Microsoft doesn't need saving, and someone should tell Mr. Plotkin that Willie Nelson didn't write "Poncho and Lefty".
Microsoft executives have admitted, under oath, that they use an "embrace, extend, and extinguish" attack on any software that doesn't let M$ muscle in and claim a cut of the profit.
Let's predict how it'll go:
Embrace: M$ issues a Linux version, heavily advertised not as "Microsoft Linux", just "LINUX" and carefully isolated from the Winblows product.
Support turns out to be a joke, just like the current crap they push out the door.
Extend: Viruses, bugs, and bad press appear faster than options at an M$ hiring hall.
Extinguish: Institutional and business users quickly abandon all versions of "LINUX".
Yeah, that sounds about right. Look for "LINUX" to be heavily trumpeted just before the next "release" of Winblows 200x/NT/whatever.
This looks like a slick box, but I'm in need of help with an older machine that runs the MediaGX.
I have a GCT-MGX running Samba, but haven't been able to get X running on it. It's a 5510 chipset, with 32MB RAM and the MediaGX running at 200MHz, and the manufacturer is belly-up.
Well, I guess it's fun: I don't like being the nay-sayer in the face of/.'s "gee wiz" futurism, but it's nice to pass along the lessons I learned in the trenches.
I'll weigh in One Last Time(TM), and then go back to my mundane life:
I suspect within twenty years we will have small distance-limited data-carrying nodes connected by wire/fiber to backbone carriers.
Could be - but the article was about a wireless network, not an extension of the cellular one we already have.
You talk about a home-by-home deployment of shared-node transceivers. Ten to twenty years from now if current trends proceed the average home will either possess or be within range of such a device.
Well, I have to disagree. The drivers aren't there for that kind of a technology:
Spectrum space costs big bucks: the federal government's budget was balanced by the spectrum auctions, and they ain't gonna turn their backs on that kind of a revenue stream. Those that pay will want control, and that obviates a decentralized paradigm.
Labor costs are born by the home owner with cable or dsl installations: radio requires much more expertise and equipment.
Installation costs ditto.
Wire is more reliable.
Wire has higher bandwidth.
The FCC isn't involved if you use wire.
I'm not saying it's impossible, but we'd have to have very dramatic increases in wireline pricing before it would be economically viable.
Regarding compression technology and embedded processors -- I am not talking about throughput -- I am talking about compression. If I take a newspaper and compress it, it will take less time to transmit, allowing another device more time to use the same frequency. Unless I misunderstand your argument I suspect you are thinking in analog terms, but with digital compression technology we can make one byte count as ten, or twenty, or even fifty, effectively increasing by a factor of ten, twenty, or fifty the total amount of data (depending on type) that can be carried on an available frequency.
Nope, sorry, my ciruit breaker trips at this point. Compression is not limitless, and the practical limits for "home use" devices were reached years ago.
Of course, I am thinking in throughput terms: bits are bits, and the time it takes to send them from one place to another is what limits your options.
The throughput possible with radio transmission is already at the Shannon limit for "part 15" devices, and unless you're arguing that the average homeowner will pay for a coax cable and dish (To where? Aimed at what?), then added compression won't increase throughput.
If we're hypothesizing a wireless network, then either it has to be nondirectional for the nodes to talk to each other, or it has to be an MDS dirivative in order to get respectable bandwidth.
MDS won't work, for the reasons I gave before: wire is cheaper.
Cooperating nodes won't work, because they suffer from the Betamax dilemma: who's going to be first to buy something they can't use until everybody else buys one?
As I said, I worked on this when hams first got into IP and packet radio. The greatest limitation to spreading a viable network turned out to be the need for sharing: there isn't a sufficient density of sites to make non-directional transmission feasable, and getting to central nodes required more investment than most could make (time/money/expertise). These were hard core techies, too: your average homeowner is going to want it to work out of the box, not after putting a ground plane 70' up in a tree like I did.
A wireless Internet is proscribed not so much by our technology - cellular doesn't use anything new, just the old stuff in different ways - but by human nature. It is, however, a vicious circle: unless (I did not say "until") the technology advances to a point where plug 'n pray devices can work at exurban distances, it won't be viable. As before, I'll be delighted if I'm wrong.
"Guerilla.net" has such a outre ring to it, doesn't it? It made me feel like buying a copy of Mother Jones, until I looked at the web site and realized that it's talking about a ham-radio network.
The idea is nice: an independent, censor-proof wireless alternative to the Internet. It has a nice, warm, "Whole Earth Epilogue" kind of feel to it. It's the sort of idea that makes one think of a future where everyone is free to speak their mind and rail away against politicians of all stripes without fear of reprisal or judgement.
In other words, Usenet, but without the reliability and without most of the audience.
