Dr. Dobb's Journal June 2000
This month, in a spirit of paradigmatic breadth and as a way of saying at least something about a whole slew of topics, I offer an inch-deep view of the computing landscape. I concentrate on operating systems, the Internet, and certain technologies and visions that border on the science fictional. Those are not really three topics: Operating systems and the Internet are part of the science-fiction world in which we live today. Or more colorfully, they are the amniotic fluid and the oxygen, respectively, of the world we have been living in and the new world we are entering. Some of the items to follow have to do with the birth pangs of that transition.
A few points: Moore's Law is not a law in the same sense as Heisenberg's uncertainty principle or Einstein's speed limit. But a sufficiently outrageous violation of Moore's Law could lead to some outrageous sci-fi scenarios. Subverting Heisenberg or Einstein is canonical science fiction, but this, too, is starting to look possible, or at least less impossible.
A computer that acts human enough to pass the Turing test would apparently not violate fundamental laws of physics. Computers that merely play chess better than most humans are now mundane. But a computer that solves chess problems and is built out of RNA would represent an attack on the Turing test from both flanks: exhibiting human intelligence while being built from the same materials as humans.
Trippy and spooky though it might be, there's no obvious barrier to creating a planet-wide distributed computer system with computational power several orders of magnitude beyond individual computers. It's fitting that the most highly publicized, globally distributed computation is the search for other intelligent life in the universe. It would be a nice irony, and only a slight variation on a classic science-fiction story, if the answer to SETI@home's question, "Is there other intelligent life in the universe?" came back, "There is now."
Little by little, we begin to blur the line between human and computer. What was once a room-sized machine became a desktop office appliance, then a shirt-pocket gadget, now apparel, soon an implant. A teen-aged Bill Gates once gave up computers for a year, but future teen-agers might be unable to do that without resorting to surgery.
Step by step, we are creating the worlds of science-fiction writers. Some of those worlds, it's worth recalling, are dystopian nightmares.
Quantum computers, if there were any, would exploit the fact that a quantum particle prior to being observed exists in a weighted superposition of states: It is both 0 and 1. N quantum bits, or qubits, can thus hold all 2 to the nth power possible n-bit values simultaneously, and a computer built to exploit this capacity could perform multiple computations simultaneously on the same qubits, making it possible to solve classically intractable problems and, in particular, reinventing the field of cryptography.
Building a quantum computer is a very hard problem, though. Even creating a quantum logic gate is a remarkable achievement. This spring, Rainer Blatt of the University of Innsbruck reported on a technique that allowed him to realize a 4-qubit logic gate, a technique that he says would scale well to a lot of particles. "If we get to that level," said Christopher Monroe, a researcher at the National Institute of Standards and Technology (NIST), "We'll not only bring the strangest feature of quantum mechanics closer to the macroscopic world...but we may also have a quantum computer."
Nanotechnology is the field that attempts to realize artificial devices made up of a few molecules or even a few atoms. I say "realize" rather than "build" because many nanoresearchers see self-assembly as the key to success in nanomanufacturing. Manually placing all those molecules is not cost effective, but a nanofactory that sucks molecular building materials out of the air or somewhere and stamps out nanoscale lathes or drug-delivery devices or computers could be.
New advances in nanotech and nanocomputing are being announced practically every day, but there's still a long way to go. When those nanocomputers arrive, though, IBM will have the disk drives for them.
IBM scientists this spring discovered chemical reactions that cause tiny magnetic particles to align themselves precisely at precise intervals. This discovery has the potential to push storage densities up to 150 gigabits per square inch, close to the theoretical limit for magnetic storage at room temperature.
Laura Landweber and her colleagues at Princeton recently built a simple computer out of RNA molecules. They gave it the classic chess problem of positioning knights on the board so that they do not threaten one another. The computer solved the problem. What is interesting, however, is simply this evidence that tools are being developed to produce biocomputers, which blur the line between us and our creations.
Another line-blurrer is computers as clothing. There are now regular conferences on wearable computers. For the computers-as-clothing set, the footprint of the Palm, you'll pardon the pun, is too large. They want the thing to go away and just be there.
