Composition Forum 49, Summer 2022
http://compositionforum.com/issue/49/

The Discourse-Based Interview on Twitch: Methods for Studying the Tacit Knowledge of Game Developers

Rich Shivener, Jessica Oliveira Da Silva, and Anika Rahman

Abstract: In this essay, we argue that Twitch is an incredible platform for cultivating discourse-based interviews (DBIs) and has yet to be addressed in DBI research involving digital tools. To demonstrate that argument, we detail the methods behind collaborative research project between two undergraduates and a faculty studying game developers on the platform. Our collaborative approach to studying game developers on Twitch is framed as a 2022 update to Odell, Goswami, and Herrington’s landmark essay The Discourse-Based Interview: A Procedure for Exploring the Tacit Knowledge of Writers in Nonacademic Settings. After providing an overview of Twitch and recent scholarship, our essay describes three major challenges associated with cultivating DBIs from the platform: recruiting participants, managing files ethically, and scaling the project. Our focus on two interviews with one game developer reveals how a DBI on Twitch allows for participant agency. Based on that experience, we close with two recommendations for future DBIs that turn to Twitch: keep the project small, and go deep.

Introduction

Layered with student-narrated videos, this essay focuses on challenges and recommendations for researchers studying content creators, such as game developers, who compose live on Twitch, a livestreaming platform with more than 2 million broadcasters. Especially during times when rhetorical fieldwork is limited to screens, Twitch is an incredible platform for cultivating discourse-based interviews (DBIs) and has yet to be addressed in DBI research involving digital tools (e.g., Bay and Sullivan; Olinger; Reid; Swarts; Van Kooten). For writing studies scholars including undergraduate researchers, Twitch is a significant platform and useful entry point into practicing interview methodologies because creators display their composing practices and surrounding discourse in real time, meaning they compose text, animations and code while talking aloud about those materials with audiences. As a result, Twitch affords a writing studies research team myriad opportunities to study writing and composing in action rapidly without premeditated intervention, using the public discourse generated from livestreams to study, recruit, and interview creators.

Our approach to studying the work of game developers on Twitch is framed as a 2022 update to Lee Odell, Dixie Goswami, and Anne Herrington’s landmark essay The Discourse-Based Interview: A Procedure for Exploring the Tacit Knowledge of Writers in Nonacademic Settings, first published in 1983. Presented as a sort of vertical timeline (i.e., with subheads) that includes four short videos, this essay discusses those opportunities as well as challenges and recommendations for conducting such research. Our intended audiences are both teacher-scholars and undergraduate researchers because it was a collaborative project between the two. The accompanying videos were designed specifically with undergraduates in mind, keeping them short and focused on the experiences of the two undergraduates who were on the research team. Consequently, this essay is designed to be pedagogical and invitational for future collaborators who might wish to adapt or scale such a research project.

Project Context and Challenge #1, May 2021: Preparing a DARE-ing Twitch Research Project amid Pandemic Restrictions

This section begins with some background on our research project: studying the rhetorical practices of game developers who livestream on social media platforms such as YouTube and Twitch. This exploratory project was a thread of Rich’s comparative studies of digital media writing and feelings across disciplines and funded by a research mentorship program at York University. In winter 2021, the university’s dean’s office for the Faculty of Liberal Arts and Professional Studies released a call for proposals for the Dean’s Award for Research Excellence (DARE) for undergraduate students. Under the DARE program, researchers submit project descriptions and calls for research assistants, and students apply directly to the researcher. (See the Appendix A of this article for Rich’s project description.) When a researcher selects a student (and in Rich’s case, two students!), the team submits letters of recommendation and statements of project goals along with other background information to the dean’s office, which makes the final decision on funding. In May 2021, Jessica and Anika were selected as participants of the DARE program and were offered a stipend by the university to work with Rich from May to August 2021. In fact, Jessica was a returning student to DARE and served as a peer mentor to Anika, who was new to qualitative research methods. Both students were asked to work between 15 hours a week on what was described as low-level and high-level tasks for the project, from assisting with curating a database of contact information of creators to interviewing research participants.

This DARE project sought to accord with the CCCC Position Statement on Undergraduate Research in Writing, first published in 2017. It calls for faculty researchers to help undergraduate researchers “obtain knowledge of writing that can be learned only through direct participation in full-fledged creative or critical inquiries,” and it argues that “faculty, staff, and graduate students who teach and mentor undergraduate writing researchers gain distinctive opportunities for student-centered instruction, collaboration (e.g., coresearch, coauthorship), and professional development.” Through mentorship on qualitative methods, Rich’s end goal was to publish an article with—and not about—undergraduate writers, which have been the subject of DBIs (see Lancaster; Jomaa and Bidin; and Eriksson and Nordrum among a bevy of discourse-based interviews and analyses about undergraduates). However, this special issue of Composition Forum was an opportunity to take a “both/and” approach to a DBI article, commenting and reflecting on our methods, as explained by and with undergraduate researchers.

