Archive for Algorithmic Culture

"The Shannon and Weaver Model"

February 20, 2012 Algorithmic Culture, Related Work 3 comments
Email, RSS Follow

First things first: some housekeeping.  Last week I launched a Facebook page for The Late Age of Print.   Because so many of my readers are presumably Facebook users, I thought it might be nice to create a “one-stop shop” for updates about new blog content, tweets, and anything else related to my work on the relationship between print media and algorithmic culture.  Please check out the page and, if you’re so inclined, give it a like.

Okay…on to matters at hand.

This week I thought it might be fun to open with a little blast from the past.  Below is a picture of the first page of my notebook from my first collegiate communication course.  I was an eighteen year-old beginning my second semester at the University of New Hampshire, and I had the good fortune of enrolling in Professor W—-‘s introductory “Communication and the Social Order” course, CMN 402.  It wouldn’t be an overstatement to call the experience life changing, since the class essentially started me on my career path.

What interests me (beyond the hilariously grumpy-looking doodle in the margin) is a diagram appearing toward the bottom of the page.  It’s an adaptation of what I would later be told was the “Shannon and Weaver” model of communication, named for the electrical engineer Claude Shannon and the mathematician Warren Weaver.

CMN 402 - UNH Jan. 28, 1992

Note what I jotted down immediately below the diagram: “1.) this model is false (limited) because comm is only one way (linear); 2.) & assumes that sender is active & receiver is passive; & 3.) ignores the fact that sender & receiver interact w/ one another.”  Here’s what the model looks like in its original form, as published in Shannon and Weaver’s Mathematical Theory of Communication (1949, based on a paper Shannon published in 1948).

Shannon & Weaver Model of Communication, 1948/1949

Such was the lesson from day one of just about every communication theory course I subsequently took and, later on, taught.  Shannon and Weaver were wrong.  They were scientists who didn’t understand people, much less how we communicate.  They reduced communication to a mere instrument and, in the process, stripped it of its deeply humane, world-building dimensions.  In graduate school I discovered that if you really wanted to pull the rug out from under another communication scholar’s work, you accused them of premising their argument on the Shannon and Weaver model.  It was the ultimate trump-card.

So the upshot was, Shannon and Weaver’s view of communication was worth lingering on only long enough to reject it.  Twenty years later, I see something more compelling in it.

A couple of things started me down this path.  Several years ago I read Tiziana Terranova’s wonderful book Network Culture: Politics for the Information Age (Pluto Press, 2004), which contains an extended reflection on Shannon and Weaver’s work.  Most importantly she takes it seriously, thinking through its relevance to contemporary information ecosystems.  Second, I happened across an article in the July 2010 issue of Wired magazine called “Sergey’s Search,” about Google co-founder Sergey Brin’s use of big data to find a cure for Parkinson’s Disease, for which he is genetically predisposed.  This passage in particular made me sit up and take notice:

In epidemiology, this is known as syndromic surveillance, and it usually involves checking drugstores for purchases of cold medicines, doctor’s offices for diagnoses, and so forth. But because acquiring timely data can be difficult, syndromic surveillance has always worked better in theory than in practice. By looking at search queries, though, Google researchers were able to analyze data in near real time. Indeed, Flu Trends can point to a potential flu outbreak two weeks faster than the CDC’s conventional methods, with comparable accuracy. “It’s amazing that you can get that kind of signal out of very noisy data,” Brin says. “It just goes to show that when you apply our newfound computational power to large amounts of data—and sometimes it’s not perfect data—it can be very powerful.” The same, Brin argues, would hold with patient histories. “Even if any given individual’s information is not of that great quality, the quantity can make a big difference. Patterns can emerge.”

Here was my aha! moment.  A Google search initiates a process of filtering the web, which, according to Brin, starts out as a thick soup of noisy data.  Its algorithms ferret out the signal amid all this noise, probabilistically, yielding the rank-ordered results you end up seeing on screen.

It’s textbook Shannon and Weaver.  And here it is, at the heart of a service that handles three billion searches per day — which is to say nothing of Google’s numerous other products, let alone those of its competitors, that behave accordingly.

So how was it, I wondered, that my discipline, Communication Studies, could have so completely missed the boat on this?  Why do we persist in dismissing the Shannon and Weaver model, when it’s had such uptake in and application to the real world?

The answer has to do with how one understands the purposes of theory.  Should theory provide a framework for understanding how the world actually works?  Or should it help people to think differently about their world and how it could work?  James Carey puts it more eloquently in Communication as Culture: Essays on Media and Society: “Models of communication are…not merely representations of communication but representations for communication: templates that guide, unavailing or not, concrete processes of human interaction, mass and interpersonal” (p. 32).

The genius of Shanon’s original paper from 1948 and its subsequent popularization by Weaver lies in many things, among them, their having formulated a model of communication located on the threshold of these two understandings of theory.  As a scientist Shannon surely felt accountable to the empirical world, and his work reflects that.  Yet, it also seems clear that Shannon and Weaver’s work has, over the last 60 years or so, taken on a life of its own, feeding back into the reality they first set about describing.  Shannon and Weaver didn’t merely model the world; they ended up enlarging it, changing it, and making it over in the image of their research.

