James Drogan

Re: Conversations wih Dave

So what? Rage doesn't help. Constructive criticism helps. This might be an article that inflames the masses, but it does little else.

This is far, far beneath a person of Gelernter's accomplishment and insight. I doubt I would accept this as a paper from an undergraduate, and certainly wouldn't accept it from a graduate.

I've embedded other comments in the article.

I suspect that Gelernter wrote this by hand and submitted by the postal service to the Wall Street Journal. Too cynical? You think?

Rage Against the Machine

By DAVID GELERNTER

October 6, 2005; Page A14

I hate our computers. Our core software tools are old; not only are they old but they're obsolete; and not only are they obsolete but they were never all that great to begin with.

Where was Gelernter all this time? What was he doing and has he been doing to ameliorate the situation?

Machines should adjust to people, not people to machines. But today our core software -- the file system, windows interface, email programs and many other fundamental tools -- comes from the '60s and '70s, pre-dating the era in which computers became powerful enough to adjust to users. Our antique system represents a long-ago way of thinking about computers. Even the word "computer" is archaic. Hardly anyone uses a computer for computing. Most of us use computers mainly for managing or accessing information.

I grant Gelernter's argument here. He misses, I think, a couple of important factors. First, not only are the computer more powerful today, but the adopters and adapters of technology have also become more powerful. See Notes from the IT Frontline October 1, 2003 for Drogan's brief history of time.

Second, there is a huge sunk investment, not only in money, but also in ideas and, perhaps most importantly, prestige and ego. This is not easily overcome. Perhaps he needs to be reminded of Machiavelli.

The stuck-in-timeness of our software and our thinking has important consequences, vital for both the industry and the world economy. When people buy a new personal computer today, chances are there is nothing important the new PC can do that the old one couldn't do just as well. And there is no law that says computers have to be more important -- culturally, economically or scientifically -- than any other technology that's been around for awhile. Some day, computers might be exactly as exciting as oil burners or toaster ovens. When that happens, it will be the end of the computer industry as we know it. The economic consequences for the world -- and especially for the U.S. -- will be tremendous.

A bold ending sentence which makes one yearn for additional details.

* * *

Of course, old doesn't meant obsolete. But here are four of the worst brilliant ideas in technology history: the file system, the mailer, the Web and the desktop interface (windows, menus, icons, mouse). Each one has a fatal flaw built in, fundamental. Some of them have several built-in fatal flaws, for failproof redundancy.

The file system, for instance, is so bad that for many people, once something falls off the desktop, it's lost forever. Even for all sorts of intelligent people, the file system is a black hole. It is so bad that, for many people, the storage capacity of the computer -- of the multigigabyte modern computer -- is limited by the size of the piece of glass that defines the screen.

Perhaps Gelernter hasn't heard of desk top search tools, tags, searching the web by concept, and the absence of folders in gmail.

And consider email. People have gotten junk mail through the U.S. Post Office for years: Why do they shrug it off, whereas electronic spam is quite literally a federal case? Why isn't spam easier to deal with than junk mail? (It should be easier to deal with electronic than physical documents; but in the case of email, it's easier in practice to deal with physical ones.)

It seem to me that the answer to this is obvious. People are much more capable of pattern recognition than machines. People are much more adaptive to new stimuli than machines. However, people do not understand how they do pattern recognition and how they are adaptable to new stimuli. Without this understanding, people can't write rules. If you can't write rules, then you can't instruct machines.

Mailers shouldn't show us (in effect) the envelopes. A good secretary opens the envelopes; in the same way, a good mailer should show us a stack of letters, not a list of unopened envelopes.

Sounds like Outlook to me.

Or consider the Internet. Think of flipping through a magazine -- you can take in an extraordinary amount of information amazingly fast by using the flip-through method of browsing. Why can't I flip through a Web site?

