Thursday, January 20, 2005
Grassroots robotics and the (coming) failure of virtual reality
For the past couple of entries I have been looking at the current revival of slobbering belief in the articles of faith in the Internet 1990s - in particular, the belief that cyberspace is about to become all-important and surpass the real world (which is just another cyberspace, according to this belief system). In the next couple of posting I will recount the attempts of roboticists to create viable world models, and how their complete failure motivated a different approach largely responsible for 'new era' robotics. This failure of world models does not bode well for a world in which cyberspace and the real world begin to blend, as imagined by many last-revolution pundits. The drive to suck video from every streetcorner of the world into the web and put RFID tags on every device, even "give every lightbulb and IP number" is about to run smack into the law of unintended consequences as well as the brittleness problem.
First off however, a salute to several new robots and robot stories. First up is the announcement made by Toyota last week concerning use of robots in their factories. According to one of many sources in Asia (Business Day Journal):
"...TOKYO – Toyota Motor will introduce robots which can work as well or better than humans at all 12 of its factories in Japan to cut costs and deal with a looming labour shortage as the country ages, a report said last week.
The robots would be able to carry out multiple tasks simultaneously with their two arms, achieving efficiency unseen in human workers and matching the cheap wages of Chinese laborers, the Nihon Keizai Shimbun said..."
So Japan, under pressure from Chinese-based manufacturing, is electing to keep the "jobs" in the country - by replacing human workers with robots. These are not mobile or autonomous robots, but advanced industrial robots which finally merit the name, since they make extensive use of computer vision and other sensors to accomplish their tasks. The reason for the move is, as is often cited in the Japanese press, a declining population and resistance to filling low-paid jobs with "guest workers" from other countries.
Second up is an interesting article at automatedbuildings.com concerning security robots by Lloyd Spencer of Coroware, a unique consultancy/system integrator which (among other things) helps business integrate robots into the workplace. Besides describing the benefits of robots versus a "wired" building, he notes that three factors - off-the shelf hardware, embedded operating systems and emerging telecommunications standards like JAUS are helping make security robots more practical. Similar analysis has been posted over at Activmedia, who survived the "robot winter" of the 1990s and is currently developing their robots for security. I agree - while I love the current "hijack" mentality of robot developers who steal from the computer/Internet industry to make their novel creations, a robotics components industry will be needed to drive robots into wider markets.
This said, the grassroots robotics world is still fun - and it is changing. For about a decade the typical robot, even at university level, was a trash can on wheels with a bump sensor. Today, a fascinating article from Omynews in Korea interviews Oh Jun Ho, the creator of the recently announced Hubo robot. While the HUBO superficially looks like Honda's Asimo, it was developed in only 2 years (as opposed to the 15 for Asimo) and cost only 1 million US (as opposed to > $300 million for Asimo).
Another article out indirectly concerns robotic hardware. Conventional PC designs have reached "heat limits" in that we can't have 100 GHz or even 10 GHz computers without pumping so much power through the CPU that it vaporizes. Companies like Intel have taken to touting "multicore" processors - in effect, two or more CPUs on the same chip. But an article by Shahin Khan, VP/CMO of Azul Systems tells us to prepare for the age of "kilocore" computers, parallellng the vast advances in RAM and disk storage. The company has been hard at work developing "compute" network appliances with hundreds of Vega processors, making (Java) computing power a resource like memory and storage.
Good quote:
"...In 2005, Azul Systems will ship compute pools with as many as 1,200 CPUs per a single standard rack (1.2 kilo cores! - I like the sound of that!)..."
As I've discussed before, multicore are useful for the server market, almost useless for most other PC/network tasks, and extremely useful for robotics. A 1 kilocore robot could handle tens of thousands of sensors, compared to the dozens found on robots today. This is because sensor processing, unlike many other computer tasks, can easily and natuarally be split among multiple processors. I've said before that rising sensor density is the real "Moore's Law" driving robotics - possibly it is linked very significantly to the rise of multicore. Time for a group with "modest" (1-5 million?) means to "hijack" this technology for an advanced robot!