And this is why, twenty years ago, I was taught to reject their thinking.  My colleagues in Communication Studies believed Shannon and Weaver were trying to model communication as it really existed.  Maybe they were.  But what they were also doing was pointing in the direction of a nascent way of conceptualizing communication, one that’s had more practical uptake than any comparable framework Communication Studies has thus far managed to produce.

Of course, in 1992 the World Wide Web was still in its infancy; Sergey Brin and Larry Page were, like me, just starting college; and Google wouldn’t appear on the scene for another six years.  I can’t blame Professor W—- for misinterpreting the Shannon and Weaver model.  If anything, all I can do is say “thank you” to her for introducing me to ideas so rich that I’ve wrestled with them for two decades.

Share

How Publishers Misunderstand Kindle

February 13, 2012 Algorithmic Culture, Digital Humanities, Electronic Reading 2 comments
Email, RSS Follow

Last week, in a post entitled “The Book Industry’s Moneyball,” I blogged about the origins of my interest in algorithmic culture — the use of computational processes to sort, classify, and hierarchize people, places, objects, and ideas.  There I discussed a study published in 1932, the so-called “Cheney Report,” which imagined a highly networked book industry whose decisions were driven exclusively by “facts,” or in contemporary terms, “information.”

It occurred to me, in thinking through the matter more this week, that the Cheney Report wasn’t the only way in which I stumbled on to the topic of algorithmic culture.  Something else led me there was well — something more present-day.  I’m talking about the Amazon Kindle, which I wrote about in a scholarly essay published in the journal Communication and Critical/Cultural Studies (CCCS) back in 2010.  The title is “The Abuses of Literacy: Amazon Kindle and the Right to Read.”  (You can read a precis of the piece here.)

The CCCS essay focused on privacy issues related to devices like the Kindle, Nook, and iPad, which quietly relay information about what and how you’ve been reading back to their respective corporate custodians.  Since it appeared that’s become a fairly widespread concern, and I’d like to think my piece had something to do with nudging the conversation in that direction.

Anyway, in prepping to write the essay, a good friend of mine, M—-, suggested I read Adam Greenfield’s Everyware: The Dawning Age of Ubiquitous Computing (New Riders, 2006).   It’s an astonishingly good book, one I would recommend highly to anyone who writes about digital technologies.

Greenfield - Everyware

I didn’t really know much about algorithms or information when I first read Everyware.  Of course, that didn’t stop me from quoting Greenfield in “The Abuses of Literacy,” where I made a passing reference to what he calls “ambient informatics.”  This refers to the idea that almost every aspect our world is giving off some type of information.  People interested in ubiquitous computing, or ubicomp, want to figure out ways to detect, process, and in some cases exploit that information.  With any number of mobile technologies, from smart phones to Kindle, ubicomp is fast becoming an everyday part of our reality.

The phrase “ambient informatics” has stuck with me ever since I first quoted it, and on Wednesday of last week it hit me again like a lightning bolt.  A friend and I were talking about Google Voice, which, he reminded me, may look like a telephone service from the perspective of its users, but it’s so much more from the perspective of Google.  Voice gives Google access to hours upon hours of spoken conversation that it can then use to train its natural language processing systems — systems that are essential to improving speech-to-text recognition, voiced-based searching, and any number of other vox-based services.  Its a weird kind of switcheroo, one that most of us don’t even realize is happening.

So what would it mean, I wondered, to think about Kindle not from the vantage point of its users but instead from that of Amazon.com?  As soon as you ask this question, it soon becomes apparent that Kindle is only nominally an e-reader.  It is, like Google Voice, a means to some other, data-driven end: specifically, the end of apprehending the “ambient informatics” of reading.  In this scenario Kindle books become a hook whose purpose is to get us to tell Amazon.com more about who we are, where we go, and what we do.

Imagine what Amazon must know about people’s reading habits — and who knows what else?!  And imagine how valuable that information could be!

What’s interesting to me, beyond the privacy concerns I’ve addressed elsewhere, is how, with Kindle, book publishers now seem to be confusing means with ends.  It’s understandable, really.  As literary people they’re disposed to think about books as ends in themselves — as items people acquire for purposes of reading.  Indeed, this has long been the “being” of books, especially physical ones. With Kindle, however, books are in the process of getting an existential makeover.  Today they’re becoming prompts for all sorts of personal and ambient information, much of which then goes on to become proprietary to Amazon.com.

I would venture to speculate that, despite the success of the Nook, Barnes & Noble has yet to fully wake up to this fact as well.  For more than a century the company has fancied itself a bookseller — this in contrast to Amazon, which CEO Jeff Bezos once described as “a technology company at its core” (Advertising Age, June 1, 2005).  The one sells books, the other bandies in information (which is to say nothing of all the physical stuff Amazon sells).  The difference is fundamental.

Where does all this leave us, then?  First and foremost, publishers need to begin recognizing the dual existence of their Kindle books: that is, as both means and ends.  I suppose they should also press Amazon for some type of “cut” — informational, financial, or otherwise — since Amazon is in a manner of speaking free-riding on the publishers’ products.

This last point I raise with some trepidation, though; the humanist in me feels a compulsion to pull back.  Indeed it’s here that I begin to glimpse the realization of O. H. Cheney’s world, where matters of the heart are anathema and reason, guided by information, dictates virtually all publishing decisions.  I say this in the thick of the Kindle edition of Walter Isaacson’s biography of Steve Jobs, where I’ve learned that intuition, even unbridled emotion, guided much of Jobs’ decision making.

