EMERGING TECHNOLOGIES



TABLE OF CONTENTS



  1. INTRODUCTION
  2. THE PDA AS EMERGING TECH?
  3. BIOMETRICS
  4. POWER SUPPLIES AND ROBOTS
  5. HAPTIC TECHNOLOGY
  6. AUTONOMOUS VEHICLES
  7. SURVEILLANCE IN AN AGE OF PARANOIA
  8. FROM RFID TO SMART DUST
  9. LED LIGHTING
  10. NANOTECHNOLOGY
  11. E-VOTING
  12. 3D PRINTING
  13. AUGMENTED REALITY
  14. MIND-MACHINE INTERFACES
  15. COMPUTER TRANSLATION
  16. VOIP--INTERNET CALLING
  17. THE WAR AGAINST SPAM




INTRODUCTION



Back to Top


Every year, I teach a course in "Emerging Technologies." It runs in "hybrid" mode, meaning that I post the lectures and discussion questions and assign papers. Attendance is optional (click here for the syllabus), meaning that I am in the classroom and students who wish to discuss the material can show up. Those who don't choose to show up discuss the material only online. They get less out of the course, but it seems to work. And I get to post the lectures in my blog too, and then package them as this e-book. Note that the links were live when I first posted the lectures. A few of them may have gone dead since then.

So what is an "emerging technology"?

One might think that an emerging technology is just a brand-new technology. Someone has just emerged from a lab with a new invention. The next step is to turn it into a product, start to sell it, and garner great wealth. Surely that must have been the dream of Charles Babbage, who in 1821 invented the first genuine computer, the "Difference Engine."

But it is hardly that simple. A great many inventions never reach the product stage; they may, like Babbage's machine, require the invention of additional technologies (such as electronics) before they can become practical. In addition, many have been turned into products that failed to sell (or failed to sell well) or to continue to sell. Who now remembers eight-track audio tape? (Click here for the eight-track story.) The failures to succeed--or to emerge--can happen for many reasons. Among those reasons are: The Segway Human Transporter is a product built on the underlying technology of computer-controlled gyroscopic stabilization. Will it succeed or emerge? As a product, it's nifty enough, but it has problems, for legal authorities are banning it from both roadways and sidewalks! On the other hand, legal authorities (police and security forces) are also finding uses for it. The underlying technology is also being used for a stair-climbing wheelchair, and Bombardier is thinking about a one-wheeled motorcycle called the Embrio. So even if the Segway itself (or its consumer version) does not emerge, the technology may.

What kinds of technologies are we going to look at here? Some are fresh from the lab. Some have been around for awhile, perhaps waiting for the development of other "enabling" technologies. Most have appeared on one or more of the several lists of up-and-coming technologies that appear each year.

Some of those lists are shorter than others. In May 2003, David Pescovitz wrote "Six Technologies that Will Change the World" for Business 2.0. One of his six, "God's Ink-Jet," is essentially an ink-jet printer that uses instead of ink a mix of cells, growth factors, and gel and can lay down multiple layers to generate a three-dimensional organ-type structure. If it works as described, it could be very valuable in medicine. But it is not entirely new, for it is a variation on existing devices, 3D printers, used for "rapid prototyping." These devices are already successful in industrial settings. They are too expensive for home use, but prices are dropping. Once they are cheap enough, new uses will be developed, which invites us to imagine what we might do with one if we had one at home. 3D printers will be considered later on.

Pescovitz also mentions "Robots you can relate to," meaning robots with facial expressions such as Kismet, faster airplanes, tiny fuel cells for PDAs and cell phones, electronic paper for thin, flexible computer displays, and swarms of tiny sensors for tracking both goods and people.

A second list comes out every year from Technology Review. The latest is "10 Emerging Technologies" (March/April 2006): Where the Pescovitz list is product-oriented, this one is less specific, focusing on the technologies that make products possible. For our purposes here, we will pay more attention to products and we will not get into the biological areas.

Intriguingly, neither list includes a technology that most of us use every day and that might be considered very well established. As discussed by Wade Roush in "Search Beyond Google" (Technology Review, March 2004), this is the technology of the computer search engine, which lies behind Google, Yahoo Search, and other search "products." The basic idea is fairly simple: A list of everything to be found on the Internet, suitably sorted and indexed. Someone types "search technology" into the Google box, and they promptly get a list of web sites that contain both the words, "search" and "technology," including Google's own PigeonRank spoof of its PageRank technology. However, the amount of material available on the Internet grows with such extraordinary speed that several generations of search technology have already proven inadequate to the task and been replaced. The current reigning technology is Google's PageRank, but it does fail to find everything one might want in a search, and it very often fails to return the most desirable results first. Many researchers--including Google's own--are therefore struggling to develop new methods. See Charles Ferguson's "Google and the Coming Search Wars, Revisited," Technology Review, April 2005, and Javed Mostafa's "Seeking Better Web Searches," Scientific American, February 2005. Most recently, Microsoft has proposed to field its own search engine, with claims that it will outdo Google and increase Microsoft's ad revenue.

It may be impossible to develop a technology that can find everything out there, for there is just too much. The new methods differ chiefly in how they prioritize what they find (how often key words appear on a page, how many other pages link to a page, which other pages link to a page) or arrange the results on the screen (lists or clusters) or let users pose questions (key words or sentences).

Google is not only trying to improve search technology to stay on top of the heap. In terms of our list of why technologies fail to emerge, it is addressing the second item by adding features. It is looking for new kinds of information to search: Look at Google's Options page for Google Scholar. Or Maps, News, Froogle (shopping), and more. Try Book Search to see how Google has been scanning millions of library books to make their contents available for searching and printing; it has signed partnerships with several major university libraries (Oxford, Harvard, Stanford, and the University of Michigan) and the New York Public Library. The potential benefits are huge, for the project promises to make available to all materials which have in the past required visits to distant repositories. But there are problems as well, including potential copyright infringement, finding a way to make the process pay, and the process's impact on the nature of libraries. See Wade Roush's "The Infinite Library," Technology Review (May 2005).

So far everything mentioned has dealt with words, and there are a great many images that people want to be able to search for. One solution to the problem is attaching descriptive tags, but another involves image-recognition software. According to Gary Stix, "A Farewell to Keywords," Scientific American (July 2006), researchers are developing software that will let one take a picture with a cell phone, send the image file to a server via the Internet, and get back web pages with information on what is in the picture. Walk down the street, snap a pic of a restaurant, and a moment later be looking at the menu or a review of the food and service. Go to a museum, snap a pic of a statue or painting, and read all about it. Look at the fifth item on the list of reasons why technologies may not emerge and consider that this is the sort of technology that might make some applications of "augmented reality" (see Nov 7 lecture) non-starters.

The following list will take you to several of today's competing search engines. Will Google's PageRank stay on top? Or will a new technology emerge to replace it? When you examine the results so far, what do you think? Is there an obvious choice? Or is something more needed? Now: Where is it all going? Everything mentioned above is pretty short-term stuff. But technology does not stand still. Every time I teach my Emerging Technologies course, new technologies demand to be covered. And over the years they add up to a lot more than a better search engine or cell phone.

Many people welcome progress, for it brings exciting new products, toys, and abilities. But many people worry about progress too. It destroys jobs, puts companies and industries out of business, threatens morality, and even threatens to destroy "human nature." This is the topic of Joel Garreau's Radical Evolution (Doubleday, 2005). He notes that nanotechnology, artificial intelligence, robotics, brain chips that let people control prosthetic limbs and machinery, mental uploads, memory boosters, life extension, genetic engineering, and more offer to change human life drastically. He calls genetic, robotic, information and nano technologies the GRIN technologies and says that they are about to enable engineered humans with such startlingly new capabilities that they may no longer be "human" in any traditional sense. The consequences may be quite utopian or quite catastrophic; Bill Joy ("Why the Future Doesn't Need Us," Wired, April 2000) has written that robotics, nanotechnology, and genetic engineering threaten to make humanity extinct and that research into these areas should therefore be cut short immediately.

People fear potential catastrophes. But the idea of transcending human nature really gives them the willies. The idea that humans might turn themselves into something that isn't really human anymore is frightening. So is the idea of people becoming somehow unnatural, which has driven protests against vaccines, antibiotics, organ transplants, and assisted reproduction, among other technological developments that go against the traditional "natural order." It provides the rhetoric being used against the idea of changing the body with such things as computer implants and genetic engineering. Yet, says Garreau, human nature is not just a matter of doing things the same way we always have. It is human nature to search for meaning, to better ourselves, to be creative, and to devise rituals to validate our actions. Given this, whatever we do with the GRIN (and other) technologies is human nature.

We might also note that in the search for whatever it is that makes humans uniquely human in a world full of our animal cousins, people have suggested communication, speech, tool-using, laughter, and several other things, all of which soon turned out to have parallels in animal behavior. The differences are of degree, not kind. But there is one thing we do that other animals don't: If we have a tool, a language, a religion, a costume, a recipe, a political system, we tinker with it. We change it. We do not leave it alone. Thus, if we wind up changing human nature, well, that's human nature.

THE PDA AS EMERGING TECH?



Back to Top


Many journalists remember the 1983 Tandy 100 fondly. It had 8K of RAM (later upgraded to 24K), a 3 MHz processor, a text editor, and a 300-baud modem. That doesn't sound like much today, but it weighed only 4 pounds, could stand being dropped, had a full-size keyboard, ran for 20 hours on 4 AA batteries, and did what it did quite well. Perhaps it should be no surprise that some are still being used to write small text documents (such as a journalist's notes), transfer the documents to more modern desktops or laptops, and even send email.

It also represented a major step in the evolution of the personal computer toward the small, modern Personal Digital Assistant or PDA. When the Tandy 100 came on the market, desktop machines were about the same size as today, although much less capable. Hard drives were tiny (10-20 Megs) and expensive, and 5.25" floppy disks held at most 360K of data. Portable computers were better known as "luggables," since they were the size of small suitcases. That all changed, but so did our idea of what a portable computer should be able to do. Today we want our portables to do word processing, spreadsheets, images, contact lists, email, games, and more. We want greater storage capacity and speed. We want wireless connectivity. And thanks to advances in chip design and manufacture, we can fit it all in a tiny package. (One popular brand is called the "Palm" for good reason.)

So we do. And many are not happy with the result, for the keyboard is too small to use, writing on the screen with a stylus is chancy, the screen is tiny, and the batteries don't last nearly as long (all those added functions require more power). The PDA industry sold about 10 million PDAs in 2003 (down from 12 million in 2002). The decline is credited to the availability of cell phones with many PDA functions, which indeed seems to reflect a trend to merge the two devices--as well as to add functions such as the ability to take and email digital photos. According to PCWorld, the trend was still going strong late in 2004. Betanews says the trend continued in 2005, though other sources had more rosy statistics. Click here for a 2006 report.

Surely sales of 10 million PDAs per year must mean the PDA has emerged? But the figures are for the world market, and in the US alone, over half of all households have at least one cell phone. Globally, cell phones sell well over half a billion units every year.