Starting in May 2021, then, we began meeting remotely to plan the Twitch project that would study, recruit, and interview game developers. Due to pandemic-related restrictions in Ontario, Canada and many parts of the world, our project was limited to remote-only meetings and research activities. (In fact, as of finishing revisions to this essay in summer 2022, we still haven’t met each other in person!) Thus, we decided we would explore the discourse and practices of game developers who livestream on Twitch. With its emphasis on gaming and public participation, Twitch has been an optimal choice for our project, because it is a platform that has only gotten more robust with creators and viewership while many of us were relegated to our homes during the first two years of the pandemic (Lim). It started in 2011 and has millions of viewers each month. The service is operated by Amazon, which offers incentives to those who subscribe to the Amazon Prime delivery and distribution service. Twitch’s creators and audiences mainly come out of gaming and game development but also include communities of visual art, writing, and “just chatting” streams. In fact, United States of America President Joe Biden’s 2021 inauguration was livestreamed on Twitch, further cementing the service’s prowess for distributing public content to thousands of viewers with a diverse range of interests (see Riddick and Shivener). What’s more, game developers who livestream on Twitch often archive their work publicly for audiences, meaning researchers have access to at least two weeks’ worth of on-demand content for review. Twitch, in a way, is a dream for writing studies researchers because content creators are self-selecting methods that support studies of process and the tacit knowledge of writers: they talk aloud about draft material, even doing so for multiple sessions over months and even years. For DBIs in particular, game developers on Twitch accord with the “repertoire of research strategies” that Odell, Goswami, & Herrington called for in 1983. Allow us to update it: Game developers engage in “a repertoire that includes interviews [with audiences who ask questions in chat], composing aloud [about practices].... and videotaping [themselves and screens] while they are writing” (235; brackets added). In sum, game developers carry out these strategies on their own, providing plenty of follow-up interviews and subsequent analyses

While YouTube has myriad livestreamers across content areas, Twitch has received more scholarly attention as a platform for livestreaming gameplay and game development. Recent research on Twitch has studied women content creators (London et al.; Ruberg and Brewster; Ruberg), game-player streaming (Taylor; Johnson and Woodcock), and audiences of streams writ large (Sjöblom and Hamari; Kobs et al.). The scholarship demonstrates that Twitch is an emergent confluence of creators, audiences, and feelings that circulate between them, all on public display. As Jamie Woodcock and Mark Johnson argue, “it is the most gregarious, emotionally engaged, and outgoing individuals who find success in live streaming” on Twitch (818), emphasizing that interaction is crucial for streamers. However, London et al. demonstrate that women who are popular streamers are “regularly engaged in a balancing act between fostering their communities and managing disruptive behaviors, which can undermine the very content of their channels” (53). Whether derived from interviews (Woodcock and Johnson) or distant observations (London et al. ), these findings about Twitch streamers are useful for understanding the rhetorical and emotional stakes of livestreaming.

This essay builds on such scholarship in two ways: further integrating Twitch into writing studies and explicitly demonstrating the affordances and challenges of a DBI methodology applied to the platform. In addition to the expansive body of DBI research in writing studies, Twitch speaks to the field’s research areas such as process studies (Rule; Lockridge and Van Ittersum) and circulation and audience studies (Silvestro; Bradshaw; Riddick) that value recording and analyzing writers working in social and public realms. However, Twitch has not been explored extensively by the field of writing studies and aligns well with the aims of researchers who investigate textual composing as much as talk, audiences and ephemeral artifacts of writing situations. A focus on Twitch affords a DBI approach similar to studies by qualitative researchers such as Jason Swarts, Hannah Rule, and Stacey Pigg who have used screen-recordings to track the multimodal discourse of writers they also interviewed. In fact, a goal of this essay is demonstrate the platform’s usefulness for researchers who study the practices and tacit knowledge therein of writers and multimodal creators. Invoking Odell, Goswami, & Herrington, we argue that Twitch is a means to “determin[ing] what assumptions writers made or what background knowledge they had concerning the audience, the topic, and the strategies that might be appropriate for achieving their assignment purpose with a given audience” (222). As demonstrated by this essay’s focus on a game developer, Twitch streamers often explicate their tacit knowledge for public audiences. In short, this methodological retrospective on our Twitch project is an argument and model for the field to further consider Twitch as a viable space for DBIs and complementary qualitative methods.

Addressing the First Challenge of Studying Twitch

Given that Twitch has more than 2 million broadcasters and thousands of game developers within, we spent a month exploring Twitch streams and deciding who to interview. This practice was great for getting acquainted with rhetorical commonplaces on Twitch (e.g., screen sharing, chatting in real time with audiences). However, many game developers’ streams run anywhere from two to four hours, which was a time sink in the early stages of research. To give ourselves some boundaries, we eventually limited our study to developers in the Greater Toronto Area. At the time of the starting the project, such a pseudo-geographic boundary had not been established in scholarship centered on Twitch, and it seemed important due to Toronto’s prominence as a hub for major-studio and independent game development in Canada. (Ubisoft has a major office in Toronto, for example, and the city has numerous non-profit organizations, such as Dames Making Games and the Hand Eye Society, that support independent game development.) Furthermore, we decided to recruit only game developers who shared both their screen and webcam as they composed live with audiences on Twitch. This attention to the whole livestream reflects what Jennifer Bay and Patricia Sullivan have called studying embodied sensemaking. Similarly, as Andrea Olinger writes of studying and interviewing videotaped participants, “their embodied actions seemed to be co-constructing the verbal utterances and providing information about their experiences that was not solely available through their talk” (177; emphasis added). For our remote project, analyzing the developer speaking on-stream about their displayed practices reflected the in-person interviews and composing-aloud protocols of Odell, Goswami, & Herrington’s study. As they write about studying welfare workers, interviewing and asking participants to compose aloud “can be an excellent way at getting at the generating, planning, and organizing activities that make up a large part of a writer’s composing process” (231). While preparing for interviews was a challenge, finding streamers who compose aloud or display their work was not because it’s a core expectation and practice of the Twitch community

Our experiences in May 2021 are illuminated by the following video narrated by Anika.

