Let's see - if Network Solutions hangs onto names after they expire, they're cybersquatting, plus they're failing to protect the trademarks if any associated with the names, so they're Bad People.
But if they let them go the second they expire, cybersquatters will snap up the good names, so NS are Bad People also.
And if they give the previous name owners lots of extra slack on renewing their non-trademarked names, they're favoritism-showing Bad People that way also.
And I thought everybody hated them just because of their customer service and pricing:-)
A couple of years ago, when S.1618 was proposed, there was discussion about it on some mailing lists that was interesting. That's long over with, and every piece of email I receive that refers to it is SPAM. So if you set your email filters to trash email referring to "[Ss].1618" and "[Ss].1618" you'll only trash spam, not real messages.
The classic meaning of the term is that not only do packets from blackhole-perp.org to target.org vanish without a trace, but that blackhole-perp.org is advertising "Hello, world, I'm a great way to reach ip_address_of_target.org", so that packets floating nearby destined for target.org get sucked into the black hole.
I have entirely no knowledge of what events occurred between the two groups or whether they were fighting over a network and its routes, but that's the usual meaning.
If the historians 500 years from now are writing anything resembling the modern book, I'll be quite surprised*.
* (er, well, depending on your views of theology and/or nanotech life extension, I'll probably actually be dead, and if I'm not dead, then probably there will have been enough big surprises that this one won't jump way out of the pack, plus I'll have had 500 years of watching the evolution of communications, but nonetheless, I certainly don't expect now that what historians will be doing 500 years from now will necessarily resemble current books.)
A Psion3a or 3c has free VT100 software available, good enough screen resolution (not stunning, but good enough, and infinitely better than Palm), speaks RS232, and you can get them on E-Bay for $75-150. Newer Psion models have better screens, and some can take builtin modem cards. The Psion 5 can do TCP/IP happily.
I think somebody's done TCP/IP for the 3c and maybe the 3a also, but I'm not certain ; otherwise you'll need to dial in to an async login session somewhere (e.g. a modem on your PC, or a few friendly ISPs.)
The Psions are really solid machines, but after 5 years of dropping on the floor occasionally, the case hinges gave out and killed the connection to the LCD screen. I replaced it with a Palm, which I guess I like, though the screen's too small and Graffiti is half as fast as 2-finger touch typing.
I agree with ketilf. They're going to get huge amounts of flak whatever they do, and they're much better off starting with lame names that nobody cares about too much so they can fight their fights on those. They only get one chance to introduce.inc,.mp3, and.xxx, and they need a couple of practice rounds first.
The main controversial one is.biz, since various Alternate Root groups have been using that for a while.
ICANN has done a minor power grab in their insistence on getting and publishing True Names in the whois records. They're mixing several very different uses of that information, which have different requirements and appropriatenesses:
Technical Contact When Things Go Wrong: Sometimes the DNS provider needs a technical contact when things go wrong. A working email address is good enough (it helps to have it on some machine not in the domain, because you're most likely to need to contact the Tech when it's broken.) Phone numbers and names are nice too, but not critical. It's nice if this is also available to the public, because sometimes other people have technical issues that need to be addressed, like machines spewing bad bits.
Administrative Address - This needs to be a workable contact, to deal with policy issues, name ownership disputes, spammer complaints, etc. Again, no need for True Names, but working contacts are important.
Billing contact for the DNS registrar to contact the owner of the name. Again, this doesn't need to be a True Name, and a working email is fine, though it's nice to provide the registrar with enough contact information that your name doesn't just vanish some day because of a billing problem. When NSI was the only DNS Registrar, they should have kept this private, not public, and it was only their own convenience that justified publishing it. With multiple registrars I suspect the same is still true, though perhaps there's a good reason I haven't thought of for doing otherwise.
Owner's True Name, ICBM Address, and Subpoena-Serving Address - IMHO, this is Nobody's Business*, but ICANN strongly believes otherwise - they want to be able to deal with legal disputes like trademark conflicts over domain names by suing or subpoenaing the owner. This one's outright wrong, and the most serious privacy violation of the lot. The alternative is that if the dispute can't be resolved using the Administrative Contact (email or whatever), that the plaintiff should deal with the Name Registrar to see about seizing the name, and if the current user (whether Wrongfully Accused Legitimate Owner or Sleazy Cybersquatter) prefers to remain more private and not respond, then they're at more risk of losing their name, but that should be their choice. Again, IMHO, ICANN's positiion is a combination of control-freakism by some members and wanting to keep the name registrar out of disputes that they don't want to be involved with (and I sympathize - a $50 or $10 name registration fee doesn't leave lots of spare money for lawsuit defenses or even clerical dispute resolution, but that's just tough.)
