TCS Daily

The Deconstruction of Argument

By Robert McHenry - June 5, 2006 12:00 AM

Thank heaven for dreamers and visionaries. I guess. One has to admire their energy and imagination even when one wants to throttle them, or at least give them a good shaking. I have in mind Kevin Kelly, whose article -- drab noun! ... call it rather a tractate -- titled "Scan This Book!" appeared recently in the New York Times magazine.

Kelly anticipates, can't hardly wait for, the time in the immediate future when all books, articles, journals, movies, television shows, recorded music -- "[i]n short, the entire works of humankind, from the beginning of recorded history, in all languages" -- will be made available on the Internet. When this happens, and it is inevitable, and when all these myriad items of culture are indexed, linked, and tagged to a fare-thee-well, then... well, it's hard to tell exactly what will happen then, but it will be swell, trust him.

I snark, of course. That is because I am not a visionary, or not much of one.

To be fair, Kelly offers in the second half of his article a lucid and, I am willing to assume, accurate account of the evolution of copyright and the issues raised by such projects as Google's plan to scan and index the contents of five major research libraries and make those contents available to the world. And I hasten to add that I am all in favor of such projects, having had a little to do with some abortive ones early on.

But in the first half of the piece, setting up his argument, Kelly soars to such rhapsodic heights that it makes me wonder if he hasn't lost sight of some simple matters, such as what books actually are. He nods dutifully to the book, meaning the codex form, and credits it with the usual virtues -- ease of copying, ease of use, foundation of Western civilization, etc. But he writes as though all books were commonplace books, as though they were mere compilations of more or less interesting utterances. On the Web, in his vision, these physical compilations will be dissolved into their atomic components:

[O]nce digitized, books can be unraveled into single pages or be reduced further, into snippets of a page. These snippets will be remixed into reordered books and virtual bookshelves.

With all the linking and tagging across the former boundaries of books, "no book will be an island." What will it be, then? An archipelago? Worse, a constellation, an accidental grouping of components, mnemonically convenient but with no intrinsic logic? Worse still, a memory?

A book, a real book, is not a compilation, not even an ordered compilation, of sentences. It is an argument, a setting forth of some thinking from premises to conclusions. It is the work of a mind wholly engaged with a problem or a challenge. Take it apart, and the individual paragraphs and sentences are as useful as a handful of randomly selected widgets from the hardware store.

What makes Kelly's rapture-inducing future possible is the Search Engine, which is not so much the Holy Grail as the Golden Calf in this particular religion. It is the Search Engine, tracking down all those links and tags and indices, that pulls these floating snippets, these disconnected declamations, into the nonce-text, that assembles a momentary concatenation of supposedly relevant bits and pieces into some sort of response to an inquiry. Now, search engines are dumb. They are pattern matchers, and they differ only in the complexity of the patterns they have been designed to detect. One and all, they are just machinery, grinding away at their algorithms. Kelly will have it that they "harness the power of relationships," but in truth they do no more than point to resemblances, which may be substantial or merely superficial. The one thing they do not point to, or harness, or recognize, is truth.

If, one day, there is a genuine artificial intelligence capable of doing the job Kelly has in mind, I suspect it will not be content to sit about answering stupid questions from humans.

A few days after "Scan This Book!" appeared, John Updike offered some comments on Kelly's vision at a booksellers' convention. Updike was rather hard on Kelly's ideas about how, in a transcopyright world, authors would be compensated or even recognized for their efforts. Kelly's suggestions amounted to signing autographs. No reasonable person should be surprised that authors and publishers who rely on copyright to secure the monetary rewards of their work are unenthusiastic about this particular vision.

But it's not just about the money. No one who has written a real book could be charmed by the idea of its being disassembled into bits and pieces for the convenience of blunt-force search engineering. No thinker whose wrestling with ideas has become embodied in a carefully constructed sequence of carefully crafted sentences is likely to welcome the notion that some ruck of links and tags will permit the world's "readers" to call up isolated fragments and imagine that they have benefited themselves or somehow honored the author. The connections between and among books that such a scheme would reveal might be as provocative and productive as Kelly imagines, but if they come at the cost of the connections within books, it is too high a price.

There is more. As a way of heightening the contrast between his dynamic vision and the realm of books as it has existed heretofore, Kelly repeatedly depicts the latter as one of utter deadness. "The static world of book knowledge," he says, will give way to the Age of Relationship. But while it is true that the book sitting on the shelf seems inert, it is merely in its resting state, not unlike the computer that has not yet been turned on. It becomes its full self precisely in relationship -- relationship to the curious, eager, sympathetic reader. It is in that interaction, and only there, that any form of stored knowledge becomes more than nonrandom signals. It is there that signification is transformed into meaning. It is that relationship, should it survive, that will keep us out of Kelly's Attention Deficit universe.