What's the difference? Phones start with smaller keypads, they're voice-oriented, and they need less memory. The basic orientation is communications, not data. This remains true when a phone adds PDA and camera functions, although then the keyboard and screen issues become important again. Resolving those issues might well mean a complete merger of PDA and phone and a true emergence of "personal electronics" technology.

Folding keyboards have been available for PDAs for years, but they are an excellent example of "OMDT" (One More Darn Thing) syndrome, which Claire Tristram, in "Handhelds of Tomorrow" (Technology Review, April 2002), alludes to when describing a gearhead festooned with gizmos. The article suggests that the proliferation of different devices is a problem. What is needed is a single device that meets all the needs of its users. It would combine PDA and phone functions, have a large screen such as the flexible one described, be able to survive dropping on a floor, and have a thumb keyboard.

Today, just two years later, the basic needs have not changed, though the specific form of the solution has. The screen, for instance, may be realized as electronic paper in the near future. Another possibility is the head-mounted display. Still another is the "fog screen" demonstrated in 2003 and already the basis of a new business, FogScreen,com. The keyboard may take the form of the "virtual keyboard" described by Tomasi et al. in "Full-Size Projection Keyboard for Handheld Devices," Communications of the ACM, July 2003). The latter two both minimize OMDT syndrome and promise a simple device that, in a sense, expands as needed. If they work as their fans expect, the PDA may emerge to the point of replacing cell phones, laptops, and even desktops with a single "personal electronics" package. The only way to improve on it might be to give the package a direct brain interface--which is actually being worked on; we'll get to it later on.

BUT... It just may not be fair to call the PDA an "emerging" technology. The sales figures cited above indicate that "failing" or "declining" or "outmoded" technology might be more like it. On March 2, 2006, CNET.com.au published "HP: Pen-Based PDA Market on Death Bed," by Asher Moses, which begins with: "The traditional pen-based PDA market will evaporate within the next four years without significant product innovation, according to Hewlett-Packard (HP)[HP's Vice-President for Consumer Products and Mobile Business Group in the Asia-Pacific region, Chin-Teik See, interviewed in Hong Kong]. The company will therefore continue to focus the majority of its handheld efforts on converged smart phone devices, relegating its traditional PDAs to the entry-level consumer and SMB [small-medium business] markets." The remaining market is small, and it is shrinking 30% per year.

So the PDA is dead, or nearly so. Smart phones, on the other hand, are going strong and adding new functions almost every month. HP's own iPAQ rw6800 and hw6900 do email and "the former is being marketed as a portable entertainment device... Some of the standout multimedia features on the rw6800 include a built-in 2.0 mega-pixel camera, FM tuner, MP3 playback software and dual stereo speakers."

BIOMETRICS



Back to Top


How do you prove you are who you say you are? Once upon a time (think of all the old movies you've seen), it was enough to be able to show a letter from someone of repute. Later, various kinds of ID cards came into play, eventually with photographs (think of driver's licenses). Tattoos have been used. If you were charged with a crime, photographs again came into play (mug shots), as did fingerprints and--before that--the sizes of body parts (the Bertillion system).

The advent of computers gave the basic question a new urgency, and the standard answer so far has been the use of usernames and passwords. Even "universal authentication," on the list of Technology Review's ten emerging technologies for 2006, does not get beyond this. But people are lazy and write their passwords down (often on the backs of keyboards!). Or they choose passwords that are too easy to guess. It has long been realized that something better was needed, preferably something that was unique to the individual and could not be stolen, imitated, or faked. As Paul Bleicher, "Biometrics Comes of Age," Applied Clinical Trials (December 2005), tells us, in the 1990s a number of approaches--using the patterns in the eye's iris and retina, palm prints, fingerprint detectors (with sensors to make sure the finger hadn't been chopped off the hand), voiceprints, handwriting, and facial configurations--had reached the demonstration stage. People in the field were expecting one or more of these approaches to soon play a major role in computer technology. It didn't happen, for several reasons. The technology can be expensive, and it is not as accurate or as consistent as many people would like. That is, because of dirt on fingers or scanners, changes in lighting, noise levels, and so on, sometimes it says you aren't who you really are (false negative or FN) and sometimes it says you are when you aren't (false positive or FP, and just what a hacker would love to see).

But the technology remains promising, its use is increasing steadily if slowly, and it has continued to advance, especially after the War on Terror and the rising tide of identity theft added even more urgency to the need for secure identification. You can now buy fingerprint readers for your personal computer for less than $50. And governments are increasingly looking to use biometrics in connection with passports, airport security, and immigration. Jane Dudman, "Biometric Vision," Computer Weekly (July 19, 2005), notes that "Like other countries Great Britain is looking to introduce biometric identifiers into passports as a way of increasing control and in line with demands from the U.S., although its deadline has slipped. Already under way in Great Britain is the Iris Recognition Immigration System being run by French firm Sagem. This project is being implemented initially at Heathrow's terminals 2 and 4, where regular travellers who are not European Union passport holders will be able to bypass queues by registering their iris patterns."

The questions of accuracy and reliability have led some researchers to work on systems that exploit multiple unique body features. Robert Buderi, "Me, Myself, and Eye," Technology Review (February 2005), reports on Anil Jain of Michigan State University, who is developing three different multifactor systems. "In one of Jain's systems, a pair of cameras gauges a subject's height. A close-up is then taken of her face, and software analyzes it to determine her gender and eye color and classify her ethnicity as either Asian or non-Asian. This data is then combined with that from a primary biometric [such as voiceprint, fingerprint, or iris scan]. Most face recognition systems use 2-D images and can be foiled by changes in illumination, head position, and expression." Jain's second system uses a laser to scan a person's face and construct a 3D model that can be matched, using "information about anchor points such as the tip of the nose and corners of the eye to conduct a preliminary search of database images. ... Because it captures surface geometry, the 3-D system works independently of lighting changes and head position. Jain's third system combines hand geometry, face recognition, and fingerprint matching."

I wonder if anyone is thinking of building a fingerprint scanner that could be incorporated into the buttons of a keyboard? Then one could type username and password as usual, and the system could check to make sure they actually matched the typist. An interesting wrinkle on this is that though different touch typists will use the same finger-key combinations, those of us who are not touch typists are likely to use different finger-key combinations, which could become part of the metric. However, this would be a hardware-intensive solution to the problem. A software-intensive alternative is to use typing patterns or "keystroke biometrics."

For the very latest, visit the Biometric Consortium, which "serves as a focal point for research, development, testing, evaluation, and application of biometric-based personal identification/verification technology." Its September 2006 conference is focused on "Biometric Technologies for Homeland Security, Identity Management, Border Crossing, Electronic Commerce, and other applications."

POWER SUPPLIES AND ROBOTS



Back to Top


As we saw a few posts ago, early laptop computers such as the Tandy 100 could run for 20 hours on 4 AA batteries. The laptop on which I am writing this can get 4 hours out of its considerably larger battery. The reason is that the modern laptop demands much more power in order to do the many things the makers of the Tandy 100 could only dream of (2.66 GHz, 512 Megs of Ram, 40 gig hard drive, color screen, etc.). That battery is rechargeable, but it can only be recharged a limited number of times. And replacing it will cost on the order of $100.

There's gotta be a better way!

In "Power to the Portables: Battery Technologies," Computerworld, April 22, 2002, Matt Hamblen says that "Users would like a free, infinite power source weighing next to nothing; reality requires a removable power 'brick' that can take hours to recharge... Devices are getting smaller, but faster CPUs and color displays drain power. So device makers are becoming increasingly reliant on software and hardware to lessen battery drain." Researchers are also developing improved batteries, including some that use nanotechnology to improve performance. Hamblen dismisses fuel cells, but Charles Murray, in "Fuel Cells Hold Promise as Power Source for Portables," EE Times, April 27, 2001, is more optimistic. They could use alcohol, hydrogen, natural gas, butane (as in lighters), or other fuels to generate the electricity needed to recharge a device's batteries, or to run the device directly. Potential problems include heat generation and water emission. But by 2005, CNet News could report on Toshiba's plans to field a fuel cell power source for iPods in the near future. On March 2, 2006, the word was that a CD-ROM drive-sized fuel cell that fits in a laptop's media bay, weighs 3.7 pounds, and can power a laptop for up to 8 hours should be available from HP and/or Dell in the second half of 2006.

Will fuel cells "emerge"? The answer must depend on whether whatever improvement they offer strikes the consumer as worth the trouble of buying new equipment. Expense will be one factor, as will reliability and convenience.

Of course, improved power storage capabilities are not needed only for PDAs, laptops, cell phones, and other handheld devices. Robots are finally, after many years of hope and hype, entering the market, and they too must carry their power with them. Some robots are large enough to carry large batteries (e.g., robotic lawn mowers), but the one getting the most attention lately is pretty small. It's the iRobot Roomba, a robotic vacuum cleaner. The same company is also developing a "PackBot" for military use. Both demand either better batteries or fuel cells.

The same thing is true--only more so--of robots that use legs instead of wheels or tracks. They will need power for multiple motors and for additional computation (legged locomotion, including balance and navigation, say many robotics researchers, is among the most challenging of computational tasks). The demands get even worse if we expect robots to interact with us in any remotely human way. As Rodney Brooks notes in "Humanoid Robots," Communications of the ACM, March 2002, the Kismet head robot has 17 separate motors for neck, eyes, eyebrows, ears, lips, and jaw, all requiring power. In the lab, of course, power is not a problem; since Kismet isn't going anywhere, it can be plugged into a wall socket.

Will truly humanoid robots ever "emerge"? It depends, says Brooks. The research will continue, partly because of its intrinsic interest and partly for its spinoff benefits (computerized legs will be useful for people who need prosthetics). Humanoid robots will be built. But they will not show up in homes and on the streets unless they offer advantages over nonhumanoid robots. For now, robots tend to be built for single purposes (such as lawn mowing, floor vacuuming, and pack carrying), which reduces their complexity and power demand.

Many of the most successful robots are not even mobile. As Wade Roush notes in "Immobots Take Control," Technology Review, December 2002-January 2003, many modern devices (from photocopiers to spacecraft) have a large amount of built-in "intelligence." They are designed to guide their users through complex procedures, including diagnosing and repairing problems. They may even be able to handle diagnoses and some repairs on their own. The trick is to build into the software a model of the device that the software can use to think about what is going on (or going wrong). This approach is being developed for controlling some automobile functions and a prototype is being tested in a Brazilian water treatment plant. In the future it may find applications in many other areas. As Roush says, "immobots could become pervasive, helping to control some of our most important infrastructure technologies. In air traffic management, for example, ... immobotic systems [could] use sensor data from satellites and ground stations to assess local weather conditions, automatically identify and select the safest, most efficient flight paths, and redirect air traffic. [And} the only way to make infrastructure technologies autonomous without increasing the risk of massive software-related failures may be to program them with distinctly human qualities, such as the ability to plan ahead and solve problems creatively. And in a world that seems increasingly dangerous, knowing that immobots are looking out for themselves—and for us—could be a source of comfort."

