Nobody in the US refers to Wi-Fi as a "GSM Killer" because GSM is already dead here (CDMA has 6x the market share).
What you're referring to is the comparison of 3G technologies (such as WCDMA or CDMA2000) to Wi-Fi, and here again you're still wrong. It's becoming very clear that the rapid growth of 802.11x hotspots will spoil the market for 3G, since they're dramatically less expensive to deploy, and they cover the spots where people are most likey to use a rich content mobile device. And spare us the "car centric" baloney as well; commuter rail has been on a huge tear in the US during the past 15 years, and yes, you can cover rail lines with Wi-Fi by using directive antennas -- and you'll most likely be more spectally efficient than 3G.
The great thing about Wi-Fi's low power approach is that it forces people to use spectrum conservatively, using directive arrays to put RF where's needed instead of spraying it around indiscriminately.
Too bad Microsoft decided to use BlueBalls^h^h^h^hTooth instead of Wi-Fi, probably saved themselves a few bucks with the inferior frequency-toad-hopping instead of real spread spectrum, which is what the market is asking for...
So the suits at COMDEX only want to put on a show for other suits. Thus concludes the transition to crying boredom that began a few years ago when they booted out the adult entertainment industry -- which is always the first adopter for any new communications technology.
Isn't it about time that someone started a new trade show, one for us? It would be in Vegas (or maybe Reno, just for spite). Tons of raw technology. Booth babes. And a tradeshow trinket requirement. Heck, have a "best of show trinket" award, they get their booth space fee back! Now that would be worth the trip.
Old News, It's Called a Vibroplex, Invented 1890s
on
One-Thumb Keyboard
·
· Score: 4, Interesting
One-thumb keying devices have been around for more than a century; the Vibroplex company still sells them to ham radio operators.
Note that the minimum speed on these puppies is 20 words per minute, and trained ops have hit speeds of over 70 wpm. Sometimes with a ham sandwich in the other fist...
When I was a kid, I always got dinged for lousy handwriting. Why? I was always trying to write too fast -- hand couldn't keep up doing all those cursive moves and curlycues. So, not only was cursive too slow, but heck, I couldn't read my own writing.
Finally, I started typing as a teenager, and the instructor couldn't slow me down. I do 40+ wpm with bursts way above that. Heck, when I learned morse code it was faster than cursive! And when I learned block lettering in drafting -- fast, economical and legibile -- I stopped doing cursive except for my signature, which is still too %^$@ slow!
We should be happy that, in the near future, no child will be hamstrung and tortured by this horse-and-buggy leftover from the days of one-room schools...
How about some history? Not that bland, boring crapola that gets poured on high school students, but real history with controversy, human personality, high stakes, and... technology?
Empire Express is the history of the first US transcontinental railroad (from Omaha to Sacramento), but it's so much more than a timeline. Bain introduces us to the key figures, then animates their personalities so that we get to know them. Well researched and factual, yet it reads like Clancy.
Railroading was the high tech of its day, and the technical, financial and political interplay will resonate with anyone who was involved in the Dotcom boom. It's all there: a tech genius with a vision ("crazy" Theodore Judah) begs politicians and financiers to build the Great Thing, then is cut out and dies trying to get back into the game. Financiers float inflated bonds and other worthless bits of paper, siphoning off tidy profits for themselves so immense that they kicked off the "Gilded Age". The despiration boring tunnels through Sierra granite that leads to the use of nitroglycerine (and an accident that vaporizes a ship in the San Francisco harbor). And all made possible by hordes of low-paid domestic and immigrant labor. Sound familiar?
One warning: it will take you the entire summer to read through all 700+ pages. But well worth it.
Al Haig's company (yes, he really is In Charge) Sky Station has been talking about this for years now. Might sound a bit hokey, but it's just an extension of the Aerostat technology that's been used for the past decade to provide better radar coverage for the Gulf of Mexico. And, it's one heck of a lot cheaper to put up another balloon to replace a failure than to launch another satellite...
