Uncovering Seams in Distributed Playof Tabletop Role-Playing Games

Andrew M. WebbCWIAmsterdam, Netherlandsamwebb@acm.org

Pablo CesarCWIDelft University of TechnologyAmsterdam, NetherlandsP.S.Cesar@cwi.nl

Permission to make digital or hard copies of part or all of this work for personal orclassroom use is granted without fee provided that copies are not made or distributedfor profit or commercial advantage and that copies bear this notice and the full citationon the first page. Copyrights for third-party components of this work must be honored.For all other uses, contact the Owner/Author.

CHI PLAY EA ’19, October 22–25, 2019, Barcelona, Spain.© 2019 Copyright is held by the owner/author(s).ACM ISBN 978-1-4503-6871-1/19/10.https://doi.org/10.1145/3341215.3356253

AbstractWe uncover how geographically distributed players of table-top role-playing games engage narrative, ludic, and socialaspects of play. Our existing understandings of tabletoprole-playing games are centered around co-located play onphysical tabletops. Yet, online play is increasingly popular.We interviewed 14 players, experienced with online virtualtabletops. Our findings reveal the seams—points wheremedia, activities, and technology intersect—within virtualtabletop environments that enable distributed players toshift among collaborative storytelling, applying game rulesand mechanics, and socially interacting with each other.

CCS Concepts•Human-centered computing → Ethnographic studies;

Author KeywordsDungeons & Dragons; narrativity; virtual tabletop; Roll20

IntroductionTabletop role-playing games, such as Dungeons & Drag-ons, are seeing a resurgence in popular culture [2]. In thesegames, players take on the roles of fictional characters asthey collaboratively create stories while sitting around atable. Players interact with various physical objects, suchas dice, character sheets, books, and maps. They engagein social interactions with each other, such as speaking

in-character, discussing game rules, and making jokes.Tabletop role-playing games originated with and evolvedfor decades around co-located players sitting at physicaltabletops [6, 8, 10, 13]. Today, geographically distributedplayers from all of the world join together to play thesegames via virtual tabletops, such as Roll20 [17] and Fan-tasy Grounds [7]. While co-located play is well studied, welack understandings of how virtual tabletops and distributedplayers affect play experiences.

This research conducts an ethnographic study to betterunderstand how technologies and media support distributedplay of tabletop role-playing games. We first discuss relatedwork, and then present our methodology. We report initialfindings from interviews with active, regular players thatuse virtual tabletops for distributed play. We conclude withimplications of our findings and future work.

Figure 1: The Roll20 interface:tool palette in top-left; virtualtabletop in middle; and chat onright with automated dice rollresults, character ability cards, andplayer messages.

BackgroundWe ground our work in prior ethnographic studies of table-top role-playing games [6, 8, 10, 14]. We discuss HCI re-search that augments experiences of co-located play withtechnology [4, 12, 15], and detail features of Roll20, a com-mercial virtual tabletop used by our participants.

Frames of Experience in Tabletop Role-Playing GamesFine [8] identifies three frames of experience in tabletoprole-playing games: (1) the primary, social framework thatpeople use to make sense of the real world, such as chat-ting about life; (2) the game context governed by gamerules and mechanics, including discussing interpretationof rules or asking players to roll dice; and (3) the gameworld in which players’ characters act, including speakingin-character or describing a character’s actions.

In considering aspects of narrativity in tabletop role-playinggames, Mackay [13] further breaks down the game world

frame into three separate frames based on narrative, con-stative, and performative types of speech. Cover [6] arguesthat constative and performative speech are both used toform the narrative, and therefore Mackay’s three framesconstitute a single narrative frame.

Building on Fine and Mackay, Cover identifies three framesof experience with differing (decreasing) levels of narrativity:

• Narrative frame: “players create textual narrative world.”

• Game frame: “players engage in game play and are im-mersed in the game world.”

• Social frame: “players interact in a social setting.”