Information may be the order of the day, but that’s no reason to overlook what Jobs so successfully grasped.  Technology alone isn’t enough.  It’s best when “married” to the liberal arts and humanities.

Share

The Book Industry's Moneyball

February 6, 2012 About the Book, Algorithmic Culture, Related Work No comments
Email, RSS Follow

Some folks have asked me how I came to the idea of algorithmic culture, the subject of my next book as well as many of my blog posts of late.  I usually respond by pointing them in the direction of chapter three of The Late Age of Print, which focuses on Amazon.com, product coding, and the rise digital communications in business.

It occurs to me, though, that Amazon wasn’t exactly what inspired me to begin writing about algorithms, computational processes, and the broader application of principles of scientific reason to the book world.  My real inspiration came from someone you’ve probably never heard of before (unless, of course, you’ve read The Late Age of Print). I’m talking about Orion Howard (O. H.) Cheney, a banker and business consultant whose ideas did more to lay the groundwork for today’s book industry than perhaps anyone’s.

Cheney was born in 1869 in Bloomington, Illinois.  For much of his adult life he lived and worked in New York State, where, from 1909-1911, he served as the State Superintendent of Banks and later as a high level executive in the banking industry.  In 1932 he published what was to be the first comprehensive study of the book business in the United States, the Economic Survey of the Book Industry, 1930-1931.  It almost immediately came to be known as the “Cheney Report” due to the author’s refusal to soft-peddle his criticisms of, well, pretty much anyone who had anything to do with promoting books in the United States — from authors and publishers on down to librarians and school teachers, and everyone else in between.

In essence, Cheney wanted to fundamentally rethink the game of publishing.  His notorious report was the book industry equivalent of Moneyball.

If you haven’t read Michael Lewis’ Moneyball: The Art of Winning an Unfair Game (2003), you should.  It’s about how the Oakland A’s, one of the most poorly financed teams in Major League Baseball, used computer algorithms (so-called “Sabermetrics“) to build a successful franchise by identifying highly skilled yet undervalued players.  The protagonists of Moneyball, A’s General Manager Billy Bean and Assistant GM Paul DePodesta, did everything in their power to purge gut feeling from the game.  Indeed, one of the book’s central claims is that assessments of player performance have long been driven by unexamined assumptions about how ball players ought to look, move, and behave, usually to a team’s detriment.

The A’s method for identifying talent and devising on-field strategy raised the ire of practically all baseball traditionalists.  It yielded insights that were so far afield of the conventional wisdom that its proponents were apt to seem crazy, even after they started winning big.

It’s the same story with The Cheney Report.  Consider this passage, where Cheney faults the book industry for operating on experience and intuition instead of a statistically sound “fact basis”:

Facts are the only basis for management in publishing, as they must be in any field.  In that respect, the book industry is painfully behind many others — both in facts relating to the industry as a whole and in facts of individual [publishing] houses….”Luck”; waiting for a best-seller; intuitive publishing by a “born publisher” — these must give way as the basis for the industry, for the sake of the industry and everybody in it….In too many publishing operations the theory seems to be that learning from experience means learning how to do a thing right by continuing to do it wrong (pp. 167-68).

This, more than 70 years before Moneyball!  And, like Beane and DePodesta, Cheney was raked over the coals by almost everyone in the industry he was criticizing.  They refused to listen to him, despite the fact that, in the throes of the Great Depression, most everything that had worked in the book industry didn’t seem to be working so well anymore.

Well, it’s almost the same story. Beane and DePodesta have enjoyed excellent careers in Major League Baseball, despite the heresy of their ideas.  They’ve been fortunate to have lived at a time when algorithms and computational mathematics are enough the norm that at least some can recognize the value of what they’ve brought to the game.

The Cheney Report, in contrast, had almost no immediate effect on the book industry.  The Report suffered due to its — and Cheney’s own — untimeliness.  The cybernetics revolution was still more than a decade off, and so the idea of imagining the book industry as a complexly communicative ecosystem was all but unimaginable to most.  This was true even with Cheney, who, in his insistence on ascertaining the “facts,” was fumbling around for what would later come to be known as “information.”

Today we live in O. H. Cheney’s vision for the book world, or, at least, some semblance of it.  People wonder why Amazon.com has so shaken up all facets of the industry.  It’s an aggressive competitor, to be sure, but its success is premised more on its having fundamentally rethought the game.  And for this Jeff Bezos owes a serious thank you to a grumpy old banker who, in the 1930s, wrote the first draft of what would go on to become publishing’s new playbook.

Share

What is an Algorithm?

January 30, 2012 Algorithmic Culture, Related Work One comment
Email, RSS Follow

For close to two years now I’ve been blogging about “algorithmic culture” — the use of computational processes to sort, classify, and hierarchize people, places, objects, and ideas.  Since I began there’s been something of a blossoming of work on the topic, including a recent special issue of the journal Theory, Culture and Society on codes and codings (see, in particular, the pieces by Amoore on derivatives trading and Cheney-Lippold on algorithmic identity). There’s also some excellent work developing around the idea of “algorithmic literacies,” most notably by Tarleton Gillespie and Cathy Davidson.  Needless to say, I’m pleased to have found some fellow travelers.

