The next big issue for computer circuitry is going to be 3d layout. At present we have 2.5d: you can layer some circuitry on top of other circuitry, but no more.
Full 3d is going to require nano-fabrication techniques: no form of lithography will cut the mustard. While painting of this kind is not the full solution, its definitely a step in the right direction.
There are of course many other challenges in 3d circuitry, cooling not least.
ISTR some work done by IBM with tiny balls that had a few hundred devices on the surface of each and were packed into a cubic array. Does anyone know where that went?
Bluetooth looks like being an even bigger headache because lots of bluetooth devices cannot be turned off, except by removing the battery. They have low power standby states, but can be woken up, or might even wake up on their own schedule.
Sure... some yahoo will probably make a point about how the two are totally differemt.
Well... they are totally different.
Consumer RFI is not going to throw a 747 into a spin. The danger is that it will interfere with navigation sufficiently to cause an accident. For instance, missing the runway on an ILS approach.
A lighting strike is going to produce enough RFI, and even EMP, to interfere with pretty much anything electronic unless it is very well shielded. But its over quickly. Consumer RFI can last the entire flight.
I wonder if the massive expense involved in the recent move to put miniature tvs on the back of all seats in third class [...] is due to the need to ensure minimal leakage?
Some years ago I was involved in a project looking at this kind of set up.
The problems are huge, and EMI is actually one of the more minor problems. You mostly solve it by using fibre or good co-ax for the cable runs and faraday cages for the electronics.
The big danger is fire. Every piece of equipment has to be certified to ensure that it won't start a fire, and if one does start that the equipment won't make things worse. Cabling is the big headache here: a cable conduit full of PTFE makes a wonderful channel for fire to spread and also creates lots of poisonous smoke. Cables need to be specially rated, as do the connectors.
Then the equipment must be safe in a crash. No broken glass, even when a passenger's head hits it. It also must not be able to fly out of the seat and hit the person in front.
It has to be cooled, even though the seatback component is surrounded by a good insulator. But at the same time it has to withstand Junior pouring his orange juice over it, and curious passengers with pocket toolkits (hello, you know who you are...)
It has to withstand vibration, pressure changes and temperature extremes (aircraft may be left parked in the hot sun or freezing cold). Components are rated for this sort of thing of course, but aircraft operation tends to put you in a corner of the envelope, and failures are therefore much more likely.
Finally you are on a strict power and weight budget.
UK contract law takes a very pragmatic approach to contracts, including "click-through" licenses. For most contracts (real estate has different rules) what counts is the state of mind of the parties to the contract. Basically, to enforce a contract you have to show that the other person understood the terms of the contract and agreed to them. So anything "unusual or onerous" must be flagged up, and the more unusual or onerous the terms, the bolder the flagging must be. The standard precedent is case where a photo library sued a customer for thousands of pounds because some pictures were a couple of weeks overdue. The penalties were in paragraph 2 of the fine print, and something like five pounds per day per item. The customer had not read this fine print, and was therefore held not to have agreed to it. In addition there are other protections for consumers which prevent you from signing away consumer rights under any circumstances. As a result, UK consumers can pretty much ignore the fine print in contracts, because if it says something unusual or onerous then it will not be upheld in a court of law.
Having said that, I always look through it anyway, just to be on the safe side. You also get asked to sign things to say "I have read and understand the page of fine print overleaf". I don't know what the legal standing of such a declaration is. Whenever asked to sign one of these I make the salesdroid wait while I read it. Maybe I should read it aloud.
pay off the local.gov and you have a rather isolated test lab with $0 pay labor
The problem isn't Monsanto, its the corrupt government that lets its officials be brought. Fix the government and the corps will behave themselves.
yes there may have always been cancer - but I don't think it has ever been so wide spread.
The reason that more people get cancer these days is merely that they are failing to die of other things, like cholera, polio, smallpox, war, famine and childbirth.
Go and look at some trading standards cases from a few hundred years back. Chalk dust in flour was common, and sweets were dyed with heavy metal salts. I kid you not. Fancy a return to those days? Whereas these days we have the luxury of getting annoyed about GM pollen.
Any news on the expedited move to the Supreme Court? Its obvious that this case is going to wind up there sooner or later (unless Linux happens to make it moot). ISTR past talk about an option for the Judge to bump it up there directly, bypassing the Appeals court. This has two important effects:
It avoids a court known to be more friendly to MS than Judge Jackson
It shortens the process by a year or two.