Fine, Mackay, and Cover point out how players fluidly shiftamong these different frames. At times, frames are short-lived, multiple shifts occurring even within the same sen-tence. Other times, frames are stable, such as when play-ers, using their characters’ voices, engage in conversa-tion that forms part of the narrative. Their discussions offrames center around spoken language. Our study uncov-ers aspects of how technology and media help shape theseframes and the shifts between them in distributed play.

Augmenting Tabletop Role-Playing GamesPrior work on augmenting play experiences develops hy-brid, mixed-reality interfaces [1] that weave together phys-ical and virtual representations for co-located play. TheSTARS platform combines digital devices that show as-pects of the virtual world, such as maps or character de-tails, with physical artifacts already present in these games,such as dice and figurines [15]. Lindley and Eladhari con-ceive trans-reality role-playing games that support com-bined play across live action, tabletop, and computer-basedgames [12]. WEARPG incorporates wearables to supportmovement-based play in tabletop role-playing games [4].

Roll20: Virtual TabletopWe selected the popular virtual tabletop, Roll20, for the fo-cus of our investigation. The basic Roll20 interface includesa gridded 2D canvas representing the ‘tabletop’, a text chat,and a tool palette (Figure 1). The canvas provides layersfor compositing visual media, such as maps and charactertokens. Text chat contents include outcomes for automateddice rolls, cards describing character abilities, and playermessages. Players can emote their characters state or pri-vately message another player using commands in the textchat. The tool palette provides a set of tools for interactingwith the canvas, including panning and zooming, movingtokens, measuring distances, and drawing annotations.

Id GM orPC

Games

P1 GM D&D 5e,Stars WithoutNumber

P2 PC D&D 5e

P3 GM D&D 5e

P4 Both D&D 5e

P5 Both D&D 5e

P6 GM Shadowrun

P7 Both Shadowrun,Call ofCthulhu

P8 GM D&D 5e

P9 Both D&D 5e

P10 GM D&D 5e,Werewolf

P11 Both D&D 5e

P12 GM Pathfinder

P13 Both D&D 5e

P14 PC D&D 5e

Table 1: The play roles ofparticipants—game master (GM),player character (PC), orboth—and the games discussed.

MethodWe conducted a qualitative study, interviewing players oftabletop role-playing games with regular experience (atleast once every 1–2 weeks) in distributed play.

ParticipantsWe recruited fourteen participants (Table 1) through a sur-vey posted on the Roll20 subreddit [18]. Twelve playersreported playing as the game master or GM—a specificrole in popular tabletop role-playing games where the playeracts as referee and orchestrator of the collaborative sto-rytelling experience, setting scenes, creating tension, andreacting to the actions of other players. Additionally, six ofthose twelve played in games as an individual characterrather than the GM. Two players never played as the GM.

ProcedureWe conducted fourteen individual interviews. We usedMackay’s [14] variation of Flanagan’s critical incident inter-view technique [9], in which we ask participants to recounttheir most recent play session to draw out specific instancesof their experiences. This technique favors detailed descrip-

tions of defined situations over global statements aboutgeneral use, grounding the participants in their experiencesrather than asking them to speak generally or abstractlyabout their practice. Interviews lasted 55–105 minutes.

Data CollectionInterviews took place using participants preferred telecom-munication software (e.g., Skype, Discord). We recordedvideo and audio of the interview using OBS [16]. We col-lected approximately 17 hours of video in total.

Data AnalysisWe transcribed interviews. We then performed an initialopen coding for thematic analysis [3] of the transcriptions.While a number of codes emerged, we selectively reporton those related to frames of experience and the shifts be-tween frames as mediated by computing technology.