David was the other half of the Metcalfe-Boggs "team" that made Ethernet a reality. Dave is an honest-to-nuts Friend of the Electron electrical engineer, who had to crack the problem of making CSMA/CD work in the Real World; in particular, how do your hear a transmitter 500 meters away when you're sitting right next to your own?
Boggs invented the first (of many) hardware circuit techniques to do collision detection, and other elements of transceiver design. If Dave hadn't picked up a soldering iron, we'd probably be doing DATAKIT or some other telco hack.
While any industrialized society sans nuclear power will produce CO2, don't forget that the same society can compensate by planting trees and other growing things.
Of the industrialized nations, the US is alone in having reforested after centuries of clearcutting. Better yet, we've reforested with rapid growing species of trees, which consume even greater amounts of CO2.
In the initial Kyoto rounds, the US was supposed to have received one-for-one credit for reforestation. This provision was taken out at the insistence of the Europeans, effectively punishing the US for its environmentally correct policy. The US promptly -- and correctly -- bailed out.
Actually I was primarily referring to all of the cruft that usually sits outside the kernel. Microsoft loves to tout the interdependence of all of its OS components, but this often means that in order to load a feature I need, I have to load 15 modules that I don't want around.
As for the kernel, I guess that depends on what you mean by "microkernel". The basic BSD kernel is less that a thousand lines of code, which seems pretty micro to me. And it seems to me that the vast majority of BSODs I've encountered have been in the Windows Kernel.
Solaris isn't really Unix (and not exactly "Open" either). I guess I come from a generation of low level hardware-software hacks who see Unix as Open by definition, and anything not Open is not Unix.
-=N9FZX
Reliable Hardware Platforms Deserve Solid Software
on
Military Grade Laptops
·
· Score: 5, Insightful
Reliable hardware is of course a Good Thing, but without some improvements in software reliability, you're buying a HumVee when a Yugo would do. The old Bell Labs standard called for 1/3 of system faults to be ascribed to hardware, 1/3 to software, and 1/3 to operator error. Most available operating systems clearly aren't there yet, and can't even match the reliability of existing off-the-shelf motherboards and power supplies.
Having said that, it's also clear that some operating systems, owing to their monlithic architecture, will never improve. At least with Unix, you can discard what you don't need, reducing the volume of code that has to be checked. That's a major reason why Microsoft's "Trustworthy Computing" initiative is such a joke -- you can't get rid of the crapola, heck you can't even see the crapola!
None of the claims that I've seen coming out of the major players (Intel, TDI, etc) has violated Shannon's Law. The problem here lies in the way that most people interpret the Law, and their stake in the existing wavelenth coordiation scheme.
In the wavelength scheme, one uses the average, or peak information rate to determine spectral occupancy, as one cannot completely predetermine the stochastic nature of the information transfer. As a result, you occupy spectrum even if you're not using it, because you might use it. In contrast, when a UWB transmitter is not in use, it doesn't emit RF, thereby decreasing the noise floor, and thus increasing the available information rate for other stations.
Aside from being a better match to the stochastics of the information, there's a true RF advantage to UWB -- the elimination of Rayleigh Fading due to multipath. Any narrowband link has to take into account destructive interference resulting in multiple RF wave paths; the resulting increase in required power reduces the spectral efficiency. However, in a time-based system, where the pulse length is shorter than the path difference, the receiver is able to easily reject reflections which arrive outside the time window.
Sorry, but in thise case, the trolls are on the other side of the bridge...
A receiver can separate two signals based on time, wavelength, polarity, or spatial diversity. Reed seems to have missed the last two, but then he's not really a radio guy. For more info on signal separation and spectral efficiency, have a look at the paper that I wrote 16 years ago...
Basically, the history of radio is the history of our practical ability to coordinate multiple stations. In the beginning, radio signals were generated by spark gaps; the resulting impluse occupied the entire longwave spectrum, propagating by groundwave. Separation was accomplished by time, and stations scheduled their transmissions by the clock. This held sway until the invention of the triode vacuum tube by DeForest, which enabled coherent, narrowband transmission of information, and thus coordiation by wavelength. The government then got involved as a third party coordination body.
