong Guo Shang Lun [Chinese Business Review], 42(6), 143-147. DOI: 10.19699/j.cnki.issn2096- 0298.2024.06.143 Healy, J., Nicholson, D., & Pekarek, A. (2017). Should we take the gig economy seriously?. Labour & Industry: a journal of the social and economic relations of work, 27(3), 232-248. Hernández, Á. D. H. (2018). The anime industry, networks of participation, and

environments for the management of content in Japan. In Arts (Vol. 7, No. 3, p. 42). MDPI. Hichibe, N., & Tanaka, E. (2016). Content Production Fields and Doujin Game Developers in Japan: Non-economic Rewards as Drivers of Variety in Games. Transnational Contexts of Culture, Gender, Class, and Colonialism in Play: Video Games in East Asia, 43-80. Ichikohji, T., & Katsumata, S. (2016). The Relationship between Content Creation and Monetization by Consumers Amateur Manga (Doujinshi) and Music in Japan. Annals of Business Administrative Science, 15(2), 89-103. Information Center, Ministry of Human Resources and Social Security of the PRC. (2024). 2023 zhong guo xin xing ling huo jiu ye bao gao (2023 nian 11 yue 23 ri) [2023 China New Flexible Employment Report (November 23, 2023)]. Retrieved June 26, 2024 from: https://www.hrssit.cn/info/3146.html Jabagi, N., Croteau, A. M., Audebrand, L. K., & Marsan, J. (2019). Gig-workers' motivation: thinking beyond carrots and sticks. Journal of Managerial Psychology, 34(4), 192-213. Kenney, M., & Zysman, J. (2019). Work and value creation in the platform economy. In Work and labor in the digital age (Vol. 33, pp. 13-41). Emerald Publishing Limited. Knobel, M., Lankshear, C., & Lewis, M. (2010). AMV Remix: Do-it-yourself anime music videos. DIY Media: Creating, Sharing and Learning with New Technologies, 205-230.

Lei, D. M., Niu, Y. X., Zhang, X. R., & Jiang, Y. J. (2018, May). Study on the Countermeasures of Perfecting Flexible Employment in China in the Internet Era. In 4th Annual International Conference on Management, Economics and Social Development (ICMESD 2018) (pp. 134-138). Atlantis Press. Lin, L., Liu, F., & He, M. (2024). Yuan gong wei shen me zuo fu ye? Mu biao nei rong li lun shi jiao xia de fu ye xing wei qu dong ji zhi yan jiu [Why do employees engage in side jobs? A target-content theory perspective on the driving mechanisms of side job behavior]. Human Resources Development of China, 41(1), 82-93. DOI: 10.16471/j.cnki.11-2822/c.2024.1.006. Lin, Z. (2021). Commercialization of creative videos in China in the digital platform age. Television & New Media, 22(8), 878-895. Maffie, M. D. (2020). The role of digital communities in organizing gig workers. Industrial Relations: A Journal of Economy and Society, 59(1), 123-149. Malik, R., Visvizi, A., & Skrzek-Lubasińska, M. (2021). The gig economy: Current issues, the debate, and the new avenues of research. Sustainability, 13(9), 5023. Manyika, J., Lund, S., Bughin, J., Robinson, K., Mischke, J., & Mahajan, D. (2016). Independent-Work-Choice-necessity-and-the-gig-economy. McKinsey Global Institute. Mieruch, Y., & McFarlane, D. (2023). Gig economy riders on social media in Thailand: Contested identities and emergent civil society organisations. VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations,

34(6), 1232-1242. Nakagawa, K. and Nakagawa Hirokazu. (2008). 「MAD movie」ni miru eizou seisaku bunka no itikousatu [A study on the video production culture of MAD Video]. Kurashiki Geijutsu Kagaku Daigaku Kiyou [Bulletin of Kurashiki University of Science and the Arts], (13), 15-19. National Development and Reform Commission, PRC. (2021). Shu zi jing ji shi dai xia yao rang ling huo jiu ye geng jia gui fan fa zhan [In the digital economy era, development of flexible employment should be more standardized]. Retrieved June 30, 2024 from: https://www.ndrc.gov.cn/fggz/jyysr/jysrsbxf/ 202108/t20210818_1293928.html Parker, G. G., Van Alstyne, M. W., & Choudary, S. P. (2016). Platform revolution: How networked markets are transforming the economy and how to make them work for you. WW Norton & Company. PwC Legal. (2022). Gig Economy 2022. Retrieved June 15, 2024 from: https://www.pwc.be/en/fy22/documents/Gig-economy-2022.pdf Qiyang, Z., & Jung, H. (2019). Learning and sharing creative skills with short videos: A case study of user behavior in tiktok and bilibili. International association of societies of design research (IASDR), design revolution. Sharif, I. and Qiang, Z. (2023). The Promise and Peril of Online Gig Work in Developing Countries. World Bank Blogs. Retrieved June 24, 2024 from: https://blogs.worldbank.org/en/voices/promise-and-peril-online-gig-work- developing-countries

State Council Office of the PRC. (2020). Guo wu yuan ban gong ting guan yu zhi chi duo qu dao lin huo jiu ye de yi jian ['Opinions on supporting flexible employment through multiple channels'], State Council Office, 2020(27). Retrieved June 08, 2024 from: https://www.gov.cn/zhengce/content/2020-07/ 31/content_5531613.htm State Internet Information Office of PRC. (2021). Zhong guo gong xiang jing ji fa zhan bao gao (2021) [China's Sharing Economy Development Report (2021)]. Retrieved June 06, 2024 from: https://www.ndrc.gov.cn/xxgk/jd/wsdwhfz/ 202102/P020210222307942136007.pdf State Internet Information Office of PRC. (2023). Shu zi zhong guo fa zhan bao gao (2022) [Digital China Development Report (2022)]. Retrieved June 07, 2024 from: https://www.cac.gov.cn/2023-05/22/c_1686402318492248.htm. The State Council Information Office of the PRC. (2022). Guo xin ban ju xing 2021 nian guo min jing ji yun xing qing kuang xin wen fa bu hui tu wen shi lu [The State Council Information Office held a press conference on the operation of the national economy in 2021]. Retrieved June 13, 2024 from: http://www. scio.gov.cn/xwfb/gwyxwbgsxwfbh/wqfbh_2284/2022n_2285/2022n01y17r/t wzb_2320/202208/t20220808_315960.html Todolí-Signes, A. (2017). The 'gig economy': employee, self-employed or the need for a special employment regulation?. Transfer: European Review of Labour and Research, 23(2), 193-205. Tsinghua University Institute of Economics, and Beijing ByteDance Public Policy

Research Institute. (2020). Hu lian wang shi dai ling gong jing ji de fa zhan xian zhuang, she hui ying xiang ji qi zheng ce jian yi [Internet Gig Economy: Development Status, Social Impact, and Policy Recommendations]. World Bank. (2023). Gong zuo wu guo jie - Zai xian ling gong gong zuo de qian li yu feng xian: Gai shu (Chinese) [Work without borders - the potential and risks of online gig work: Overview (Chinese)]. Washington, D.C.: World Bank Group. Retrieved June 11, 2024 from: http://documents.worldbank.org/curated/en/ 099071923114585564/P17730204e67320870b82302bc88dcc4a05 Yang, G., & Lu, Y. (2024). Cong chuan tong ling gong dao she qun wang yue gong: Ping taijing ji bei jing xia de “bian” yu “bu bian” [From traditional gig workers to community-based gig workers: Changes and continuities in the platform economy]. China Collective Economy, 4, 17-20.