Reply to Lanier
by David Gelernter
January 18th, 2006
Jaron’s piece is, as usual, a combination of statements that make me want to say “Hallelujah!” and others that strike me as dead wrong. Which explains why he remains, as far as I’m concerned, one of the few cyberintellectuals who is always worth reading. First I’ll comment on areas of disagreement, then on points where I think he’s exactly right.
Jaron writes that the Internet is “above all else an elaboration of the structure of computer software.” I believe he’s referring to the Internet as users see it—to the “software internet,” which is mainly the Web. But the Web isn’t based on the structure of our software; just the opposite, so to speak. The Web is based on the structure of our hardware. Just as the internet consists of a bunch of machines wired together essentially at random, the Web consists of a bunch of sites linked together essentially at random. Search-engines are successful because they overcome hypertext as the main organizing (or disorganizing) metaphor of the Web. Users who rely on search engines don’t need to worry about URLs, links or hypertext—which is why search engines are wildly successful.
(And, although it’s true that small changes in software are often catastrophic, small changes to the Web are usually irrelevant. A site goes down, wholly or partially; many sites go down; who cares? Usually, almost no one. Of course, it depends on the site!)
When Jaron says that “Files have become too fundamental to reconsider,” he’s right and wrong. For concreteness, I’ll focus on our own research. The “lifestreams” software Eric Freeman and I built at Yale, starting in 1995, was designed specifically to make files obsolete; to allow users to name their documents only where names seemed appropriate, and to dispense with directories completely. (”Directories” were created as needed, on the fly.) In 2001 I gave a talk specifically on the obsoleteness of files at a large commercial meeting in New York (John Schwartz wrote up the talk in the New York Times). In 2002, I published my own piece in the Times (11/7/02) called “Forget the Files and the Folders: Let Your Screen Reflect Life.” In 2005, the Economist magazine published a longish piece about the encroaching obsolescence of files, mentioning many projects—not just ours. So the message is getting through, albeit slowly. But Jaron is certainly right that the industry is in no hurry to re-examine the file idea.
He says that “A voting system in which there is absolute protection of voter privacy has never before existed. It is a digital phenomenon.” But his comments here are too implementation-dependent. He seems to assume that non-electronic voting depends on paper ballots (or at any rate, not on voting machines). But voting machines have been widely used for at least 40 years—especially (but not only) in the Northeast. In a voting machine booth, you step in, pull the handle and an opaque curtain closes, screening you off until you’ve finished—whereupon you pull the handle in the opposite direction and the curtain opens. These machines don’t supply “absolute protection of voter privacy.” But they do provide all the protection the average voter wants or needs. Furthermore, voting machines create no paper trails or stash of marked ballots to reassure people. But they’ve been used reliably for many decades.
Jaron is right, of course, that “China wants to build the physical computers that hold the bits.” But we ought to remember that China also wants to make sure that subversive bits (ones that encode talk about liberty, equality, and democracy) are screened out, so that Chinese Web-browsers never read them. And large American technology companies have been happy to help the Chinese dictatorship censor the Web-—-one of the grimmest incidents in American commercial history.
“Now we enter the endgame feared by the Luddites,” Jaron writes, “in which technology becomes so efficient that there aren’t any more jobs for people.” But of course the Luddites have been expecting this end game for several centuries. It hasn’t come so far, and computers won’t make it come. Because, in the end, people need and want physical, not electronic, stuff. Whether they’re poor or rich, whether they live in Bangladesh or Switzerland, people need to eat real (not virtual) food, wear real (not virtual) clothes, live in real (not virtual) houses. They need real furnaces to keep warm, fed by real fuel. And they need real computers to browse the Web. Rich people want fancy cars, homes, jewelry, toys. No brand-new cyberbillionaire ever used his billions to buy information, any more than the poor have ever worried about not having enough money to keep their families well-informed. So Luddites have nothing to worry about. Producing, hauling, selling, servicing, fueling, and junking all this real, physical stuff will provide plenty of work for as long as human beings exist.
“A program without a computer to run it,” Jaron says, “does not exist.” A fascinating claim, but on balance I don’t believe it. Suppose I take the word-processor I’m using right now and copy it to some sort of disk. Then I smash my computer so the program has no computer to run it. Later this afternoon, I buy another computer, same as the old one. Did my program stop existing during the interim, before I bought the new computer? I can’t see why. Granted, the bits on the disk (or the physical artifacts that record the presence of bits) became useless. But gasoline, by the same token, is useless unless I have a gasoline engine on hand (or some other use for gasoline—a dirty object to clean, say). Which doesn’t mean that gasoline doesn’t exist unless I have a gasoline engine on hand.
It’s true that (as Jaron says), reading old data is a classically hard problem. But it’s getting easier. I now carry around perhaps 5 years worth of readable files. In the future, I’ll carry around more. Each computer user is a comet streaking through the Cybersphere with a lengthening tail of information stretched out behind. Today NASA can’t read tapes from 40 years ago; but 40 years hence, chances are good that NASA will be able to read today’s tapes because lots of time and attention has been invested in the problem in the meantime. Whether NASA has time to read them, and whether there’s any point in reading them are, of course, different questions!)
Jaron writes that “the digital economy could be said to resemble a slave economy in the abstract.” I agree that the software world relies on enormous amounts of unpaid labor. This is true and important. But it’s very different from saying that the digital economy resembles a slave economy, even in the abstract. No slave ever picked cotton because he felt like it. But the thousands of programmers who kept UNIX alive by maintaining it, refining it, porting it, and expanding it (until it became as huge and ugly as any other operating system) did it mainly because they wanted to. Good programmers tend to like programming. Better programmers tend to like it even more. The best programmers like it most. And the best programmers are by far the most productive; so the digital economy depends on free labor donated by people who are, by and large, enjoying themselves.
I don’t understand Jaron’s antithesis between culture and capitalism. Shakespeare wrote plays so that a particular London theater company could earn enough money to support itself. No artist’s standards have ever been higher than Beethoven’s. But Beethoven wrote music to make a living and bent his composing (not always but sometimes) to the needs of the market.
I’ve run out of words, but, for the record, I strongly agree with Jaron that UNIX is pathetically obsolete (it was new and daring in 1977!). I agree that a programming “language” is not a language in any sense whatsoever. (It’s a specification for a machine—specifically for a machine built out of software.) I agree that browsing is crucial to the way human beings deal with information, and that software developers still don’t understand browsing. I agree that “a little leakage” can be hugely important, and it’s a tremendously important concept and a useful phrase. Google relies on the Internet (and the free labor that keeps it running) like an old-time carnival barker relied on the show; no show, no barker. The amount of time we all spend keeping our software and computers operating is huge and growing.
The underlying message I read in Jaron’s piece is: the software world doesn’t understand itself clearly enough—doesn’t understand where it’s been, where it is, and where it’s going. Which strikes me as true, and hugely important.