As more stations went on the air, technological development was aimed at expanding the useable spectrum beyond longwave -- first medium wave (300 kHz to 3MHz) then shortwave (3-30MHz) then VHF (30-300MHz).
WWII advanced the pace of development in UHF (300-900 MHz) and microwaves (above 900 MHz). With those developments came the ability to use polar and spatial diversity. But the latter really took off with the development of microprocessor controlled radios, which enabled spatial diversity by cell -- cellular radio.
However, even with all of the spectrum that these techniques have enabled, the fact remains that, owing to propagation differences, some parts of the spectrum are inherently more valuable than others, a scarcity that leads to economic realities that agencies like the ITU and FCC have been exploiting for decades.
Quietly, however, which these developments were taking place in wavelength coordination, our ability to coordinate transmissions in time has caught up -- first with spread spectrum (not that funny frequency hopping junk) and now individual pulse trains for Ultra Wide Band. UWB in particular holds the promise of ending the economics of scarcity found in wireless. Aside from a thousandfold increase in spectral efficiency, it also maps well to the bursty nature of information -- you don't need a channel all the time, but thanks to coordination by wavelength, you sit on it anyway.
Needless to say, when you challenge the economics of the status quo, you're not going to be too popular in certain political circles.
The history of the Apple Newton Messagepad is worthy of study by technophiles who think that Good Technology sells itself. Basically, Apple made all of the classical marketing/strategy mistakes that one could make on a single platform, including:
Unleashing a new product category before the product itself was ready.
Ignoring customer feedback on design issues and pricing.
Vague and inexplicit value proposition.
Getting nailed by a competitor who listened to customers, built a more portable platform, hit the market target price, and understood the core value proposition of a PDA.
I was a big fan of the Newton from the Messagepad 120 on -- with the 120 and Newton 2.0, Apple had finally delivered on the promise of the PDA. Unfortunately, after years of fumbling, overpromising and under-delivering, the market had moved on, and simply would not believe anything that Apple had to say -- something I'll dub the Newton Effect. The Messagepad 2000, which was ultimately more capable (thanks to DEC's StrongARM), was also ultimately a step in the opposite direction of what the market was asking for: smaller and cheaper, not bigger and over $1000!
I stil have my Messagepad 120, and it still comes out of the case every now and then to remind me that bad business decisions can and will kill superior technology.
Sometimes, visibility is a bad thing...
Since some folks just love to harp on the BSD license, this legal plink is an indelicate reminder that anything can be considered a legal loophole. Try to craft the ideal software license, and you'll spend the rest of your life swimming with lawyers instead of productively writing code.
Before posting yet another Linux-is-ghod rant, why not consider this: We are all so lucky to have more than one freeware Unix to choose from. That choice provides the needed competition to force both variants to improve in meaningful ways.
Without that competition, Unix would eventually stagnate. Or worse, innovation would be driven into the same kind of useless creeping featurism we've come to expect from the folks in Redmond.
1. Science has always been political. Most scientific questions have at least some controversy and the answers are neither clear or clean cut. Scientific "truths" are few and far between.
2. The funding of science is even more political. Especially if your funding comes from a political body.
Unfortunately, liberal atheists often see science as their religion, so when conservatives have any input to science, they tend to cry apostate. Never mind that the Clinton Administration, for example, heavily politicized NIST grants for liberal pet causes...
Same article was in the WSJ *Six Years Ago*
on
Googling For Dates?
·
· Score: 2, Funny
This is really old news; The Wall Street Journal ran a similar article in 1996, only then 1) it was AltaVista, and 2) it was actually news.
Time to haul out my trusty Collins Space Systems Slide Rule Calculator (circa 1974):
A perfectly fed 2ft. diamater dish at 2.4 GHz has a gain of 24db, but even a Pringles can feed will give you a dish with 21db of gain. More importanly, the spatial extent of the signal will be minimized, allowing for cellular reuse.
Personally, at 2.4 GHz I'd go with a loop Yagi-Uda array instead, as they're still cheap, and much more wind resistant, something you care about when pointing matters. Make enough of them and they'll be as cheap or cheaper than the Yagis used for UHF TV reception.