One of the things that strikes me about so much of the work on algorithmic culture, however rigorous and inspiring it may be, is the extent to which the word algorithm goes undefined.  It is as if the meaning of the word were plainly apparent: it’s just procedural math, right, mostly statistical in nature and focused on large data sets?  Well, sure it is, but to leave the word algorithm at that is to resign ourselves to living with a mystified abstraction.  I’m not willing to do that. To understand what algorithms do to culture, and the emerging culture of algorithms, it makes sense to spend some time figuring out what an algorithm is.

Even before getting into semantics, however, it’s worth thinking about just how prevalent the word algorithm is.  Why even bother if it’s just some odd term circulating on the fringes of language?  What’s interesting about algorithm is that, until about 1960 or so, it was exactly that type of word.  Here’s a frame grab from a search I ran recently on the Google Books Ngram Viewer, which allows you to chart the frequency of word usage in the data giant’s books database.

(Yes, I realize the irony in using the tools of algorithmic culture to study algorithmic culture.  We are always hopelessly complicit.)

What does this graph tell us?  First, algorithm remains a fairly specialized word, even to this day.  At its peak (circa 1995 or so) its frequency was just a notch over 0.0024%; compare that to the word the, which accounts for about 5% of all English language words appearing in the Google Books database.  More intriguing to me, though, is the fact that the word algorithm almost doesn’t register at all until about 1900, and that it’s a word whose stock has clearly been on the rise ever since 1960.  Indeed, the sharp pitch of the curve since then is striking, suggesting its circulation well beyond the somewhat insular confines of mathematics.

Should we assume that the word algorithm is new, then?  Not at all.  It is, in fact, a fairly old word, derived from the name of the 9th century Perian mathematician al-Khwārizmī, who developed some of the first principles of algebra.  Even more intriguing to me, though, is the fact that the word algorithm was not, until about 1960, the only form of the word in use.  Before then one could also speak of an algorism, with an “s” instead of a “th” in the middle.

Based on the numbers, algorism has never achieved particularly wide circulation, although its fortunes did start to rise around 1900.  Interestingly, it reaches its peak usage (as determined by Google) long about 1960, which is to say right around the same time algorithm starts to achieve broader usage.  Here’s what the two terms look like when charted together:

Google Ngram | Algorithm, Algorism

Where does all this leave us, then?  Before even attempting to broach the issue of semantics, or the meaning of the word algorithm, we first have to untangle a series of historical knots.

  • Why are there two forms of the “same” word?
  • Why does usage of algorithm take off around 1960?
  • Why does algorism fade after 1960, following a modest but steady 60 year rise?

I have answers to each of these questions, but the history’s so dense that it’s probably not best to share it in short form here on the blog.  (I give talks on the subject, however, and the material all will eventually appear in the book.)  For now, suffice it to say that any consideration of algorithms or algorithmic culture ought to begin not from the myopia of the present day but instead from the vantage point of history.

Indeed, it may be that the question, “what is an algorithm?” is the wrong one to ask — or, at least, the wrong one to ask, first.  Through what historical twists and turns did we arrive at today’s preferred senses of the word algorithm?  That seems to me a more pressing and pertinent question, because it compels us to look into the cultural gymnastics by which a word with virtually no cachet has grown into one whose referents increasingly play a decisive role in our lives.

Share

WordPress

October 24, 2011 Algorithmic Culture, Site Info No comments
Email, RSS Follow

Lest there be any confusion, yes, indeed, you’re reading The Late Age of Print blog, still authored by me, Ted Striphas.  The last time you visited, the site was probably red, white, black, and gray.  Now it’s not.  I imagine you’re wondering what prompted the change.ir-leasing.ru

The short answer is: a hack.  The longer answer is: algorithmic culture.polvam.ru

At some point in the recent past, and unbeknownst to me, The Late Age of Print got hacked.  Since then I’ve been receiving sporadic reports from readers telling me that their safe browsing software was alerting them to a potential issue with the site.  Responsible digital citizen that I am, I ran numerous malware scans using multiple scanning services.  Only one out of twenty-three of those services ever returned a “suspicious” result, and so I figured, with those odds, that the one positive must be an anomaly.  It was the same service that the readers who’d contacted me also happened to be using.

Well, last week, Facebook implemented a new partnership with an internet security company called Websense.  The latter checks links shared on the social networking site for malware and the like.  A friend alerted me that an update I’d posted linking to Late Age came up as “abusive.”  That was enough; I knew something must be wrong.  I contacted my web hosting service and asked them to scan my site.  Sure enough, they found some malicious code hiding in the back-end.

Here’s the good news: as far as my host and I can tell, the code — which, rest assured, I’ve cleaned — had no effect on readers of Late Age or your computers.  (Having said that, it never hurts to run an anti-virus/malware scan.)  It was intended only for Google and other search engines, and its effects were visible only to them.  The screen capture, below, shows how Google was “seeing” Late Age before the cleanup.  Neither you nor I ever saw anything out of the ordinary around here.

