The epistemic privilege of platforms

By Mark Carrigan

What is the relationship between platforms and their users? I’ve been thinking about this all morning while reading The Know‑It‑Alls: The Rise of Silicon Valley as a Political Powerhouse and Social Wrecking Ball, by Noam Cohen. On loc 277 he writes:

In fact, tech companies believe that through artificial intelligence tools they understand their users’ state of mind in a way few other companies can, and far better than any regulator. They can track, measure, and analyze the billions of decisions their users make, and they can detect even the most minor feature that may be turning them off. And rather than wait for problems, these companies can compel their users to express a preference by staging so-called A/ B testing, which involves showing groups of users slightly different versions of the site and measuring which group stays longer and is thus happier with the experience. Google famously went so far as to prepare forty-one shades of blue to test which was the best color for displaying links in its Gmail service.

This epistemic privilege is inflated but it nonetheless has to be taken seriously. There are forms of knowledge about users which platforms have unique access to, discerning real-time behaviour (including responses to planned stimuli) with a degree of granularity that would be difficult to match in any other context. What matters is how this epistemic relation is raised into a political claim: if we know our users better than any external party, how could regulation be anything other than incompetent?

This relies on a reduction of the salient characteristics of the user to their actions which register within the confines of the platform, representing the core of what I’ve written about in an upcoming chapter as the evisceration of the human: the reduction of real agency to its empirical traces. Furthermore, it is bound up with the conviction of transcending the murky mess of self-interpretation, offering apparent insight into what OK Cupid data scientist Christian Rudder memorably described as Who We Are (When We Think No One’s Looking) in the subtitle to his book Dataclysm. This is bound up in a political economy which Mark Andrejevic identifies on loc 870 of his InfoGlut:

In this regard the digital era opens up a new form of digital divide: that between those with access to the databases and those without. For those with access, the way in which data is understood and used will be fundamentally transformed. There will be no attempt to read and comprehend all of the available data – the task would be all but impossible. Correlations can be unearthed and acted upon, but only by those with access to the database and the processing power. Two different information cultures will come to exist side by side: on the one hand, the familiar, “old- fashioned” one in which people attempt to make sense of the world based on the information they can access: news reports, blog posts, the words of others and the evidence of their own experience. On the other hand, computers equipped with algorithms that can “teach” themselves will advance the instrumental pragmatics of the database: the ability to use tremendous amounts of data without understanding it.

Does anyone know of ethnographic work which looks at how this epistemic relation is talked about in everyday labour within these firms? It must presumably be invoked constantly, in an everyday manner, during user interface design and similar activities. This could help elucidate the micro-structure for the inflation of epistemic privilege which I suspect Cohen is correct to identify as one source of hostility to regulation.