Video #1: Getting Familiar with the Platform

Audio	Visual
Anika: Getting Familiar Audio track: Sayera by Blue Dot Sessions	Text: Getting Familiar with the Platform
Anika: We identified and selected a number of game developers who livestream their process on Twitch. Anika: We spent a month exploring Twitch streams and deciding who to interview.	Video: A grid of screens showing the Twitch platform and developers we observed. The visual is animated so the screens show up left to right.
Anika: One of the challenges I faced at the beginning was getting used to the language of the online world.	Text: First Challenge Graphic: planets Animation: messaging online
Anika: Studying different platforms, you’re diving into a completely different world of online interactions. I remember watching the first couple of Twitch streams...	Text: “Online Languages” and “Requires studying the interaction and learning the jargon” Video: A screen recording of a game developer sharing his screen with a game and text. The text reads ”Run Game.“ Used with permission by Mikhail Nilov from Pexels under a Creative Commons license.
Anika: And there was a lot of jargon being used, such as “bits” and “raids” and at first I had no idea what the viewers in chat were talking about.	Animation: people talking with messages popping up around them. Graphics: question marks flying on screen.
Anika: But by spending more time on these platforms and interacting and engaging...	Video: Hands typing on a computer. Used with permission by Dimitar Dimitrov from Pexels under a Creative Commons license.
Anika: we tend to pick up the language pretty easily.	Text: Solutions and Interacting + Engaging Animation: Clock with spinning hands.
Anika: After having watched a couple of streams, and getting used to the nature of Twitch, I was definitely a lot more comfortable when it came to understanding it.	Video: A woman working at a computer, with their back to the camera. Used with permission by Ron Lach from Pexels under a Creative Commons license.
	Text: This practice was great for getting acquainted with rhetorical commonplaces on Twitch.

Challenge #2, June 2021: Managing the Temporal Limitations of Twitch and Files for Analysis

Before we recruited game developers for interviews, we viewed and analyzed developers’ public Twitch livestreams for moments of interaction between developers and audience in the live chat. As a team of three, we started by selecting three developers each and discussing their streams in our weekly meetings. We paid close attention to what developers did on screen as well as how they interacted with live chats, annotating live-chat transcripts with the streamers’ actions and on-screen work. As Crystal VanKooten writes of video-based interviews for understanding first-year writers’ meta-awareness, “notes about participants’ body movements and facial expressions were added to the transcripts when significant” (Identifying). Streamers’ actions, as displayed on webcams and shared screens, are significant for the transcript and subsequent interview; we were concerned with the ways in which their actions shifted as they spoke aloud and interacted with audiences. To invoke Odell, Goswami, & Herrington, “this procedure [of studying writers as they compose aloud] elicits certain types of information that cannot be obtained through the interview procedure... (232). In other words, the interview is a means to elucidating the information obtained through video recordings.

Table 1 shows our note-taking practice for studying a stream in which a Toronto game developer was testing a game during the annual Toronto Game Jam (TOJam). Even though this stream was broadcast in public on Twitch for two weeks and thus did not require informed consent by our institutional review board (i.e., because it didn’t require a sign-in and was public), we have still removed any identifying information of the participants because the broadcast was eventually deleted.

Table 1. Sample Note-Taking Grid for Studying Game Developers on Twitch

Stream title	Summary of Interaction between Streamer and Chat	Chat Transcript Excerpt (paste here)	Streamer Action/Response (summarize or paraphrase what was said)
[Toronto Game Jam event]	Tech difficulties	20:13 [User]: I hear music 20:25 [User]: i hear really loud music lol 20:52 [User]: sounds good to me! 21:29 [User]: I love the jump sounds 21:33 [User]: Ah, this is a lot better with sound!	Apologizes while grimacing; continues to play with volume a bit; talks about the frustrations with lagging on stream.

After taking notes on a number of developers’ streams, we began contacting developers for interviews. As this was a rather short summer project that was exploratory in nature, we emailed five developers and completed three interviews in July and August. Any developers we recruited for interviews gave us informed consent to archive their streams and reference their work and streams in future projects.

Working Through a Threefold Challenge of Twitch’s Archival Approach

When studying textual-visual data like that of Twitch livestreams for later recall, we faced the temporal limitations of past livestreams archived as “Video on Demand” on the platform. Unlike YouTube, Twitch archives streams as VOD for two weeks, unless the broadcaster disables the feature or extends the archive timeline to 60 days because they are a Twitch “partner” or Amazon Prime member (see Video on Demand and Twitch Partner Program). In order to reference past streams throughout the summer as we conducted interviews based on the discourse stemming from those streams, then, we used a third-party platform called TwitchDownloader to download VODs and any associated live-chat transcripts from the streams.

This archival approach might strike some readers as toeing the line between ethical and unethical data collection, and for that we offer a disclaimer. We made the participating developers aware that we were referencing their streams, and we had no intention of profiting from or recirculating any past streams in private or public contexts. Working on the typical two-week timeline of the VOD archive would diminish the goal of studying discourse-rich streaming practices before, during, and after an interview with a developer. The research timeline for DBIs can and should unfold over several sessions and/or months to allow for iterative coding and review. In the conclusion of Update Culture and the Afterlife of Digital Writing, John Gallagher comments on similar temporal constraints when he studied digital writers. “Frustrating at times for me, participants changed their text without any evidence of the change. Interfaces also changed over the course of the project, with new ambient affordances arising and older ones fading into the dustbin of digital history” (161). His solutions for capturing those changes? Screenshots and screen recordings.

However, the consequence of archiving hours-long VOD is that they produce many gigabytes of video and large text files. VODs ranged from three to nine gigabytes apiece. As a research team working remotely, we found it challenging to share those files without clogging up local hard drives and free Google Drive accounts—let alone downloading them over networks. Seeing the effects on the laptops and Google accounts of Jessica and Anika, Rich invested in extra storage space and used the open-source program Handbrake to reduce the file sizes of VODs. But here’s the interesting part: we eventually realized that downloading and watching a VOD in full was another time sink, especially for Jessica and Anika’s hourly commitments to the project. An optimal solution was to scan the VOD live-chat transcripts for moments of audience-development interaction and then refer to the video segment for any effects the comment had on the developer’s production.