But I can't see anything in any of the news stories on this.
you are obviously a clueless admin and should not be let near an NT box. Runnign you normal account ith admin privs is wrong.
Well, I don't normally bother responding to flamebait, but on this occasion...
I'm not totally dumb. In addition the corporate firewall I also run AtGuard (now part of Norton IIRC) which keeps an eye on any active content from the Web, and should also trap anything unauthorised that tries to send information back out. Also I don't run Internet Explorer, so the worst idiocies of ActiveX are not an issue for me.
On top of that we also have a clued in administrator (I am not an administrator, I just look after my desktop box) who keeps our virus checker up to date and does threat monitoring on the servers. ILOVEYOU didn't get in here.
Finally, I spent about six months trying to run with separate user and admin IDs, and believe me it just wasn't worth the hassle. There are so many little jobs, from defragging the hard drive to updating the IP configuration, that have to be done by an administrator. Its just too much trouble.
Sure, it would be better practice to keep separate IDs. But this brings me back to my original point: the fact that I can't do "su" or equivalent means that NT is less secure than it might be because human beings (I am one you know) have better things to do with their time than save all their work, log out, log in, wait for Outlook to fire up, wait for Netscape to fire up, do whatever is needed, repeat.
So, mister clueless pratt, what are you going to do now?
So once again we have another VBS virus. But everyone on/. keeps ranting about the evils of VBS, the same thing could be done in Perl, or any other unix scripting lang.
The point is not the scripting language, its the fact that emailed executables can be run without the user's intent, and with full user privileges. On a W98 machine that means full root privilege because there isn't anything else, and even on my NT machine I have given my normal user ID full admin priviledge because it takes about 3 minutes to log out and log back in again, and there is no equivalent to "su".
Compare this with, say, the Gnome 1.2 install. To initiate the installation I have to be logged on as root (and because "su" is so quick and easy its practical for me to stay in user mode most of the time). Then I have to cut and paste a long command line from the web page, because on Unix boxen there is simply no other way of getting the program run outside a sandbox. On Windows it would be "click here" to run a VBS script.
This makes trojans and their relatives much harder to produce in Unix than in Windows.
I'd have thought that class action suits would be an effective deterrent. Its fairly easy to track misuse of marketing data if you try: just give a false name, and log who you give it to. So once someone has evidence that his name has been sold on in violation of the rules, it follows that lots of other people have as well. These people are a Class in whose name a class action suit can be brought. And the defendent has a handy list of their names and addresses too...
In effect this privatises the enforcement side. All it takes is a few lawyers who make a practice of signing up for things under false names and tracking the resulting spam. When they find a violation they can sue and pocket a fee.
This leaves open two issues:
Establishing damage. Its hard to argue that you are damaged by receiving snail spam. You might be able to sue for the cost of reading it, deciding its not important, and disposing of it. But if you price your time at $20/hour (probably too optimistic), and it takes you 1 minute to open the envelope, scan it and decide you don't want it, then thats just 33 cents per "victim". Even with punitive damages its still only a dollar. This probably requires a separate law.
Setting rules which can then be enforced. This could be done by government, but it could also be done by market pressure if we can just educate people. The people to do this educating are the same lawyers who will be bringing the class action suits. They just need to tell people to look for the TrustE logo on a web site, or whatever.
I notice the article doesn't say anything specific about the amount of time and energy required to write to this stuff.
AIUI the reading is done by simply measuring the resistance of a tiny cell of this amorphous material, and that would be pretty fast and easy. But writing to it requires a phase change induced by heating. That sounds slow and expensive by comparison.
Add this to the problem of the memory "wearing out" after a few weeks of being toggled at 10MHz and it looks more like a replacement for Flash memory than for DRAM.
One thing about the OS390 port is that now you can say to management
Linux scales from a single floppy disk right up to an OS390 mainframe, and can do useful work at both ends. A Linux-based router or simple firewall will fit on a floppy, and IBM now support Linux on OS390.
If so then I don't think this is the way to do it. MS have pitched this issue on plain copyright law, where they are (sorry everyone) squarely in the right. They wrote the Kerberos extensions spec, and they own the copyright. Anyone who makes unauthorised duplications is therefore in the wrong.