Essentially the code grafted invisible links to specious online pharmacies onto the legitimate links appearing in many of my posts.  The point of the attack, when implemented widely enough, is to game the system of search.  The victim sites all look as if they’re pointing to whatever website the hacker is trying to promote. And with thousands of incoming links, that site is almost guaranteed to come out as a top result whenever someone runs a search query for popular pharma terms.

So, in case you were wondering, I haven’t given up writing and teaching for a career hocking drugs to combat male-pattern baldness and E.D.

This experience has been something of an object lesson for me in the seedier side of algorithmic culture.  I’ve been critical of Google, Amazon, Facebook, and other such sites for the opacity of the systems by which they determine the relevance of products, services, knowledge, and associations.  Those criticisms remain, but now I’m beginning to see another layer of the problem.  The hack has shown me just how vulnerable those systems are to manipulation, and how, then, the frameworks of trust, reputation, and relevance that exist online are deeply — maybe even fundamentally — flawed.

In a more philosophical vein, the algorithms about which I’ve blogged over the last several weeks and months attempt to model “the real.”  They leverage crowd wisdom — information coming in the form of feedback — in an attempt to determine what the world thinks or how it feels about x.  The problem is, the digital real doesn’t exist “out there” waiting to be discovered; it is a work in progress, and much like The Matrix, there are those who understand far better than most how to twist, bend, and mold it to suit their own ends.  They’re out in front of the digital real, as it were, and their actions demonstrate how the results we see on Google, Amazon, Facebook, and elsewhere suffer from what Meaghan Morris has called, in another context, “reality lag.”  They’re not the real; they’re an afterimage.

The other, related issue here concerns the fact that, increasingly, we’re placing the job of determining the digital real in the hands of a small group of authorities.  The irony is that the internet has long been understood to be a decentralized network and lauded, then, for its capacity to endure even when parts of it get compromised.  What the hack of my site has underscored for me, however, is the extent to which the internet has become territorialized of late and thus subject to many of the same types of vulnerabilities it was once thought to have thwarted.  Algorithmic culture is the new mass culture.

Moving on, I’d rather not have spent a good chunk of my week cleaning up after another person’s mischief, but at least the attack gave me an excuse to do something I’d wanted to do for a while now: give Late Age a makeover.  For awhile I’ve been feeling as if the site looked dated, and so I’m happy to give it a fresher look.  I’m not yet used to it, admittedly, but of course feeling comfortable in new style of anything takes time.

The other major change I made was to optimize Late Age for viewing on mobile devices.  Now, if you’re visiting using your smart phone or tablet computer, you’ll see the same content but in significantly streamlined form.  I’m not one to believe that the PC is dead — at least, not yet — but for better or for worse it’s clear that mobile is very much at the center of the internet’s future.  In any case, if you’re using a mobile device and want to see the normal Late Age site, there’s a link at the bottom of the screen allowing you to switch back.

I’d be delighted to hear your feedback about the new Late Age of Print.  Drop me a line, and thanks to all of you who wrote in to let me know something was up with the old site.

 

Share

Algorithmic Literacies

October 17, 2011 Algorithmic Culture, Digital Humanities, Related Work 8 comments
Email, RSS Follow

I’ve spent the last few weeks here auditioning ideas for my next book, on the topic of  “algorithmic culture.”  By this I mean the use of computers and complex mathematical routines to sort, classify, and create hierarchies for our many forms of human expression and association.dekor-okno.ru

I’ve been amazed by the reception of these posts, not to mention the extent of their circulation.  Even more to the point, the feedback I’ve been receiving has already prompted me to address some of the gaps in the argument — among them, the nagging question of “what is to be done?”

I should be clear that however much I may criticize Google, Facebook, Netflix, Amazon, and other leaders in the tech industry, I’m a regular user of their products and services.  When I get lost driving, I’m happy that Google Maps is there to save the day.  Facebook has helped me to reconnect with friends whom I thought were lost forever.  And in a city with inadequate bookstores, I’m pleased, for the most part, to have Amazon make suggestions about which titles I ought to know about.

In other words, I don’t mean to suggest that life would be better off without algorithmic culture.  Likewise, I don’t mean to sound as if I’m waxing nostalgic for the “good old days” when small circles of élites got to determine “the best that has been thought and said.”  The question for me is, how might we begin to forge a better algorithmic culture, one that provides for more meaningful participation in the production of our collective life?

It’s this question that’s brought me to the idea of algorithmic literacies, which is something Eli Pariser also talks about in the conclusion of The Filter Bubble. 

I’ve mentioned in previous posts that one of my chief concerns with algorithmic culture has to do with its mysteriousness.  Unless you’re a computer scientist with a Ph.D. in computational mathematics, you probably don’t have a good sense of how algorithmic decision-making actually works.  (I count myself among the latter group.)  Now, I don’t mean to suggest that everyone needs to study computational mathematics, although some basic understanding of the subject couldn’t hurt.  I do mean to suggest, however, that someone needs to begin developing strategies by which to interpret both the processes and products of algorithmic culture, critically.  That’s what I mean, in a very broad sense, by “algorithmic literacies.”

In this I join two friends and colleagues who’ve made related calls.  Siva Vaidhyanathan has coined the phrase “Critical Information Studies” to describe an emerging “transfield” concerned with (among other things) “the rights and abilities of users (or consumers or citizens) to alter the means and techniques through which cultural texts and information are rendered, displayed, and distributed.”  Similarly, Eszter Hargittai has pointed to the inadequacy of the notion of the “digital divide” and has suggested that people instead talk about the uneven distribution of competencies in digital environments.