Finally, in the "another day, another robot" category, Sega has developed a music-playing "robot dog" wich some of the same features as the Sony Aibo. Interesting, though I suspect the "music composing" capabilities of this not very doglike robot are realling DJing existing music tracks. Also, the sensory capability of the Sega system is extremely limited, touch only. Not sure if this is a real robot or another useless gadget. I'd be really surprised if Sega, a company whose lifeblood is the virtual world of games, understands robots at all.
Some might want to cite a "Moore's Law" for robotic development, but the truth is more interesting - humanoid robots are cheap to build if you deliberately try to do it on a shoestring. The problem is the culture around these projects. At Korea Advanced Institute of Science & Technology (KAIST) there was little money for creating the KHR robot series, so Ho used a mix of students and volunteers. Instead of hiring an army of engineers, he designed the system himself and did the research for components (e.g. motors) over the Internet. And every time a student told him that this or that graduate thesis said you could'n't us a particular (cheap) motor, Ho told the student to figure it out anyway.
Ho's comments are very relevant to the US robotics community. At robot centers like CMU and MIT the "thesis" mentailty spoken of in this article is very strong. A long string of completely impractical robots have been built for 30 years or more simply to make sure that grad students get their doctorate. While good for education, this approach has stifled the community. In contrast, KAIST threw out that 30 years of thesis work and jumped in and built a humanoid robot on the cheap.
A great quote:
What, technologically, was the most difficult stage?
Firstly, I couldn't approach the problem with the existing paradigm. It would have taken me 20 years had I done so. All the theses in the world are written according to that paradigm. There were no theses for me to refer to. More difficult was persuading my students. At first, they didn't believe a word I said. No matter what I said, they would bring me this thesis and that thesis that said what I wanted to do was impossible. I tore up whatever thesis they brought me. It took from 6 months up to a year to convince them.
For example, if I told them to use motor B instead of motor A, the students would tell me it was impossible, citing some thesis or simulation result. But the use of motor B was breaking existing blueprint concepts. Nobody would imagine that you could use that motor. Even specifications like its size, power and weight were adoptable according to the old plans. As this is now known, I say give everything a try.
And the approach - a small team of 5 workers, one project leader, and 1 million dollars - worked. the HUBO can't walk as well as the Asimo, but it is the first humanoid robot to have fully articulated fingers that move independently.
Hmmmm, this sounds suspiciously like the size of the "teams" that are currently entering the DARPA Grand Challenge. In this light, it is not such a surprise that a tiny startup like Digital Auto Drive spending $50,000 almost managed to beat the $5 million robot car entry from CMU in 2004.
We are at that exact moment in history, comparable to the Homebrew Computer Club at the end of the 1970s, when a small robotic group can blow away established "leaders" in the field. It also means that small startup robot companies creating (gasp) consumer products have a real shot of becoming the next "Microsoft" or "Apple" (Microsoft's rising interest in robots nonwithstanding).
Note that this doesn't mean that CMU or Honda are "bad" because they have created big-budget robots - but it does mean that the groaning pile of theses from past graduate students is as much a hinderance as a help to new-era robotics. It is possible that a small robotics group, unencumbered by the huge pile of PhD theses in the engineering department, can follow novel paths that the big ones decide won't
Ho makes a final point in the HUBO which is very interesting - he says that HUBO is motivitated by artistic, rather than purely engineering considering. This guy is a Robots That Jump hero all the way!
For the past couple of entries I have been looking at the current revival of slobbering belief in the articles of faith in the Internet 1990s - in particular, the belief that cyberspace is about to become all-important and surpass the real world (which is just another cyberspace, according to this belief system). In the next couple of posting I will recount the attempts of roboticists to create viable world models, and how their complete failure motivated a different approach largely responsible for 'new era' robotics. This failure of world models does not bode well for a world in which cyberspace and the real world begin to blend, as imagined by many last-revolution pundits. The drive to suck video from every streetcorner of the world into the web and put RFID tags on every device, even "give every lightbulb and IP number" is about to run smack into the law of unintended consequences as well as the brittleness problem.