On trade secret law their position is much less firm. A judge might allow their claim that something can be made public to anyone who bothers to look but still be called "secret", but I rather doubt it. Therefore this is not the place where the software industry is going to make a stand for UCITA. Too much risk of losing. If you want a test case you make it out of cast iron, with yourselves unambiguously in the role of goodies. Make it a software pirate with stacks of copied CDs in a police raid, and have the evidence for wrongdoing rest on a clickthrough license. Don't do a David vs Goliath act, because the judge is apt to side with David.
First, all those flaming on about free speech are missing the point. The MS letter makes this a copyright issue, and there is no doubt that Slashdot is in violation of copyright by hosting and distributing copies of the document. Whether MS would be entitled to any significant damages is another question, and it is possible that in the absence of significant damages a lawsuit would be ruled vexatious. But I am not a lawyer.
However, this is orthogonal to the original issue, which was that the contents of the document were a trade secret. Trade secret protection extends beyond copyright, and its pretty clear that MS want to effectively prevent anyone from interworking with this aspect of their Kerberos service without their permission. Whether they can make TS status stick on a click-through license is another question.
Meantime, where does this leave a hypothetical future version of Samba which would interwork with MS Kerberos? If MS can claim trade secret status then true open source would be impossible. But I suspect that if you wrapped the source code up in a click-through license then MS would not be able to claim violation of the trade secret. Copyright might be a bit trickier: most of the MS document appears to be source code from a.h file, so simply copying that code into your own.h file would violate their copyright. However if you rephrased their source code, using different names but equivalent data structures, then I think that would be OK.
The thing about English is that it is already the de-facto world standard language, and has been for some years. I'm embarrased about this because I am English, as well as anglophone, so I don't have to learn another language. I'm also relieved because I am very bad at learning other languages. I do fine until I hit the irregular bits, and then my mind rebels. However thats by the by.
The thing that kills any other language, including nice regular ones like Esperanto and Lojban, is simply the network effect. Learn Esperanto and you can speak to a few (tens of?) thousand like-minded enthusiasts around the world. Learn English and you can get by just about anywhere with a capitalist economy because the locals all learn English as their second language. I once tried learning Esperanto, but gave up because there was simply nothing out there to read in it beyond newsgroups written by other Esperanto enthusiasts.
Back to network effects: we all know why MS Windows and Office have become the de-facto standard systems on 90+% of the worlds PCs. English will become the world language for exactly the same reason.
"DX.exe has caused a general protection fault is module kernel32.dll" - not very friendly. Or helpful. It then goes on to say that if the problem continues, contact the vendor. Not helpful, as I don't really know what went wrong.
This is simple buck-passing. The "vendor" is not usually MS, its the poor guy who sold you the computer. MS OEM licenses explicitly prohibit you from passing MS telephone numbers to the customer, so the customer can only telephone the guy who sold the system, and will consequently tend to aim the blame that way as well.
The article does not mention FreeBSD or TrustedBSD. Both of these make a big thing of security, including reviews of software. TrustedBSD is even going for Orange Book B1 certification.
ISP's have lost the fight and can now be held responsible for information found on their servers. Only after they know it is there. The theory is that if you knowingly publish defamatory information then you are contributing to the damage done, and hence should be liable along with everyone else doing the same thing. OTOH if you merely carry a stack of books to a bookshop with no knowledge of their contents then you are not liable.
Accordingly, ISPs are only liable for distribution which happens after they have been informed of the defamation. Once they are aware of it they can take a decision, based on the facts, as to whether to carry on distributing the information or to pull it.
In practice of course the ISPs are in no position to make such a decision, and have to pull almost anything that anyone complains about in simple self defence.
It could be worse though. AIUI bookshops are assumed to be aware of every word in every book they sell, and hence are liable for damage done even before they have been notified. ISPs have escaped this fate, but it was a close thing.
Whilst I agree that the libel law situation here is dumb, its not actually having much effect on the web. Saying that UK libel law makes the Internet "the easiest medium to censor ever" is certainly overstating it. The Slashdot article itself said that the solution for targetted sites is easy: just up sticks and move to the US. You may need a new URL (and your old ISP won't host a pointer because that makes it part of the publication), but thats the only real problem. Apart from that, its just a matter of juristiction shopping.