The title of Kelly's rodomontade recalls, of course, Abbie Hoffman's Steal This Book. The fact that there is no apparent reason for the allusion only underlines the flood of accidental association that awaits us in a future where, as in a neutron star, the immense pressure of measureless content has quashed rational structure out of existence and we are left with a mush of words. Give the Yipster credit: he at least felt a book might be worth having.

Robert McHenry is Former Editor in Chief, the Encyclopedia Britannica, and author of How to Know (, 2004).



The Future of Learning
The value of digitized information cannot be fully realized by current search engine technologies. I just “googled” the exact phrase: “dark matter”. I received 7.2 million links. If my intent was to understand the subject of “dark matter”, I would have much work to do. Such is the status of current search technology. However, search technology is in its infant stage. By 2020 (at the latest), the search for “dark matter” could result in a research summary as opposed to just links. The search engine will not just find material on a subject, but actually “read and analyze” that material and present the inquirer with information as well as supporting links. Such capability is not true computer AI (creativity and awareness are not included or required). What is required and what is possible are a system of algorithms that can read, comprehend and summarize facts about a topic. Such a product could be called an “Intelligent Assistant [IA]” or an “Intelligent Interface”. As a practical matter, most IA inquiries will result in follow-up questions and a “conversation” of sorts that will continue until the user is satisfied. Competent and motivated inquirers will be able to nearly master any subject in a fraction of the time and at a fraction of the cost that exists today.

But this potential will not be fully realized until and unless all information-to-date is available to the search systems. Today, the complete digitations of known knowledge will yield useful references but not direct education. But in the future, the digitation of known information coupled with enhanced search technologies could lead to an exponential increase in the potential for learning.

truth versus nonsense
Your comment about the truth is very important. Search engines just search. They don't discriminate between truth, lies, hate, propaganda, etc. They treat them all equally.

Dream on
taBonfils: You stated: "The search engine will not just find material on a subject, but actually “read and analyze” that material and present the inquirer with information as well as supporting links." And how will it analyze? Will it be conservative, liberal, Christian, humanist, atheists, agnostic, scientific, or philosophical? This is simply dreaming because the computer will never stand up to the immaterial intellect of humans in analyzing. You disagree? Dream on.

No Subject
"You disagree?"

Thats an affirmative.

An illustration...I searched for "dark matter" at Clusty organized its 39.5 million responses by catagory as well as highlighting what it considered to be the most relevant links. This is the beginning of an "intelligent response" approach. And it will improve rapidly in the next two decades.

Dream on? Not Really.
I searched for "dark matter" at Clusty found 39.5 million links, displayed what it considered the most relevant on the first page AND catagorized (an intial analytical step) several of the references. This approach is the beginning of an "intelligent response" model. Given the competition and the advertising dollars at stake in the search engine business, I am confident the "intelligent response" model will improve dramatically (possibly exponentially) in the next 2 decades.

The Future of the Tyranny of Democracy
In the first place, "mastery of a subject" hardly comes from a single conversation, no matter how long or well-informed with raw data. Mastery comes from repeated practice and long, long thinking.

To those of us who know how to think, the Internet is nothing short of miraculous. But for the vast majority, it only makes them more misinformed about more things. Which is precisely why, for example, in the AI saturated future (of which I am a proponent), merely implanting information-receiving microchips into people's brains won't work to let them learn faster or deeper. Learning without thought is perilous.

In the second place, an individual work is more than a vast sum of snippets. It is a pattern, and some patterns are better than others. How do you propose to compensate, and compensate fairly, the massive efforts of those who undertake to write a great book with this vile, foul, and slimy stroke democratization? Bob Dylan was once asked why he deserved to make the money he did from his songs. He replied, "because for every stanza I've written there is a pool of blood on the floor."

“You must drop all your democracy. You must not believe in ''the people'.' One class is no better than another. It must be a case of Wisdom, or Truth. Let the working classes be working classes. That is the truth. There must be an aristocracy of people who have wisdom, and there must be a Ruler: a Kaiser: no Presidents and democracies,” said DH Lawrence.

Are you a dualist?
Immaterial intellect? It's only "immaterial" as long as it is a emergent feature from a complex system. Can "immaterial" things interact with material things in any way? If they can, then they're not so "immaterial".

About the "conservative, liberal, Christian..." point, it should be as intelligent as possible. That condition excludes the "liberal" and the "Christian" adjetives.

Intelligence (I'm not talking about "conscience", since we don't know yet what is it) is just an algebra that acts on representations of knowledge. I see no fundamental problem in implementing intelligent agents, except complexity.

TCS Daily Archives