HAPTIC TECHNOLOGY



Back to Top


It's relatively easy to turn light and sound into electrical signals. We did it first with analog radio, microphones, recordings, TV, and camcorders. Now analog has been thoroughly replaced by digital, but we're still pretty much limited to light and sound.

Sure, people have played with adding smell to web pages. The trick is a variation on the color ink-jet printer, only instead of ink, it outputs a mix of stinks that matches a particular odor. See How Stuff Works for details. The basic idea has definite appeal to companies that sell stinky things such as perfume, flowers, wine, or food online, but it does require that the consumer buy one more darned thing, which is perhaps the main reason why it hasn't caught on. Not that there aren't other drawbacks too: I shiver at the thought of teen boys competing to see who can produce the most putrid web page! Or of what hackers might do with it!

Sight, hearing, and smell. Taste isn't likely to find a home on web pages, but touch could, and indeed people have been working on ways to make a mouse or other device vibrate when run over a "texture" zone on a web page. This has clear benefits for the blind, for instance, as do pinpads that can display Braille. More ambitious is the sort of thing that HapTex is working on: representing fabrics both visually and tactilely in such a way that an online person can explore the texture of the fabric with a fingertip. According to New Scientist, "Detailed measurements of a fabric's stress, strain and deformation properties are fed into a computer, recreating it virtually. Two new physical interfaces then allow users to interact with these virtual fabrics – an exoskeleton glove with a powered mechanical control system attached to the back and an array of moving pins under each finger. The 'haptic' glove exerts a force on the wearer's fingers to provide the sensation of manipulating the fabric, while the 'touching' pins convey a tactile sense of the material's texture." A prototype will supposedly be ready for showing off by the end of 2007.

It's easy to imagine applications in ecommerce. More important may be telework, for although it is possible for someone to remote-control a machine or use a robotic arm to pick up objects, it is much easier if one can actually feel what one is doing, in which case you have telepresence, the feeling that you are really there. Essential here is the concept of force-feedback, meaning that a force gets send (fed) back to the user proportional to how hard the user is touching the object at the other end. Lacking force-feedback, it would be difficult to use a robotic arm to pick up an egg, change a baby, or do surgery (telesurgery), at least not without doing damage. Force-feedback devices are already found in joysticks and other gaming devices.

Something similar will be necessary if robots are ever to be given a sense of touch, but here there is an added difficulty. Everything above involves a human in the system and uses electronic sensors to generate signals that can be relayed to the human's own sensors in such a way that the human brain is convinced that it is feeling something real. With a robot, we need to give it the sensors and then process the information in such a way that the robot can make sense of and function in the world. People are working on it, but so far most of the progress is on the sensor end. For instance, Vladimir Lumelsky at NASA Goddard is working on sensor-containing skin for robots. So far the sensors are big and clunky and they're only infra-red proximity sensors. But Lumelsky says that in the future the sensors will be smaller. If we want roboskin that does a good job of mimicking human skin, it'll also need multiple types of sensors—pressure, vibration, heat, cold, pain, and perhaps more.

Where will it all end? Well, for years, people have been dreaming of the full-body virtual-reality suit for gaming, telework, even sex. Here's a nice clean link, though it's a bit old. If you want something raunchier, it won't be hard to find. And it may not be too long before it's on the market. There are many applications of force-feedback, haptic interfaces in ecommerce, telesurgery, telepresence, and robotics. Add the digital smell, and maybe we will finally see the teletourism of science fiction. Log in, go to the beach, feel the sand on your butt, smell the sea, feel wet, feel the jellyfish sting your legs…

AUTONOMOUS VEHICLES



Back to Top


We often think of robots as the sorts of things we see in science fiction movies and TV shows: Shaped roughly like humans or other animals, with two or more legs, a head, and massive arms equipped with claws or cannons. But it has been said that over the last fifty years of computer science research, it has proved much easier to make computers think (play chess, solve logic problems, do math) than to make them walk.

Walking is difficult partly because a leg has multiple joints, and movement at each joint must be controlled in such a way as to make the leg function as a whole. It is also difficult because it requires some understanding of the world outside the robot. The robot must know where it is and where it is going, what obstacles look like and how to get around them, how to identify pathways and how to stay on them, and so on.

It's possible to make mechanical legs. The Dante volcano explorer and the robolobster offer two examples. Asimo's legs look more like ours. However, it is also possible to dodge this part of the problem and focus on the navigation issue. Airplanes make it look easy, for autopilots have been in use for many years; today planes can even land with no human hand on the controls. It's admittedly easier for a robot plane to navigate, since there are fewer obstacles in the air than on the ground and runways are designed to be obstacle-free, but the navigation problem is still not trivial.

If we don't need legs, what are the options for a robot on the ground? Wheels are the obvious choice, partly because they come attached to packages including engines, brakes, steering, etc., with room for the computers that will handle the navigating. The Defense Department's Defense Advanced Research Projects Agency (DARPA) has been doing its best to stimulate the development of autonomous wheeled vehicles. The first DARPA Grand Challenge race, in 2004, set robot cars (and other wheeled vehicles) loose on a desert course (Barstow, CA, to Las Vegas, NV) and offered a million dollars to the team whose car finished the course first. But there was no winner; the best performer, from Carnegie Mellon University, made it only 7.4 miles.

In 2005, the prize was two million dollars, and five vehicles finished the 132-mile course through the Mojave Desert, with four making it in less than ten hours on October 9. The winner, Stanley, a VW Touareg from Stanford University, made it in a bit under seven hours. See Joshua Davis, "Say Hello to Stanley, Robot Race Car Champion of the World," Wired (January 2006). The secret of Stanley's success lay partly in its programming and partly in its array of sensors for keeping track of its location (GPS units and odometers) and finding a path clear of obstacles ahead of it (laser range finder and video camera).

Other contenders used similar sensors, plus such others as a "ground mouse" that measured movement relative to the ground under the vehicle. Overall, says W. Wayt Gibbs in "Innovations from a Robot Rally," Scientific American (January 2006), the Grand Challenge has produced a number of innovations that may find use both in robots and in safer human-driven vehicles.

What is next? The Defense Department's goal is military vehicles that can deliver supplies or even fight without putting human crews at risk. Military technologies often find lucrative civilian applications, and this will surely be no exception. One 2005 competitor (William "Red" Whittaker of Carnegie Mellon University, responsible for the two Hummers in the race) wants to see one tending fence on his farm and is already putting together proposals for a lunar mission. The team behind Stanley, the 2005 winner, is talking about driving from LA to San Francisco--with no human intervention--within the next two years. Click here for the full story. "Cars that drive themselves will be a market reality at some point in the not-too-distant future. 'It is only a matter of time until consumers have self-driving cars.'"

May 15-18, 2006, saw the European equivalent of the DARPA race, the Land-Robot Trial. It tested vehicles in both urban and non-urban settings, as well as in landmine detection and removal. ELROB 2008 will take place from June 30, 2008, to July 3, 2008, at the Infantry School in Hammelburg, Germany.According to the May 1, 2006, press release, the third DARPA Grand Challenge will be held on November 3, 2007 over a 60-mile simulated urban course. "DARPA will award prizes for the top three autonomous ground vehicles that compete in a final event where they must safely complete a 60-mile urban area course in fewer than six hours. First prize is $2 million, second prize is $500,000 and third prize is $250,000. To succeed, vehicles must autonomously obey traffic laws while merging into moving traffic, navigating traffic circles, negotiating busy intersections and avoiding obstacles." The results will be of great interest to the Defense Department, for "The DARPA Urban Challenge features autonomous ground vehicles conducting simulated military supply missions in a mock urban area. Safe operation in traffic is essential to U.S. military plans to use autonomous ground vehicles to conduct important missions." For a New England connection, MIT is entering a car in the new Urban Challenge (it sat out the earlier Challenges).

Navigating 130 miles of rough desert track with no human hand on the wheel is an impressive accomplishment, but not very commercial. If we wish to forecast the future we need to use our imaginations, and as robotic driving moves onto city streets, that becomes easier. It might be nice to have a car that could drive us home when we're too drunk to touch the wheel. The military would love trucks that can be blown up without killing soldiers. But I am sure the trucking and insurance industries have a very close eye on what is going on. A robot truck could be on the road 24/7, with no need to worry that a sleepy driver might run over someone on the shoulder. Increased productivity and decreased liability mean money.

Would society accept robot trucks on the highway? They'll need demonstrations first, which is why I think one future milestone will be a transcontinental DARPA Grand Challenge (perhaps within five years, considering how much progress happened between 2004 and 2005). Initial acceptance will be for Interstate highways only, and trucks will be required to have a human driver once off the highway. This will ease the problem of unemployment for truckers and perhaps even improve their working conditions (since they will be able to spend nights at home). But before long, robot trucks will even be able to navigate through urban traffic. Human truckers will no longer be needed at all.

But will people ever accept robot buses? Taxis? Self-driving cars have already been announced for the near future. See also the Honda Accord ADAS system. But will cars ever be made without steering wheels and other human-accessible controls?

And don't worry about how a robot car pumps gas for itself. That problem was solved almost a decade ago, with the Smart Pump.

As for the future--where else might autonomous vehicle technology prove useful? How about self-driving wheelchairs and hospital gurneys, among other things? Or let's put the sensors and navigation into a unit small enough to wear on a belt. Equipped with speech synthesis and earbuds, it could talk a blind person around town as easily as it could steer a car.

SURVEILLANCE IN AN AGE OF PARANOIA



Back to Top


After September 11, 2001, the US government felt it appropriate to propose studying the feasibility of sifting through (data-mining) all the data available on American citizens in search of patterns of behavior that might suggest terrorist plots. This effort was called at first "Total Information Awareness" and later, when privacy advocates and civil libertarians objected, renamed "Terrorism Information Awareness," both abbreviated TIA.

The data to be sifted included purchasing records (databases maintained by credit card companies, online sellers such as Amazon, and more), court records, phone records, travel records, tax records, and a great deal more. Objections centered on the danger of "false positives" (people whose activities looked suspicious but were in fact innocent; should such people be arrested or barred from air travel as some have just because of a similarity of name?) and on the fact that the Fourth Amendment to the U.S. Constitution established the right of private citizens to be secure against unreasonable searches and seizures. "Unreasonable" has come to mean "without a search warrant" for physical searches of homes and offices, and "without a court order" for interceptions of mail and wiretappings of phone conversations. Surely, said the critics, it must include the TIA, despite the fact that the post-9/11 PATRIOT Act gives law enforcement greatly expanded search powers. See also this summary of the fuss over warrantless surveillance.

The basic shape of the debate seemed simple: The government insists that private citizens do not have the right to act in such a way that they cannot be watched, supervised, and punished if government deems it necessary. The Electronic Frontier Foundation (EFF), the Electronic Privacy Information Center (EPIC), and numerous other groups and individuals insist equally strenuously that the right to privacy must come first. For EPIC's TIA site, see http://www.epic.org/privacy/profiling/tia/.