Of course the site authors are still just as vulnerable to lawsuits as ever, but in practice the plaintiff has to find them first. If you go to a US host in the first place then that would mean extracting customer records from the US company.
Overall its the same old mechanism: censorship is damage, so route around it.
Last time I was at the cinema, watching Toy Story 2, they had a trailer for a new Aardman feature film called "The Chicken Run" about a load of factory-farm chickens planning a Great Escape (shot of chicken lying down on a little cart in a tunnel and tapping it twice with a trowel). Paul.
Mattel might sue the hackers for breach of contract. This puts them back in court, which is where both hackers and Mattel obviously don't want to be.
Mattel might have difficulty in arguing that the hackers are in breach of contract. Mattel were perfectly well aware of the wide distribution of the software, and if they have read it they should have been aware of the terms it was distributed under. Hence a court might (IANAL) hold that there was an implied term excepting the GPL licenses from that declaration. At the very least, the lack of due care on the part of Mattel to check this out this should count against any award for damages for breach of contract.
I don't know about US law, but UK law tends to be biased against booby traps in the small print of contracts. What counts is what was in the minds of the parties at the time they signed. Written contracts are merely evidence about what this state of mind was, rather than an absolute definition. Courts are quite free to ignore the terms of a written contract if there is evidence that one of the parties did not know about the terms, and would not have agreed if they did. I don't know if US contract law works the same way.
I think this is a good idea, provided that limits are placed on this non-disclosure. For example an embargo of 48 hours to give the security people a head start over the script kiddies could be a very good idea.
On the other hand if the script kiddies already have an exploit for a hole then not telling sysadmins about the problem is obviously counterproductive.
So, limit it to 48 hours, and only apply the embargo if the knowledge is not already available in the cracker community.
That network - bequeathed to Telstra for free - was paid for by the Australian taxpayer, an investment of billions.
Hang on a minute, you can't sell something twice. At present (AIUI) Telestra is a government-owned telco. The government hasn't given away the infrastructure, its just packaged up a bit of itself and called that package Telestra. Telestra will then be sold, and at that point the government gets the money.
We issue - quarterly, say - "bandwidth bonds". A bandwidth bond entitles the holder to a certain percentage of network carrying capacity over a certain time period.
Thats an interesting idea. The problem is going to be in administering it. What happens when everybody wants to use their bandwidth over one link? Sell off bandiwdth over particular links as individual commodities? Maybe. Take a look at what Band-X are doing with this on international routes.
Slashdot Mirror
Full 3d is going to require nano-fabrication techniques: no form of lithography will cut the mustard. While painting of this kind is not the full solution, its definitely a step in the right direction.
There are of course many other challenges in 3d circuitry, cooling not least.
ISTR some work done by IBM with tiny balls that had a few hundred devices on the surface of each and were packed into a cubic array. Does anyone know where that went?
Paul.
Paul.
Well... they are totally different.
Consumer RFI is not going to throw a 747 into a spin. The danger is that it will interfere with navigation sufficiently to cause an accident. For instance, missing the runway on an ILS approach.
A lighting strike is going to produce enough RFI, and even EMP, to interfere with pretty much anything electronic unless it is very well shielded. But its over quickly. Consumer RFI can last the entire flight.
Paul.
Some years ago I was involved in a project looking at this kind of set up.
The problems are huge, and EMI is actually one of the more minor problems. You mostly solve it by using fibre or good co-ax for the cable runs and faraday cages for the electronics.
The big danger is fire. Every piece of equipment has to be certified to ensure that it won't start a fire, and if one does start that the equipment won't make things worse. Cabling is the big headache here: a cable conduit full of PTFE makes a wonderful channel for fire to spread and also creates lots of poisonous smoke. Cables need to be specially rated, as do the connectors.
Then the equipment must be safe in a crash. No broken glass, even when a passenger's head hits it. It also must not be able to fly out of the seat and hit the person in front.
It has to be cooled, even though the seatback component is surrounded by a good insulator. But at the same time it has to withstand Junior pouring his orange juice over it, and curious passengers with pocket toolkits (hello, you know who you are...)
It has to withstand vibration, pressure changes and temperature extremes (aircraft may be left parked in the hot sun or freezing cold). Components are rated for this sort of thing of course, but aircraft operation tends to put you in a corner of the envelope, and failures are therefore much more likely.
Finally you are on a strict power and weight budget.