First off however, a salute to several new robots and robot stories. First up is the announcement made by Toyota last week concerning use of robots in their factories. According to one of many sources in Asia (Business Day Journal):
"...TOKYO – Toyota Motor will introduce robots which can work as well or better than humans at all 12 of its factories in Japan to cut costs and deal with a looming labour shortage as the country ages, a report said last week.
The robots would be able to carry out multiple tasks simultaneously with their two arms, achieving efficiency unseen in human workers and matching the cheap wages of Chinese laborers, the Nihon Keizai Shimbun said..."
So Japan, under pressure from Chinese-based manufacturing, is electing to keep the "jobs" in the country - by replacing human workers with robots. These are not mobile or autonomous robots, but advanced industrial robots which finally merit the name, since they make extensive use of computer vision and other sensors to accomplish their tasks. The reason for the move is, as is often cited in the Japanese press, a declining population and resistance to filling low-paid jobs with "guest workers" from other countries.
Second up is an interesting article at automatedbuildings.com concerning security robots by Lloyd Spencer of Coroware, a unique consultancy/system integrator which (among other things) helps business integrate robots into the workplace. Besides describing the benefits of robots versus a "wired" building, he notes that three factors - off-the shelf hardware, embedded operating systems and emerging telecommunications standards like JAUS are helping make security robots more practical. Similar analysis has been posted over at Activmedia, who survived the "robot winter" of the 1990s and is currently developing their robots for security. I agree - while I love the current "hijack" mentality of robot developers who steal from the computer/Internet industry to make their novel creations, a robotics components industry will be needed to drive robots into wider markets.
This said, the grassroots robotics world is still fun - and it is changing. For about a decade the typical robot, even at university level, was a trash can on wheels with a bump sensor. Today, a fascinating article from Omynews in Korea interviews Oh Jun Ho, the creator of the recently announced Hubo robot. While the HUBO superficially looks like Honda's Asimo, it was developed in only 2 years (as opposed to the 15 for Asimo) and cost only 1 million US (as opposed to > $300 million for Asimo).
Another article out indirectly concerns robotic hardware. Conventional PC designs have reached "heat limits" in that we can't have 100 GHz or even 10 GHz computers without pumping so much power through the CPU that it vaporizes. Companies like Intel have taken to touting "multicore" processors - in effect, two or more CPUs on the same chip. But an article by Shahin Khan, VP/CMO of Azul Systems tells us to prepare for the age of "kilocore" computers, parallellng the vast advances in RAM and disk storage. The company has been hard at work developing "compute" network appliances with hundreds of Vega processors, making (Java) computing power a resource like memory and storage.
Good quote:
"...In 2005, Azul Systems will ship compute pools with as many as 1,200 CPUs per a single standard rack (1.2 kilo cores! - I like the sound of that!)..."
As I've discussed before, multicore are useful for the server market, almost useless for most other PC/network tasks, and extremely useful for robotics. A 1 kilocore robot could handle tens of thousands of sensors, compared to the dozens found on robots today. This is because sensor processing, unlike many other computer tasks, can easily and natuarally be split among multiple processors. I've said before that rising sensor density is the real "Moore's Law" driving robotics - possibly it is linked very significantly to the rise of multicore. Time for a group with "modest" (1-5 million?) means to "hijack" this technology for an advanced robot!
Finally, in the "another day, another robot" category, Sega has developed a music-playing "robot dog" wich some of the same features as the Sony Aibo. Interesting, though I suspect the "music composing" capabilities of this not very doglike robot are realling DJing existing music tracks. Also, the sensory capability of the Sega system is extremely limited, touch only. Not sure if this is a real robot or another useless gadget. I'd be really surprised if Sega, a company whose lifeblood is the virtual world of games, understands robots at all.
Some might want to cite a "Moore's Law" for robotic development, but the truth is more interesting - humanoid robots are cheap to build if you deliberately try to do it on a shoestring. The problem is the culture around these projects. At Korea Advanced Institute of Science & Technology (KAIST) there was little money for creating the KHR robot series, so Ho used a mix of students and volunteers. Instead of hiring an army of engineers, he designed the system himself and did the research for components (e.g. motors) over the Internet. And every time a student told him that this or that graduate thesis said you could'n't us a particular (cheap) motor, Ho told the student to figure it out anyway.