Bottom line: There's nothing inherently expensive about gain antennas, and they're the cheapest way to improve the link equation.
More proof that a cheap $2 aluminum dish will beat a $200 silicon power amplifier any day -- and it uses the spectrum more efficiently!
Seriously, however, broadcast medium networks like 802.11b are best used for distribution, not long distance point-to-point links (fiber is ultimately cheaper on a bit-for-bit basis), but this demonstrates that you can build a really cheap 802.11b distribtuion network to solve the Last Mile Problem. Another nail in the coffin of Ma Bell...
Real AltaVista History
on
Altavista Renewed
·
· Score: 3, Interesting
It always amazes me how history gets munged in the retelling. For example, the lunch meeting I had with Louis Monier and Joella Paquette (Left at Albuquerque) was actually the second or third meeting I'd had with Joella about "Alto Vista", and it was the first time I'd met Louis. There was no napkin involved. Or that the name came from a half-erased chalkboard: actually, it came from the award plaqques in Joella's office; when she asked me to code the name the project, I looked at the placques, chose the word "Alto" from "Palo Alto" and first "View" from "Mountain View", which we immediately changed to the Spanish "Vista". Louis' wife corrected the feminisation error a few weeks later, and the project settled into "Alta Vista". The whole idea was to build a search engine to demonstrate that DEC could do things with Alpha and the Internet that nobody else could.
Louis did the crawler code (known now as "scooter") and was the prefect person to do the job right, as he's a graph theorist by nature and had just finished working on a massive threads debugging tool. Chuck Thacker then suggested that we talk to Mike Burrows over at SRC, who had a wonderful full text database, which Louis and I concluded would work far better than my original idea (using Oracle). So Mike did the database code. I did the first (crude) web-based UI for Mike's code, and even with Louis' first crawl, it was amazing what we could do (relative to the other seach tools of the time). My other chore as "hardware guy" was to spec out the first AlphaServer 8400 that we would get to run the demo. There was a huge backlog of 8400 orders at the time, and only about a half dozen of DEC's techs were trained and authorized to work on them.
AltaVista's initial triumph was simple -- the database held ten times more pages than anything else, and also indexed all of the words in the pages. And yet the response time was nearly instantaneous. Keeping it that way for the first few weeks required a DEC VP to drive several CPU cards through a Boston blizzard to be Fedexed out to Palo Alto, as well as a lot of long hours by the team to diagnose and defend against a number of attacks.
Two things ultimately kept AltaVista from leveraging its early successes. First, DEC wouldn't part with the necessary capital -- as it turned out later, they were negotiating to be bought by Compaq. And secondly, when DEC was finally bought by Compaq, the latter had no idea what to do with AltaVista. The "portal" strategy was designed to maximize the IPO valuation, exactly what investors wanted in 1999. Large amounts of cash were spent on that strategy, only to have the DotCom Bomb go off a week before the IPO.
It's remarkable and I'm gratified to see that AltaVista managed to survive and transition to its roots.
Not sure what you're doing wrong, but the database itself supports Unicode as well as all of the Asian encodings (JIS, etc.). Make sure that you're using the appropriate front end for the encoding -- note the drop-down menu in the upper right hand corner.
-=paulf
Voting Software vs. Slot Machine Software
on
Indecision 2002
·
· Score: 1
The failure of voting software is particularly galling when you consider the fact that the Nevada Gaming Commission has had software testing standards for things like computerized slot machines for over ten years now. There is no excuse for a voting machine to be less reliable than your typical Nevada one-armed bandit!
In most developing countries, the government, not private industry, runs the state monopoly telecom carrier, and is the sole economic beneficiary of its operation. The cream of telephony for these governments has always been international long distance, something that they've always jealously protected from any competitor, even ham radio. [Want to know why morse code lasted so long in amateur radio? Hint: It's not easy to learn, it presents a barrier to entry, and every developing country has a vote at the ITU.]
Even if they're using a contractor like C&W, the government reaps most of the profit from those $5 per minute calls to overseas families. Ah, the evils of Socialism!