In 2003, Congress withdrew the funding for TIA, but much of the research behind the TIA continued, and the goal remains the same: to be able to extract signs of danger from the vast amounts of data available. See Shane Harris, "TIA Lives On," National Journal (February 23, 2006). The National Security Agency (NSA) has been in the news lately for collecting records of phone calls made by U.S. citizens. Other programs monitor money transfers to try to detect terrorist activity. Tracking purchases can detect when people buy untraceable cell phones in suspicious quantity, but it is worth noting that when in August 2006 the British arrested terrorists plotting to use liquid bombs to blow up transatlantic flights, they were reportedly tipped off by an informant. ("MI5 used a mole from within the Muslim community to infiltrate the alleged plot.")

A great deal of surveillance technique and equipment are also being developed by the private sector. Small video cameras are available for household use, and people are in fact using them to keep an eye on the babysitter, to be sure that Grammy is okay in the nursing home, to detect burglars when the house is empty, and even to record the policeman on the doorstep (although the cops don't seem too happy about that!). Similar cameras are being installed in a great many commercial and outdoor locations, as described at the beginning of our first reading, Dan Farmer's and Charles C. Mann's "Surveillance Nation I," Technology Review, April 2003. Noah Shachtman's "Spycam Force" Wired, May 2005, makes the value of the technology in fighting crime crystal clear. Radio frequency identification (RFID) tags are being used to track products (we will consider them in their own right soon). Cell phones are now legally required to be trackable (using built-in GPS locators) during 911 calls, and law enforcement is using the capability to an alarming extent. An interesting application uses cell-phone tracking to produce traffic reports. Cell-phone tracking and other surveillance technologies can clearly be useful, but privacy is quickly vanishing.

Is privacy important to you? Many people, note Dan Farmer and Charles C. Mann in "Surveillance Nation II," Technology Review, May 2003, are not aware of how much data is available about them and how available it is to others. As small examples, try Googling on your own name, look up a friend's address with Yahoo's People Search and get driving directions to their house with Mapquest, get an aerial photo of your house (or a friend's), or (in Maine) use the Maine Marriages Index to look up yourself or a friend.

The technologies necessary for tracking virtually every move we make are being rapidly developed. The March 2004 issue of Communications of the ACM (a major journal in the computer science field) centered on "Emerging Technologies for Homeland Security," with papers titled "Countering Terrorism through Information Technology" and "Proactive Information Gathering for Homeland Security Teams." As the readings show, some of the technology has already "emerged." Much more is on the way, and there seems little question that it will find wide use.

We have already seen that the "public" supposedly has a role in determining whether a technology succeeds. A better word might well be "market." In the case of surveillance technology, the general public does offer a market, but government and corporate interests offer a much larger market. Can the general public have any say at all in such a case? As Farmer and Mann note, "The rapidity of the advances in surveillance technology ... means that society has much less time to confront the trade-offs between security and privacy. The moment for debate and conversation is now, while the technology is still in its adolescence." Only if we do not fail to consider what is happening can we hope to exert any control over the technology, as by restricting access, requiring that we are informed when someone accesses data on us, or giving us the ability to opt out of universal surveillance."

Meanwhile, we can wonder whether surveillance technologies should be available to private citizens as well as law enforcement agencies? Some are, of course, even "cell phone tracking". We can also wonder whether we can do anything about the possibility of too many people having access to data about us. One possibility is setting things up so we are notified anytime someone looks at information about us. This and other measures are discussed in David Brin's The Transparent Society; an excerpt is available here.)

FROM RFID TO SMART DUST



Back to Top


Over the past few years, a great deal has been written about "radio frequency ID tags" or RFID tags. Initially, the message was "They're coming, and they'll be very useful." Then it was, "They're here, but they're big and clunky and expensive. But soon they'll be small enough to hide under a product label and they will cost only pennies."

The alarmists did not hesitate to express concern about the negative effect on privacy, but before we get into that, we should consider just what an RFID tag is.

The basic idea is rooted in the physics of antennae. When a radio or TV signal hits an antenna, some of the energy in the electromagnetic wave that is the signal transfers to the antenna. In a radio or TV, this energy becomes an electrical current that is amplified to modulate a speaker and/or picture tube. The same basic process is what is going on with a wireless computer connection.

An RFID tag is a small computer chip with a tiny antenna built in. When a radio signal hits the antenna, an electrical current is generated. That current then powers the chip to do whatever it is designed to do. In an RFID tag, the chip reads information in the radio signal, and if that information says the signal comes from an authorized source, the chip then generates a signal of its own, carrying information such as an identification number. Since the signal is quite weak, it can be detected only by nearby detectors, or by very sensitive and somewhat further away detectors. Bigger chips, with bigger antennae, amd sometimes with batteries, can generate stronger signals. Bigger chips can also carry more information than just an ID number.

So what are they good for? The early (large, expensive) RFID tags were fastened to cargo containers and pallets of goods. When queried by a suitable radio signal, they could report ID numbers (corresponding, perhaps, to invoice numbers), origin, and destination. They are the heart of highway automatic toll systems (Transpass and E-ZPass). Customers of some gasoline companies can wave a gadget at the pump that automatically debits their account. I have a bank debit card that works similarly in some stores. Similar devices form the heart of systems you can buy now to track stolen cars. People in the field are now talking of including sensors for temperature that would enable a chip to report whether a food shipment was likely to be spoiled.

Current tiny tags can be used to replace bar codes on labels. Instead of using a laser scanner, a cashier would activate a radio signal that induced everything in a shopping cart to report its identity and price. Checkout could be greatly speeded up. Taking inventory would become an easy task. So would logging new shipments into a warehouse. And such gains promise considerable cost savings for businesses. But businesses that have proposed using them in that way have met public objections. So far Wal-Mart (for instance) is restricting their use to inventory and keeping them away from the checkout stand; in "Wal-Mart Turns to Smart Tags," the Associated Press (Wired News, Friday, April 30, 2004) reported that for Wal-Mart,
"The radio frequency information, or RFID, tags provide automatic tracking of pallets and cases of goods. Eight suppliers are participating, using 21 products to be tracked. Wal-Mart said Thursday that it will have more than 100 suppliers using the tags by January. ... In a backshop retail environment, the tags will contain the details of what is in a case or on a pallet of goods. Rather than have a worker with a handheld scanner logging in barcodes, the system will let a computer system use a radio signal to log the goods as they arrive at the loading dock.

"The tags can also be used in the manufacturing process, which Dillman said can help suppliers become more efficient, and the tags will help companies on both ends know where their products are at all times.

"Wal-Mart says the tags will help reduce theft and counterfeiting, the latter particularly affecting prescription medicines.

"Dave Hogan, chief information officer for the National Retail Federation, said the RFID tags could gain an important place rather quickly. He said barcodes will likely be around for quite a while and that he expects them to be used in concert with RFID tags even when the new technology moves to store shelves."
In August 2006, Wal-Mart's new CIO, Rollin Ford, said that though less than 10% of their 6600+ worldwide stores are RFID-equipped, they do plan to continue with their RFID rollout.

Other current uses include inserting them under the skin to track livestock and identify lost pets. ID tags are even being sold for people; the idea is to track lost or kidnapped kids or record medical history data in a way that cannot be lost (as can a medical alert bracelet).

To a technophile, this all sounds rather nice. It's a gadget with a host of beneficial uses. But there is enough resistance to keep commercial applications largely restricted to the supply chain. This resistance is based on the potential for someone with a reader to read the contents of your cupboards from outside your house. Such data could be used in many ways, not all of which would make us happy. The idea of embedded personal ID tags also alarms people, for--quite aside from concern over such things as national ID cards--such tags might enable moment-by-moment tracking (as discussed in the unit on Surveillance--see "Whereware"). This is already becoming a feature of cell phones, but at least one can leave a cell phone home. Concern over such things prompted Simson Garfinkel, writing in the October 2002 Technology Review, to propose an RFID Bill of Rights, saying that
"Consumers should have "I see these not necessarily as the basis for new law, but as a framework for voluntary guidelines that companies wishing to deploy this technology can publicly adopt. Consumers could then boycott companies that violate these principles.

"Of course, some of these 'rights' could easily be curbed or otherwise limited by federal regulation. For example, the U.S. Department of Transportation could require certain safety-critical parts inside a car to have radio tags to aid in recalls. But for the overwhelming majority of applications, these rights make sense. Manufacturers have no business playing hide-and-seek with radio tags when consumer privacy is at issue. Likewise, they shouldn’t be able to require that consumers choose between participating in tomorrow’s economy and preserving their privacy. For example, this spring [2002] the Massachusetts Turnpike Authority started giving discounts to state residents who pay tolls with electronic transponders—a policy that is both discriminatory and coercive."
I can easily imagine a device for sale to consumers that would detect and destroy RFID tags attached to products. But even if such a thing never shows up in stores, I suspect it is only a matter of time before RFID tags are as ubiquitous as bar codes. It is also only a matter of time before RFID tags start getting hacked.

As we noted earlier, technologies can fail to emerge because they are supplanted by better alternatives. On July 17, 2006, Hewlett Packard announced a new chip the size of a tomato seed. "The Memory Spot has a 10 megabits-per-second data-transfer rate and can store up to 4 megabits of data." Reading it requires physical contact, so no more scan-from-a-distance privacy issues. Because it contains so much more memory, new applications become possible, such as attaching sound clips to the pictures in a (hardcopy) family album. HP says it may hit the market in two to five years.

This seems something to keep an eye on. Will it replace RFID? Perhaps not, for some uses of RFID do depend on being able to detect and scan the chip from a distance. I suspect it will find its own niche, based on the new and unique applications it makes possible.

How about a memory necklace? Each bead on the necklace has a Memory Spot, and you can review your data by running the scanner over the necklace. Or a children's book that uses the scanner to play video and sound clips to go with the narrative?

It is worth noting that the basic RFID idea has extensions about which most people have not heard even hints. "Smart dust" was first proposed in 1997 and promptly funded by DARPA (see Jessica Jones, "Dust in the Wind," Government Technology, March 2004). Writing in the March 23, 2003, Computerworld, Thomas Hoffman describes
"'Smart dust' devices [as] tiny wireless microelectromechanical sensors (MEMS) that can detect everything from light to vibrations. Thanks to recent breakthroughs in silicon and fabrication techniques, these 'motes' could eventually be the size of a grain of sand, though each would contain sensors, computing circuits, bidirectional wireless communications technology and a power supply. Motes would gather scads of data, run computations and communicate that information using two-way band radio between motes at distances approaching 1,000 feet."
Some depictions of smart dust suggest the power supply could be handled the same way as for RFID tags, by inducing a current in an antenna with an external radio signal. Given that the chips are larger than most RFID tag chips and they contain sensors and computational capacity, they may be able to accomplish surprising things, beginning with the simple capacity to pass information from mote to mote until it reaches a final destination (motes scattered on a battlefield could report on enemy movements, and feed targetting data to missile batteries or bombers). They might also collect weather data and then--functioning as a large distributed (networked) computer, generate weather forecasts of unprecedented accuracy.