Algorithmic literacies would proceed from the assumption that computational processes increasingly influence how we perceive, talk about, and act in the world.  Marxists used to call this type of effect “ideology,” although I’m not convinced of the adequacy of a term that still harbors connotations of false consciousness.  Maybe Fredric Jameson’s notion of “cognitive mapping” is more appropriate, given the many ways in which algorithms help us to get our bearings in world abuzz with information.  In any case, we need to start developing a  vocabulary, one that would provide better theoretical tools with which to make sense of the epistemological, communicative, and practical entailments of algorithmic culture.

Relatedly, algorithmic literacies would be concerned with the ways in which individuals, institutions, and technologies game the system of life online. Search engine optimization, reputation management, planted product reviews, content farms — today there are a host of ways to exploit vulnerabilities in the algorithms charged with sifting through culture.  What we need, first of all, is to identify the actors chiefly responsible for these types of malicious activities, for they often operate in the shadows.  But we also need to develop reading strategies that would help people to recognize instances in which someone is attempting to game the system.  Where literary studies teaches students how to read for tone, so, too, would those of us invested in algorithmic literacies begin teaching how to read for evidence of this type of manipulation.

Finally, we need to undertake comparative work in an effort to reverse engineer Google, Facebook, and Amazon, et al.’s proprietary algorithms.  One of the many intriguing parts of The Googlization of Everything is the moment where Vaidhyanathan compares and contrasts the Google search results that are presented to him in different national contexts.  A search for the word “Jew,” for example, yields very different outcomes on the US’s version of Google than it does on Germany’s, where anti-Semitic material is banned.  The point of the exercise isn’t to show that Google is different in different places; the company doesn’t hide that fact at all.  The point, rather, is to use the comparisons to draw inferences about the biases — the politics — that are built into the algorithms people routinely use.

This is only a start.  Weigh in, please.  Clearly there’s major work left to do.

Share

The Conversation of Culture

October 10, 2011 Algorithmic Culture, Hype 2 comments
Email, RSS Follow

Last week I was interviewed on probably the best talk radio program about culture and technology, the CBC’s Spark. The interview grew out of my recent series of blog posts on the topic of algorithmic culture.  You can listen to the complete interview, which lasts about fifteen minutes, by following the link on the Spark website.  If you want to cut right to the chase and download an mp3 file of the complete interview, just click here.focuz.ru

The hallmark of a good interviewer is the ability to draw something out of an interviewee that she or he didn’t quite realize was there.  That’s exactly what the host of Spark, Nora Young, did for me.  She posed a question that got me thinking about the process of feedback as it relates to algorithmic culture — something I’ve been faulted on, rightly, in the conversations I’ve been having about my blog posts and scholarly research on the subject.  She asked something to the effect of, “Hasn’t culture always been a black box?”  The implication was: hasn’t the process of determining what’s culturally worthwhile always been mysterious, and if so, then what’s so new about algorithmic culture?

The answer, I believe, has everything to do with the way in which search engine algorithms, product and friend recommendation systems, personalized news feeds, and so forth incorporate our voices into their determinations of what we’ll be exposed to online.

They rely, first of all, on signals, or what you might call latent feedback.  This idea refers to the information about our online activities that’s recorded in the background, as it were, in a manner akin to eavesdropping.  Take Facebook, for example.  Assuming you’re logged in, Facebook registers not only your activities on its own site but also every movement you make across websites with an embedded “like” button.

Then there’s something you might call direct feedback, which refers to the information we voluntarily give up about ourselves and our preferences.  When Amazon.com asks if a product it’s recommended appeals to you, and you click “no,” you’ve explicitly told the company it got that one wrong.

So where’s the problem in that?  Isn’t it the case that these systems are inherently democratic, in that they actively seek and incorporate our feedback?  Well, yes…and no.  The issue here has to do with the way in which they model a conversation about the cultural goods that surround us, and indeed about culture more generally.

The work of culture has long happened inside of a black box, to be sure.  For generations it was chiefly the responsibility of a small circle of white guys who made it their business to determine, in Matthew Arnold’s famous words, “the best that has been thought and said.”

Only the black box wasn’t totally opaque.  The arguments and judgments of these individuals were never beyond question.  They debated fiercely among themselves, often quite publicly; people outside of their circles debated them equally fiercely, if not more so.  That’s why, today, we teach Toni Morrison’s work in our English classes in addition to that of William Shakespeare.

The question I raised near the end of the Spark interview is the one I want to raise here: how do you argue with Google?  Or, to take a related example, what does clicking “not interested” on an Amazon product recommendation actually communicate, beyond the vaguest sense of distaste?  There’s no subtlety or justification there.  You just don’t like it.  Period.  End of story.  This isn’t communication as much as the conveyance of decontextualized information, and it reduces culture from a series of arguments to a series of statements.

Then again, that may not be entirely accurate.  There’s still an argument going on where the algorithmic processing of culture is concerned — it just takes place somewhere deep in the bowels of a server farm, where all of our movements and preferences are aggregated and then filtered.  You can’t argue with Google, Amazon, or Facebook, but it’s not because they’re incapable of argument.  It’s because their systems perform the argument for us, algorithmically.  They obviate the need to justify our preferences to one another, and indeed, before one another.