Overall, a challenging collection of issues.
Paul.
Having said that, I always look through it anyway, just to be on the safe side. You also get asked to sign things to say "I have read and understand the page of fine print overleaf". I don't know what the legal standing of such a declaration is. Whenever asked to sign one of these I make the salesdroid wait while I read it. Maybe I should read it aloud.
Disclaimer: I am not a lawyer.
Paul.
The problem isn't Monsanto, its the corrupt government that lets its officials be brought. Fix the government and the corps will behave themselves.
yes there may have always been cancer - but I don't think it has ever been so wide spread.
The reason that more people get cancer these days is merely that they are failing to die of other things, like cholera, polio, smallpox, war, famine and childbirth.
Go and look at some trading standards cases from a few hundred years back. Chalk dust in flour was common, and sweets were dyed with heavy metal salts. I kid you not. Fancy a return to those days? Whereas these days we have the luxury of getting annoyed about GM pollen.
Paul.
- It avoids a court known to be more friendly to MS than Judge Jackson
- It shortens the process by a year or two.
But I can't see anything in any of the news stories on this.Paul.
Well, I don't normally bother responding to flamebait, but on this occasion...
I'm not totally dumb. In addition the corporate firewall I also run AtGuard (now part of Norton IIRC) which keeps an eye on any active content from the Web, and should also trap anything unauthorised that tries to send information back out. Also I don't run Internet Explorer, so the worst idiocies of ActiveX are not an issue for me.
On top of that we also have a clued in administrator (I am not an administrator, I just look after my desktop box) who keeps our virus checker up to date and does threat monitoring on the servers. ILOVEYOU didn't get in here.
Finally, I spent about six months trying to run with separate user and admin IDs, and believe me it just wasn't worth the hassle. There are so many little jobs, from defragging the hard drive to updating the IP configuration, that have to be done by an administrator. Its just too much trouble.
Sure, it would be better practice to keep separate IDs. But this brings me back to my original point: the fact that I can't do "su" or equivalent means that NT is less secure than it might be because human beings (I am one you know) have better things to do with their time than save all their work, log out, log in, wait for Outlook to fire up, wait for Netscape to fire up, do whatever is needed, repeat.
So, mister clueless pratt, what are you going to do now?
Paul.
The point is not the scripting language, its the fact that emailed executables can be run without the user's intent, and with full user privileges. On a W98 machine that means full root privilege because there isn't anything else, and even on my NT machine I have given my normal user ID full admin priviledge because it takes about 3 minutes to log out and log back in again, and there is no equivalent to "su".
Compare this with, say, the Gnome 1.2 install. To initiate the installation I have to be logged on as root (and because "su" is so quick and easy its practical for me to stay in user mode most of the time). Then I have to cut and paste a long command line from the web page, because on Unix boxen there is simply no other way of getting the program run outside a sandbox. On Windows it would be "click here" to run a VBS script.
This makes trojans and their relatives much harder to produce in Unix than in Windows.
Paul.
Yeah, but it also mucks up your perception of distance and velocity. This isn't going to do much for your gaming.
(Now awaiting a slew of "I got all my high scores while high" postings).
Paul.
In effect this privatises the enforcement side. All it takes is a few lawyers who make a practice of signing up for things under false names and tracking the resulting spam. When they find a violation they can sue and pocket a fee.
This leaves open two issues:
Paul.
AIUI the reading is done by simply measuring the resistance of a tiny cell of this amorphous material, and that would be pretty fast and easy. But writing to it requires a phase change induced by heating. That sounds slow and expensive by comparison.
Add this to the problem of the memory "wearing out" after a few weeks of being toggled at 10MHz and it looks more like a replacement for Flash memory than for DRAM.
Nevertheless, it sounds pretty cool.
Paul.
On trade secret law their position is much less firm. A judge might allow their claim that something can be made public to anyone who bothers to look but still be called "secret", but I rather doubt it. Therefore this is not the place where the software industry is going to make a stand for UCITA. Too much risk of losing. If you want a test case you make it out of cast iron, with yourselves unambiguously in the role of goodies. Make it a software pirate with stacks of copied CDs in a police raid, and have the evidence for wrongdoing rest on a clickthrough license. Don't do a David vs Goliath act, because the judge is apt to side with David.
Paul.