Or they may prove invaluable as environmental monitors. William J. Broad, "A Web of Sensors, Taking Earth's Pulse," New York Times (May 10, 2005), reports that "ecologists are planning to set up more than $1 billion worth of sensor web technology to study diverse environments with an eye toward saving the planet. Dr. Deborah Estrin with UCLA's Center for Embedded Network Sensing says the goal of such deployments is to create the ecological equivalent of MRI or CAT scans... Factors driving the sensor web wave include the support of institutions such as the NSF and the Defense Department, which have respectively financed planning and research into new sensor network deployments and the miniaturization of electronics to yield technologies such as motes and smart dust."

Will "smart dust" pose privacy problems? The sensors it can carry mean that it is not capable only of reporting, like an RFID, what product you bought or own. It can take pictures, record sound, sniff for drugs, and more. It can be scattered in a student's dorm room or clothing. It can be embedded in paint. It is actually much more alarming than RFID tags!

Some people are speculating on what you get if you give smart dust tiny grippers or other movable parts. The name for it now is utility fog, and if RFID tags can be hacked, so can this. And here we have the stuff of nightmares!

LED LIGHTING



Back to Top


The history of human civilization is in part the history of human efforts to stay up later at night. We have moved from torches to oil lamps, from candles to whale oil and kerosene lanterns, from incandescent light bulbs to compact fluorescents. Along the way we have suffered from unfortunate side effects of each technology, from a tendency to burn down the tent or fill the air with smoke, to the near extinction of whales, to strip mining, air pollution, and high electric bills. But there's a clear trend visible, for each technology gets more light out of the same amount of fuel. This is easily apparent if one but compares a 100-Watt incandescent bulb (10% efficient at turning electricity into light; the rest becomes heat) and a 23-Watt compact fluorescent light bulb (70% efficient), which produces the same amount of light. The compact fluorescent uses less energy and lasts longer, so that even though it costs more to buy, the owner saves money over the life of the bulb. The US government's Energy-Star site says, "If every household in the U.S. replaced one light bulb with an ENERGY STAR qualified compact fluorescent light bulb (CFL), it would prevent enough pollution to equal removing one million cars from the road."

That's impressive. But the trend is hardly over, for now electronics technology has given us the ability to make silicon emit light. The basic device is the Light-Emitting Diode, which has a theoretical efficiency about triple that of the fluorescent light (imagine the equivalent of a 100 Watt incandescent bulb running on 7 Watts!). Already, colored LEDs are replacing traffic lights; they use a tenth as much energy and last five times as long as incandescent bulbs. White-light LEDs are under development and can already be bought in various forms, including flashlights (here is a sample site). Neil Savage, in "Turning on LEDs," Technology Review (January 11, 2006), says that "If the technology can be improved so that half of all lighting is solid-state by 2025, it will cut worldwide power use by 120 gigawatts, saving $100 billion a year and reducing carbon dioxide emissions from power plants by 350 megatons a year." This projection appears fairly reasonable because white-light LED technology is on track "to produce 150 lumens of illumination per watt of input power by 2012, up from just 25 lumens/watt in 2002. That’s 10 times the efficiency of an incandescent bulb and substantially more than the 50-100 lumens/watt from a fluorescent bulb."

David Talbot's "LEDs vs. the Lightbulb," Technology Review (May 2003), describes both silicon-based and organic LEDs (OLEDs), saying the latter are an important line of development that is not yet as ready for the market. Robert F. Service, "Organic LEDs Look Forward to a Bright, White Future," Science (December 16, 2005), notes that OLEDs are very promising. Janet Raloff, in "Illuminating Changes," Science News (May 20, 2006), says these lights may soon make traditional incandescent and fluorescent lights obsolete and save large amounts of energy in the process, while also reducing emissions of carbon dioxide, which is largely responsible for global warming. OLED lighting holds hope for easy (cheap) manufacture and new forms, such as flexible sheets that could be hung from walls or made part of clothing or furniture. (The same technology is also the key to flexible screens; see Michael Kenellos, "Samsung Unveils Largest Flexible LCD," CNET News.com, November 28, 2005.)

Will LED lights emerge as have compact fluorescents? Initial applications are likely to be for commercial displays and difficult-to-maintain sites (such as traffic lights). They offer distinct advantages in energy consumption and long-term costs, but their own costs will have to come down a bit before most people will be willing to stand the up-front purchase cost. But that will happen, as it did with the compact fluorescent lights. I am sure that it is only a matter of time before most people have more LED lights than incandescent lights.

Actually--they already do. All those little red and green indicator lights on appliances, computers, TVs, etc., are LEDs. So I need to adjust that prediction to: "I am sure that it is only a matter of time before most people have more LED lights bright enough to read by than incandescent lights."

Note that lighting uses a significant fraction of all the electricity generated. According to the International Energy Agency, "For the industrialized countries national lighting electricity use ranges from 5 % to 15 %, while in developing countries the value can be as high as 86 % of the total electricity use. The corresponding carbon dioxide emissions were 1775 million tonnes, of which approximately 511 million tonnes was attributable to the IEA member countries." (A tonne is a metric ton, 1000 kg, or 2200 lbs.) Improving efficiency can either make more electricity available for other uses or decrease the amount of electricity produced, and therefore also reduce carbon dioxide released and impact on global warming.

NANOTECHNOLOGY



Back to Top


The prefix "nano" means one billionth. In "nanotechnology," it means devices built on a scale of billionths of a meter, which is the size range for viruses.

These devices have been imagined in many forms. We have mentioned "utility fog." As described by Eric Drexler in his 1986 book, Engines of Creation, they would be able to manipulate and position single atoms and molecules. For a time, enthusiasts talked of the devices as self-reproducing robots that needed only suitable programming to manufacture practically anything from dirt, air, and water, or to disassemble anything into its component atoms. Consumer goods--from steaks to cars--would be essentially free! Furthermore, nanomachines would repair wounds, destroy cancer, and rotoroot the cholesterol out of our arteries. It sounded like magic, and it stirred debate over the possibility that out-of control nanobeasties might turn everything into gray goo. But despite the enthusiasm, progress has been slow and tiny manufacturing and disassembly robots now seem unlikely. However, the National Heart, Lung, and Blood Institute published a report in 2003 (Denis B. Buxton, et al., "Recommendations of the National Heart, Lung, and Blood Institute Nanotechnology Working Group," Circulation, 108, pp. 2737-2742) that called the medical prospects encouraging and called for increased research effort and funding. Enthusiasts such as the Foresight Institute remain optimistic.

Chuck Lenatti, in "Nanotech's First Blockbusters?" Technology Review (March 2004), reports that the effort to learn how to make tiny things is having some practical payoffs already. No one is building tiny robots of any kind, but some companies are making tiny components (such as "nanowires") from which they hope to build marketable photovoltaic cells, LEDs, flexible circuitry, electronic devices, and so on.

How long might it take to go from this sort of thing to the tiny manufacturing and disassembling robots? Most people have thought that if the step is possible at all, it will take decades. But in summer 2003, the Center for Responsible Nanotechnology concluded that it could be a matter of weeks, for even simple nanomanufacturing, combined with computer-aided design and manufacturing techniques, would enable extraordinarily rapid progress. (See Mike Treder, "Molecular Nanotech: Benefits and Risks," Futurist, February 2004.) And while the benefits may seem enticing, the hazards strike some as so fearful that society should ban the development of nanotechnology. Bill Joy, "Why the future doesn't need us," Wired (April 2000), even extended the ban to robotics and genetic engineering on the grounds that these three technologies threaten to make humans an endangered species.

Clearly, something is going on here that is not covered by our list of reasons why a technology may not emerge. Perhaps we should add "Society decides that the new product or technology is too hazardous to accept." But, since the list is rooted in reasons why products or technologies have failed in the past, our new entry has never before been used. Has it? We can discuss nuclear power and other debates (over genetic engineering and cloning, for instance), but perhaps we also need to consider whether our new list entry is ever likely to be needed. That is, can society choose to block a technology?

E-VOTING



Back to Top


E-Voting is an excellent example of an enchanting technology with acceptance problems.

The basic idea is simple enough: A ballot is like a thousand forms that we fill out on computer screens all the time. We are used to the process, and we rather like the idea that we might not have to go to a specific place to fill out the form. So why not use computers? Touch-screen or even mouse-and-keyboard machines at a polling place could replace paper ballots with the advantage of instant vote-counting. Or we could use the Internet, which would be a special boon for absentee voters--soldiers overseas, home-bound elderly, commuters who aren't home when the polls are open, and so on.

And it's not as if traditional paper ballots can't be fiddled. In the 2000 Presidential election, we learned about punched card ballots (with hanging chads), butterfly ballots, and other problems. Ballot boxes can be stuffed. Dead people can vote. Even voting machines can be rigged.

Is there any voting method that can't be rigged? The idea of e-voting is as old as the Internet, but it gained new impetus after the 2000 election. Since then a number of versions have developed. eBallot is a software approach. Diebold Election Systems offers a hardware-based system that records and tabulates votes. There are others as well.

But the old paper-based systems have some desirable features: On a computer: The first two of these problems, says Simson Garfinkel in "The E-Vote Campaign," The Net Effect (September 3, 2003), are why many computer science professionals oppose ATM-style (direct recording or DRE) voting machines. Yet "Politicians ... have been hacking elections in America for more than 200 years. The geeks are focusing on the abilities of hackers to steal elections by reprogramming DREs because electronic attacks are what these folks understand. But if your goal is truly better elections, he says, the DREs can do more good than harm." Some experts believe DREs are the wave of the future. However, there is strong pressure for making the machines able to provide a "paper trail" that permits comparing votes cast against votes recorded and allows recounts.

Yet a paper trail does not help much if people are casting "absentee" votes over the Internet. Here tampering may both be easier and be harder to detect. Aviel D. Rubin, "Security Considerations for Remote Electronic Voting," Communications of the ACM (December 2002), discusses the nature of the problem and available solutions. However, he says, there is currently no way to guarantee security of the system. The stakes are so high that national enemies might devote great resources to hacking a remote voting system and controlling the outcome of an election. Even a simple Distributed Denial of Service attack, such as happens all too frequently with current computer viruses and worms, could shut an election down.

In preparation for the 2006 elections, many states chose to implement e-voting. However, according to Patrick O'Driscoll, "Several Lawsuits Target E-Voting," USA Today (June 5, 2006), "Lawsuits have been filed in at least six states... to block the purchase or use of computerized machines. ... Most of the suits argue that the machines are vulnerable to software tampering, don't keep an easily recountable printed record and may miscount, switch or not record votes and even add phantom votes." Andrea Stone, "Analysis Finds E-Voting Machines Vulnerable," USA Today (June 27, 2006), reports that though there are no reported cases of electronic voting machines being hacked, there is a genuine risk that demands paper trails and periodic auditing, as well as removing all wireless connectivity (which can facilitate hacking). However, the HBO documentary "Hacking Democracy" paints a much more pessimistic tale, "exposing secrecy, votes in the trash, hackable software and election officials rigging the [2004] presidential recount."

The Information Technology and Innovation Foundation says that a paper trail would not really improve security, though it would increase costs. As of October 2007, one electronic California election has been thrown out, and 26 states have pending legislation requiring paper trails. The U.S. Government Accounting Office is insisting on testing Florida voting machines because of problems reported by voters. "The Dutch government decided [recently] to pull the plug on the e-voting venture, citing the lack of a paper trail as its biggest shortcoming." In Ireland, voting machines have been mothballed because of concerns that hackers could "take over" an election.