Our experiences in June 2021 are illuminated by the following video narrated by Rich and Jessica.

Video #2: File Management and Analytics

Audio	Visual
Jessica: File Management and Analytics Audio track: Sayera by Blue Dot Sessions	Text: File Management and Analytics /York University / Dean’s Award of Research Excellence
Rich: Before we recruited creators for interviews, we analyzed Twitch streams for moments of interaction and invention between creators and viewers in the live chat.	Video: A Twitch streamer working on a game.
Rich: However, unlike YouTube, Twitch streams are typically archived for two weeks and then evaporate.	Text: Temporal Challenge for discourse-based research YouTube = content stays on site until the creator removes it. Twitch = content only available for 2 weeks post-publication Graphics: YouTube logo and Twitch logo
Rich: They’re gone.	Blank Background
Rich: Our workaround was to screen capture videos and download live chat transcripts to generate interview questions	Text: Solutions / Step 1 - Screen Capture Video / Step 2 - Download Transcript / Step 3 - Come Up with Interview Questions / as was approved by our university’s ethics board Graphics: simple line depictions of video, the person in a headset, and documentation with a gear at the bottom right corner
Rich: But that meant storing and sharing gigabytes on gigabytes of videos as well as thousands of words.	Text: Storage / Google sheets / Google documents / video downloads + uploads Graphics: two people at a desk looking at multimedia platforms
Rich: Together they clogged our local computers and free cloud drives.	Video: coding
Rich: On my end, as the faculty, I invested in extra storage space and eventually found a program that downgraded videos to less than 200 megabytes apiece.	Text: Investing in storage / Downloading helpful programs Animations: money in hand / downloading symbol from cloud
Jessica: Coding and organizing all of the data from these live streams was a bit tedious. I found the best way to collect meaningful notes was to read through the entire chat transcript,	Text: Coding / Step 1 - read through entire transcript / Step 2 - highlight possible moments of interaction / Step 3 - observing time-stamped video Graphic: computer screen with moving elements
Jessica: look for possible moments of interaction and pull up the timestamp of the video to see what effect the comment had on the production or outcome.	Video: screen recording of looking through document and highlighting + looking for that moment in the video
Rich: Despite these material challenges, Twitch was still an incredible platform for studying rhetorical practices.	Text: Despite these material challenges, Twitch was still an incredible platform for studying rhetorical practices.

Challenge (or Recommendation?) #3, July 2021: Adapting and Scaling the Project for Time and Developer Agency

In this section, we discuss our behind-the-scenes work for a DBI with Attila “Gabriel” Branyiczky, who livestreams his game development practices under the name BluishGreenProductions on Twitch. In July 2021, Branyiczky was livestreaming the development of his game Worlds Within Worlds, a two-dimensional “platformer” in which you move through worlds to “collect Golden Leaves to restore Trees” (Bluish-Green Productions). (The game is free to play, but donations help future developments.) To anticipate our previous challenges of archiving and studying VODs remotely, and to prepare Branyiczky for discussing and re-considering his rhetorical choices while on the livestream, we did an introductory interview outside of his Twitch streaming hours and then a second interview during one of his livestreams. (See Appendix 2 for sample questions asked during the introductory interview.) Anika and Rich interviewed Branyiczky over Zoom, and then Rich joined Branyiczky a few days later on his Twitch stream. Jessica and Anika watched the stream’s video on demand as the project progressed.

The first interview was a chance to give Branyiczky a better sense of how DBIs have worked for the research team. To begin, we sent a recruitment email, noting that we have been following his stream for two weeks. For the team, a DBI is dependent on gathering public background information (e.g., streams) about possible participants before recruitment. This first step informs our language and ethos in the recruitment email, and it helps build context for future DBIs with livestreamers. Once he agreed to the interview, we prepared a questionnaire about general composing practices and practices specific to his work. During the interview, we asked him about his practices as displayed on recent VODs. Listen as Anika and Branyiczky talk about his rationale for engaging with live chat on his stream versus working on Worlds Within Worlds. His switching between the two tasks is something we wanted to learn more about—something that wasn’t quite clear by observing Branyiczky’s stream alone. Put differently, this switching was Branyiczky’s “inexplicit functional knowledge” that he draws on when he streams (Odell et al. 221). It was a means to understanding how critical the audience engagement is to his stream, and why he chooses not to simply compose on screen in lieu of chatting with audiences.

Audio Clip

Anika: Do you think that, as a creator who livestreams their work, the ability to multitask is necessary?

