Twitch.tv represents something perhaps none of us imagined ten or so years ago: the means of compulsively watching other people play videogames. The now Amazon-owned livestreaming platform is the most popular venue for watching videogame play in most regions of the world. Loading up Twitch as we write this, there’s upwards of 40,000 people watching an esports tournament broadcast from Germany, several thousand watching a days-long World of Warcraft marathon, and a few hundred watching a woman outfitted in cosplay attire playing a musical rhythm game.

However, it’s not just the passive act of watching this feed that makes Twitch so compelling. Like other kinds of networked participatory media before it, this new channel is so appealing because it gives users the ability to do more than just watch. Attention to a Twitch stream is characterised by direct interactions between audiences, other audience members and the streamer. The primary mechanism for this exchange is Twitch Chat – an Internet Relay Chat filled with text, emoji-style images (most users can send upwards of twenty messages every thirty seconds) and notifications about financial pledges made to streamers. Twitch Chat makes watching less about viewing in isolation and more about participation – something that gives streams a kind of interest-excitement they would otherwise lack. As the company’s CEO, Emmett Shear puts it – Twitch might be best understood as ‘multiplayer entertainment’.

As media and cultural studies research tells us, if we’re getting something for free, it’s the consumer who’s the product. In most cases, what’s valuable is our attention, which in one way or another turns a profit for media industries. Writing about old cultural industries like broadcast television and radio, Dallas Smythe famously coined the term ‘audience commodity’ to conceptualise how audience activity can be understood as a lucrative source of economic value. In the model of old broadcast media, viewer attention to advertisements is exchanged for the ‘free lunch’ of television programmes. In this way, the act of watching television constitutes a form of productive labour.

A range of writings since – such as those deriving from the Italian autonomist Marxist movement – has considered modern, digitally mediated audience activities, which generate value, in terms of ‘digital labour’. Broadly construed, this refers to the surplus value generated by media audiences in using, watching, or engaging with media.

Twitch is an exemplary case of how audience activities continue to be commodified in the context of online digital media. Unlike with traditional televisual media, a key way that this value is generated is through the interactivity performed by viewers, enabled by the Twitch platform. What features like Chat enable is the transmission of a whole lot that’s fairly unsubstantial content (ASCII images of ‘doge’ to emotes of various flavours of Pepe the frog), yet potent and palpable in terms of its affect – or what the media scholar Raymond Williams would describe as its ‘structures of feeling’. Twitch mediates and intensifies affectivity, creating a kind of undulating rhythm and vivacity. This is at the heart of the spectacle that makes Twitch so compelling to watch. The work of watching is the work that these audiences do in creating and sustaining captivating structures of feeling.

So, how exactly is this labour mediated by Twitch? A viewer’s basic act of opening up a stream and watching it contributes to that particular stream’s view count – a small numerical aggregate of viewers displayed at the bottom of each stream, as well as on the Twitch stream directory. Viewercount is the default way that Twitch sorts streams for viewers. The viewer’s simple watching of a stream, then, contributes to a collective framing of how people feel, think about and engage with content. A larger viewer count might frame a stream as more lively, whereas a smaller viewercount might frame it as more intimate.

Twitch’s chat function is another way that audiences interact. A very busy chatroom might work similarly to the noise of a sports crowd in creating an atmosphere of excitement. The crowd of Twitch Chat creates a sense of ‘happening’, of vitality – of activity, excitement, frustration, elation, etc. This ‘noise’ is often layered into the broadcast by being superimposed over the top of the screen. The experience of watching a popular stream – such as an esports match – derives not only from the content of the stream, but also from the visibility of these audience practices, which allow the viewer to be fully taken up in the here and now of the event.

Related to Twitch’s chat system are Twitch’s emotes – a key mechanism of exchange between user, audience and streamer, and for the circulation of affect. As with emojis, users can select from a range of pre-selected emotes or by typing a particular phrase to post a small, text-sized image in the chat window. For instance, entering “osfrog” into the chat window results in the posting of the OSFrog emote, displayed in chat as an image of a green frog. These emotes are generally meant to convey a feeling or affect in response to the events of the Twitch stream.

In many cases, the use of emotes is the result of various game-specific vernaculars and memes. Through emotes, Twitch audiences develop systems of meaning in and around the games and communities they’re involved in, often different from their original intended meaning. For instance, the OSFrog emote was originally part of a 2015 Old Spice deodorant advertising campaign, but quickly became used as an emote for the game Dota 2, in reference to the game’s lead developer ‘IceFrog’. Within Twitch Chat, the OSFrog emote is used in response to an ability or strategy or moment in gameplay that is taken to be unfair, overpowered or poorly designed. Much like how a crowd collectively despairs at a missed penalty shot in football, the collective OSfrogging in a Dota 2 conveys the e-sports audiences’ disbelief or frustration at a questionable play, or that a team has succeeded due to a technical glitch or bug.

When we think about desirable atmospheres and the value they generate, it is important we ask who these atmospheres are desirable for, and how they are experienced asymmetrically across Twitch. A clear case in point is the use of Twitch Emotes by gaming’s noxious subcultures, comprised of extremely online young men, to perpetuate systemic racism, particularly toward black streamers. A key example comes from a promotional crossover between PlayerUnknown’s Battlegrounds and KFC (and the release of a KFC chicken bucket emote), which was often posted alongside emotes of black streamers (for instance TriHex). In this way, emotes come to militate against the participation of individuals and reinforce many of the same asymmetrical power structures and uneven experiences on Twitch that have plagued videogaming since its inception.

As the sociologist TL Taylor writes in her book Watch me Play: Twitch and the Rise of Game Live Streaming, there’s no Twitch without broadcasters. Taking this further, there’s no broadcast without audiences. As we occupy a central space in the political economy of Twitch, much of the reason why we’re all such captivated subjects in watching lies in our own collective interactions on the platform itself. Chatting, emoting, and simply watching are all activities that we ourselves perform on the platform, yet that are put in service of capitalism and are what ultimately keeps many of us hooked. More than a simple progression of capital’s encroachment upon leisure time, much of the commercial viability of livestreaming is reliant upon actively capturing and harnessing users’ sociability and affect. Writing of social media in Capital is Dead, McKenzie Wark notes a ‘desire to share … is making someone else very, very rich’ – a statement just as true of the Twitch audience as it is the streamer themselves.