A recent study from New York University School of Law (see Marc L. Songini, "Concerns about Fraud Potential Continue to Plague Users of Electronic Voting Machines," Computerworld, July 3, 2006) reports that "half of the manual voting systems in the country have been replaced with electronic devices" and they may be vulnerable to external attack. Countermeasures are needed. Nevertheless, says the study, at least until external attacks actually happen and prove successful, e-voting looks like the wave of the future. Despite debate and lawsuits, it is being adopted, security is being addressed, and standards are being developed. However, in 2007, as voting officials prepare for the 2008 elections, there remains a great deal of skepticism and concern. We may not yet be ready to shift to e-voting with confidence that our votes will be counted accurately.

3D PRINTING



Back to Top


Once upon a time, the only way to make a three-dimensional object was by hand. One could carve wood or bone, shape clay, blow glass, and pound metal into shape. Unfortunately, while such techniques are extraordinarily versatile, they are not good at making many identical duplicates of a single object. Molding and casting can make duplicates that seem identical to the human eye, but they nevertheless vary on the microscopic scale, which can be quite important if the duplicates are intended to function as interchangeable parts in complex machines. Machine tools such as lathes improve the situation enough to make interchangeable parts for relatively simple machines such as rifles, and indeed Eli Whitney's development of interchangeable parts for muskets helped create mass production and make America's westward expansion possible in the nineteenth century.

But early machine tools were human-controlled. It took a great deal of expertise to use them, and a fair amount of time to make each duplicate. In the late 1940s, John Parsons (head of a company that produced helicopter rotors) devised a way to make punch-card-operated electromechanical calculators generate templates for human-operated machine tools to follow. He then envisioned an extension of the system that would have automated machine tools follow the templates on their own. He became known as the father of numerical control technology and was awarded the National Medal of Technology in 1985.

As computers developed, numerical control became quite sophisticated. It even became possible to design an object on a computer and feed the design to automated equipment to make the object; this is CAD/CAM (computer-aided design/computer-aided manufacturing). But the process remained expensive--"machine tools" meant drills and lathes capable of working hard metal--and it remained impossible to make hollow objects in one piece (two or more pieces had to be glued or welded or bolted together).

Both of these problems are now being addressed. "Rapid-prototyping" and "3D-printing" tools have been built from the basic idea that a thin layer of powder (plastic, metal, or ceramic) can be fused with a laser, or a thin film of liquid (plastic) can be polymerized (hardened) with a laser. A second layer can then be laid down atop the first and similarly fused or hardened. The trick is to build a machine that can produce and harden layers on demand. When such layers amount to slices through a 3D object (even ones with interior spaces), the accumulation of layers produces the object, as described in Ivan Amato, "Instant Manufacturing," Technology Review (November 2003). It is already finding much use in building prototypes (rapid prototyping) of items to be manufactured by more conventional means, as well as special items such as gears and machine parts, bone implants, and form-fitting items such as hearing aids. The equipment tends to be expensive--in the fall of 2006, the lowest prices were in the neighborhood of $20,000. But that seems about to change, drastically. In January 2007, New Scientist reported that a team at Carnegie Mellon has developed an open-source 3D-printer or "fabber" kit that should cost about $2,400 for the parts. You put it together with a soldering iron and screwdriver.
The standard version of their Freeform fabricator – or "fabber" – is about the size of a microwave oven and can be assembled for around $2400 (£1200). It can generate 3D objects from plastic and various other materials. Full documentation on how to build and operate the machine, along with all the software required, are available on the Fab@Home website, and all designs, documents and software have been released for free.
Want to bet on these? I've talked with people about these fabbers, and I hear a lot of "I want one!" comments. The Fab@Home folks have tested the device with silicone, plaster, play-doh, frosting, and chocolate. Sculptors are eager to try using wax, for a fabber would make lost-wax casting much easier.

So far, the fabber is at about the same stage of development as the PC in the late 1960s, when you had to buy a kit and put it together yourself. But the PC developed rapidly and became enormously more powerful, as well as cheaper. I don't think it will be long before someone is selling pre-assembled fabbers and the price is under $1,000.

In 2004, the latest version of the technology was essentially an inkjet printer that could build up small shapes. Another version offered a 3D printer than could almost reproduce itself! (See Celeste Biever, "3D Printer to Churn Out Copies of Itself," New Scientist news service, March 18, 2005.) Still another version "printed" a biodegradable gel and distributed cells; the aim was to produce custom-designed tissues and organs for use in transplants. Researchers have already begun to develop techniques for "printing" skin and blood vessels, and the future should be very interesting. See Rebecca Camber, "Tailor-Made Skin from 'Ink' Printer," Manchester University (January 19, 2005), and Kate Green, "Printing Blood Vessels," Technology Review (January 20, 2006).

A 3D printer prints CAD/CAM files prepared with CAD/CAM software. But that's not the only way to make CAD/CAM files, just as a keyboard and word-processor is not the only way to prepare a text file for a 2D printer. We can also use a scanner.

Is there such a thing as a 3D scanner? In a word, yes. But they aren't for home use. The cheapest I see is priced at $2,495! (It gets lots worse!) Or you can make your own, which must appeal to anyone DIY enough to want the 3D printer kit.

And then there's DAVID:
"DAVID is a freeware software for laser range scanning. All you need is a PC, a camera (e.g. a webcam), a background corner, and a laser which projects a line onto the object you want to scan. So everyone can use it to scan objects without high costs; this is the big advantage over commercial solutions which are rather expensive.

"DAVID has been developed by the computer scientists Dr. Simon Winkelbach and Sven Molkenstruck from the Institute for Robotics and Process Control, Technical University of Braunschweig, Germany."
But everything I'm seeing here seems limited to scanning the exterior contours of an object. Since 3D printing can reproduce objects with empty spaces totally surrounded by solid material, what we need is a desktop CT (X-ray) or ultrasound scanner to pick up the interior details. Both are commonly found in medical settings, and there's one, the Echoblaster, that might come close to being usable for our purposes. Price not given, so that may mean the basic idea of an ultrasound scanner would need a different implementation.

Or maybe you just cut the object you're scanning open, which makes the inside outside, and use the regular 3D scanner.

Note that there's no real limit on how big a 3D printer can be. Engineers in both California and England are reportedly constructing arrangements of concrete-spraying robots mounted on a framework that lets them move like the computer-controlled syringe in the Fab@Home fabber to build a house 200 times as fast and at a fifth the price of a conventional house. Click here for the details. This is seriously cool! But an important question is what else can be built using a similar approach. How about highways? What would an artist use it for? And of course, if you can make 3D printers the size of microwave ovens and airplane hangars, you should certainly be able to build them any size in between those extremes.

Will 3D printing "emerge"? It already has a place in industry, so one can say it has already done so in one sense. But as Amato mentions, there is enormous potential in the realm of "on-demand manufacturing," personal customization (think of case-modding!), and so on, if the equipment can be brought down enough in price. If this happens, a great many products may no longer be sold in physical form, but as computerized design files. Consumers will "print" the files to obtain the physical products.

So what's happening to the price? You can now--in September 2007--buy the Fab@Home kits in a package for about $3,000 (assembly still required). If you have a little more disposable income--on the order of what loads of folks plunk down for a snowmobile or motorboat or even one of the bigger plasma TVs--there is Idealab's Desktop Factory for just under $5,000! The company says:
Within three years, Desktop Factory's 3D printers will be affordable even for home use. Imagine your child being able to select a toy from a catalog or even design his or her own and create it right at home. Think about the benefits of making your own parts for sprinkler systems or small appliances at your desk. The long-range possibilities are endless. Desktop Factory's 3D printers are set to unleash a new wave of talent and productivity in a three-dimensional world.
So where can you buy it? In May, Engadget said they were due to ship later in 2007. So it won't be long, folks. This has the potential to be as disruptive a technology as we have ever seen, and it is developing with astonishing speed. And it isn't just a matter of making toys and sprinkler parts at home. As the gadgets get better and cheaper, you will be able to make any small part you need, saving money and of course obtaining small parts that are no longer made (such as parts for classic cars). Current businesses that make small parts (and small toys) are in for a shock, for their sales will drop drastically. It is already time for them to start preparing CAD-CAM files to sell online to users of 3D printers. Before long, that may be about all they have to sell. Note, of course, that they will be facing intense competition from hobbyists with their own files for sale, not to mention file-sharers.

Will consumers choose to use the technology? If they do, there will be negative effects on all those companies that make and distribute the objects that can now be sold in the new form. There will also be positive effects will be on new companies that generate and distribute design files, as well as on consumers. The net effect is what will determine the emergence of the technology as a consumer technology.

Does that stimulate your thinking? Well, the technology does not stop here! Nanotechnology advocates have been talking for some time now about the "nanofactory," which will be able to manufacture on demand almost anything from a basic supply of atoms and molecules. If nanofactories are even possible (and some do question their practicality), they are much further off than home 3D printers.

AUGMENTED REALITY



Back to Top


We know what "virtual reality" is: A computer-synthesized version of reality. The first consumer version of it--almost as old as color graphics--appeared in games such as "Flight Simulator," which let a person pretend the screen of a computer monitor was what an airplane pilot could see from his or her seat, complete with altimeter and other gauges and a stylized window-view. The person pretended to fly the plane using a joystick or other control. The simulation was not very accurate, but it was convincing enough to make some users throw up on their keyboards. Current versions of "Flight Simulator," are much better. (More elaborate simulators were and still are used in pilot training.)

The ultimate version will surely be what we get when we are finally able--as Hans Moravec promises us--to upload a human mind into a computer. Such a downloaded mind could live forever (as long as no one pulls the plug or the hard drive crashes) in a synthesized world. It could also be the "control system" for a mechanical body--a robot, a truck, a submarine,a spaceship, you name it.

Current--and more moderate--versions of virtual reality replace the monitor with a head-mounted viewer as shown in the graphic on the opening page of Steve Ditlea's "Reality Redefined," Computer Graphics World (August 2002). What the user sees is still computer-synthesized, although the quality of the graphics is much improved. Computer synthesized sound is also available, and many researchers are working on adding touch (haptics) and even smell to the computer-synthesized sensations. Ultimately, virtual reality users may wear a special "body suit" that stimulates every nerve of the body just the way the real world does. Not surprisingly, some people take this thought all the way to teledildonics!

But despite the enthusiasm of VR proponents, there really aren't that many users out there. The reason may be that the illusion (the graphics, etc.) just isn't convincing enough yet. Or it might be that VR still tends to make some people feel pretty woozy.

Yet even if VR dies, the research into presenting personalized views will not. It is already giving rise to a field called "augmented reality" or "annotated reality." Ditlea describes the basics very well in his article: If you wear suitable gear, you will be able to look at a restaurant and see the menu, or a review, overlaid on the real-world view. How does the gear manage this? It has to track the wearer's position in its environment, using inertial tracking and/or GPS. It has to track which way the wearer is looking. It has to be able to find information that matches whatever the wearer is looking at (call it the target). This can be provided by a radio signal emitted by a device attached to the target, or it can be stored in a database which the gear can access as soon as it has identified the target.

