This is cross-posted from the HIIG Science Blog, and is part of a series on metaphors and digital society hosted by Christian Katzenbach and Stefan Larsson. I recommend the other essays as well: Nik John on sharing, Noam Tirosh on revolution, and Christian Djeffal on artificial intelligence. 

Sometimes a metaphor settles into everyday use so comfortably, it can be picked back up to extend its meaning away from what it now describes, a metaphor doing metaphorical service. Platform has certainly done that. When I first wrote about the term in 2010, social media companies like YouTube and Facebook were beginning to use the term to describe their web 2.0 services, to their users, to advertisers and investors, and to themselves. Now social media companies have embraced the term fully, and have extended it to services that broker the exchange not just of content or sociality but rides (Uber), apartments (AirBnB), and labor (Taskrabbit). The term so comfortably describes these services that critics and commentators can draw on the word to extend out for the purposes of argument. The past few years have witnessed a “platform revolution”, (Parker, van Alstyne, and Choudary) the rise of “platform capitalism” (Srnicek) driven by “platform strategy” (Reillier and Reillier), with the possibility of “platform cooperativism” (Scholz) all part of “the platform society” (van Dijck, Poell, and DeWaal) These books need not even be referring to the same platforms (they all have their favorite examples, somewhat overlapping), their readers know what they’re referring to.

From programmability to opportunity

As platform first took root in the lexicography of social media, it was both leaning on and jettisoning a more specific computational meaning: a programmable infrastructure upon which other software can be built and run, like the operating systems in our computers and gaming consoles, or information services that provide APIs so developers can design additional layers of functionality. The new use shed the sense of programmability, instead drawing on older meanings of the word (which the computational definition itself had drawn on): an architecture from which to speak or act, like a train platform or a political stage. Now Twitter or Instagram could be a platform simply by providing an opportunity from which to speak, socialize, and participate.

At the time, some suggested that the term should be constrained to its computational meaning, but it’s too late, platform has been widely accepted in this new sense – by users, by the press, by regulators, and by the platform providers themselves. I argued then that the term was particularly useful because it helped social media companies appeal to several different stakeholders of interest to them. Calling themselves platforms promised users an open playing field for free and unencumbered participation, promised advertisers a wide space in which to link their products to popular content, and promised regulators that they were a fair and impartial conduit for user activity, needing further regulation.

This is what metaphors do. They propose a way of understanding something in the terms of another; the analogy distorts the phenomenon being described, by highlighting those features most aligned with what it is being compared to. Platform lent social media services a particular form, highlighted certain features, naturalized certain presumed relations, and set expectations for their use, impact, and responsibility. Figuratively, a platform is flat, open, sturdy. In its connotations, a platform offers the opportunity to act, connect, or speak in ways that are powerful and effective: catching the train, drilling for oil, proclaiming one’s beliefs. And a platform lifts that person above everything else, gives them a vantage point from which to act powerfully, a raised place to stand.

What metaphors hide

Metaphors don’t only highlight; they also downplay aspects that are not captured by the metaphor. “A metaphorical concept can keep us from focusing on other aspects of the concept that are inconsistent with that metaphor.” (Lakoff and Johnson, 10) We might think of this as incidental or unavoidable, in that any comparison highlights some aspects and thereby leaves others aside. Or we could think of it as strategic, in that those deploying a metaphor have something to gain in the comparison it makes, presumably over other comparisons that might highlight different aspects.

By highlighting similarities – social media services are like platforms – metaphors can have a structural impact on the way we think about and act upon the world. At the same time, metaphor cannot be only about similarity – otherwise the ideal metaphor would be tautological, “X is like X.” Metaphor also depends on the difference between the two phenomena; the construction of similarity is powerful only if it bridges a significant semantic gap. Steven Johnson points out that “the crucial element in this formula is the difference that exists between ‘the thing’ and the ‘something else.’ What makes a metaphor powerful is the gap between the two poles of the equation.” (58-59) Phil Agre goes further, suggesting that “metaphors operate as a ‘medium of exchange’” (37) between distinct semantic fields, negotiating a tension between elements that are, at least in some ways, incompatible. This structural bridge constructed by metaphor depends on choosing aspects of comparison that will be salient and rendering others insignificant. The platform metaphor does a great deal of work, not only in what it emphasizes, but in what it hides:

1. Platform downplays the fact that these services are not flat. Their central service is to organize, structure, and channel information, according both to arrangements established by the platform (news feed algorithms, featured partner arrangements, front pages, categories) and arrangements built by the user, though structured or measured by the platform (friend or follower networks, trending lists). Platforms are not flat, open spaces where people speak or exchange, they are intricate and multi-layered landscapes, with complex features above and dense warrens below. Information moves in and around them, shaped both by the contours provided by the platform and by the accretions of users and their activity – all of which can change at the whim of the designers. The metaphor of platform captures none of this, implying that all activity is equally and meritocratically available, visible, public, and potentially viral. It does not prepare us, for example, for the ability of trolls to organize in private spaces and then swoop together as a brigade to harass users in a coordinated way, in places where the suddenness and publicness of the attack is a further form of harm.
2. The platform metaphor also obscures the fact that platforms are populated by many, diverse, sometimes overlapping, and sometimes contentious communities. It is absurd to talk about Facebook users, as if two billion people can be a single group of anything; talk about the Twitter community only papers over the tension and conflict that has been fundamental, and sometimes destructive to how Twitter is actually used. As Jessa Lingel argues, social media platforms are in fact full of communities that turn to social media for specific purposes, often with ambivalent or competing needs around visibility, pseudonymity, and collectivity; then they struggle with how the platforms actually work and their sometimes ill fit with the aims of that community. When we think not of ‘Facebook users’ but a group of Brooklyn drag queens, the relationship between users and platform is not an abstract one of opportunity, but a contentious one about identity and purpose.
3. Platform also helps elide questions about platforms’ responsibility for their public footprint. Train platforms are not responsible for the passengers. Like other metaphors like conduit and media and network, platform suggests an impartial between-ness that policymakers in the U.S. are eager to preserve – unlike European policymakers, where there is more political will to push responsibility onto platforms, though in a variety of untested ways. When, as Napoli and Caplan point out, Facebook refuses to call itself a media company, they are disavowing the kind of public and policy expectations imposed on media. They’re merely a platform. In the meantime, they have each built up a complex apparatus of content moderation and user governance to enforce their own guidelines, yet these interventions are opaque and overlooked.
4. Finally, platform hides all of the labor necessary to produce and maintain these services. The audience is not supposed to see the director or the set decorators or the stagehands, only the actors in the spotlight. Underneath a platform is an empty, dusty space – it’s just there. Social media platforms are in fact the product of an immense amount of human labor, whether it be designing the algorithms or policing away prohibited content. When we do get a glimpse of the work and the workers involved, it is culturally unexpected and contentious: the revelation, for example, that Facebook’s Trending Topics might have been curated by a team of journalism school grads, working like machines. (1, 2) What if they make mistakes? What if they are politically biased? How are humans involved, and why does that matter? Platform discourages us from asking these questions, by leaving the labor out of the picture.

We need not discard the term, just to swap in another metaphor in its place. It is not as if it’s impossible to think about these obscured aspects of platforms; the metaphor can downplay them, but cannot erase them. But we have to either struggle upstream against the discursive power of the term, or playful subvert it. A platform may hide the labor it requires, but in a different framework it could be asked to shelter that labor, protect it. If a platform lifts up its users, then there may be some manner of responsibility for lifting some people up over others. We might also play with other metaphors: are platforms also shopping malls, or bazaars? amusement parks, or vending machines? nests, or hives? pyramids, or human pyramids? But mostly, we can scrutinize the metaphor in order to identify what it fails to highlight, how that may serve the interest of the metaphor’s practitioners, and what design interventions and obligations might best attend to these gaps and obscurities. And, as Kuhn notes about scientific paradigms, any frame of understanding works to coalesce the phenomenon by leaving off aspects that do not fit – and these discarded aspects can return to challenge to that frame, and sometimes tear it down. Platforms downplay these aspects at their own peril.