Branyiczky: I think that if you are going to be livestreaming your work, the whole point of doing a livestream is to engage with an audience. I don’t find myself particularly engaged when a livestreamer is not paying attention to the chat. It’s the same way if you’ve ever been at a gathering and maybe there’s a bunch of people around and everyone’s kind of having like a little bit of a conversation. You hear something that someone says and you say, “Oh, I’ll make some funny comment,” and then no one reacts and it just feels awful, like “I’ll just pretend that I wasn’t here, then.” I don’t know if it’s the exact same thing, but it’s the first thing that came to my mind. It’s this feeling you get when you are putting yourself out there. I’m comfortable as a straight white male to have myself broadcast on webcam at all times, but I know that a webcam presence is not strictly necessary and I can understand why people might not be comfortable for a variety of reasons to have their image up on the stream. But, I think that the engagement with the audience is much more important than having a webcam feed. Being on top of the conversation that you’re participating invites people to converse with you and that’s the whole point of the Twitch platform. If you’re doing a livestream, I think that’s just sort of part and parcel with it because otherwise, you could just be recording yourself and uploading archives of it to YouTube. The whole point of these live streams is to be engaging with an audience. That’s my take on it, at least, and it’s what I find the most engaging. Maybe there are going to be other people who are just cool with seeing the process, but again, in that case, I'd be just as happy watching the stream archive or the backlog on YouTube down the line. I understand why it’s not always possible to keep on top of those things and I know that sometimes I get into a conversation with someone in the chat and completely neglect what I’m doing. I could be working on my game ostensibly but we end up having a five-minute conversation about what our favorite enemy is and I get absolutely nothing done on the screen. Nothing’s happening, because I’m just here engaging with the person in the chat. (Branyiczky, Zoom)

Branyiczky’s conversation with Anika helped us understand the significance that he places on audience interaction as a rhetorical practice. Without prior observation of his composing aloud on stream, we likely wouldn’t have brought out his tacit knowledge and associated feelings about audiences. Indeed, this two-interview approach layered with observation proved to be quite effective. During the interview on Twitch a few days later, Branyiczky shared his webcam and his screen, showing code, design documents, and gameplay, all the while chatting with and being interrupted by audiences before, during, and after the interview.

Methodologically speaking, this approach to the DBI was optimal because it collected production artifacts and multimodal interview data in-situ on a livestream Branyiczky controlled. Here is a video clip of Rich and Branyiczky discussing his archive of levels scraped from the game. The archive is a source of invention, a source Branyiczky refers to on the stream. (Readers will notice that the audience chat is not included in this recording because they didn’t consent to the interview being recorded. However, we will encourage readers to check out Branyiczky’s recent streams for chat-based conversations and his interviews with fellow developers.)

Video Clip

Rich: Do you have levels that you’ve archived, but haven’t actually made it into the game? I’m just thinking about... if you have so much fun iterating in this level design. Is that part of your process? For me, I work in [the writing program] Scrivener a lot of times and I just have scraps of chapters that are just probably won’t go anywhere unless someone hacks my computer. But I find that iteration... and I like to keep it. It’s almost the garbage I like to think of, and it could be productive at some point.

Gabriel: Yeah, absolutely. And I do, I have said this as well on stream, but this what I have open here is, and it’s obscured by my webcam in the corner there, but there’s a massive list of all the levels that I have in the game that are in the engine, I should say. And some of them are scrap. Some of them are original versions of others that I’ve redone since. But there is a solid, I think I’ve had built a full 20 levels in this game already. And I think some of those are great levels, but I’m holding myself to a different standard now that I’m rebuilding them for Worlds Within Worlds. So, I’m literally taking ideas from these original levels. (Branyiczky, Twitch)

Overall, this approach to conducting a two-part DBI before and during Branyiczky’s stream addresses Olinger’s concerns of consent in the video-interview process. As she writes, “participants cannot make informed decisions if researchers do not communicate why they prefer video-recording, do not show how images and videos can be anonymized, and do not discuss the different options with participants” (196). Branyiczky’s options for the Twitch interview were vetted well before we went live on his stream. Furthermore, it meant that we had established a relationship with the developer and his audiences, and the topics we discussed were perhaps relevant to audiences who are interested in game development. In addition, opting to do two interviews meant that Jessica and Anika could participate in the DBI without doing a live, uncut interview in public on Twitch and outside of typical working hours (i.e., Branyiczky streams most evenings after 5:00 pm). While Branyiczky’s audience was small at the time of the interview, it was public in the sense that anyone could join the stream and comment nefariously in a way that has affected many women and content creators in general (see London et al.; Ruberg). A secondary but important goal of this project was to give Jessica and Anika time and practice for conducting interviews and analyzing public data on Twitch—not working as a streamer.

Our experiences in July 2021 are illuminated by the following video narrated by Jessica. Since our interview in July 2021, conversations between Rich and Branyiczky have continued on Twitch, with Rich contributing under the username “theerhetoric.” In fact, video #3 demonstrates another follow-up conversation with Branyiczky in which Rich presents an alternative idea about instructions for his game Infinite Descent. Branyiczky, in response, refers to the game’s speed and history to justify the lack of instructions on screen.

Video #3: Interviewing a Game Developer

Audio	Visual
Jessica: “Gabriel” Branyiczky case study Audio track: Sayera by Blue Dot Sessions	Text: Attila “Gabriel” Branyiczky Case Study
Jessica: Gabriel live streams games under the name BluishGreenProductions on Twitch.	Video: Gabriel saying hi and opening his stream
Jessica: We thought Gabriel was a good candidate to interview because while he was hosting the TO Jam 2021 livestream...	Video: A video of Gabriel displaying code and games on his screen.
Jessica: he was able to communicate with the live chat while also commenting on the game and offering the developer tips.	Graphics: Branyiczky depicting engagement and idea making
Jessica: We did an introductory interview outside of his Twitch streaming hours and then a second interview during one of his live streams.	Text: Methods First interview: outside of his Twitch streaming hour Second Interview: during one of his live streams Video: Gabriel during Twitch interview.
Jessica: The latter proved to be quite effective, as Branyiczky shared his screen during the interview, showing code, design documents, and gameplay, all the while chatting with viewers. His tacit knowledge was on full display and open to public inquiries and conversations.	Video: from the live interview, showing some of the code/design documents/etc.
Jessica: Since our interview, Branyiczky has continued interacting regularly with viewers on his livestream. Here is another example of Rich interacting with Branyiczky about another game, Infinite Descent. Branyiczky is presented with an idea for adding instructions at the beginning of the game, then he explains his decision for not doing so. For privacy purposes, we’ve excluded other comments by viewers.	Video: from the live interview, showing some of the code and webpage for game Infinite Descent. The left side of the side shows a question from a viewer and reads: “Random thought: Does Infinite Descent need instructions for playing? It has a lot of space in that first fall. Maybe text there?”
Jessica: Because there are in-real-time comments and questions coming from viewers, the researcher can get a better understanding of the developer’s tacit knowledge.	Text: In-real-time comments and questions coming from viewers, the researcher can get a better understanding of the topic at hand.