Long story short: don't confuse the medium with the message. Anonymity is not equal to freedom, but even if it were, it's very easy to stay anonymous on Usenet, and very hard on a radio.
Bill, W1AC
P.S. I am not over 50, I don't smoke, and I don't have a pot belly.
I think what will make a ubiquitous wireless environment possible is an increase in the processing power of embedded systems.
The systems' processing power is moot: current chips are able to process much more data than they can receive, and even in a network of allied peers where each carries some of the data for others, the processing power has been available for years. It's not a question of how quickly we can process the data, but how quickly we can move it from point to point.
We will obviously not have more frequencies available, but I suspect we will be able to use alloted frequencies more wisely, and possibly devise some more robust low power transmission schemes.
The scheme we would need for a wireless internet is one that will place a cooperating shared-node transceiver into every home in the nation. The limiting factor is distance.
[snip]
We have not dumped more wattage into cell transmission technology. We certainly have not dramatically increased the amount of alloted frequency available to place a single cell call. What we have done is to employ technologies like
realtime compression and high-speed channel switching/frequency hopping.:)
No, what we've done is put a lot of small cell transceivers into every village, hamlet, suburb, town, city, and building in the nation - and they're all hooked up by wire. You're making my argument for me here: the infrastructure required for cellular distribution is several orders of magnitude greater than the existing wire plant.
On a similar note we now cram two hundred plus channels of high quality video into the same amount of bandwidth used by twenty analog video channels -- using advanced compression technology. The set top DSS boxes used to weigh five kilograms and cost several hundred dollars. Now they weigh under a kilogram and cost less than fifty.
Well, I'll believe it when I see it working with typical TV transmit powers and rabbit-ears at each receiver, but I digress. We're talking about a bidirectional network here, not a broadcast system with 1/2 meter dishes. Even with asymetric bandwidth, the "haulback" problem remains, and most systems solve it by using wire.
As for compression: we could compress the daily newspaper onto a single floppy, but the problem of how to deliver it would still remain. Compression techniques are usefull only where there's enough data going to a single receiver to make it worthwhile. For the average net user, it doesn't track: while compression might be useful between a "community" server and an Internet peering point, the wireless dilemma reamins one of transporting bits to home users without requiring a cellular network that would cost a lot more than wire. We can compress the data all we want, but we've still got to deliver it, and that's where wireless falls off the curve: even at the abysmally low bit rates hams used, there are too many trees in the way between the server and the user.
Wireless is often pitched as a seductive "last mile" solution to allow bypass of the PSTN or cable. It's been tried before, and the companies that did are now footnotes in business school textbooks about what does and doesn't work. There just aren't enough viable central distribution points to allow plug 'n pray wireless networks, and absent an as-yet-undiscovered breakthrough that lowers the path loss of the typical microwave signal, we're stuck with wire.
Technology marches forward.:)
The technology may march forward, but the square-law limit remains. A wireless network for the Internet is not a realistic possibility at this time.
Zeus Transceivers operate through walls and floors up to 1,500 feet in most buildings, and line-of-sight to the horizon outdoors. For longer distances, networks of Zeus Transceivers can relay data packets from otherwise out-of-range devices to significantly extend the range of a wireless network.
As I said, I'll be delighted if I'm wrong, but this doesn't sound like the solution. It's nice to be buzzword-compliant, but the RF decks I worked on as a ham had longer range than this, and they still wouldn't work.
You have three problems here:
Terrain.
"Line of sight" sounds great - until you go outside and actually look. I'll bet most/. users will see pretty much what I do: other buildings, trees, hills, maybe a water tower, and sky.
To make a "line of sight" system work, you've got to get a central collection point transceiver at a location all potential users can "see". Trust me: the supply of high places that you can mount an antenna on is getting scarcer by the millisecond, and nobody gives that space away anymore.
Absent a central point that's within your site distance, you're limited to relaying through other users. In the average suburb, 1500 feet is less than a street length, and I doubt you'd get even that far. In any case, you need everybody to keep their boxes turned on all the time.
Node density: you have to have enough nodes within your circle of influence to relay your packets at all times, and home users are just too spread out for that.
Protocol. The Zeus website has a lot of breezy marketdroid-speak, but the "128 bits" you allude to isn't mentioned, and neither is any compliance to an IEEE standard. In any case, a distributed, shared-routing network with independent nodes would need a new protocol to allow routing without "hidden transmitter syndrome" and to satisfy FCC ID requirements.
Sorry, this just isn't the panacea. I didn't see any mention of throughput on the site, but I doubt it's any significant fraction of LAN speed, and Zeus clearly states that their transceivers are intended to hook up serial devices to a computer. In other words, they're ready to trickle data, not "pump" it.