The chip under the skin has already arrived; deeper implants and nanobots swimming in the bloodstream aren't far off.
SETI@home is the most visible distributed computing project, setting over a million people -- or the unused cycles on their personal computers, anyway -- to work analyzing the raw data collected by the Arecibo Observatory in Puerto Rico, data that could confirm the existence of somebody else out there (SETI stands for "search for extraterrestrial intelligence").
But there are other cases of using the unused cycles on computers to tackle tough problems that can be appropriately decomposed. If Adam Beberg has anything to say about it, there will be many more. He and his colleagues are creating Cosm, a set of tools and protocols that they hope will lead to the mainstreaming of such collaborative computing. The possibilities for this paradigm include companies paying individuals for cycles, a marketplace developing in which people can buy and sell cycles and buy into large-scale projects such as weather forecasting, and the gradual emergence of the kind of planet-wide computer system envisioned in many science-fiction stories.
A patent was recently granted for a faster-than-light device. The Hyper Light Speed Antenna, according to the patent application, generates "opposing magnetic fields each having a plane of maximum force running perpendicular to a longitudinal axis of the respective magnetic field [yadda yadda yadda], thereby sending and receiving the communication signal at a speed faster than a known speed of light."
A neat trick. Getting the patent, I mean.
The rules of the game of physics still make it impossible to move matter or information faster than the speed of light, the U.S. Patent Office notwithstanding. Nevertheless, the April 2000 issue of Scientific American contains an optimistic article on the possibility of instantaneous quantum teleportation. Rudimentary teleporters have been built and demonstrated. These devices, using entangled particles, seem to produce something like instantaneous communication, except that the information transferred is unavailable until information sent via a conventional light-speed channel indicates how to get at it. Or something like that.
One conclusion of that article, by Anton Zeilinger of the University of Vienna, is that what it would make most sense to teleport via such a device are qubits. In fact, IBM researchers have recently proven that a general-purpose quantum computer can be built out of single-qubit gates, entangled particles, and quantum teleporters. So far, they don't seem to have filed a patent application for such a computer, but it's probably only a matter of time.
Bill Joy, the chief scientist of Sun Microsystems, has a long cautionary article in the April 2000 issue of Wired magazine. Past technological dangers, like the H-bomb, he argues, were held by only a few people. Future technological horrors, the nanotechnological, the genetic, and the robotic, will be in the hands of vastly more people. "Will we be able to survive the kinds of technoterrorism that lies ahead?" Joy wonders.
Perhaps this explains why Sun hasn't made more progress on its Jini technology. Caution born of social responsibility?
That's a set of snapshots of a few of the science-fiction technologies currently under development, and one scenario that I hope is science fiction. Closer to the ground is the question of where computer software architecture is headed. The operating system is, after all, only an invention and may be superseded, circumvented, or obsoleted one day.
Is this that day?
A company called "Aestiva" claims to have invented a web-based operating system. The press release on the product is fairly confusing: It acknowledges that this isn't an operating system in the sense that you could choose it instead of Windows or UNIX. It sounds like a development system centering around an HTML editor, designed for creating web sites, but that wouldn't be as dramatic a claim as the claim to have produced the first WWW-based OS.
Aestiva (http://www.aestiva.com/) can't even claim to have the first claim of a web-based OS: That would be Netscape's assertion a web-generation ago that its browser was going to make desktop operating systems irrelevant. Although there is some logic to the idea, the operating system arena is vibrant and alive today.
Granted, it's the operating systems that you don't have to pay for that are getting all the press. Doubtless, that's why Be Inc. isn't being written up in the press as a dead issue. After failing as a hardware company, failing as a commercial operating systems company, vacillating between Intel and Motorola, courting Apple (up to this point, I could be talking about NeXT), Be has decided to give away its OS.
That is, it will give away one of its operating systems. In a sane market, this would mean that the entire value of the company depends on its still-commercial embedded OS, Stinger. In this world, though, Be could probably take the free OS public and have the Be share of BeFree be worth more than all of Be Inc.