Overall Recommendations: Staying Small and Going Deep

After narrating such practices and associated challenges behind our Twitch research project, we close with two recommendations for future undergraduate researchers and faculty researchers mentoring them: stay small and go deep.

Admittedly, Rich went into this summer project about Twitch with little to go on other than exploratory questions that would shed light on his ongoing research. Twitch streams and streamers’ practices are multivalent and complex, so future summer projects might start smaller and with more direction. Before the summer DARE project began, for example, Rich could have identified and recruited one or two regular Twitch streamers who develop games, and then he could have invited Jessica and Anika to start following those streamers and their livestream discourse for a month before we began contacting streamers. It would have allowed Jessica and Anika to ask a number of questions about Twitch’s features and streamer-specific discourse while developing questions for potential interviews.

As Jeff Grabill and Stacey Pigg have argued, the early stages of (exploratory) qualitative research on online writing can be messy “because the interactions (as text) are persistent in time and space and non-linear in terms of when and how participants engage” (99). Perhaps scaling down a summer project to just two case studies that unfold over three months could help researchers better put some boundaries around messy discourse and the participants involved. As Olinger demonstrates in her study of two faculty writers’ contradictory views on “good writing,” focused case studies can lead to deep analyses—in her case, analyses that “reveal cracks in the autonomous model of literacy that would not have emerged otherwise” (21). From Rich’s perspective, undergraduate researchers need not worry about producing generalizable results from DBIs like our aforementioned study of Branyiczky’s work. In a summer project, they should come away with a solid, experiential-based understanding of how DBIs reveal “writing beyond text and inscribed discourse and... writing as something fluid and changeable” (Gallagher 162) in digital environments. After all, in 1983, Odell, Goswami, & Herrington acknowledged “[i]t is unlikely that a single methodology—in effect, a single perspective—will ever tell us all we need to know” (234). In the context of Twitch, DBI approaches that mix interviews and composing aloud protocols (i.e., commenting on their work in some way) might embrace the fluidity of the platform’s communities. That is, future researchers who combine these methods might trade generalizable results for deep case studies that question “the strategies a [streamer] uses to the solve the unique problem presented by each [composing] task, more specifically the way context-specific knowledge is combined with more global-[composing] strategies to solve these problems (Odell et al. 235; brackets added). In fact, Odell, Goswami, & Herrington posit that such research methods could support inexperienced workers because the methods help foreground their tacit knowledge. More importantly, though, we think small DBI projects associated with Twitch could support novice researchers as they explore rhetorical practices in nonacademic settings. This essay has demonstrated what students like Jessica and Anika could learn. For a final refrain, we encourage future researchers to consider Twitch as a low-barrier entry point to developing “a repertoire that includes interviews, composing aloud, analyses of written products, and videotaping writers while they are writing” (Odell et al. 235) over long periods of time.

Our final thoughts are best illuminated and summarized by the following video narrated by the research team.

Video #4: Final Thoughts

Audio	Visual
Jessica: Final Thoughts on Twitch and Discourse-based Interviews Audio track: Sayera by Blue Dot Sessions	Text: Final Thoughts
Rich: As the faculty researcher supervising this project, I want to start with the first of two recommendations for future researchers working with Twitch and conducting discourse-based interviews: stay small.	Text: Faculty Researcher Recommendation: Stay Small Video: Fish swimming
Rich: Twitch streamers are multivalent and complex, so future projects might start smaller—and earlier.	Graphics: The words “Smaller” above a pound sign, and the words “Earlier” a calendar icon
Rich: Before the summer DARE project began, I could have identified two, maybe three, regular Twitch streamers who compose and develop games live, and then I could have invited [students 1 and 2] to follow those streamers for a month or so.	Text: Solution Step 1: Identified a couple of regular Twitch streamers who compose and develop games live Step 2: Invite Jessica and Anika to follow those streamers Graphics: A computer tablet with blue and white shapes A laptop screen with shapes that look like a website.
Rich: The earlier stages of research can be messy, so perhaps scaling down a summer project could help researchers better navigate that messiness and go deeper with case studies.	Video: A messy table with coffee and crumpled paper.
Jessica: Having been a research assistant for Rich for a similar project observing feelings and emotions of digital creators,	Video: Screen recording from last summer/previous work with Rich Text: DARE 2020 (or year/month)
Jessica: I can definitely see a pattern in doing this kind of qualitative research.	Video: screen recording form this year Text: Month
Jessica: You start just collecting and observing and taking a ton of notes.	Video: People working together/lots going on around them
Jessica: From all this data and notes, you need to find the meaningful bit	Graphic: diamond popping
and get deeper into those aspects.	Graphic: digging
Anika: Being a part of this research project gave me hands-on experience with the information overload that is inherent within digital media.	Video: person working at a computer
Anika: We were able to collect so much data right from the jump,	Graphic: race car starting
Anika: but we didn’t really know what we were looking for yet. Without having a clear focus, we spent a lot of time taking a “blind research” approach.	Graphic: Animated figures talking and asking questions
Anika: The best approach for anyone studying the digital world is to identify a small number of research focuses right at the beginning and hone in on them.	Text: Studying digital worlds Graphic: Magnifier