In practice, ICANN's Data Grabbing isn't accomplishing its positive goals - When I've wanted to hunt down a spammer using Whois, it's generally not very practical - the Supposed True Name info is bogus, or it's a mailbox from a mailbox vendor, or it's outside the US in some jurisdiction where I don't know the alphabet, much less the legal code, and the email contract addresses either get you a black hole, or bounce, or sell your email to other spammers. On the other hand, people have supposedly been stalked, and lots of people have been spammed using this information, and it's Nobody's Business.
* Technically, I'm probably not allowed to use the phrase "Nobody's Business" here in California, because there's a store by that name in Mendocino County, so it'd be name-squatting or trademark dilution or something:-). It's owned by Wavy Gravy, aka Hugh Romney, who runs the "Nobody For President" campaign. So far, Nobody's winning the election, Nobody's leading the country, and Nobody's going to do a great job!
OK, so it's a cheap exploitation, and its authors lack the depth of vision that Frank Herbert had. But if you're *going* to do a cheap exploitation, they picked a pretty good place to hang it on the story lines. Trying to write Dune: Volumes 7,8,9 would have been an utter failure - even Frank Herbert was running pretty dry on stories by the end. Picking a random thread and going sideways would be pretty lame, though perhaps Frank left around some notes they could use; writing Other Children Of Dune to try to make up for the relatively lame Children Of Dune would only emphasize that Frank was the better writer, since they'd be lamer than it was. Doing a prequel at least gives you *some* chance to think about "where did these characters come from? What makes them tick?" which wasn't in the original. Worst you can do is come out like the lasest Star Wars I: A New Prequel.
This *is* Silicon Valley, after all. OK, I'm not currently working at a startup, but many of my friends are, and sometimes I've got hot projects. If you've got a professional job, you get your work done when you need to get it done, and you make yourself available to the people you need to be available to. If the project requires concentrated grinding away, like coding, often that's mostly solitary work; if it's more interactive, you need to spend more time communicating, but that doesn't mean you need to sit at a desk from 9-5, it means you need email and cell phones and pagers and similar tools.
The way to deal with the occasional anal-retentive clock-watching manager is to frequently send mail at 8pm saying "Hi, I need some advice on how to do Foo", and followons at midnight saying "Haven't heard from you, so I'm doing it This Way, and meanwhile I did That, Those, and These".
A decade and a half ago, Bell Labs was changing our work hours from 7.5 per day to 8, and the original pre-announcement blurb didn't mention flextime. Anybody I knew who worked with computers immediately responded "If they make us come in by 8:00, we'll all leave by 5:01.", because none of us worked rigid hours. The real announcement said we had flex-time, and we resumed normal working conditions, which gradually became more flexible, and the world has radically changed since then.
Some jobs are really communications-intensive, and they're a bit different. If you're doing phone support, you're doing phone support, and you need to be there when the customers call, but you still need a workforce that spreads out the hours so they support the customers doing evening applications in California as well as morning work in London. Again, you really need good communications tools, and if that means working from the train by cellphone, that's fine.
I've got a test/demo lab with a bunch of routers and switches and firewalls and things. This is the cheapest mostly-full-functionality box I've seen - you can plug it into an Ethernet and ping it, and if you use the hacking instructions described here to burn a fancier CD, you can telnet and ftp to it. I do have some cheaper solutions (we've got leftover Pentium 60 desktops), but these are much smaller, can probably run keyboardless(?), and do the job.
Cross-country pipes in 1996 tended to be skinny things, T3 at best, and you really needed them. Today, most of the big sites use caching services such as Akamai (used by Slashdot, though not its victims:-) and AT&T (handled the Democratic National Convention just fine) which spread the data out to geographically diverse caches, so Easterners get their copies from servers on the East Coast and Westerners get theirs from the West. Plus those T3 backbones are now OC48 backbones (with occasional OC192, but most of it will really be OC48 or smaller until the routers mature a bit more) - that's 48 times as large,
plus there are a lot more OC48s than there used to be T3s.