Results and DiscussionOur findings identify seams occurring as virtual tabletopfunctionality, media, and player appropriations of the twointersect with frames of experience. Chalmers defines aseam as a “break, gap, or ‘loss in translation’ in a numberof tools or media designed for use together as a uniformlyand unproblematically experienced whole” [5]. While manyof the tools and media that participants report using werenot designed for use together, participants created ‘setups’that enabled combined use of the tools and media to sup-port holistic play experiences. Perhaps more appropriately,Kuznetsov et al. [11] define seams as “points at which dif-ferent materials, practices, categories, etc., intersect, some-times in unexpected ways.”

Narrative FrameIn distributed play, we find that the narrative frame connectsacross voice communication, music, text chat, and virtualspaces. All participants report using both voice and text

to speak in-character. Most participants (10) report usingDiscord, a free communication tool for gamers that com-bines calling features of Skype with text chatting featuresof Slack. While play typically occurs in a channel with allplayers, on certain occasions, several players may sepa-rate from the group into a private channel. For example,half the participants, as a GM, report taking a player asidewhen that player’s character goes on a solo endeavour,which is outside the purview of the other players’ charac-ters. This separation keeps the solo endeavour a secretfrom the other players, and it allows the other players tochat with each other without disrupting play. This actioncreates a hard seam in audio communication and narrativewhere players are separated across channels without anyfeedback of what is happening in other channels.

Figure 2: Example of playersintegrating memes into spell orability cards presented in Roll20text chat.

Most participants (8) report using sound in their games.They describe using Roll20’s Jukebox, which enables syn-chronized playback of music and sound effects. Participantsreport how music and sound effects add to the immersiveexperience. They find the integrated ability to play musicand sound effects in Roll20 fairly seamless. Several partic-ipants (3) express dissatisfaction when a convenient audiosharing service was recently removed from the platform.Music and sound effects can serve the narrative, drawingplayers attention towards aspects of the story.

P3: “There were some undead fighting theparty, and there was a mage that was hurt andbleeding behind them. And so, when they werefighting, at some point I played a sound overthe music, ‘Please, help me! Please, help me!’So, they basically heard this. They were like,‘Holy shit! What’s going on?’ Then, they looka little further, and they saw there was a manon the ground. That’s why the sound system inRoll20 is amazing!”

All participants report using the text chat within the virtualtabletop. They role play their characters in this text chat.They use commands (e.g., /em in Roll20) that allow play-ers to emote (in third-person) about their character’s actionsor attitudes towards other characters. By using these com-mands, players separate narrative discourse from socialmessages and rules discussion in the text chat.

P6: “Some of my players are ardent text roleplayers, so when the they needed to conveythat their character was sending a text mes-sage to somebody, they would do /em and thensay, character name, colon, the text that theyare sending, sent to x character. And it justlooks significantly nicer than just putting it in theregular chat. And plus, then it’s actually beenuseful for me, as a GM. Having multiple char-acters be texted by multiple other characters,I can /em and show that this is an in-universething that is happening.”

All participants describe using private messages to com-municate information directly to other players. GMs usethis to reveal details to one player that their character wouldknow without informing the other players. Players use it tokeep their plans hidden from the GM. Roll20 supports pri-vate messaging via a chat command (/whisper) where theplayer must specify the target and message. Most partici-pants (10) report using the whisper command. It requiresissuing the command with every message. Alternatively,Discord supports private conversations, which participantsfavor when having lengthier discussions.

Participants report creating multiple text channels (sup-ported in Discord) for role playing, including ones for speak-ing in-character, and discussing the lore of the game world.

P2: “We’ve got an in-game chat text channel,where if we are not actually playing the gameand somebody wants to say something in char-acter, we do it within that chat channel. And,that captures stuff like, ‘Hey, when we wakeup from our short rest here, how about we alltravel to this next village.’. . . Another channel islore questions. In here, we had questions aboutwho the rulers of the area were [and] seeinga lot of troops on the road. It’s not really in-character, but it’s specific to what the GM hasput before that we should be able to figure out.”

