Algorithms Are Decision Systems

Ted Striphas research and teaching interests include media history, theory, and criticism; the history of technology; and cultural studies. He is an award winning teacher and scholar. He's the author of "The Late Age of Print" and "An Infernal Culture Machine".
 
 
Blog | @striphas
 
 
As you said, "An Infernal Culture Machine", your last essay, is prompted by the question «What is culture today?». Which do you think are the main drivers of change in culture, today?

There are surely many drivers in the realm of culture today, as there have been, always. In terms of how we access and make sense of culture, however, digital – or rather computational – technologies increasingly hold sway.

Consider Google, for example. Most everyone who uses its search service seems to appreciate how it has something important to do with adjudicating "the best which has been thought and said," as Matthew Arnold defined "culture" back in 1869. Yet, where in the definitions of the word is there any strong connection to computing technology? Most English-language dictionaries define "culture" almost exclusively in 19th century terms, linking its meaning to ideas like civilization, aesthetics, world-views, ways of life, and material artifacts; some dictionaries also include the pre-modern sense of the term as "tending toward natural growth." What I am trying to reconcile right now in my research is the gap between the standard dictionary definitions of "culture" and our actual experiences with it, the latter of which are increasingly shot through with technology. This is exactly where algorithms come it.

The Algorithmic Culture is one of the main topics you face in your works. How do you think algorithms are crucial to understand our new relationship with culture?

Culture has long been about argument and reconciliation: argument in the sense that groups of people have ongoing debates, whether explicit or implicit, about their norms of thought, conduct, and expression; and reconciliation in the sense that virtually all societies have some type of mechanism in place – always political – by which to decide whose arguments ultimately will hold sway. You might think of culture as an ongoing conversation that a society has about how its members ought to comport themselves.

Increasingly today, computational technologies are tasked with the work of reconciliation, and algorithms are a principal means to that end. Algorithms are essentially decision systems – sets of procedures that specify how someone or something ought to proceed given a particular set of circumstances. Their job is to consider, or weigh, the significance of all of the arguments or information floating around online (and even offline) and then to determine which among those arguments is the most important or worthy. Another way of putting this would be to say that algorithms aggregate a conversation about culture that, thanks to technologies like the internet, has become ever more diffuse and disaggregated.

On sites like Google, Facebook, Amazon.com, and others, all of this decision-making is handled mathematically, of course. On Facebook, for example, your friends are assigned particular weights, or values, based on the degree to which you interact with them on the site, how much they interact with you, the nature of your connection to them (e.g, whether you're a relative or a causal acquaintance), and more. What they're talking about – a particular product, say, or a major life event – also factors in here as well. Your news feed gets prioritized on these and other bases, which are quantified and computed.

I mention this not because I am scared of numbers – far from it. But it is important to recognize how these behind-the-scenes determinations about how an algorithm will work are, as in earlier moments in culture, political decisions. Reconciling culture – establishing norms – is not a neutral process. Status updates from a cousin of mine, with whom I am not particularly close, constantly appear in my Facebook news feed, presumably because we have identified ourselves as cousins. Note the value that's both implicit and reinforced here: that kinship is more important than other types of relationships. Why should this be the presumption?

My concern with algorithms like the one driving Facebook has less to do with fact that they may be wrong than with our lack of knowledge about how exactly they work. Most such algorithms are protected by patent laws, trade secret laws, and other legal and technical instruments, which make it difficult if not impossible to determine which values – which weights – are programmed in to them, and why. Algorithmic culture renders opaque the reconciliation part of conversation about culture.

In "The Late Age of Print" you talk about how the book industry has adapted to changes in twentieth-century print culture. Could you list three aspects publishers should look at to evolve rapidly in the digital age?

If a publisher wants to evolve rapidly, then my advice would be:

1. shift your current paper publications and backlist to print-on-demand, and contract with Amazon to sell electronic editions of your books;
2. chunk books up into small pieces and sell the parts, perhaps in connection with full or even extended paper or digital editions;
3. make digital books cheap, as customers twenty years ago caught on to the fact that bits don't cost as much as atoms.

Of course, that's a completely crass, bottom-line approach that sees the value in books almost exclusively in economic terms. In that sense, then, if one really wants to evolve as a publisher, one ought to consider the broader context of books and reading in society. I won't say much more here beyond suggesting to your readers that they check out the preface to the paperback edition of The Late Age of Print, where I develop these concerns at length. What does it mean, for example, that Amazon and other sellers of e-books can essentially eavesdrop while people are reading them, or delete certain titles from people's electronic libraries without their consent? There are significant ethical concerns that are too easily glossed over in the book industry's headlong rush into the digital age.

About Letizia Sechi

Social Media Manager at 40k. Twitter @LetiziaSechi