Some cable and DSL boxes work as routers, some as bridges, some as NAT boxes. If you're using a bridge-flavored box, it's your PC's MAC that matters. But those guys are probably not going to switch to IPv6 until Cisco and the Tier 1 ISPs make it easy, ICANN stops their current predatory pricing which is designed to prevent IPv6 adoption, and cheap DSL and cable routers support IPv6.
Hey, it's kind of a cool hack being reminded that 2**32 isn't a very big number, and that you really *can* ping everybody on the outer intranet. Of course, many of us live at addresses like 10.116.16.1 or 192.168.1.100 which don't resolve so well, or at 127.0.0.1 when we're in a solipsistic mood. If you don't live behind a firewall, you can always use www.anonymizer.com or Publius or Zero Knowledge to delocalize where you are, and as marketing continues to take over everything, it'll be increasingly worthwhile to do that. Meanwhile, it's the middle of the night, and I'm not really in New Jersey, but my firewall is (&!^$#@# censorware won't let me connect to the anonymizer from there, though:-)
On the technical side, besides the "we tracerouted everybody" hack, if they did use traceroute, they're also getting a lot of correlation information on what's connected to what, and on how long those distances are. And most of their connections are going to go through the NAPs, or through their ISP's peering relationships with other carriers, which are usually in a small number of cities, so they get a lot of correlation on locations they can exploit (they could even get fancy and reduce their traceroute load by taking advantage of serial searches.)
The main problem the article addresses is not the supply of IP addresses, but the rapid increase in the number of BGP AS numbers, which increases the amount of memory and CPU that routers need to track and calculate routes. We've largely fixed the problem of regular IP addresses, between CIDR, RFC1918 10.x addresses behind firewalls, and virtual hosting for web sites. So why do people need their own BGP addresses? It's not just for ISPs any more - there are about 5-10,000 ISPs but 100,000 BGP addresses in use.
I think the answer is that, as IP connectivity from the outside world becomes mission-critical for business applications, businesses often want to deal with more than one ISP, or at least more than one technology (e.g. cable modem plus DSL) so that their customers can reach them even if their primary ISP is down, and to improve performance. To some extent, you fix this by using reliable ISPs and hosting services, or by using fancy DNS tricks to make it easy to find the connections that aren't down or that will give the fastest connections. But ultimately, you get yourself a BGP number and advertise your routes diversely so you can get diversity.
How do we find alternatives to this? Either ISPs need to come up with ways to handle it for their customers, or routers need to get bigger and faster, or we need alternative protocols that make it easier to avoid BGP. A good local ISP can provide this - buying service from a couple of big carriers, and providing enough transparency and responsiveness that customers trust them, and enough customers that their one BGP number supports multiple customers. Hosting centers also do the same thing, and let their customers avoid access circuits as well. But it's tougher to make it work for customers who have offices in multiple locations.
D.A.R.E. was founded by Daryl Gates, the infamous Los Angeles police chief. It's a huge profit-making business - T-Shirts, bumper stickers, classroom material. That business is theoretically separate from the police departments, who also get to collect lots of money in police overtime for teaching school D.A.R.E. classes. Yes, your schools are spending their education money funding cops instead of trained teachers teaching about drugs - is that a good idea in general, much less because police are in the warring-against-drugs business instead of the education business? And do you think kids are going to ask cops potentially incriminating questions, like "my friend tried some marijuana and was stoned for a couple of days - is that normal?" Tough enough getting them to ask teachers.
You've probably seen the T-shirt "D.A.R.E. - I turned in my parents and all I got was this lousy T-Shirt"? Orange County CA cops busted a local hemp store for selling them, and confiscated all their shirts, claiming it was a trademark violation. So much for Supreme Court cases on parody and First Amendment protection.
Here's a Northernlights search URL for "Parents Against D.A.R.E., a parents group opposing this scam.
In this situation, your email account can be anywhere on the net, so you've got a lot of choices. Most of them are small ISPs, because that's who offers shell accounts and security flexibility with SSH, but you're trading off smallness vs. redundancy a bit. I've been quite pleased with idiom.com , and other people have mentioned Illuminati Online. Another place to look is Anonymizer.com, if they offer shell accounts. Or you could check out XS4ALL.NL, in the Netherlands, if they do shell.
Finding a provider who won't roll over on subpoenas is tough - just about anybody big enough to be incorporated (you wanted reliability) will respond, though some will go out of their way to help anybody official-sounding who asks, while others will insist on seeing court orders on paper first. Non-US / Non-UK providers may have some advantages, since most people don't want to bother getting a Finnish court order just to yell at you about something you posted on Usenet that they didn't like.