OK, let me apologize right up front for throwing cold water on this idea.
It's a great idea. Really. The thought of freeing people from the wired network is great.
Unfortunately, we need wires to make the Internet work. The problem isn't bandwidth, but rather bit rate: in order to get close to the transfer rates we now enjoy in the wired world, we'd need a large infrastructure of radio relay sites, which would, in turn, need a large infrastructure of point-to-point links in between them.
I worked (a lot) on this problem as a ham operator, when the Internet first became popular and hams were trying to implement a ham-radio version of IP. Pioneers such as Phil Karn (now at Qualcom) put out shareware that allowed hams to connect PC's to "modems", which in turn were connected to radios.
It didn't work. The limitation was distance: in order to get a signal from a typical user to a "server", the user often had to have his signal realyed through three or four other hams before he even reached the server, and the bit rate (try not to faint...) was seldom over 1,200 bps.
Yes, that's right: 1200 bps, or about 1/46th of what your 56k modem delivers. Now, I know that the new wireless cards have much higher bit rates, but they also have dramatically shorter distances.
Unless we're all willing to put a microwave dish on our tower - uh, you do have a tower, right? - it's not going to be possible with current paradigms.
For distributed wireless to work, it would need these advances:
Large blocks of spectrum dedicated to home wireless: business users can't coexist, since they use much lower power levels.
FCC action to limit the liability of nodes that relay anything illegal, immoral, or fattening:
The ten words you can't say on television.
Copyrighted material.
Anonymous politcal speech.
Anything having to do with codes or ciphers.
A "critical mass" of users (who's first?) in order to make relaying possible
Extensions to the wireless protocols to allow forwarding at the mac layer - check out AX.25 for an example - and an assignment mechanism for it.
It's a great idea, and I was it's biggest fan, but there's too much distance to cover and you can't break the laws of physics. I'll be delighted if I'm wrong.
Slashdot Mirror
Microsoft doesn't need saving, and someone should tell Mr. Plotkin that Willie Nelson didn't write "Poncho and Lefty".
Microsoft executives have admitted, under oath, that they use an "embrace, extend, and extinguish" attack on any software that doesn't let M$ muscle in and claim a cut of the profit.
Let's predict how it'll go:
Yeah, that sounds about right. Look for "LINUX" to be heavily trumpeted just before the next "release" of Winblows 200x/NT/whatever.
Bellhead
I have a GCT-MGX running Samba, but haven't been able to get X running on it. It's a 5510 chipset, with 32MB RAM and the MediaGX running at 200MHz, and the manufacturer is belly-up.
All help appreciated. mailto:redeye@banet.net
I'll weigh in One Last Time(TM), and then go back to my mundane life:
Could be - but the article was about a wireless network, not an extension of the cellular one we already have. Well, I have to disagree. The drivers aren't there for that kind of a technology:- Spectrum space costs big bucks: the federal government's budget was balanced by the spectrum auctions, and they ain't gonna turn their backs on that kind of a revenue stream. Those that pay will want control, and that obviates a decentralized paradigm.
- Labor costs are born by the home owner with cable or dsl installations: radio requires much more expertise and equipment.
- Installation costs ditto.
- Wire is more reliable.
- Wire has higher bandwidth.
- The FCC isn't involved if you use wire.
I'm not saying it's impossible, but we'd have to have very dramatic increases in wireline pricing before it would be economically viable. Nope, sorry, my ciruit breaker trips at this point. Compression is not limitless, and the practical limits for "home use" devices were reached years ago.Of course, I am thinking in throughput terms: bits are bits, and the time it takes to send them from one place to another is what limits your options.
The throughput possible with radio transmission is already at the Shannon limit for "part 15" devices, and unless you're arguing that the average homeowner will pay for a coax cable and dish (To where? Aimed at what?), then added compression won't increase throughput.
If we're hypothesizing a wireless network, then either it has to be nondirectional for the nodes to talk to each other, or it has to be an MDS dirivative in order to get respectable bandwidth.
- MDS won't work, for the reasons I gave before: wire is cheaper.
- Cooperating nodes won't work, because they suffer from the Betamax dilemma: who's going to be first to buy something they can't use until everybody else buys one?
As I said, I worked on this when hams first got into IP and packet radio. The greatest limitation to spreading a viable network turned out to be the need for sharing: there isn't a sufficient density of sites to make non-directional transmission feasable, and getting to central nodes required more investment than most could make (time/money/expertise). These were hard core techies, too: your average homeowner is going to want it to work out of the box, not after putting a ground plane 70' up in a tree like I did.A wireless Internet is proscribed not so much by our technology - cellular doesn't use anything new, just the old stuff in different ways - but by human nature. It is, however, a vicious circle: unless (I did not say "until") the technology advances to a point where plug 'n pray devices can work at exurban distances, it won't be viable. As before, I'll be delighted if I'm wrong.