Appendices

1. Appendix 1: Shivener’s Project Description for the Dean’s Awards for Research Excellence 2021 Program
2. Appendix 2: Sample Recruitment Email and Interview Questions for a Game Developer

Appendix 1: Shivener’s Project Description for the Dean’s Awards for Research Excellence 2021 Program

Welcoming an undergraduate into remote, qualitative research, this DARE project is part two of studying digital content creators and feelings that imbue their production processes. Digital writing encompasses authors who experiment with digital media (sound, video, code) to produce interactive projects: multimedia-rich books, video games, infinite-scrolling comics, to name a few. Last summer, part one examined the processes of independent game developers through interviews and textual anaylsis. This summer picks up where we left off and addresses the affordances and constraints of remote production, focusing more on the work of independent creators who live-stream content and build communities on YouTube, Twitch, Discord, etc. in light of COVID-19. Of particular interest are creators whose work addresses mental health, anti-racism, and other topics overshadowed by mainstream studios. Approved by the university’s Research Ethics Board, this ongoing study is significant because it foregrounds practices and feelings often relegated to the background of an interactive project (i.e., such items aren’t discussed or reflected upon in the project itself). The visibility of such process work is significant in light of rising publications (The Atavist, Digital Humanities Quarterly) and activist movements (#BlackLivesMatter) that have leveraged digital media to distribute and circulate texts. We need more “practitioner stories” (Ridolfo) of how media-rich texts are composed and the feelings that authors work through from inception to final delivery of a text. By helping make visible the practices and feelings vital to interactive projects, the research assistant will contribute to rhetorical studies scholarship and build on their digital writing practices.

In consultation with the professor, the student will receive training and assist with low- and high-level responsibilities related to an ongoing research project on the rhetorical practices and feelings of creators (writers, composers) of digital texts. Designed to help the student develop their research skills, these responsibilities will build on each other and are dependent on the student’s research interests:

Low-level

• Assisting with a database of contact information of writers and creators

• Viewing and archiving audience commentaries in public settings (YouTube, Twitch).

• Assisting with scheduling of interviews and archiving of interviews (for transcription)

• Cleaning up transcriptions produced by the research team and/or artificial intelligence (e.g., Otter AI)

• Conducting literature searches and uploading articles and texts (through shared folder) related to interactive storytelling, live-streaming, and rhetorical invention

High-level

• Composing a comprehensive literature review

• Analyzing online audience commentary, with particular attention to invention (helping a creator with an idea, offering feedback, etc.)

• Drafting a presentation, and, if possible, an article manuscript at the end of the DARE term

Appendix 2: Sample Recruitment Email and Interview Questions for a Game Developer

My name is Rich Shivener, a professor at York University in Toronto, Canada and Jessica Da Silva, an undergraduate working this summer on games research. I’m working on a series of academic articles and a book on independent games creators’ practices and reflections. Would you be willing to speak with me about your work on [game] and streaming it live on Twitch? I’m working on this project over the next several months and through the summer. Essentially, I’m conducting interviews in hopes of highlighting cool work by diverse games creators, also connecting academia to industry.

Sample Questions

• If your work has a public URL, please list it here:

• What sorts of digital media work do you do/have you done?

• Could you talk a bit about the project we have selected for this interview?

• What genre or form of writing did you work on most during TO Jam (Toronto Game Jam)? Why?

• How much time have you spent on this project (from conception to working on final publication)?

• If this text was reviewed by playtesters, beta testers and audiences before publication, what did the reviewers say about drafts?

• Related: How much value do you place on streaming live on Twitch, using Discord, etc?

• What influenced you to begin live-streaming your work?

• You mentioned in one of your streams that though you did work in teams before, you do not do so anymore lately. Is that because of the pandemic? How would you say remote work environments have changed your work processes - for the better, for worse, or a combination of both?

Works Cited

Bay, Jennifer, and Patricia Sullivan. Researching Home-Based Technical and Professional Communication: Emerging Structures and Methods. Journal of Business and Technical Communication, vol. 35, no. 1, 2021, pp. 167-173.

Bluish-Green Productions. Worlds Within Worlds by BluishGreenPro. Itch.io, https://bluishgreenpro.itch.io/www

Bradshaw, Jonathan L. Slow Circulation: The Ethics of Speed and Rhetorical Persistence. Rhetoric Society Quarterly, vol. 48, no. 5, 2018, pp. 479-498.

Branyiczky, Attila Gabriel. Zoom interview. Conducted by Anika Rahman and Rich Shivener, 5 July 2021.

---. Twitch interview. Conducted by Rich Shivener, 7 July 2021.

CCCC Position Statement on Undergraduate Research in Writing: Principles and Best Practices. Conference on College Composition and Communication, March 2017, https://cccc.ncte.org/cccc/resources/positions/undergraduate-research.

Eriksson, Andreas, and Lene Nordrum. Unpacking Challenges of Data Commentary Writing in Master’s Thesis Projects: An Insider Perspective from Chemical Engineering. Research in Science & Technological Education, vol. 36, no. 4, 2018, pp. 499-520.

Gallagher, John R. Update Culture and The Afterlife of Digital Writing. University Press of Colorado, 2020.

Grabill, Jeffrey T., and Stacey Pigg. Messy Rhetoric: Identity Performance as Rhetorical Agency in Online Public Forums. Rhetoric Society Quarterly, vol. 42, no. 2, 2012, pp. 99-119.

Handbrake. Handbrake. GitHub, https://github.com/HandBrake/HandBrake.