Slashdot Mirror
What you're referring to is the comparison of 3G technologies (such as WCDMA or CDMA2000) to Wi-Fi, and here again you're still wrong. It's becoming very clear that the rapid growth of 802.11x hotspots will spoil the market for 3G, since they're dramatically less expensive to deploy, and they cover the spots where people are most likey to use a rich content mobile device. And spare us the "car centric" baloney as well; commuter rail has been on a huge tear in the US during the past 15 years, and yes, you can cover rail lines with Wi-Fi by using directive antennas -- and you'll most likely be more spectally efficient than 3G.
The great thing about Wi-Fi's low power approach is that it forces people to use spectrum conservatively, using directive arrays to put RF where's needed instead of spraying it around indiscriminately.
Too bad Microsoft decided to use BlueBalls^h^h^h^hTooth instead of Wi-Fi, probably saved themselves a few bucks with the inferior frequency-toad-hopping instead of real spread spectrum, which is what the market is asking for. ..
Isn't it about time that someone started a new trade show, one for us? It would be in Vegas (or maybe Reno, just for spite). Tons of raw technology. Booth babes. And a tradeshow trinket requirement. Heck, have a "best of show trinket" award, they get their booth space fee back! Now that would be worth the trip.
Note that the minimum speed on these puppies is 20 words per minute, and trained ops have hit speeds of over 70 wpm. Sometimes with a ham sandwich in the other fist...
Finally, I started typing as a teenager, and the instructor couldn't slow me down. I do 40+ wpm with bursts way above that. Heck, when I learned morse code it was faster than cursive! And when I learned block lettering in drafting -- fast, economical and legibile -- I stopped doing cursive except for my signature, which is still too %^$@ slow!
We should be happy that, in the near future, no child will be hamstrung and tortured by this horse-and-buggy leftover from the days of one-room schools...
Empire Express is the history of the first US transcontinental railroad (from Omaha to Sacramento), but it's so much more than a timeline. Bain introduces us to the key figures, then animates their personalities so that we get to know them. Well researched and factual, yet it reads like Clancy.
Railroading was the high tech of its day, and the technical, financial and political interplay will resonate with anyone who was involved in the Dotcom boom. It's all there: a tech genius with a vision ("crazy" Theodore Judah) begs politicians and financiers to build the Great Thing, then is cut out and dies trying to get back into the game. Financiers float inflated bonds and other worthless bits of paper, siphoning off tidy profits for themselves so immense that they kicked off the "Gilded Age". The despiration boring tunnels through Sierra granite that leads to the use of nitroglycerine (and an accident that vaporizes a ship in the San Francisco harbor). And all made possible by hordes of low-paid domestic and immigrant labor. Sound familiar?
One warning: it will take you the entire summer to read through all 700+ pages. But well worth it.
Al Haig's company (yes, he really is In Charge) Sky Station has been talking about this for years now. Might sound a bit hokey, but it's just an extension of the Aerostat technology that's been used for the past decade to provide better radar coverage for the Gulf of Mexico. And, it's one heck of a lot cheaper to put up another balloon to replace a failure than to launch another satellite...
Boggs invented the first (of many) hardware circuit techniques to do collision detection, and other elements of transceiver design. If Dave hadn't picked up a soldering iron, we'd probably be doing DATAKIT or some other telco hack.
Anybody else remember the little ditty courtesy of Mad Magazine and that evil genious William Gaines????
Of the industrialized nations, the US is alone in having reforested after centuries of clearcutting. Better yet, we've reforested with rapid growing species of trees, which consume even greater amounts of CO2.
In the initial Kyoto rounds, the US was supposed to have received one-for-one credit for reforestation. This provision was taken out at the insistence of the Europeans, effectively punishing the US for its environmentally correct policy. The US promptly -- and correctly -- bailed out.
As for the kernel, I guess that depends on what you mean by "microkernel". The basic BSD kernel is less that a thousand lines of code, which seems pretty micro to me. And it seems to me that the vast majority of BSODs I've encountered have been in the Windows Kernel.