UUNET has a bunch of DS3 trunks shown in their networks. That doesn't mean that they're real T3 copper coax pairs. It means they're a 45Mbps signal, which is almost always carried on fiber except for inside wiring and a few funky places where it's radio. Most long-haul fiber is OC48 (multiple wavelengths of it using DWDM), though OC192 is starting to be mature enough to for newer applications, and a T3 channel is just data riding on an OC1. Similarly, the T1 to Hong Kong wouldn't be copper under the ocean, it'd be just another 1.544 Mbps bit channel muxed into SONET, probably in a T3 or E3, though maybe the cable carrier they're using uses VT1.5 instead.
There is real copper T1 in the phone networks, out in the last-few-miles side of things, as well as analog voice, though much of that is also carried on fiber loop carrier equipment or voice muxed onto T1s. The wires that have been used for analog phones can often be cranked up for DSL, so carry higher-speed signals on the same old crappy wire, but it's T1 or below, not T3.
The interesting new copper out there isn't backbones, it's cable TV, which typically does hybrid fiber-coax systems - copper coax down your block, fiber networks feeding the copper, and subdividing the networks any time there's enough load to make it worth adding more fiber. On the other hand, there's also a lot of fiber direct to businesses, some of it run by cable TV companies, and some by access providers (including telcos and competitors.)
Because that's what the web was designed for :-)
on
Does P = NP?
·
· Score: 2
The initial user community that the web was designed for were physicists who needed a faster and more convenient method for distributing preprints than the academic journal process. Of course, one of the reasons you distribute preprints is so that your peers (or betters:-) can see and review your work, and find errors before you get into the official journal review process.
capacity=bandwidth*log(1+Signal/Noise) ; I forget which base the logarithm is. Those same copper wires can do 384kbps or other high speeds using DSL (depends on distance, loss, capacitance, etc), pumping higher frequency signals through them and doing appropriate numbers of bits per hertz. However, modems will never do better than 64kbps, because the signals get digitized to 64kbps at the phone company (and robbed-bit signalling generally reduces this to 56kbps.) We used to say that modems would never do better than about 33kbps because of signal/noise ratios on typical phone lines, but 56kbps modems cheat by knowing the effects of digitization instead of just doing analog.
One limit you may be thinking of is that the number of samples per second for pulse-amplitude modulation needs to be twice the frequency of the sound waveform you're trying to carry (so a 4kHz audio signal needs 8k samples/second, and typically that gets digitized to 8 bits/sample using a non-linear quantization.)
Re:I want an ear-sized two-way radio
on
Ready-To-Wear PCs
·
· Score: 2
Some of the third-party headsets available for Nokia cellphones have the microphone built into the earpiece (only sticks out an inch or so), and many have the mike on a small lightweight frob in the cord. I assume that similar technology is available in standard (everybody but *^&()*& Nokia) 2.5mm cellphone headsets, and I wouldn't be surprised if there's some in the 3.5mm or whatever size it is that PCs and Walkmans use.
You can certainly find lots of Family Radio Service cheap radios out there; probably some use headsets (however, they tend to be push-to-talk.) One catch is battery life - if you leave them on full time, you'll want to get rechargeable batteries, using something other than NiCd.
Slashdot Mirror
And I thought everybody hated them just because of their customer service and pricing
7 years in prison does seem a bit harsh. On the other hand
"I had to pay $50 and pick up the garbage."
sounds a bit light.
A couple of years ago, when S.1618 was proposed, there was discussion about it on some mailing lists that was interesting. That's long over with, and every piece of email I receive that refers to it is SPAM. So if you set your email filters to trash email referring to "[Ss].1618" and "[Ss] .1618" you'll only trash spam, not real messages.
I have entirely no knowledge of what events occurred between the two groups or whether they were fighting over a network and its routes, but that's the usual meaning.
* (er, well, depending on your views of theology and/or nanotech life extension, I'll probably actually be dead, and if I'm not dead, then probably there will have been enough big surprises that this one won't jump way out of the pack, plus I'll have had 500 years of watching the evolution of communications, but nonetheless, I certainly don't expect now that what historians will be doing 500 years from now will necessarily resemble current books.)