Ho's comments are very relevant to the US robotics community. At robot centers like CMU and MIT the "thesis" mentailty spoken of in this article is very strong. A long string of completely impractical robots have been built for 30 years or more simply to make sure that grad students get their doctorate. While good for education, this approach has stifled the community. In contrast, KAIST threw out that 30 years of thesis work and jumped in and built a humanoid robot on the cheap.
A great quote:
What, technologically, was the most difficult stage?
Firstly, I couldn't approach the problem with the existing paradigm. It would have taken me 20 years had I done so. All the theses in the world are written according to that paradigm. There were no theses for me to refer to. More difficult was persuading my students. At first, they didn't believe a word I said. No matter what I said, they would bring me this thesis and that thesis that said what I wanted to do was impossible. I tore up whatever thesis they brought me. It took from 6 months up to a year to convince them.
For example, if I told them to use motor B instead of motor A, the students would tell me it was impossible, citing some thesis or simulation result. But the use of motor B was breaking existing blueprint concepts. Nobody would imagine that you could use that motor. Even specifications like its size, power and weight were adoptable according to the old plans. As this is now known, I say give everything a try.
And the approach - a small team of 5 workers, one project leader, and 1 million dollars - worked. the HUBO can't walk as well as the Asimo, but it is the first humanoid robot to have fully articulated fingers that move independently.
Hmmmm, this sounds suspiciously like the size of the "teams" that are currently entering the DARPA Grand Challenge. In this light, it is not such a surprise that a tiny startup like Digital Auto Drive spending $50,000 almost managed to beat the $5 million robot car entry from CMU in 2004.
We are at that exact moment in history, comparable to the Homebrew Computer Club at the end of the 1970s, when a small robotic group can blow away established "leaders" in the field. It also means that small startup robot companies creating (gasp) consumer products have a real shot of becoming the next "Microsoft" or "Apple" (Microsoft's rising interest in robots nonwithstanding).
Note that this doesn't mean that CMU or Honda are "bad" because they have created big-budget robots - but it does mean that the groaning pile of theses from past graduate students is as much a hinderance as a help to new-era robotics. It is possible that a small robotics group, unencumbered by the huge pile of PhD theses in the engineering department, can follow novel paths that the big ones decide won't
Ho makes a final point in the HUBO which is very interesting - he says that HUBO is motivitated by artistic, rather than purely engineering considering. This guy is a Robots That Jump hero all the way!
Monday, January 17, 2005
More on why networks, wireless, gadgets, etc are not the future - they're the past
After the great response to my last post on the posturing of techie websites/blogs with respect to the "last revolution" I thought I would add more based on examples from history.
At our current point in time, the Internet, wireless, flatscreens, bluetooth gadgets, etc. are like looking in the rear-view mirror - they are not "the future" but the past. The big revolutions which created the PC and Internet are decades old, and these technologies have entered their "finishing touches" stage. They are not new, and supposed "new" ideas one sees touted regularly on the sites listed in Daily Rotation, for example, are actually many years old. The obsession with, say Microsoft vs. Apple ignores the real picture - computer operating systems are not advancing, and are about as interesting as my wall socket. How "old" is the cyberspace revolution today? Read on - more than you think...
The worship of the PC/Internet paradigm common today is preventing people from seeing the next revolution in robotics, which is in its early, explosive startup stage. Unlike PCs and the Internet, we're seeing a wave of genuine new ideas in robotics, coupled with rapid technological advancement of robot bodies, and a crop of garage (or backyard if your robot is big like Neogentronyx) startups like the homebrew computers of the 1970s.
One interesting complaint I've heard about robots is that they are "retro". It is true that movies like I, Robot had an interesting 1950s feel to them. The upcoming animated feature Robots uses robot bodies that look straight out of the 1940s. I saw a guy carrying a model of one of the characters in the Robots movie over at Digital Domain in Santa Monica the other day - it really had that 1940s vibe! Analogies to home robots invariably mention The Jetsons from the early 1960s. It is as if robots lie in the past, not the future.