Solaris isn't really Unix (and not exactly "Open" either). I guess I come from a generation of low level hardware-software hacks who see Unix as Open by definition, and anything not Open is not Unix. -=N9FZX
Having said that, it's also clear that some operating systems, owing to their monlithic architecture, will never improve. At least with Unix, you can discard what you don't need, reducing the volume of code that has to be checked. That's a major reason why Microsoft's "Trustworthy Computing" initiative is such a joke -- you can't get rid of the crapola, heck you can't even see the crapola!
None of the claims that I've seen coming out of the major players (Intel, TDI, etc) has violated Shannon's Law. The problem here lies in the way that most people interpret the Law, and their stake in the existing wavelenth coordiation scheme.
In the wavelength scheme, one uses the average, or peak information rate to determine spectral occupancy, as one cannot completely predetermine the stochastic nature of the information transfer. As a result, you occupy spectrum even if you're not using it, because you might use it. In contrast, when a UWB transmitter is not in use, it doesn't emit RF, thereby decreasing the noise floor, and thus increasing the available information rate for other stations.
Aside from being a better match to the stochastics of the information, there's a true RF advantage to UWB -- the elimination of Rayleigh Fading due to multipath. Any narrowband link has to take into account destructive interference resulting in multiple RF wave paths; the resulting increase in required power reduces the spectral efficiency. However, in a time-based system, where the pulse length is shorter than the path difference, the receiver is able to easily reject reflections which arrive outside the time window.
Sorry, but in thise case, the trolls are on the other side of the bridge...
Basically, the history of radio is the history of our practical ability to coordinate multiple stations. In the beginning, radio signals were generated by spark gaps; the resulting impluse occupied the entire longwave spectrum, propagating by groundwave. Separation was accomplished by time, and stations scheduled their transmissions by the clock. This held sway until the invention of the triode vacuum tube by DeForest, which enabled coherent, narrowband transmission of information, and thus coordiation by wavelength. The government then got involved as a third party coordination body.
As more stations went on the air, technological development was aimed at expanding the useable spectrum beyond longwave -- first medium wave (300 kHz to 3MHz) then shortwave (3-30MHz) then VHF (30-300MHz).
WWII advanced the pace of development in UHF (300-900 MHz) and microwaves (above 900 MHz). With those developments came the ability to use polar and spatial diversity. But the latter really took off with the development of microprocessor controlled radios, which enabled spatial diversity by cell -- cellular radio.
However, even with all of the spectrum that these techniques have enabled, the fact remains that, owing to propagation differences, some parts of the spectrum are inherently more valuable than others, a scarcity that leads to economic realities that agencies like the ITU and FCC have been exploiting for decades.
Quietly, however, which these developments were taking place in wavelength coordination, our ability to coordinate transmissions in time has caught up -- first with spread spectrum (not that funny frequency hopping junk) and now individual pulse trains for Ultra Wide Band. UWB in particular holds the promise of ending the economics of scarcity found in wireless. Aside from a thousandfold increase in spectral efficiency, it also maps well to the bursty nature of information -- you don't need a channel all the time, but thanks to coordination by wavelength, you sit on it anyway.
Needless to say, when you challenge the economics of the status quo, you're not going to be too popular in certain political circles.
- Unleashing a new product category before the product itself was ready.
- Ignoring customer feedback on design issues and pricing.
- Vague and inexplicit value proposition.
- Getting nailed by a competitor who listened to customers, built a more portable platform, hit the market target price, and understood the core value proposition of a PDA.
I was a big fan of the Newton from the Messagepad 120 on -- with the 120 and Newton 2.0, Apple had finally delivered on the promise of the PDA. Unfortunately, after years of fumbling, overpromising and under-delivering, the market had moved on, and simply would not believe anything that Apple had to say -- something I'll dub the Newton Effect. The Messagepad 2000, which was ultimately more capable (thanks to DEC's StrongARM), was also ultimately a step in the opposite direction of what the market was asking for: smaller and cheaper, not bigger and over $1000!I stil have my Messagepad 120, and it still comes out of the case every now and then to remind me that bad business decisions can and will kill superior technology.