Bill, W1AC
"Guerilla.net" has such a outre ring to it, doesn't it? It made me feel like buying a copy of Mother Jones, until I looked at the web site and realized that it's talking about a ham-radio network.
The idea is nice: an independent, censor-proof wireless alternative to the Internet. It has a nice, warm, "Whole Earth Epilogue" kind of feel to it. It's the sort of idea that makes one think of a future where everyone is free to speak their mind and rail away against politicians of all stripes without fear of reprisal or judgement.
In other words, Usenet, but without the reliability and without most of the audience.
Long story short: don't confuse the medium with the message. Anonymity is not equal to freedom, but even if it were, it's very easy to stay anonymous on Usenet, and very hard on a radio.
Bill, W1AC
P.S. I am not over 50, I don't smoke, and I don't have a pot belly.
The systems' processing power is moot: current chips are able to process much more data than they can receive, and even in a network of allied peers where each carries some of the data for others, the processing power has been available for years. It's not a question of how quickly we can process the data, but how quickly we can move it from point to point.
The scheme we would need for a wireless internet is one that will place a cooperating shared-node transceiver into every home in the nation. The limiting factor is distance.
No, what we've done is put a lot of small cell transceivers into every village, hamlet, suburb, town, city, and building in the nation - and they're all hooked up by wire. You're making my argument for me here: the infrastructure required for cellular distribution is several orders of magnitude greater than the existing wire plant.
Well, I'll believe it when I see it working with typical TV transmit powers and rabbit-ears at each receiver, but I digress. We're talking about a bidirectional network here, not a broadcast system with 1/2 meter dishes. Even with asymetric bandwidth, the "haulback" problem remains, and most systems solve it by using wire.
As for compression: we could compress the daily newspaper onto a single floppy, but the problem of how to deliver it would still remain. Compression techniques are usefull only where there's enough data going to a single receiver to make it worthwhile. For the average net user, it doesn't track: while compression might be useful between a "community" server and an Internet peering point, the wireless dilemma reamins one of transporting bits to home users without requiring a cellular network that would cost a lot more than wire. We can compress the data all we want, but we've still got to deliver it, and that's where wireless falls off the curve: even at the abysmally low bit rates hams used, there are too many trees in the way between the server and the user.
Wireless is often pitched as a seductive "last mile" solution to allow bypass of the PSTN or cable. It's been tried before, and the companies that did are now footnotes in business school textbooks about what does and doesn't work. There just aren't enough viable central distribution points to allow plug 'n pray wireless networks, and absent an as-yet-undiscovered breakthrough that lowers the path loss of the typical microwave signal, we're stuck with wire.
The technology may march forward, but the square-law limit remains. A wireless network for the Internet is not a realistic possibility at this time.
Bill
As I said, I'll be delighted if I'm wrong, but this doesn't sound like the solution. It's nice to be buzzword-compliant, but the RF decks I worked on as a ham had longer range than this, and they still wouldn't work.
You have three problems here:
Sorry, this just isn't the panacea. I didn't see any mention of throughput on the site, but I doubt it's any significant fraction of LAN speed, and Zeus clearly states that their transceivers are intended to hook up serial devices to a computer. In other words, they're ready to trickle data, not "pump" it.
Bill
OK, let me apologize right up front for throwing cold water on this idea.
It's a great idea. Really. The thought of freeing people from the wired network is great.
Unfortunately, we need wires to make the Internet work. The problem isn't bandwidth, but rather bit rate: in order to get close to the transfer rates we now enjoy in the wired world, we'd need a large infrastructure of radio relay sites, which would, in turn, need a large infrastructure of point-to-point links in between them.
I worked (a lot) on this problem as a ham operator, when the Internet first became popular and hams were trying to implement a ham-radio version of IP. Pioneers such as Phil Karn (now at Qualcom) put out shareware that allowed hams to connect PC's to "modems", which in turn were connected to radios.
It didn't work. The limitation was distance: in order to get a signal from a typical user to a "server", the user often had to have his signal realyed through three or four other hams before he even reached the server, and the bit rate (try not to faint...) was seldom over 1,200 bps.
Yes, that's right: 1200 bps, or about 1/46th of what your 56k modem delivers. Now, I know that the new wireless cards have much higher bit rates, but they also have dramatically shorter distances.
Unless we're all willing to put a microwave dish on our tower - uh, you do have a tower, right? - it's not going to be possible with current paradigms.
For distributed wireless to work, it would need these advances:
It's a great idea, and I was it's biggest fan, but there's too much distance to cover and you can't break the laws of physics. I'll be delighted if I'm wrong.
Bill, W1AC