How dare they drag out this old tech! How could it be important if it doesn't have Wi-Fi or can't play DVDs? Hm, let's make fun of the funny old robot or make a Saturday morning cartoon prediction that they will "take over the world."
During the 1990s you often saw tech writers arguing that robots were, indeed, the past. Many said that the rise of "network" computer technologies like the Internet eliminated the need for robots. Robots were a past fantasy that missed the all-powerful features of networks and cyberspace. Instead of a security robot in a building, you would connect sensors throughout the building to a PC for a "smart house" - and access it via a virtual reality world on that PC. Inner (virtual) space, not outer (real) space was important.
Another group of writers said that while real-world robots were a fantasy, we should realize that we did have robots - software programs that would do work for us on the web. Examples of these "web-bots" were programs that monitored news for certain keywords and auto-emailed the articles to you. These software 'bots, rather than physical robots, were the "real" future.
Reading these things I can only wish that the techno-digeratti would spend a little more time looking at history. During the last 200 years, there have been several tech booms (frequently followed by economic depressions), all involving "network" technology. The 1990s Internet mania is just the latest of these. Here are several of these "network tech" eras:
1830s-1860s - Railroads
1850s-1880s - Telegraph/telephone
1920s-1940s - Power companies, radio, television
In each of these eras, a new technology dedicated to "networking" people and machines expanded quickly, causing a decade-long "revolution". In each case, their techno-gurus proclaimed that the network of the hour would bring technological utopia, unite humanity, end wars, blah, blah.
So you think "this time is different?" I invite you to read a great book called The Victorian Internet by Tom Standage (Berkeley Publishing, 1999, ISBN: 0425171698).
Great review quote:
"Imagine an almost instantaneous communication system that would allow people and governments all over the world to send and receive messages about politics, war, illness, and family events. The government has tried and failed to control it, and its revolutionary nature is trumpeted loudly by its backers. The Internet? Nope, the humble telegraph fit this bill way back in the 1800s. The parallels between the now-ubiquitous Internet and the telegraph are amazing, offering insight into the ways new technologies can change the very fabric of society within a single generation."
Standage's analysis gets more and more convincing, and when you get to the chapter on "telegraph cybersex" and online marriages you realize that the Internet is not so revolutionary as we imagine. It also is fascinating to realize that the telegraph is a "digital" technology that was replace by the "analog" technology of the telephone. Late-era telegraph systems incorporated mechanical relays and storage devices - the forerunner of Internet routers.
Of course, more people use the Internet today, but the difference is a matter of degree rather than fundamentals.
Network technologies like the telegraph and railroads are seductive to the imagination because of the "multiplier effect." The idea is that the value of a network increases exponentially as you connect more users to it. Therefore, growth and progress, aided by the current network of fashion will become exponential and the distant future will actually arrive tomorrow. We saw these fantasies in the "telecosm" visions of the 1990s, and today in the slobbering obsession with networked gadgets as the salvation of humanity.
However, networks have a physical basis, and their growth actually follows a sharp sigmoid curve. During the early phases they do grow very quickly. However, as the final connections are put into place they screech to a halt all the faster. And in the end we do not end wars or transform humanity to a higher state of "cyber" reality by installing a new network. Instead, the negative consequences multiply - in our case viruses, hacking, spam. It is interesting to see that people are actually giving up "broadband" Internet due to these problems. In a recent article entitled "No More Internet for Them" in the Los Angeles Times we have the following priceless quote:
"...Stephen Seemayer had the first Pong video game system on his block. A decade later, the Echo Park artist was the first in his neighborhood to get a personal computer. And in 1996, he was so inspired by the World Wide Web that he created a series of small paintings for viewing over the Internet. Now the 50-year-old Seemayer is once again on the cutting edge: Sick of spam clogging his in-box and spyware and viruses crashing his system, Seemayer yanked out his high-speed connection.'I'm not going to pay for something that I can't use,' he said."
More evidence that the PC/Internet revolution is essentially finished comes in the following quote from the article:
"In a recent survey, 31% of online shoppers said they were buying less than before because of security issues. And though more people are signing up for high-speed, commerce-friendly connections, the proportion of U.S. Internet users paying for things online barely budged in 2004 from a year earlier. It rose to 27% from 26% in 2003 after jumping from 20% the previous year, according to Harris Interactive."