Johnson, Mark R., and Jamie Woodcock. ‘And Today’s Top Donator Is’: How Live Streamers on Twitch.tv Monetize and Gamify Their Broadcasts. Social Media+ Society, vol. 5, no. 4, 2019, pp. 1-11.

Jomaa, Nayef and Siti Jamilah Bidin. Variations in the Citation Use and Perceptions in Writing the Literature Review by EFL Postgraduates. Eurasian Journal of Applied Linguistics, vol. 5, no. 3, 2019, pp. 441-460.

Kobs, Konstantin, et al. Emote-Controlled: Obtaining Implicit Viewer Feedback Through Emote-Based Sentiment Analysis on Comments of Popular Twitch.tv Channels. ACM Transactions on Social Computing, vol. 3, no. 2, 2020 pp. 1-34.

Lancaster, Zak. Using Corpus Results to Guide the Discourse-Based Interview: A Study of One Student’s Awareness of Stance in Academic Writing in Philosophy. Journal of Writing Research, vol. 8, no. 1, 2016, pp. 120-147.

Lim, Claire. Creators Who Joined Twitch in the Pandemic Plan to Stay. Wired, 12 May 2021, https://www.wired.com/story/creators-comedians-twitch-pandemic-plan-to-stay/.

Lockridge, Tim, and Van Ittersum, Derek. Writing Workflows: Beyond Word Processing. University of Michigan Press, 2020.

London, Tabitha M., et al. Finding Effective Moderation Practices on Twitch. Digital Ethics, edited by Jessica Reyman and Erika M. Sparby, Routledge, pp. 51-68.

Odell, Lee, Goswami, Dixie, and Herrington, Anne. The Discourse-Based Interview: A Procedure for Exploring the Tacit Knowledge of Writers in Nonacademic Settings. Research on Writing: Principles and Methods, edited by Peter Mosenthal, Lynne Tamor, and Sean A. Walmsley, Longman, 1983, pp. 221-236.

Olinger, Andrea R. Visual Embodied Actions in Interview-based Writing research: A Methodological Argument for Video. Written Communication, vol. 37, no. 2, 2020, pp. 167-207.

---. Self-Contradiction in Faculty’s Talk about Writing: Making and Unmaking Autonomous Models of Literacy. Literacy in Composition Studies, vol. 8, no. 2, 2021, pp. 1-38. https://doi.org/10.21623/1.8.2.2.

Pardo, Lewis [Lay295]. TwitchDownloader. GitHub, https://github.com/lay295/TwitchDownloader.

Pigg, Stacey. Coordinating Constant Invention: Social Media’s Role in Distributed Work. Technical Communication Quarterly, vol. 23, no. 2, 2014, pp. 69-87.

Riddick, Sarah. Deliberative Drifting: A Rhetorical Field Method for Audience Studies on Social Media. Computers and Composition, vol. 54, 2019, pp. 1-27.

Riddick, Sarah, and Rich Shivener. Affective Spamming on Twitch: Rhetorics of an Emote-Only Audience in a Presidential Inauguration Livestream. Computers and Composition, vol. 64, 2022. https://doi.org/10.1016/j.compcom.2022.102711.

Ruberg, Bonnie, Amanda LL. Cullen, and Kathryn Brewster. Nothing but a ‘Titty Streamer’: Legitimacy, Labor, and the Debate Over Women’s Breasts in Video Game Live Streaming. Critical Studies in Media Communication, vol. 36, no. 5, 2019, pp. 466-481.

Ruberg, Bonnie. ‘Obscene, Pornographic, or Otherwise Objectionable’: Biased Definitions of Sexual Content in Video Game Live Streaming. New Media & Society, vol. 23, no. 6, 2021, pp. 1681-1699.

Reid, Gwendolynne. Compressing, Expanding, and Attending to Scientific Meaning: Writing the Semiotic Hybrid of Science for Professional and Citizen Scientists. Written Communication, vol. 36, no. 1, 2019, pp. 68-98.

Ridolfo, Jim. Rhetorical Delivery as Strategy: Rebuilding the Fifth Canon from Practitioner Stories. Rhetoric Review, vol. 31, no. 2, 2012, pp. 117-129.

Rule, Hannah J. Situating Writing Processes. WAC Clearinghouse, 2019.

Silvestro, John. In the Loop: Articulating Our Entanglement in Others’ Digital Ethos Positions. Enculturation, vol. 28, 2019.

Sjöblom, Max, and Juho Hamari. Why Do People Watch Others Play Video Games? An Empirical Study on the Motivations of Twitch Users. Computers in Human Behavior, vol. 75, 2017, pp. 985-996.

Swarts, Jason. Recycled Writing: Assembling Actor Networks from Reusable Content. Journal of Business and Technical Communication, vol. 24, no. 2, 2010, pp 127-163.

Taylor, T. L. Watch Me Play: Twitch and the Rise of Game Live Streaming. Princeton University Press, 2018.

Twitch.tv. Video on Demand. https://help.twitch.tv/s/article/video-on-demand?language=en_US.

---. Twitch Partner Program. https://www.twitch.tv/p/en/partners/.

VanKooten, Crystal. Identifying Components of Meta-Awareness about Composition: Toward a Theory and Methodology for Writing Studies. Composition Forum., vol. 33, 2016, https://compositionforum.com/issue/33/meta-awareness.php.

---. Transfer Across Media: Using Digital Video in the Teaching of Writing., Computers and Composition Digital Press, 2020.

Woodcock, Jamie, and Johnson, Mark R. The Affective Labor and Performance of Live Streaming on Twitch.tv. Television & New Media, vol. 20, no. 8, 2019, pp. 813-823.