I think somebody's done TCP/IP for the 3c and maybe the 3a also, but I'm not certain ; otherwise you'll need to dial in to an async login session somewhere (e.g. a modem on your PC, or a few friendly ISPs.)
The Psions are really solid machines, but after 5 years of dropping on the floor occasionally, the case hinges gave out and killed the connection to the LCD screen. I replaced it with a Palm, which I guess I like, though the screen's too small and Graffiti is half as fast as 2-finger touch typing.
I agree with ketilf. They're going to get huge amounts of flak whatever they do, and they're much better off starting with lame names that nobody cares about too much so they can fight their fights on those. They only get one chance to introduce .inc, .mp3, and .xxx, and they need a couple of practice rounds first.
.biz, since various Alternate Root groups have been using that for a while.
The main controversial one is
EUDEM - The European Union in Humanitarian DEMining is another anti-landmine group, which provides a clearinghouse for technical information.
There are a number of groups working against government terrorism and torture and trying to stop the damage from wars.
Medecins sans Frontieres, aka Doctors Without Borders" is an international medical relief group. The Campaign for Access to Essential Medicines is a related organization.
In practice, ICANN's Data Grabbing isn't accomplishing its positive goals - When I've wanted to hunt down a spammer using Whois, it's generally not very practical - the Supposed True Name info is bogus, or it's a mailbox from a mailbox vendor, or it's outside the US in some jurisdiction where I don't know the alphabet, much less the legal code, and the email contract addresses either get you a black hole, or bounce, or sell your email to other spammers. On the other hand, people have supposedly been stalked, and lots of people have been spammed using this information, and it's Nobody's Business.
* Technically, I'm probably not allowed to use the phrase "Nobody's Business" here in California, because there's a store by that name in Mendocino County, so it'd be name-squatting or trademark dilution or something
OK, so it's a cheap exploitation, and its authors lack the depth of vision that Frank Herbert had. But if you're *going* to do a cheap exploitation, they picked a pretty good place to hang it on the story lines. Trying to write Dune: Volumes 7,8,9 would have been an utter failure - even Frank Herbert was running pretty dry on stories by the end. Picking a random thread and going sideways would be pretty lame, though perhaps Frank left around some notes they could use; writing Other Children Of Dune to try to make up for the relatively lame Children Of Dune would only emphasize that Frank was the better writer, since they'd be lamer than it was. Doing a prequel at least gives you *some* chance to think about "where did these characters come from? What makes them tick?" which wasn't in the original. Worst you can do is come out like the lasest Star Wars I: A New Prequel.
The way to deal with the occasional anal-retentive clock-watching manager is to frequently send mail at 8pm saying "Hi, I need some advice on how to do Foo", and followons at midnight saying "Haven't heard from you, so I'm doing it This Way, and meanwhile I did That, Those, and These".
A decade and a half ago, Bell Labs was changing our work hours from 7.5 per day to 8, and the original pre-announcement blurb didn't mention flextime. Anybody I knew who worked with computers immediately responded "If they make us come in by 8:00, we'll all leave by 5:01.", because none of us worked rigid hours. The real announcement said we had flex-time, and we resumed normal working conditions, which gradually became more flexible, and the world has radically changed since then.
Some jobs are really communications-intensive, and they're a bit different. If you're doing phone support, you're doing phone support, and you need to be there when the customers call, but you still need a workforce that spreads out the hours so they support the customers doing evening applications in California as well as morning work in London. Again, you really need good communications tools, and if that means working from the train by cellphone, that's fine.
This box isn't hiding information - you may have a software license for the box, but much of the content is covered by the GPL.
I've got a test/demo lab with a bunch of routers and switches and firewalls and things. This is the cheapest mostly-full-functionality box I've seen - you can plug it into an Ethernet and ping it, and if you use the hacking instructions described here to burn a fancier CD, you can telnet and ftp to it. I do have some cheaper solutions (we've got leftover Pentium 60 desktops), but these are much smaller, can probably run keyboardless(?), and do the job.
Cross-country pipes in 1996 tended to be skinny things, T3 at best, and you really needed them. Today, most of the big sites use caching services such as Akamai (used by Slashdot, though not its victims :-) and AT&T (handled the Democratic National Convention just fine) which spread the data out to geographically diverse caches, so Easterners get their copies from servers on the East Coast and Westerners get theirs from the West. Plus those T3 backbones are now OC48 backbones (with occasional OC192, but most of it will really be OC48 or smaller until the routers mature a bit more) - that's 48 times as large,
plus there are a lot more OC48s than there used to be T3s.
