If Herodotus can ask 2,400 years ago for immortality while striving to prevent his subject from being “forgotten in time”, then perhaps it is not too presumptuous for me, more than two millennium later, to ask you to read for five minutes. Of course Herodotus wrote with an eternal, almost sculpted, quality, his words on the Greco-Persian Wars translated and studied and assigned in countless classrooms over the centuries, his words carved into stone for what is certainly an unreasonable amount of time. I don’t quite expect a similar result, as the writing in our century, it seems, comes in sentences that are a tad more malleable, with the age of permanence pretty much at its end. So you won’t stumble upon Herodotus today. You won’t discover any ageless, fixed sentences. You won’t even uncover any unexpected books. To stroll through the modern library in your pocket is to watch the once static and eternal books around you contort and transform before your very eyes—with every page now liquidlike, shifty, calling out, adapting in realtime to your interests.
A traditional library, it is worth admitting, is rather cumbersome relative to the modern alternative. You can appreciate the oak floors and have fond memories of the darkened corridors and the hushed atmosphere, but you can’t believe, if you’re honest, that the traditional library even competes with the exhaustiveness of the modern library in your pocket. Nevertheless, a stuffy old library retains at least one advantage—its books don’t predict your desires and transform while you read. If you grab a book, walk to your seat, skim a few pages, and then return to the shelf—the remaining books haven’t learned your preferences and begun to discern what title and chapter and even story will most arouse your desires, revising the already written sentences into new forms, which is exactly what happens when you look for information on any screen.
Of course to see personal results on the screen in your pocket isn’t that dissimilar from your haphazard choice of a book from the shelf—in the reservoir of intellectual history there’s never been a central authority or arbitrator. Herodotus just happened to last through wars and plagues and the occasional book burning, and you either have or haven’t read his words. And most people are already accustomed to the subtle but still pernicious fact that search information is tailored—when you use a screen to search, the results are specific to your interests and past behaviors and location and innumerable other factors that are forever shifting because, in our world, everyone gets a personal internet. What’s new today, however, is that your decisions are beginning to shape and distort and affect the information that you seek, in a manner that’s far beyond the mere tailoring of search results, because we’re now tailoring the actual information to your tastes.
Looking closely at the news business reveals this behavior at its most pernicious. Now there’s nothing inherently noble about the news business—you don’t have to sell tobacco or alcohol or work for a streaming service to desire addicts as customers. News agencies have always targeted reader preferences, with the salacious and the vulgar and famed given the most prominent coverage. But the abundance of information about clicks and engagement and the realtime measurement of the precise words and phrases that drive the most traffic isn’t used merely to format the front page—because that information also slips into what were once journalistic decisions. These incentives are subtle yet still have a significant affect on what you read.
More than one prominent newsroom in Washington has, for instance, a realtime monitor of engagement that all journalists can view—which seems like a fairly poor way to decide what’s newsworthy. It is, in fact, the practice of journalism based on the whims of the mob, with each journalistic decision feeding back into the ever-accelerating machine. The peculiar result is more news about what’s already known and less news about what’s unknown—which seems like a rather perverse definition of news. And an even greater distortion comes from financial media, where all the major news agencies measure the success of their stories based on changes in market prices: the larger the move in prices after publication, the more successful the story. This isn’t to observe with binoculars from a distance and report what you see. This is to fiddle with what you observe—it is to be judged based on how much you change what you’re observing. If you close your eyes and truly labor, you just might conjure up a worse incentive.
Imagine walking, once again, past a news kiosk, and you catch a newspaper adjusting a headline, revising a long article into one that’s a little bit shorter, trying to entice you based on what it perceives are your interests. And that’s exactly what happens when news agencies algorithmically test various headlines—showing different populations different stories, showing you and your neighbor different versions of what was once the same news. That’s what happens when sentences in news stories are adjusted to contain just the right keywords and phrases—pouring the news into a fixed mold of consumer desires. That’s what happens when subheadings adapt automatically based on individual readers. That’s what happens when the snippets shown in search results populate from algorithms. That’s what happens when breaking news or even short articles are written with machine learning. Your newsfeed doesn’t update because there is news—it updates because you’ve begun to desire news.
Ensuring that a new book or essay or research isn’t lost amid the cavalcade of information doesn’t require a champion. There’s no editor to pitch. Nor will a good review or notice usually help. What is needed, instead, is to gauge the hidden formula that activates the screen in everyone’s pocket. What hits the algorithm just right today? What ranks highest in search results? What drives most newsfeeds? And the answer to those questions is to conform to just the right box at just the right time: to ensure that anything new fits into the already expressed old desires.
You can still stumble upon Herodotus in the library, as it will wait for you on the shelf in its durable form, but you’ll have a difficult time stumbling upon anything outside your current awareness on most screens. Even when you’re looking for classic books, or for information about history, the results that you uncover are shaped by what is perceived to be your desires. And you’ll never know what questions you’re not asking and what information you’re not seeing. Searching for something on a screen today comes remarkably close to looking into the mirror.
So we just might be a tad concerned by the effect of cataloguing our information with these methods. Right now we’re sorting the world with an architecture that automatically adapts to what we supposedly need, but the perverse incentive is to concentrate more on what we already know and to ignore what we don’t know. You’re not really seeing the world when you look—you’re seeing a personal sales pitch from your personal algorithm, and you’re limited to the questions that you're expected to ask. You might even call this world solipsistic: the old method involves walking into a library and uncovering what’s on the shelves, while the new method demands that the library conform to you. And we’re just clicking away and assuming that all this adapting and customizing and sorting is, in fact, acting in your interest.
Such an interesting read!