Walnut Creek CDROM and BSDI are merging. This is big news for the BSD world, and totally redefines the BSD space, which was beset by lawsuits around the time Linux came on the scene and has never recovered its former vigor. These two companies make up about 90 percent of the BSD UNIX implementations. The most obvious advantage of the merger is shared codebases, and the combined company will be working toward a single API for third-party vendors to port to. That will mean more applications available on BSD, and possibly a real boost to interest in and viability of BSD-related ventures.
One company that may be interested in the rise of BSD is Apple. Its upcoming MacOS X, based on a Mach kernel, is apparently full of BSD.
Linux just rolls on. With the capitulation of SCO, it has few remaining competitors in UNIX space. There is Solaris, of course, and then there's BSD, which points out how unhelpful the notion of competition is in understanding what's happening. Yes, there is competition, but the high bit is cooperation.
There's a lot of scuffling around in the area of embedded Linux right now. There's Transmeta, of course, with its minimalist Linux written by the man himself. Gmate, a South Korean start-up, unveiled at the CeBIT show a Palm-sized Linux-based device that plays MP3s, browses the Web, and handles e-mail. Lineo, the successor company to Caldera Thin Clients, is aggressively going after WinCE developers with its Lineo Embedix PDA, which includes a CE-compatible layer on top of Embedix Linux, allowing CE applications to be ported easily to the device. Lineo has a lot of OEM relationships built already from selling DOS to them, but there are others in this space, too, such as Lynx's Blue Cat and Monta Vista's Hard Hat Linux and some embedded Linux operation TurboLinux has in China. Red Hat bought device-maker Cygnus Solutions with an eye to getting into this market. There's even an embedded Linux consortium to promote embedded Linux.
Back on the desktop, what had been a pretty even race between KDE and GNOME for the GUI of choice got shaken up recently when Eazel, the company formed by some of the creators of the original Mac OS, threw its support behind Linux. Andy Hertzfeld, Mike Boich, and their pals plan to make Linux easier to use than Windows or the Mac, and they have the credentials to do it. Says open-source guru Eric Raymond, "Andy Hertzfeld and his crew showed the world how to do GUIs right back in 1984." Among the innovations Eazel is working on are the ability to view the contents of files right in the folder windows that contain them, and a new approach to icons: Icons for graphic files will just be the graphics themselves and icons for text files will show the first few words of the text.
I was going to say something about Microsoft's OS strategy, but I realize that I don't know what it is. I'll charitably assume that Microsoft does, and try to figure it out for some later column.
I also won't delve into the future of CP/M, although here's a link to one of the more creative examples of the art of benchmarking: http://www.oualline.com/ col/cpm.html.
Given the resiliency of CP/M, it seems that the imminent death of the desktop operating system is a long shot. HTML/OS may not be an operating system, but it is one piece of the abundant evidence of a movement from the desktop to the Internet.
But if the Internet is to be, as Ted Nelson put it talking about something else, the future intellectual home of mankind, then how's the building coming along?
By the time you read this, the folks behind the second-generation Internet effort will have announced winners in a contest "for the most demanding end-to-end, bandwidth-intensive Internet applications in the world." Internet2 is hoping to deliver gigabit-per-second throughput more or less planet wide.
Microsoft has dropped Usenet newsgroups from MSN due to lack of interest. More evidence that Usenet is on the way out. Now there's Freenet. It's an alternative publishing network that promises true anonymity in sharing documents and files over the Internet. Freenet has no centralized administrative infrastructure of domain name servers (DNS) and IP addresses that can be used to track users.
Then there's XML, the language of the future. Okay, that's a little strong.
Maybe it's a little strong even to call it a language. Anyway, a good idea that seems to be moving forward while other seemingly good ideas (like Jini) languish. I get several press releases every day about XML projects or products. I file them and smile. Any technology that has Microsoft, Sun, and the Linux community working in harmony is deserving of our attention. I'd mention more of the changes afoot if I had the space. Weblogs as a new phenomenon in publishing are subtly shifting the control of news and embracing a sketchier but perhaps more honest subjectivity. And how about the Amazon patents and the question of software patents generally? All of these changes will do damage, change always does. I hope the damage they do is less than the good, and that we can learn from warnings like Bill Joy's and not fear progress.
DDJ