GMs report creating virtual spaces to represent differentlocations in the game world. In Roll20, they use a featurecalled Pages. Pages are presented in Roll20’s 2D canvas.A hard seam exists between each Page. Page transitionsare initiated by the GM with a visible load screen for allplayers. These transitions can function towards narrativetension, indicating to the players that an encounter (poten-tially combat-related) or something else of importance isabout to occur (see P9-A quote).

P9-A: “We had a encounter acouple of sessions back wherea dryad was in a particular tree.During the night when theywere sleeping, I rolled that thisdryad would do something. SoI usually I just say, ‘Yeah, okay,night goes on, eventually.’ So,when I say, ‘Yeah, so we go tothis map’, this player who is justsitting watch, then they knowthat something is going to hap-pen. So I dragged them overto this map that had a forestclearing with some trees to ei-ther sides. And I had put somedryads on the GM layer that theplayers couldn’t see those inthose trees. And I just askedthem to describe what theywere doing when they were set-ting up the camp or that kind ofstuff. And eventually someonesaid something along the lines,‘Yeah, I’m going to chop somefirewood.’ So I kind of made upthis story that the dryads wereangry that they were stealingthat woods and burning it. Soduring that night, they weresitting around the campfire withone of them, keeping watch.And then I just bought thosedryads from the GM layer ontothe player layer so they couldsee them.”

Social FrameThe social frame connects across voice communication,text chat, and video. Participants report engaging in conver-sations about their lives, often at the start of a session. Dur-ing play, just as with co-located sessions, players interleavesocial discussions, such as out-of-character jokes or popculture references. This occurs in text chat as well. Play-ers type messages, such as “brb” (be right back), to conveysocial cues about their presence. Participants describe cre-ating general text channels for similar discussions, postingmemes, or scheduling upcoming play sessions.

Several participants (4) report posting images and othernon-textual media in text chat during play sessions (see

P9-B quote). Often times these postings are inspired fromevents taking place in the narrative, but shift away to asocial frame using content from the real world for humor.Since they are posted in a text chat, they do not directly dis-rupt voice communications. Participants describe insertingimages into ability cards or outcomes of automated dicerolls, such as memes for humor (Figure 2), intersecting thesocial and game frames.

Four participants report using video of faces. Only one usesvideo within Roll20 to position the faces of other playersnear his own webcam allowing him to look at the camerawhile also observing their responses. The others use videoin Discord or Skype, located in a separate window. Twoparticipants report preference towards not using video asthey feel it makes the experience more immersive whenonly hearing someone’s voice. All participants that did notuse video note that this makes them less aware of socialcues, such as the attentiveness of other players.

Game FrameThe game frame connects across voice communication,text chat, and virtual spaces. Participants discuss rules andstrategy by talking. These discussions also take place intext chat. They may occur in secret using private messag-ing, something that is harder to hide in co-located play.

P2: “If you are sitting at a table, and you arelike, "Everybody else don’t pay attention whileI whisper to this person." Everybody else islike, ‘What the hell are they whispering about?!’Where in Roll20, you can be like, ‘Hey, on thedown low, if this guy twitches, lets attack.’”

All participants report how the virtual tabletop streamlinedparts of dealing with game rules, such as rolling dice, keep-ing track of status effects, and establishing turn order. They

all use the built-in dice rollers of virtual tabletops. Resultsare displayed in the text chat. The GMs hide the resultsof their rolls. In some virtual tabletops, while results arehidden, the players are informed that dice were rolled. Tocounteract this, one GM would privately roll physical dicewhen he didn’t want the players to know a roll was happen-ing. Thus, as game rules become streamlined, players gainless control over how they are administered.

P8: “I always have a set of dice. Because whenI, as a GM, roll, it shows me roll, and you justsee. They see a like a shadow dice. They don’tactually see what it is. But they can see that I’mrolling. Sometimes I want to roll, I don’t wantthem to know that I’m rolling at all. And so I’lljust take out my dice, and I’ll roll.”

Just as participants created role playing text channels fornarrative frame, they similarly create text channels for dis-cussing game rules and questions about what is allowed.