(A die-hard BSD user since 1986)
Without that competition, Unix would eventually stagnate. Or worse, innovation would be driven into the same kind of useless creeping featurism we've come to expect from the folks in Redmond.
2. The funding of science is even more political. Especially if your funding comes from a political body.
Unfortunately, liberal atheists often see science as their religion, so when conservatives have any input to science, they tend to cry apostate. Never mind that the Clinton Administration, for example, heavily politicized NIST grants for liberal pet causes...
This is really old news; The Wall Street Journal ran a similar article in 1996, only then 1) it was AltaVista, and 2) it was actually news.
Given that dishes are usually even cheaper than linear arrays, your "guesstimate" is off by an order of magnitude!
A perfectly fed 2ft. diamater dish at 2.4 GHz has a gain of 24db, but even a Pringles can feed will give you a dish with 21db of gain. More importanly, the spatial extent of the signal will be minimized, allowing for cellular reuse.
Personally, at 2.4 GHz I'd go with a loop Yagi-Uda array instead, as they're still cheap, and much more wind resistant, something you care about when pointing matters. Make enough of them and they'll be as cheap or cheaper than the Yagis used for UHF TV reception.
Bottom line: There's nothing inherently expensive about gain antennas, and they're the cheapest way to improve the link equation.
Seriously, however, broadcast medium networks like 802.11b are best used for distribution, not long distance point-to-point links (fiber is ultimately cheaper on a bit-for-bit basis), but this demonstrates that you can build a really cheap 802.11b distribtuion network to solve the Last Mile Problem. Another nail in the coffin of Ma Bell...
Louis did the crawler code (known now as "scooter") and was the prefect person to do the job right, as he's a graph theorist by nature and had just finished working on a massive threads debugging tool. Chuck Thacker then suggested that we talk to Mike Burrows over at SRC, who had a wonderful full text database, which Louis and I concluded would work far better than my original idea (using Oracle). So Mike did the database code. I did the first (crude) web-based UI for Mike's code, and even with Louis' first crawl, it was amazing what we could do (relative to the other seach tools of the time). My other chore as "hardware guy" was to spec out the first AlphaServer 8400 that we would get to run the demo. There was a huge backlog of 8400 orders at the time, and only about a half dozen of DEC's techs were trained and authorized to work on them.
AltaVista's initial triumph was simple -- the database held ten times more pages than anything else, and also indexed all of the words in the pages. And yet the response time was nearly instantaneous. Keeping it that way for the first few weeks required a DEC VP to drive several CPU cards through a Boston blizzard to be Fedexed out to Palo Alto, as well as a lot of long hours by the team to diagnose and defend against a number of attacks.
Two things ultimately kept AltaVista from leveraging its early successes. First, DEC wouldn't part with the necessary capital -- as it turned out later, they were negotiating to be bought by Compaq. And secondly, when DEC was finally bought by Compaq, the latter had no idea what to do with AltaVista. The "portal" strategy was designed to maximize the IPO valuation, exactly what investors wanted in 1999. Large amounts of cash were spent on that strategy, only to have the DotCom Bomb go off a week before the IPO.
It's remarkable and I'm gratified to see that AltaVista managed to survive and transition to its roots.
-=paulf
Not sure what you're doing wrong, but the database itself supports Unicode as well as all of the Asian encodings (JIS, etc.). Make sure that you're using the appropriate front end for the encoding -- note the drop-down menu in the upper right hand corner.
-=paulf
-=FZX
In most developing countries, the government, not private industry, runs the state monopoly telecom carrier, and is the sole economic beneficiary of its operation. The cream of telephony for these governments has always been international long distance, something that they've always jealously protected from any competitor, even ham radio. [Want to know why morse code lasted so long in amateur radio? Hint: It's not easy to learn, it presents a barrier to entry, and every developing country has a vote at the ITU.]
Even if they're using a contractor like C&W, the government reaps most of the profit from those $5 per minute calls to overseas families. Ah, the evils of Socialism!
FZX