This is a conversation about culture, yes, but minus its moral obligations.

Share

Cultural Informatics

October 3, 2011 Algorithmic Culture No comments
Email, RSS Follow

In my previous post I addressed the question, who speaks for culture in an algorithmic age?  My claim was that humanities scholars once held significant sway over what ended up on our cultural radar screens but that, today, their authority is diminishing in importance.  The work of sorting, classifying, hierarchizing, and curating culture now falls increasingly on the shoulders of engineers, whose determinations of what counts as relevant or worthy result from computational processes.  This is what I’ve been calling, “algorithmic culture.”

The question I want to address this week is, what assumptions about culture underlie the latter approach?  How, in other words, do engineers — particularly computer scientists — seem to understand and then operationalize the culture part of algorithmic culture?

My starting point is, as is often the case, the work of cultural studies scholar Raymond Williams.  He famously observed in Keywords (1976) that culture is “one of the two or three most complicated words in the English language.”  The term is definitionally capacious, that is to say, a result of centuries of shedding and accreting meanings, as well as the broader rise and fall of its etymological fortunes.  Yet, Williams didn’t mean for this statement to be taken as merely descriptive; there was an ethic implied in it, too see this site.  Tread lightly in approaching culture.  Make good sense of it, but do well not to diminish its complexity.

Those who take an algorithmic approach to culture proceed under the assumption that culture is “expressive.”  More specifically, all the stuff we make, practices we engage in, and experiences we have cast astonishing amounts of information out into the world.  This is what I mean by “cultural informatics,” the title of this post.  Algorithmic culture operates first of all my subsuming culture under the rubric of information — by understanding culture as fundamentally, even intrinsically, informational and then operating on it accordingly.

One of the virtues of the category “information” is its ability to link any number of seemingly disparate phenomena together: the movements of an airplane, the functioning of a genome, the activities of an economy, the strategies in a card game, the changes in the weather, etc.  It is an extraordinarily powerful abstraction, one whose import I have come to appreciate, deeply, over the course of my research.

The issue I have pertains to the epistemological entailments that flow from locating culture within the framework of information.  What do you have to do with — or maybe to — culture once you commit to understanding it informationally?

The answer to this question begins with the “other” of information: entropy, or the measure of a system’s disorder.  The point of cultural informatics is, by and large, to drive out entropy — to bring order to the cultural chaos by ferreting out the signal that exists amid all the noise.  This is basically how Google works when you execute a search.  It’s also how sites like Amazon.com and Netflix recommend products to you.  The presumption here is that there’s a logic or pattern hidden within culture and that, through the application of the right mathematics, you’ll eventually come to find it.

There’s nothing fundamentally wrong with this understanding of culture.  Something like it has kept anthropologists, sociologists, literary critics, and host of others in business for well over a century.  Indeed there are cultural routines you can point to, whether or not you use computers to find them.  But having said that, it’s worth mentioning that culture consists of more than just logic and pattern.  Intrinsic to culture is, in fact, noise, or the very stuff that gets filtered out of algorithmic culture.

At least, that’s what more recent developments within the discipline of anthropology teach us.  I’m thinking of Renato Rosaldo‘s fantastic book Culture and Truth (1989), and in particular of the chapter, “Putting Culture in Motion.”  There Rosaldo argues for a more elastic understanding of culture, one that refuses to see inconsistency or disorder as something needing to be purged.  “We often improvise, learn by doing, and make things up as we go along,” he states.  He puts it even more bluntly later on: “Do our options really come down to the vexed choice between supporting cultural order or yielding to the chaos of brute idiocy?”

The informatics of culture is oddly paradoxical in that it hinges on a more and less powerful conceptualization of culture.  It is more powerful because of the way culture can be rendered equivalent, informationally speaking, with all of those phenomena (and many more) I mentioned above.  And yet, it is less powerful because of the way the livingness, the inventiveness — what Eli Pariser describes as the “serendipity” — of culture must be shed in the process of creating that equivalence.

What is culture without noise?  What is culture besides noise?  It is a domain of practice and experience diminished in its complexity.  And it is exactly the type of culture Raymond Williams warned us about, for it is one we presume to know but barely know the half of.

Share

Who Speaks for Culture?

September 26, 2011 Algorithmic Culture 9 comments
Email, RSS Follow

I’ve blogged off and on over the past 15 months about “algorithmic culture.”  The subject first came to my attention when I learned about the Amazon Kindle’s “popular highlights” feature, which aggregates data about the passages Kindle owners have deemed important enough to underline.Укладка дикого камня

Since then I’ve been doing a fair amount of algorithmic culture spotting, mostly in the form of news articles.  I’ve tweeted about a few of them.  In one case, I learned that in some institutions college roommate selection is now being determined algorithmically — often, by  matching up individuals with similar backgrounds and interests.  In another, I discovered a pilot program that recommends college courses based on a student’s “planned major, past academic performance, and data on how similar students fared in that class.”  Even scholarly trends are now beginning to be mapped algorithmically in an attempt to identify new academic disciplines and hot-spots.