P9-B: “One of my players inthat game is really into intocooking for some reason. It’s inhis backstory, so whenever westopped in a city to buy rationsor something like that, he actu-ally would spend more moneyto buy some expensive ra-tions, premium rations, stuff likethat. . . And his bag of holding isgetting a bit light. So he says,‘Yeah, I’ve only got potatoesnow.’ And then the other guyswere like, ‘Oh, but you can doso much with potatoes.’ Then,they actually posted picturesof all the different dishes youcan make with potatoes, allthat kind of stuff. So we werekind of like semi-serious invoice. And then every once ina while someone would post anew picture of hash browns orsomething like that. And thenpeople would start laughing.”

Implications for Virtual Tabletop DesignOur interviews reveal how participants use tools and mediaas they engage in distributed play. While virtual tabletops,such as Roll20, functionally support audio, video, and textcommunication, participants choose to use other commu-nication tools, which better support shifting of frames andhave fewer technical breakdowns. We identify the followingimplications for designing virtual tabletops for distributedplay of tabletop role-playing games:

Provide seamful and organizable communication chan-nels. Despite the presence of text chat in Roll20, manyparticipants opted to use Discord for out-of-character com-ments, private messaging, and lengthier in-character con-versations. Roll20 provides a single text chat where all tex-tual communication takes place. Similar to Discord, virtual

tabletops could allow multiple communication channels, ofnot only text, but additionally audio and video, of which play-ers can categorize for distinct purposes, such as ones fornarrative, game, and social frames. Feedback about playerpresence within these varied channels becomes importantfor seamful design.

Enrich narrative and social frames via seamless inte-gration of external media. Participants made use of im-ages, animated GIFs, and music. Participants valued theability to directly integrate media into the play environment,such as inserting memes into dice roll outcomes or addingsound effects in response to character actions. This en-ables fluid shifting between frames of experience, similar towhat occurs in in spoken language with these games.

Support flexible control over the seams of the gameframe. Streamlining how game rules are administered bythe virtual tabletop could help players remain more in thenarrative and social frames, which has potential value forengagement. However, a lack of control over how rules andoutcomes are presented may instead bring players’ atten-tion towards the game frame. Visibility controls are neededto allow players to decide when it is appropriate to reveal.

ConclusionOur findings reveal how frames of experience intersecttools and media in distributed play of tabletop role-playinggames. We elicited perceived experiences through inter-views. Future work will validate our findings through obser-vations of actual practice with multiple virtual tabletops. Weplan to conduct an observational study, as well as, analyzevideo content of distributed play published online via Twitchand YouTube. We foresee design implications for distributedcollaboration environments and live-streaming experiences,in which media, activities, and technology intersect as par-ticipants engage with each other.

REFERENCES1. Steve Benford, Carsten Magerkurth, and Peter

Ljungstrand. 2005. Bridging the Physical and Digital inPervasive Gaming. Commun. ACM 48, 3 (March 2005),54–57. DOI:http://dx.doi.org/10.1145/1047671.1047704

2. Geoff Boucher. 2018. Like a wily old wizard, ’Dungeons& Dragons’ shows it has enduring magic. (July 2018).https://www.latimes.com/entertainment/movies/la-et-hc-dungeons-dragons-20180730-story.html[Online; posted 31-July-2018].