Some cable and DSL boxes work as routers, some as bridges, some as NAT boxes. If you're using a bridge-flavored box, it's your PC's MAC that matters. But those guys are probably not going to switch to IPv6 until Cisco and the Tier 1 ISPs make it easy, ICANN stops their current predatory pricing which is designed to prevent IPv6 adoption, and cheap DSL and cable routers support IPv6.
On the technical side, besides the "we tracerouted everybody" hack, if they did use traceroute, they're also getting a lot of correlation information on what's connected to what, and on how long those distances are. And most of their connections are going to go through the NAPs, or through their ISP's peering relationships with other carriers, which are usually in a small number of cities, so they get a lot of correlation on locations they can exploit (they could even get fancy and reduce their traceroute load by taking advantage of serial searches.)
Film? How quaint....
I think the answer is that, as IP connectivity from the outside world becomes mission-critical for business applications, businesses often want to deal with more than one ISP, or at least more than one technology (e.g. cable modem plus DSL) so that their customers can reach them even if their primary ISP is down, and to improve performance. To some extent, you fix this by using reliable ISPs and hosting services, or by using fancy DNS tricks to make it easy to find the connections that aren't down or that will give the fastest connections. But ultimately, you get yourself a BGP number and advertise your routes diversely so you can get diversity.
How do we find alternatives to this? Either ISPs need to come up with ways to handle it for their customers, or routers need to get bigger and faster, or we need alternative protocols that make it easier to avoid BGP. A good local ISP can provide this - buying service from a couple of big carriers, and providing enough transparency and responsiveness that customers trust them, and enough customers that their one BGP number supports multiple customers. Hosting centers also do the same thing, and let their customers avoid access circuits as well. But it's tougher to make it work for customers who have offices in multiple locations.
You've probably seen the T-shirt "D.A.R.E. - I turned in my parents and all I got was this lousy T-Shirt"? Orange County CA cops busted a local hemp store for selling them, and confiscated all their shirts, claiming it was a trademark violation. So much for Supreme Court cases on parody and First Amendment protection.
Here's a Northernlights search URL for "Parents Against D.A.R.E., a parents group opposing this scam.
Finding a provider who won't roll over on subpoenas is tough - just about anybody big enough to be incorporated (you wanted reliability) will respond, though some will go out of their way to help anybody official-sounding who asks, while others will insist on seeing court orders on paper first. Non-US / Non-UK providers may have some advantages, since most people don't want to bother getting a Finnish court order just to yell at you about something you posted on Usenet that they didn't like.
There is real copper T1 in the phone networks, out in the last-few-miles side of things, as well as analog voice, though much of that is also carried on fiber loop carrier equipment or voice muxed onto T1s. The wires that have been used for analog phones can often be cranked up for DSL, so carry higher-speed signals on the same old crappy wire, but it's T1 or below, not T3.
The interesting new copper out there isn't backbones, it's cable TV, which typically does hybrid fiber-coax systems - copper coax down your block, fiber networks feeding the copper, and subdividing the networks any time there's enough load to make it worth adding more fiber. On the other hand, there's also a lot of fiber direct to businesses, some of it run by cable TV companies, and some by access providers (including telcos and competitors.)
The initial user community that the web was designed for were physicists who needed a faster and more convenient method for distributing preprints than the academic journal process. Of course, one of the reasons you distribute preprints is so that your peers (or betters :-) can see and review your work, and find errors before you get into the official journal review process.
One limit you may be thinking of is that the number of samples per second for pulse-amplitude modulation needs to be twice the frequency of the sound waveform you're trying to carry (so a 4kHz audio signal needs 8k samples/second, and typically that gets digitized to 8 bits/sample using a non-linear quantization.)
Some of the third-party headsets available for Nokia cellphones have the microphone built into the earpiece (only sticks out an inch or so), and many have the mike on a small lightweight frob in the cord. I assume that similar technology is available in standard (everybody but *^&()*& Nokia) 2.5mm cellphone headsets, and I wouldn't be surprised if there's some in the 3.5mm or whatever size it is that PCs and Walkmans use.
You can certainly find lots of Family Radio Service cheap radios out there; probably some use headsets (however, they tend to be push-to-talk.) One catch is battery life - if you leave them on full time, you'll want to get rechargeable batteries, using something other than NiCd.