There are other applications, too. Look at the sky and get a weather report. Look at a prospective date and see their last date's rating of them. Tour a campus or a museum, and see the sights explained as you go. Sign up for an AR game, and see dragons on Main Street. Use special sensors as in the sonic flashlight, and see bones inside a hand. A surgeon could see a CAT or MRI image overlaid on the body of a patient. A mechanic could look at an engine and see the appropriate pages of the manual or hear suitable instructions, thus speeding repair and assembly work (see Christoph Hammerschmidt, "'Augmented Reality' Speeding Assembly and Service Tasks," EETimes Germany, July 07, 2003). Mark Billinghurst and Hirokazu Kato, "Collaborative Augmented Reality," Communications of the ACM (July 2002), describe AR as a way for people who may not be in the same room (or even on the same continent) to collaborate, viewing images of coworkers as if they are in the same room and going "hands-on" with computer-generated images of objects the size of houses or molecules.

The potential is huge, and a great deal of work is going into making that potential real. (For a list of AR projects and related sites, click here.) Some of that potential seems bound to be realized. I think of medical and industrial applications. Collaborative AR seems likely to appear. Will it emerge into the consumer realm? This depends on such things as expense, cumbersomeness of the gear, and intrusiveness of the uses. It also depends on whether appearance of competing technologies, such as cell phone cameras and image search engines. According to Gary Stix, "A Farewell to Keywords," Scientific American (July 2006), researchers are developing software that will let one take a picture with a cell phone, send the image file to a server via the Internet, and get back web pages with information on what is in the picture. Walk down the street, snap a pic of a restaurant, and a moment later be looking at the menu or a review of the food and service. Go to a museum, snap a pic of a statue or painting, and read all about it. This clearly serves the same functions as some varieties of AR, and without requiring special location-tracking, gaze-tracking, databases, or signal-emitters on objects of potential interest (such as restaurants or museum exhibits). If it develops, it will be cheaper, more convenient, and more versatile. And it seems likely to develop! Microsoft already has in the works a version called Phone2Search.

MIND-MACHINE INTERFACES



Back to Top


For more than fifty years, neurophysiologists have been able to implant electrodes into the brains of mice, dogs, monkeys, apes, and humans. At first, the idea was to listen in on the messages nerve cells were sending each other. Later, researchers sent messages, using the electrodes to stimulate nerve cells and study what happened when the cells passed the messages on. In this way, they learned a great deal about how sensory information is processed and the commands to muscles are generated. That was enough to let some researchers dream of someday integrating computer chips and brains, adding electronic memory or personality implants, adding communications interfaces so the brain could access the Internet or other people directly (brain-to-brain Instant Messaging!), or even controls so that actions and even thoughts could be dictated by machines or other people. So far, this remains almost entirely science fiction--but not entirely. In 2002, "roborat" was displayed to the public. Rats with wires implanted in their brains could be steered right and left; the ultimate aim was "rats [that] might ... be used as scouts for search and rescue teams looking for survivors amid rubble or for sniffing out hidden land mines."

Gregory T. Huang, "Mind-Machine Merger," Technology Review (May 2003), makes it clear that current research is far beyond using just two or three electrodes. With $24 million in funding from the Defense Advanced Research Projects Agency (DARPA), researchers are using arrays of many electrodes, with the aim of producing "a new generation of electrodes, computer chips, and software that might eventually equip soldiers, for example, to control superfast artificial limbs, pilot remote vehicles, and guide mobile robots in hazardous environments, using only the power of their thoughts. Even more remarkable, such devices could enhance decision-making, upgrade memory and cognitive skills, and even allow one person’s brain to communicate wirelessly with another’s." Other aims include prostheses for the blind, the deaf, and even the brain-damaged. So far, many of these aims (like the remote-controlled rat) have not gotten past the prototype stage. But even that stage is both intensely interesting and--to some--frankly terrifying.

Consider the monkey equipped with wires that relayed the signals its brain was sending to its muscles to a robotic arm. It soon learned that it didn't have to really move its arm in order to move the robotic arm--just its thoughts could do the trick. Researchers then showed that the thoughts of a wired brain could even operate an arm hundreds of miles away, via the Internet. As Huang notes, further work could well give humans the ability to control distant machinery. When the same control technology is turned the other way--making muscles obey the mind's will when the nerves that should convey that will no longer work--it is easy to see benefits (for instance, for paraplegics). But when we imagine the control being exerted on human brain function (thoughts!)--it is easy to worry about the motives of the controllers.

By March 2005, the implanted electrodes approach was paying off in a big way. According to Richard Martin, "Mind Control," Wired (March 2005), a chip implanted in the brain of paraplegic Matt Nagle gave Nagle the ability to control a computer well enough--with just his thoughts--to play at least a simple computer game. In 2006, the same chip (with new software) was being used to give others the ability to steer a wheelchair. See href="http://www.technologyreview.com/read_article.aspx?id=17629&ch=biotech">the story here.

Jose del R. Millan, "Adaptive Brain Interfaces," Communications of the ACM (March 2003), is concerned with another method of conveying the mind's will to the outer world, without using electronic implants. (See also "Patients Put on Thinking Caps," Wired News, January 14, 2005.) He says, "Researchers and designers of human-computer interfaces are motivated by a growing interest in the use of physiological signals for communication among and operation of devices by physically handicapped people, as well as by their able-bodied counterparts. Combining neuroscience and computer science, recent experimentation has demonstrated the possibility of analyzing brainwaves online to derive information about a subject’s mental state that could then be mapped onto some external action (such as selecting a letter from a virtual keyboard or moving a robotic device). A brain-computer interface (BCI) is an alternative communication and control channel that does not depend on the brain’s normal output pathway of peripheral nerves and muscles." At present, the human side of the device looks like a cap studded with electrodes that pick up the electrical waveforms we call brain waves (electroencephalographic or EEG waves). The computer analyzes the waves to find meaningful patterns. It can then use those patterns to move a cursor on a computer screen, open or close a prosthetic hand, steer a robot, or even play a computer game. And in its present primitive state, the technology has drawn very positive reactions from disabled humans who have tried it. An early version is supposedly even ready for market! (See also Harry Yeates, "Brain-Computer Interface Device Now Commercially Viable for Use at Home," Electronics Weekly, April 12, 2006.)

Where is this technology going to go? This is a very good question. It has some distinctly beneficial uses. But it can also be used in other ways. If computers can control brains (recall the "roborat" item), then at least in principle computers could control human brains, and hence human actions or behavior. It worries me that governments would have few compunctions about using a new way to control the citizenry.

And then there's the fraud angle. The potential benefits are so great that this technology--like many others in the past--invites opportunists with a gift for bafflegab to offer buyers products that may be much more hype than hope. That is, they don't do what they promise, but they do move money from one set of pockets to another. As a case in point, in searching for recent developments in this area, I came across the Interchange Laboratories page. Their "technology is based on interactions at the subatomic level where the distinction between mind and matter disappears and effects become instantaneous over any distance." This line arouses my intense skepticism, for no one knows what the distinction between mind and matter is, much less under what conditions it disappears. And while instantaneous action at a distance may be a physics laboratory curiosity, it does not seem so far to have any real applications.

COMPUTER TRANSLATION



Back to Top


For as long as there have been computers, people have sought to program them to translate from one language to another. The reasons are obvious, for language differences must be overcome in business dealings, diplomacy, intelligence gathering, and even with pen pals. But though the problem is one that a child can master, it has not proved amenable to computerized solution. Supposedly, a CIA-sponsored program was once asked to translate "The spirit is willing but the flesh is weak" into Russian and then back again; the result was "The vodka is tempting, but the meat's a bit suspect." Click here and here to see why this may be more urban myth than truth. But that myth has a truth in it: Machine translation (MT) is a difficult task, and we have not licked it yet.

If you wish to see how one computerized translation program works, take a look at Altavista's Babelfish (named after the translation device in The Hitchhiker's Guide to the Galaxy). It is invaluable for giving you at least a rough idea of what a correspondent or web page is saying, but it is inaccurate enough that you should never get upset with a letter-writer solely on the basis of what Babelfish says that person is saying.

The basic problem is not in grammar or even in vocabulary but in the fact that most languages have a wealth of synonyms and idioms. Some languages (such as Japanese) can spell a word in many ways. An understanding of context is essential to choosing synonyms and interpreting idioms. Since computers do not (yet) understand context, "machine translation" is limited to areas such as engineering where the language is constrained. In less constrained areas, such as diplomacy or intelligence gathering, "computer-assisted translation" works better.
"Computer assisted translation is made up of a series of tools aimed at providing assistance for the translator with regards to both coherence (consistency) of his work and speed. The largest of these tools manage the specific terminology within the area of work as well as the translation memories.

"Terminology management, above all, has the computer scan each word of the source text in order to locate them in the specialised dictionaries and, whenever possible, offers an equivalent to the translator, automatically and in the target language. The efficiency of this function is therefore basically determined by the quality and the volume of the specialised dictionary. The constitution of translation memories requires the creation of equivalency tables between the source text and the target text. In order to do so, the software breaks the text to be translated into segments. At the point where the translator validates the corresponding target text, the software memorises the source segment and the target segment as being linguistic equivalents. If the source segment then appears in the text again (repetitions may be frequent in technical texts), the software automatically proposes the memorised translation. When updating the source version of a text that has already been translated, the software automatically takes the parts already translated and alerts the translator in the case of any new or modified elements. The most sophisticated types of CAT software recognise segments that are approximately identical, and alert the translator by marking the elements that differ from the memorised segment."

Until recently, says Michael Erard, "Translation in the Age of Terror," Technology Review (March 2004), the US government has supported research into machine translation. If that research were successful, the flood of documents and recordings that must be scanned for clues to crime, terrorism, and other topics of interest could be handled with computer speed. But as Babelfish shows, that research has been only partially successful. Now greatly expanded effort is going into "computer-assisted translation." The process begins with on-the-spot (perhaps in a cave in Afghanistan) scanning of documents, transmission of the digitized copies via satellite to a central translation facility where computers can do the rough work (including highlighting key words) and humans can more speedily handle details. If necessary, experts can be consulted via Internet. The hope is that a successful system will greatly aid the war against terrorism.

Will it be successful? Erard notes that government has a history of dysfunctional computer systems. It also often has trouble sharing sensitive information as would be necessary to transmit documents to outside experts. But even if the government cannot make this approach work, Erard notes that the attempt is spurring commercial work as well. And since the commercial sector moves a lot of information across translation barriers and does know how to make working systems and collaborations, "computer-assisted translation" seems likely to be a big success in the private sector. Indeed, if you Google on "computer-assisted translation," you find a number of commercial sites selling software and/or services.

Is there a solution to the computerized-translation problem? It may be only a matter of time. Consider this interesting report about a Carnegie Mellon research project developing a "people dubber." Instead of translating speech as such, it reads muscle movements and detects what one is about to say. Then it says the words in another language. The accuracy has a ways to go, but they seem to be making great progress toward a science-fiction dream, the universal translator.