3. Virginia Braun and Victoria Clarke. 2006. Usingthematic analysis in psychology. Qualitative Researchin Psychology 3, 2 (2006), 77–101. DOI:http://dx.doi.org/10.1191/1478088706qp063oa

4. Oğuz Turan Buruk and Oğuzhan Özcan. 2018.Extracting Design Guidelines for Wearables andMovement in Tabletop Role-Playing Games via aResearch Through Design Process. In Proceedings ofthe 2018 CHI Conference on Human Factors inComputing Systems (CHI ’18). ACM, New York, NY,USA, Article 513, 13 pages. DOI:http://dx.doi.org/10.1145/3173574.3174087

5. Matthew Chalmers, Marek Bell, Barry Brown, MalcolmHall, Scott Sherwood, and Paul Tennent. 2005. Gamingon the Edge: Using Seams in Ubicomp Games. InProceedings of the 2005 ACM SIGCHI InternationalConference on Advances in Computer EntertainmentTechnology (ACE ’05). ACM, New York, NY, USA,306–309. DOI:http://dx.doi.org/10.1145/1178477.1178533

6. Jennifer Grouling Cover. 2014. The Creation ofNarrative in Tabletop Role-Playing Games. McFarland& Company, Inc.

7. Fantasy Grounds. https://www.fantasygrounds.com

8. Gary Allen Fine. 1983. Shared Fantasy: Role PlayingGames as Social Worlds. University of Chicago Press.

9. John C. Flanagan. 1954. The Critical IncidentTechnique. Psychological Bulletin 51, 4 (1954), 327.

10. Antero Garcia. 2019. Gaming Literacies: Spatiality,Materiality, and Analog Learning in a Digital Age.Reading Research Quarterly 0, 0 (2019), 1–19. DOI:http://dx.doi.org/10.1002/rrq.260

11. Stacey Kuznetsov, Alex S. Taylor, Tim Regan, NicolasVillar, and Eric Paulos. 2012. At the Seams: DIYbioand Opportunities for HCI. In Proceedings of theDesigning Interactive Systems Conference (DIS ’12).ACM, New York, NY, USA, 258–267. DOI:http://dx.doi.org/10.1145/2317956.2317997

12. Craig Lindley. 2005. Narrative structure in trans-realityrole-playing games: Integrating story construction fromlive action, table top and computer-based role-playinggames. Proceedings of Digital Games ResearchAssociation (DiGRA) 2005 Conference (2005).

13. Daniel Mackay. 2001. The Fantasy Role-Playing Game:A New Performing Art. McFarland & Company, Inc.

14. Wendy E. Mackay. 2002. Using video to supportinteraction design. DVD Tutorial, CHI 2, 5 (2002).

http://dx.doi.org/10.1145/1047671.1047704https://www.latimes.com/entertainment/movies/la-et-hc-dungeons-dragons-20180730-story.htmlhttps://www.latimes.com/entertainment/movies/la-et-hc-dungeons-dragons-20180730-story.htmlhttp://dx.doi.org/10.1191/1478088706qp063oahttp://dx.doi.org/10.1145/3173574.3174087http://dx.doi.org/10.1145/1178477.1178533https://www.fantasygrounds.comhttp://dx.doi.org/10.1002/rrq.260http://dx.doi.org/10.1145/2317956.2317997

15. Carsten Magerkurth, Maral Memisoglu, Timo Engelke,and Norbert Streitz. 2004. Towards the NextGeneration of Tabletop Gaming Experiences. InProceedings of Graphics Interface 2004 (GI ’04).Canadian Human-Computer Communications Society,School of Computer Science, University of Waterloo,Waterloo, Ontario, Canada, 73–80. http:

//dl.acm.org/citation.cfm?id=1006058.1006068

16. Open Broadcaster Software. https://obsproject.com

17. Roll20. https://roll20.net

18. Roll20 subreddit. https://reddit.com/r/roll20

http://dl.acm.org/citation.cfm?id=1006058.1006068http://dl.acm.org/citation.cfm?id=1006058.1006068https://obsproject.comhttps://roll20.nethttps://reddit.com/r/roll20

IntroductionBackgroundFrames of Experience in Tabletop Role-Playing GamesAugmenting Tabletop Role-Playing GamesRoll20: Virtual Tabletop

MethodParticipantsProcedureData CollectionData Analysis

Results and DiscussionNarrative FrameSocial FrameGame Frame

Implications for Virtual Tabletop DesignConclusionREFERENCES