I grant a bit of Gelernter's argument here. For example, I scan the Wall Street Journal and New York Times every day, but I also scan the physical paper. It's easier for me to see things I wasn't looking for when I scan the physical paper. Part of the issue revolves around the size of the window the computer allows us to have on the world (Gelernter's comment on the "piece of glass") and part of the issue revolves around available bandwidth.

Is a software revolution wildly improbable? Just the opposite: It's wildly probable. The technology field revolutionizes itself regularly. There hasn't been a single decade in digital computer history so far -- not the '50s, '60s, '70s, '80s or '90s -- that hasn't seen at least one industry-transforming revolution.

The issue this raises in my mind is the potential and probable need to learn a new metaphor in order to manage knowledge. See Machiavelli (again).

Moreover, virtually all major information technologies of the last 100 years were transformed, soon after they were invented, from new technologies to new media. (At first, film was a new technology. Before long it became a new medium.) Engineers no longer run the show; artists and content producers take over. But that transition hasn't happened in computing. When it does, the field will be transformed.

I think we would have to say that this transformation is, at a minimum, in its infancy. Kartoo and Vivisimo in search engines; Google Earth and its associated interesting hacks; Kognito Solutions in the field of learning; FLICKR and del.icio.us in the area of content management and delivery are examples of the transformation I think Gelernter is yearning for and I have experienced.

But the deeper question is what will the results of this transformation look like?

The future: Everyone has a personal, private cyberpool afloat in the cybersphere like an astral body, your own personal planet which you can tune in using any computer anywhere.

Poetic, perhaps romantic, but what the hell does Gelernter mean?

Microsoft and other companies recently noticed that people tend to use more than one computer. They store copies of important files on lots of computers, and they need to keep all these files updated and identical. New software can do this by sending information transparently over network connections. But no one is going to buy it. People hate installing this sort of fancy software gadget.

I want all my files to live "out there" in the cybersphere, stored in a secure, reliable way. I want to use a computer to "tune in" my information the way my TV tunes in Fox or a telescope "tunes in" Jupiter. This is the way the multiple-computer problem will be solved. When I buy a new computer, I'd plug it in, turn it on, identify myself -- and my cyberpool would automatically be right there on my new machine. It would be like scooping a hole in beach sand near the surf-line -- you get the ocean rising, automatically, in your new sand-hole; and you'd get your cyberpool rising, automatically, in your new computer.

I agree with Gelernter on this one.

Fifteen years ago I was involved with an ad tech effort with a major railroad. The intent was to describe the technology infrastructure required to support their needs in the year 2000. Emerging from this effort were such diagrams as

Figure 4 Adaptive System 1

and

 

Figure 5 Adaptive System 2

A bit more on this project can be found at A Note on Business Drivers, Business Configuration, and Information Technology Strategy.

Autonomic computing and plug-and-play (upscaled) also apply here.

There is a lot that needs to be done here.

* * *

Today the electronic information that defines and explains our lives is spread out over millions of scattered, disconnected, rolling-everywhere beads. Computers will thread them all together onto a single necklace, one per human being. Each will be wildly heterogeneous -- our lives encompass all sorts of information -- but it will be one continuous strand. It will start from nothing and keep growing until we die. Institutions, too, have life stories, which will also be captured in these constantly-lengthening strands of cyber-coral and pearl, which will replace today's awkward Internet.

You and I have discussed this before and maybe, to some extent, we see examples of this.

What does a search return but beads of information strung together?

Or, if you prefer, you could imagine an all-inclusive superbeam encompassing all the infobeams in the world -- everyone's infobeam, and every Web site's infobeam, blended together into a single beam or "worldstream." Instead of a Web, you'd have the world, in effect, telling its story -- billions of different versions, all accumulating simultaneously, interleaved.

All you have to do is glance at your own computer to know that we can do better -- and must do better if this field is to continue growing and energizing the world economies. We can do better; we will do better.

Who will lead? And where will we be led to? It would be nice to know Gelernter's ideas of the what and how.

Mr. Gelernter is professor of computer science at Yale and senior fellow at Scientific Computing Associates in New Haven, Conn.