And yet another....
"The aggravation level has reached the point that some in the computer industry believe it threatens to undermine advances of the last decade, during which the Internet has grown from a virtually empty domain to a global community of 800 million souls. They say they need to act before the same early adopters who led mainstream Americans online lead them off."
Read The Victorian Internet and see just how closely techno-pundits today sound like their counterparts 150 years ago - it is creepy. Many of the things we imagine are the future, e.g., 24/7 connectivity - hinder rather than promote daily life. The value of networking has its limits - and we may have reached them. And we've been here before!
Even a weblog like this one is not that revolutionary. Recall that in the 19th century postal mail delivery in England was fast and efficient compared to today. People would write a letter in London in the morning and be confident that it would reach the other side of the city by afternoon - letter-writing was more like email than most suspect. Networks of informal correspondence approximated our current weblogs. Email improves this kind of communication only modestly over postal mail.
In contrast, real-world technologies are involved in creating the destinations we link networks to. Their growth is typically slower, but ultimately they do cause greater real change in daily life. They involve creating new ways of making things we can use and talk about on the networks. Examples include the development of factories run by electrical power, power tools, kitchen tools, washing machines, automobiles, airplanes, rockets, manufacturing technology - all dedicated to the real. Robots are the next step in this "turn to the real" technology. They differ significantly from earlier "real" technologies in their autonomy and sensors, promising that their long-term impact will be particularly large.
Taking both "network" and "real" technologies to an extreme cause problems. Worship of heavy iron can lead to real-world problems, witness environmental and social consequences of industrialization in the 19th and 20th centuries.
But worship of networks tends to suck us into the imaginary world of "cyberspace" and treat it as if it existed - and then apply its uncoupled-from-reality, "multiplier" growth dynamics to the world with negative results. For example, many people today believe that energy use can rise endlessly - where the truth is that we are fast approaching a worldwide "energy crisis" which will require us to find a new energy source if industrial civilization is to continue in the lifestyle to which we have become accustomed to. Deriving predictions for the real world from Internet growth models is a fatal pastime of today's tech-gurus, yet the world economy has adopted financial instruments that "grow" money independently of the real economy via network multipliers (e.g., "financial derivatives") - leading to unsustainable long-term debt at all levels of society.
Fortunately, in cultural terms, the cyberspace model has just about run course. The desperate techno-guru protests of an ever-advancing wireless future is a sign of "whistling in a graveyard." The repeated predictions that we'll watch video on our cellphones are beginning to seem quaint. As a culture, we are about to lose interest - Windows as a wall socket instead of a revolution.
Civilizations and movements build their biggest monuments at the onset of their decline - witness 1st-century Alexandria and its giant lighthouse. Cyberspace is about to go "out". Over the next decade, the public, as it has before, will turn away from the latest "network" technology to the "real" for 30 years or so. This time the "real" is robots.
So here is the answer to why robots in popular media today look "retro" - we're beginning a new era comparable to earlier eras in history. We have to derive our artistic visions (not contaminated by "Terminator" or "Matrix" fantasy) from that earlier time. We have to reach back past our current "network" era to the last "real world" era - the 1930s-1950s. In that era "things" were more important than "connectivity." Robots were a major vision of the future. And so it will be again.
After the great response to my last post on the posturing of techie websites/blogs with respect to the "last revolution" I thought I would add more based on examples from history.
At our current point in time, the Internet, wireless, flatscreens, bluetooth gadgets, etc. are like looking in the rear-view mirror - they are not "the future" but the past. The big revolutions which created the PC and Internet are decades old, and these technologies have entered their "finishing touches" stage. They are not new, and supposed "new" ideas one sees touted regularly on the sites listed in Daily Rotation, for example, are actually many years old. The obsession with, say Microsoft vs. Apple ignores the real picture - computer operating systems are not advancing, and are about as interesting as my wall socket. How "old" is the cyberspace revolution today? Read on - more than you think...