There’s much to be impressed by in these systems, both functionally and technologically.  Yet, as Eli Pariser notes in his highly engaging book The Filter Bubble, a major downside is their tendency to push people in the direction of the already known, reducing the possibility for serendipitous encounters and experiences.

When I began writing about “algorithmic culture,” I used the term mainly to describe how the sorting, classifying, hierarchizing, and curating of people, places, objects, and ideas was beginning to be given over to machine-based information processing systems.  The work of culture, I argued, was becoming increasingly algorithmic, at least in some domains of life.

As I continue my research on the topic, I see an even broader definition of algorithmic culture starting to emerge.  The preceding examples (and many others I’m happy to share) suggest that some of our most basic habits of thought, conduct, and expression — the substance of what Raymond Williams once called “culture as a whole way of life” — are coming to be affected by algorithms, too.  It’s not only that cultural work is becoming algorithmic; cultural life is as well.

The growing prevalence of algorithmic culture raises all sorts of questions.  What is the determining power of technology?  What understandings of people and culture — what “affordances” — do these systems embody? What are the implications of the tendency, at least at present, to encourage people to inhabit experiential and epistemological enclaves?

But there’s an even more fundamental issue at stake here, too: who speaks for culture?

For the last 150 years or so, the answer was fairly clear.  The humanities spoke for culture and did so almost exclusively.  Culture was both its subject and object.  For all practical purposes the humanities “owned” culture, if for no other reason than the arts, language, and literature were deemed too touchy-feely to fall within the bailiwick of scientific reason.

Today the tide seems to be shifting.  As Siva Vaidhyanathan has pointed out in The Googlization of Everything, engineers — mostly computer scientists — today hold extraordinary sway over what does or doesn’t end up on our cultural radar.  To put it differently, amid the din of our pubic conversations about culture, their voices are the ones that increasingly get heard or are perceived as authoritative.  But even this statement isn’t entirely accurate, for we almost never hear directly from these individuals.  Their voices manifest themselves in fragments of code and interface so subtle and diffuse that the computer seems to speak, and to do so without bias or predilection.

So who needs the humanities — even the so-called “digital humanities” — when your Kindle can tell you what in your reading you ought to be paying attention to?

Share

The Billion Dollar Book

May 2, 2011 Algorithmic Culture, Bookselling One comment
Email, RSS Follow

About a week ago Michael Eisen, who teaches evolutionary biology at UC Berkeley, blogged about a shocking discovery one of his postdocs had made in early April.  The discovery happened not in his lab, but of all places on Amazon.com.abisgroup.ru

While searching the site for a copy of Peter Lawrence’s book The Making of a Fly (1992), long out of print, the postdoc happened across two merchants selling secondhand editions for — get this — $1.7 million and $2.2 million respectively!  A series of price escalations ensued as Eisen returned to the product page over following days and weeks until one seller’s copy topped out at $23 million.

But that’s not the worst of it.  One of the comments Eisen received on his blog post pointed to a different secondhand book selling on Amazon for $900 million.  It wasn’t an original edition of the Gutenberg Bible from 1463, nor was it a one-of-a-kind art book, either.  What screed was worth almost $1 billion?  Why, a paperback copy of actress Lana Turner’s autobiography, published in 1991, of course!  (I suspect the price may change, so in the event that it does, here’s a screen shot showing the price on Saturday, April 30th.)

Good scientist that he is, Eisen hypothesized that something wasn’t right about the prices on the fly book.  After all, they seemed to be adjusting themselves upward each time he returned to the site, and like two countries engaged in an arms race, they always seemed to do so in relationship to each other.  Eisen crunched some numbers:

On the day we discovered the million dollar prices, the copy offered by bordeebook [one of the sellers] was 1.270589 times the price of the copy offered by profnath [the other seller].  And now the bordeebook copy was 1.270589 times profnath again. So clearly at least one of the sellers was setting their price algorithmically in response to changes in the other’s price. I continued to watch carefully and the full pattern emerged. (emphasis added)

So the culprit behind the extraordinarily high prices wasn’t a couple of greedy (or totally out of touch) booksellers.  It was, instead, the automated systems — the computer algorithms — working behind the scenes in response to perceived market dynamics.

I’ve spent the last couple of blog posts talking about algorithmic culture, and I believe what we’re seeing here — algorithmic pricing — may well be an extension of it.

It’s a bizarre development.  It’s bizarre not because computers are involved in setting prices (though in this case they could have been doing a better job of it, clearly).  It is bizarre because of the way in which algorithms are being used to disrupt and ultimately manipulate — albeit not always successfully — the informatics of markets.

Indeed, I’m becoming  convinced that algorithms (at least as I’ve been talking about them) are a response to the decentralized forms of social interaction that grew up out of, and against, the centralized forms of culture, politics, and economics that were prevalent in the second and third quarters of 20th century.  Interestingly, the thinkers who conjured up the idea of decentralized societies often turned to markets — and more specifically, to the price system — in an attempt to understand how individuals distributed far and wide could effectively coordinate their affairs absent governmental and other types of intervention.

That makes me wonder: are the algorithms being used on Amazon and elsewhere an emergent form of “government,” broadly understood?  And if so, what does a billion dollar book say about the prospects for good government in an algorithmic age?

Share
« Older Entries   Recent Entries »