"The idea is that you can mouth words in English and they will come out in Chinese or another language."

The ultimate goal, the researchers said, was to be in a position where you can just have a conversation.

Many other approaches are also being tried, including neural networks (click here for all you ever wanted to know about these things!), which have the interesting property that they learn from experience. Even though research into neural nets is an active area of computer science, they have not produced major products for the consumer market. Another approach that could be tried--"genetic algorithms"-- is still in its infancy but has already demonstrated impressive potential. Read Sam Williams' "Unnatural Selection," Technology Review (February 2005), and Jonathan Keats, "John Koza Has Built an Invention Machine," Popular Science (April 2006), for a description of genetic algorithms. The basic approach here is Darwinian: Generate a large number of candidate solutions to a problem and try them out. Then take those that do best, make a large number of random changes (you can even combine pieces of the successful ones in a sort of computer sex) and try out the new batch. Keep trying. And in the end... Williams and Keats talk about how this method comes up with patentable antenna designs and lenses and efficient fraud detectors. What will it accomplish when applied to larger problems such as computer translation?

Still another approach involves the use of statistics to link word groups or phrases to their equivalents in another language. See Gary Stix, "The Elusive Goal of Machine Translation," Scientific American (March 2006), which notes that the basic problem of translation is harder than chess. Some people think it too can be licked. Others, of course, do not agree, but on June 8, 2006, The Economist published an article titled "How to Build a Babel Fish," about how statistical techniques may bring a handheld "universal translator" to market in the next few years. Present handheld translators are hardly universal! Note the way they seem to be designed to cope with specialized vocabularies. But that may be about to change. The military is presently testing in Iraq software that converts a laptop into an Arabic-to-English translator: "When someone speaks into a microphone attached to the computer, the machine translates it into Arabic and reads that translation aloud over the PC's speakers. The software then translates the Arabic speaker's response and utters it in English."

VOIP--INTERNET CALLING



Back to Top


"Voice Over IP" (VOIP) is an excellent example of an enchanting technology that could not possibly succeed in its initial form.

It was enchanting because it let a user attach a microphone to his or her computer, establish an Internet connection with another user anywhere in the world, and talk to that second user through the computer's speakers. If the second user also had a microphone, they could hold a conversation--just as if they were on the phone. And there weren't any long-distance charges, not even for international calls.

The first version was offered by firms such as Net2Phone in 1995. I tried a later version (Dialpad) and found that it worked--but only sort of. The computer converted what one was saying into packets before sending them out onto the Internet. This is how the Internet (or Internet Protocol, IP) works; they may follow different routes through the Net and arrive after different intervals, but since they are numbered, they may be put back in order at the receiving end. Unfortunately, this means that a voice signal gets chopped up into chunks with gaps between them. Early VOIP worked, but one had to be both very patient and a dedicated cheapskate to stick with it.

By the late 1990s, the computer microphone could be replaced with a regular phone, and phones could be routed through the computer to make free calls. It was still choppy. But the technology was improving, and more importantly, more people were getting broadband (DSL and cable modems), which speeds up the flow of packets to and from the Internet tremendously. In 2003, Vonage was able to offer customers a $40 all-you-can-call service or a $25.99 plan with 500 minutes of long-distance (with a $29.99 activation fee). Vonage CEO Jeffrey Citron could brag that "We gave consumers an experience that's almost identical to what they're used to." The company had 34,000 subscribers who could use the service wherever they could plug a phone and adapter into a broadband line; the phone number stayed with the device. It was expecting to reach a million subscribers by 2006. By July 2006, it had 1.6 million. Other providers are doing well too, thanks to improved reliability, better sound quality, and lower prices. (Today, Vonage charges only $24.99 for its premium unlimited service [domestic and some international] package and $14.99 for the basic 500-minute plan.)

So far, the technology requires special hardware (click here for example) and a broadband connection. The next step is the Wi-Fi phone, and the Linksys EIP300 was announced as available for pre-order in February 2006. By July it was for sale as the WIP300.

By 2004, according to David M. Ewalt, "The New Voice Choice," InformationWeek (March 1, 2004), VOIP had become popular in corporate settings, with 80 percent of 300 surveyed business-technology executives saying their companies were using, testing or planning to deploy VOIP, largely because of the cost savings and despite some remaining problems (gaps between packets, computer crashes, Internet traffic jams, vulnerability to computer viruses, etc.).

On the consumer level, a number of cable companies (including Time Warner) are also offering or planning to offer VOIP service. Competition has emerged, with Yak advertising on prime-time TV. Skype is free for calling from computer to computer and cheap for calling regular phones. But there are regulatory issues. VOIP providers until recently had escaped the state and federal regulations (including the requirement that users be able to dial 911) and many taxes that affect the telephone industry, which thinks VOIP is enjoying an unfair advantage. Telecom lobbyists have urged that the playing field be leveled, and the Federal Communications Commission (FCC) and state and federal legislatures have debated extensively whether to extend the same or similar rules and taxes to VOIP. Many court decisions have gone in VOIP's favor, but in 2005, the FCC said VOIP companies had to make calls tappable by law enforcement and in June 2006, the FCC decided that VOIP customers had to contribute to the Universal Service Fund. Click here for the latest VOIP news. It may change rapidly. The FCC has established an Internet Policy Working Group to "identify, evaluate and address policy issues that will arise as telecommunications services move to Internet-based platforms."

As regulations and taxes are imposed, VOIP rates will surely rise. If the rates wind up the same as for standard phone service, and if the technology continues to have problems, VOIP may never emerge into anything more than a niche market, such as internal corporate communications where it is already popular. If the rates remain lower than standard service, and if the technology continues to improve, it may displace conventional phone service, largely or even completely.

On May 4, 2004, FCC Chairman Michael Powell, speaking with National Cable & Telecommunications Association President Robert Sachs at the organization's national convention, said he thinks that VOIP poses a low-cost threat to local phone markets that have long been dominated by traditional phone companies and "is going to be the very, very best and biggest breakthrough in our ambitions and dreams about competition ever. If consumers respond to it, we will have to be vigilant about not allowing the incumbent, in any anti-competitive way, to choke off that possibility." Today, however, that "incumbent" appears to be trying its best to limit the competition by arguing against "net neutrality"; VOIP providers such as Skype worry that large ISPs (such as Comcast) may block any VOIP service that they do not themselves provide. Skype plans to become so popular that such moves will not work.

For the future, keep an eye on the growing trend of delivering games, video, etc., on cell phones. As this moves into the VOIP area, we have what is being called "purple minutes" or Voice 2.0 (to go with Web 2.0), with additional content beyond plain "black and white" voice being added. A major conference on this was held October 16, 2006, in Ottawa, Canada. Of course, if you don't have broadband, forget it!

THE WAR AGAINST SPAM



Back to Top


People love their email. But over 40 percent of it is junk or spam, advertising sexual aids, porn, small-cap stocks, and fake Rolexes, and tempting the gullible with offers to share millions of illicit Nigerian dollars, and so on. Almost 13 billion spam emails are sent every day, and the traffic is increasing steadily. A surge in the last few months may be due to the increasing use of "bot-nets." Some fear that spam will kill email.

What happens if you email the spammers to demand they stop? They're delighted! Now they know you read your spam, and they will send you more. They will also sell your email address to other spammers.

That's if your message gets through. Usually it won't, for they close their email accounts as soon as they have dumped a million or so spams into the Net, or the return box gets filled so quickly with complaints that it overflows, or the return address is fictitious.

Anti-spam laws have been passed in many countries. In 2003, the US Congress passed the Can-Spam Act to require that spammers provide working return addresses and ways for consumers to say, "Lay off!" Unfortunately, the spammers have already found ways around the law (for instance, by sending spams from or through computer systems outside the US). The Federal Trade Commission is interested in the problem but has so far accomplished relatively little. Only a few cases have so far reached the courts. In May 2005, the Massachusetts Attorney General filed a lawsuit against one of the world's biggest spam gangs, and an emergency court order demanded that the gang's web sites shut down. In September 2006, "The Court of Appeals of Virginia upheld ... what is believed to be the first conviction in the nation under a state anti-spamming law that makes it a felony to send unsolicited mass e-mails."

Individual victims of spam have been inventive in devising ways to fight back. Scam-baiters have had a great deal of fun with that Nigerian scam, for instance. Andrew Conry-Murray ("Fighting the Spam Monster--and Winning," Network Magazine, April 2003) describes a number of ways that computer scientists are developing to fight spam. They include blocking email from known spam sources (such as AOL), even at the risk of blocking legitimate email; accepting email only from trusted sources; filtering out email containing key words; and others, all of which can be circumvented by ingenious spammers. Fortunately, one new anti-spam technology, Bayesian filtering, in a way learns as it goes, adjusting to the spammers' attempts to get around blocks. However, once again the spammers are innovative alternatives, such as making their messages images, not words (see "A Picture Is Worth Even More Spam," Communications of the ACM, October 2006).

Another technique, described in Evan I. Schwartz, "Spam Wars," Technology Review, July-August 2003, involves imposing costs on email. The basic idea is that of postage, but not enough for an ordinary user to notice. Consider that if you had to pay a tenth of a cent for every email you sent, you might spend a few dollars a year. But a spammer who sent a million spams a day would be out $365,000 a year. If that is not enough to slow the flow, raise the rate to a penny. Or instead of money, use time: Engineer the system so no one can send more than ten email messages per minute. Ordinary users would rarely notice, but the spammer's million daily emails would take over two months to send!

In February 2006, AOL proposed charging commercial emailers for mail sent to AOL customers. This "certified email" scheme was not well received, but both AOL and Yahoo are pushing forward.

Another new technique proposes using peer-to-peer (P2P) networks to expand any one email recipient's database of identified spam. This technique has not yet been implemented, but its inventors propose to do so soon. A similar system was in the alpha stage in 2004.

Still another technique would require authentication of senders or their domain names. Hardware-based solutions are also being tried, but PC World said in June 2006 that they were only partially effective.

A more promising tack may lie in taking a more biological approach. Living things such as our own bodies have an immune system that, in essence, defines what belongs in the body and attacks what does not. It can go wrong, but in general it works quite well. Researchers have recognized that a similar approach might work well in computers as a defense again viruses, worms, trojans, and spyware, if not spam, at least since 1997 (see Stephanie Forrest, et al., "Computer Immunology," Communications of the ACM, October 1997). A few researchers have been working on the problem, with some signs of progress recently. See, for instance, Li Tao, "An Immune Based Dynamic Intrusion Detection Model," Chinese Science Bulletin (November 2005).

As we have seen before, with translation and surveillance, for instance, we often identify an emerging technology or a cluster of related technologies by naming the problem they address. Here the problem is spam or other unwanted email. The previous paragraphs mention several technologies that may prove useful, but some have other applications as well. P2P, for instance, is used most notoriously in file-sharing and music-downloading. Some applications are legal. Some are not. Some have already been thought of. Some have not.

Back to Top