The worship of the PC/Internet paradigm common today is preventing people from seeing the next revolution in robotics, which is in its early, explosive startup stage. Unlike PCs and the Internet, we're seeing a wave of genuine new ideas in robotics, coupled with rapid technological advancement of robot bodies, and a crop of garage (or backyard if your robot is big like Neogentronyx) startups like the homebrew computers of the 1970s.
One interesting complaint I've heard about robots is that they are "retro". It is true that movies like I, Robot had an interesting 1950s feel to them. The upcoming animated feature Robots uses robot bodies that look straight out of the 1940s. I saw a guy carrying a model of one of the characters in the Robots movie over at Digital Domain in Santa Monica the other day - it really had that 1940s vibe! Analogies to home robots invariably mention The Jetsons from the early 1960s. It is as if robots lie in the past, not the future.
How dare they drag out this old tech! How could it be important if it doesn't have Wi-Fi or can't play DVDs? Hm, let's make fun of the funny old robot or make a Saturday morning cartoon prediction that they will "take over the world."
During the 1990s you often saw tech writers arguing that robots were, indeed, the past. Many said that the rise of "network" computer technologies like the Internet eliminated the need for robots. Robots were a past fantasy that missed the all-powerful features of networks and cyberspace. Instead of a security robot in a building, you would connect sensors throughout the building to a PC for a "smart house" - and access it via a virtual reality world on that PC. Inner (virtual) space, not outer (real) space was important.
Another group of writers said that while real-world robots were a fantasy, we should realize that we did have robots - software programs that would do work for us on the web. Examples of these "web-bots" were programs that monitored news for certain keywords and auto-emailed the articles to you. These software 'bots, rather than physical robots, were the "real" future.
Reading these things I can only wish that the techno-digeratti would spend a little more time looking at history. During the last 200 years, there have been several tech booms (frequently followed by economic depressions), all involving "network" technology. The 1990s Internet mania is just the latest of these. Here are several of these "network tech" eras:
1830s-1860s - Railroads
1850s-1880s - Telegraph/telephone
1920s-1940s - Power companies, radio, television
In each of these eras, a new technology dedicated to "networking" people and machines expanded quickly, causing a decade-long "revolution". In each case, their techno-gurus proclaimed that the network of the hour would bring technological utopia, unite humanity, end wars, blah, blah.
So you think "this time is different?" I invite you to read a great book called The Victorian Internet by Tom Standage (Berkeley Publishing, 1999, ISBN: 0425171698).
Great review quote:
"Imagine an almost instantaneous communication system that would allow people and governments all over the world to send and receive messages about politics, war, illness, and family events. The government has tried and failed to control it, and its revolutionary nature is trumpeted loudly by its backers. The Internet? Nope, the humble telegraph fit this bill way back in the 1800s. The parallels between the now-ubiquitous Internet and the telegraph are amazing, offering insight into the ways new technologies can change the very fabric of society within a single generation."
Standage's analysis gets more and more convincing, and when you get to the chapter on "telegraph cybersex" and online marriages you realize that the Internet is not so revolutionary as we imagine. It also is fascinating to realize that the telegraph is a "digital" technology that was replace by the "analog" technology of the telephone. Late-era telegraph systems incorporated mechanical relays and storage devices - the forerunner of Internet routers.
Of course, more people use the Internet today, but the difference is a matter of degree rather than fundamentals.
Network technologies like the telegraph and railroads are seductive to the imagination because of the "multiplier effect." The idea is that the value of a network increases exponentially as you connect more users to it. Therefore, growth and progress, aided by the current network of fashion will become exponential and the distant future will actually arrive tomorrow. We saw these fantasies in the "telecosm" visions of the 1990s, and today in the slobbering obsession with networked gadgets as the salvation of humanity.