However, this is orthogonal to the original issue, which was that the contents of the document were a trade secret. Trade secret protection extends beyond copyright, and its pretty clear that MS want to effectively prevent anyone from interworking with this aspect of their Kerberos service without their permission. Whether they can make TS status stick on a click-through license is another question.
Meantime, where does this leave a hypothetical future version of Samba which would interwork with MS Kerberos? If MS can claim trade secret status then true open source would be impossible. But I suspect that if you wrapped the source code up in a click-through license then MS would not be able to claim violation of the trade secret. Copyright might be a bit trickier: most of the MS document appears to be source code from a .h file, so simply copying that code into your own .h file would violate their copyright. However if you rephrased their source code, using different names but equivalent data structures, then I think that would be OK.
I await the legal advice with interest.
Paul.
The thing that kills any other language, including nice regular ones like Esperanto and Lojban, is simply the network effect. Learn Esperanto and you can speak to a few (tens of?) thousand like-minded enthusiasts around the world. Learn English and you can get by just about anywhere with a capitalist economy because the locals all learn English as their second language. I once tried learning Esperanto, but gave up because there was simply nothing out there to read in it beyond newsgroups written by other Esperanto enthusiasts.
Back to network effects: we all know why MS Windows and Office have become the de-facto standard systems on 90+% of the worlds PCs. English will become the world language for exactly the same reason.
Paul.
This is simple buck-passing. The "vendor" is not usually MS, its the poor guy who sold you the computer. MS OEM licenses explicitly prohibit you from passing MS telephone numbers to the customer, so the customer can only telephone the guy who sold the system, and will consequently tend to aim the blame that way as well.
Paul.
Paul.
Accordingly, ISPs are only liable for distribution which happens after they have been informed of the defamation. Once they are aware of it they can take a decision, based on the facts, as to whether to carry on distributing the information or to pull it.
In practice of course the ISPs are in no position to make such a decision, and have to pull almost anything that anyone complains about in simple self defence.
It could be worse though. AIUI bookshops are assumed to be aware of every word in every book they sell, and hence are liable for damage done even before they have been notified. ISPs have escaped this fate, but it was a close thing.
Paul.
Of course the site authors are still just as vulnerable to lawsuits as ever, but in practice the plaintiff has to find them first. If you go to a US host in the first place then that would mean extracting customer records from the US company.
Overall its the same old mechanism: censorship is damage, so route around it.
Paul.
Last time I was at the cinema, watching Toy Story 2, they had a trailer for a new Aardman feature film called "The Chicken Run" about a load of factory-farm chickens planning a Great Escape (shot of chicken lying down on a little cart in a tunnel and tapping it twice with a trowel). Paul.
Mattel might have difficulty in arguing that the hackers are in breach of contract. Mattel were perfectly well aware of the wide distribution of the software, and if they have read it they should have been aware of the terms it was distributed under. Hence a court might (IANAL) hold that there was an implied term excepting the GPL licenses from that declaration. At the very least, the lack of due care on the part of Mattel to check this out this should count against any award for damages for breach of contract.
I don't know about US law, but UK law tends to be biased against booby traps in the small print of contracts. What counts is what was in the minds of the parties at the time they signed. Written contracts are merely evidence about what this state of mind was, rather than an absolute definition. Courts are quite free to ignore the terms of a written contract if there is evidence that one of the parties did not know about the terms, and would not have agreed if they did. I don't know if US contract law works the same way.
Paul.
At http://www.zdnet.co.uk/news/2000/ 12/ns-14378.html Paul.
On the other hand if the script kiddies already have an exploit for a hole then not telling sysadmins about the problem is obviously counterproductive.
So, limit it to 48 hours, and only apply the embargo if the knowledge is not already available in the cracker community.
Paul.
Hang on a minute, you can't sell something twice. At present (AIUI) Telestra is a government-owned telco. The government hasn't given away the infrastructure, its just packaged up a bit of itself and called that package Telestra. Telestra will then be sold, and at that point the government gets the money.
We issue - quarterly, say - "bandwidth bonds". A bandwidth bond entitles the holder to a certain percentage of network carrying capacity over a certain time period.
Thats an interesting idea. The problem is going to be in administering it. What happens when everybody wants to use their bandwidth over one link? Sell off bandiwdth over particular links as individual commodities? Maybe. Take a look at what Band-X are doing with this on international routes.
Paul.