However, networks have a physical basis, and their growth actually follows a sharp sigmoid curve. During the early phases they do grow very quickly. However, as the final connections are put into place they screech to a halt all the faster. And in the end we do not end wars or transform humanity to a higher state of "cyber" reality by installing a new network. Instead, the negative consequences multiply - in our case viruses, hacking, spam. It is interesting to see that people are actually giving up "broadband" Internet due to these problems. In a recent article entitled "No More Internet for Them" in the Los Angeles Times we have the following priceless quote:
"...Stephen Seemayer had the first Pong video game system on his block. A decade later, the Echo Park artist was the first in his neighborhood to get a personal computer. And in 1996, he was so inspired by the World Wide Web that he created a series of small paintings for viewing over the Internet. Now the 50-year-old Seemayer is once again on the cutting edge: Sick of spam clogging his in-box and spyware and viruses crashing his system, Seemayer yanked out his high-speed connection.'I'm not going to pay for something that I can't use,' he said."
More evidence that the PC/Internet revolution is essentially finished comes in the following quote from the article:
"In a recent survey, 31% of online shoppers said they were buying less than before because of security issues. And though more people are signing up for high-speed, commerce-friendly connections, the proportion of U.S. Internet users paying for things online barely budged in 2004 from a year earlier. It rose to 27% from 26% in 2003 after jumping from 20% the previous year, according to Harris Interactive."
And yet another....
"The aggravation level has reached the point that some in the computer industry believe it threatens to undermine advances of the last decade, during which the Internet has grown from a virtually empty domain to a global community of 800 million souls. They say they need to act before the same early adopters who led mainstream Americans online lead them off."
Read The Victorian Internet and see just how closely techno-pundits today sound like their counterparts 150 years ago - it is creepy. Many of the things we imagine are the future, e.g., 24/7 connectivity - hinder rather than promote daily life. The value of networking has its limits - and we may have reached them. And we've been here before!
Even a weblog like this one is not that revolutionary. Recall that in the 19th century postal mail delivery in England was fast and efficient compared to today. People would write a letter in London in the morning and be confident that it would reach the other side of the city by afternoon - letter-writing was more like email than most suspect. Networks of informal correspondence approximated our current weblogs. Email improves this kind of communication only modestly over postal mail.
In contrast, real-world technologies are involved in creating the destinations we link networks to. Their growth is typically slower, but ultimately they do cause greater real change in daily life. They involve creating new ways of making things we can use and talk about on the networks. Examples include the development of factories run by electrical power, power tools, kitchen tools, washing machines, automobiles, airplanes, rockets, manufacturing technology - all dedicated to the real. Robots are the next step in this "turn to the real" technology. They differ significantly from earlier "real" technologies in their autonomy and sensors, promising that their long-term impact will be particularly large.
Taking both "network" and "real" technologies to an extreme cause problems. Worship of heavy iron can lead to real-world problems, witness environmental and social consequences of industrialization in the 19th and 20th centuries.
But worship of networks tends to suck us into the imaginary world of "cyberspace" and treat it as if it existed - and then apply its uncoupled-from-reality, "multiplier" growth dynamics to the world with negative results. For example, many people today believe that energy use can rise endlessly - where the truth is that we are fast approaching a worldwide "energy crisis" which will require us to find a new energy source if industrial civilization is to continue in the lifestyle to which we have become accustomed to. Deriving predictions for the real world from Internet growth models is a fatal pastime of today's tech-gurus, yet the world economy has adopted financial instruments that "grow" money independently of the real economy via network multipliers (e.g., "financial derivatives") - leading to unsustainable long-term debt at all levels of society.
Fortunately, in cultural terms, the cyberspace model has just about run course. The desperate techno-guru protests of an ever-advancing wireless future is a sign of "whistling in a graveyard." The repeated predictions that we'll watch video on our cellphones are beginning to seem quaint. As a culture, we are about to lose interest - Windows as a wall socket instead of a revolution.
Civilizations and movements build their biggest monuments at the onset of their decline - witness 1st-century Alexandria and its giant lighthouse. Cyberspace is about to go "out". Over the next decade, the public, as it has before, will turn away from the latest "network" technology to the "real" for 30 years or so. This time the "real" is robots.
So here is the answer to why robots in popular media today look "retro" - we're beginning a new era comparable to earlier eras in history. We have to derive our artistic visions (not contaminated by "Terminator" or "Matrix" fantasy) from that earlier time. We have to reach back past our current "network" era to the last "real world" era - the 1930s-1950s. In that era "things" were more important than "connectivity." Robots were a major vision of the future. And so it will be again.
