I was describing this idea back in 1995, and claim no originality for it. This is the NC or Network Computer model.
People talk about the TCO of a PC, such that only about 20% of the total cost of ownership is the cost of the hardware, over the life of the device.
If you break the PC down further, you end up with three functional components, without any of which the device won't operate:
CPU
RAM
HDD
If you then look at where the TCO costs break down over those three components, nearly all of them are incurred by the presence of the HDD. Specifically, it's all the futzing around with the operation system, applications and user data. Additionally, failure rates of HDDs are significantly higher than CPUs or RAM, again incurring higher costs.
Get rid of the HDD, make the PC boot off of the CD-ROM drive, or even better, off of the network, and a significant portion of the TCO disappears.
Some people will argue that you must have a HDD in case the network goes down. Well, what you do is spend some of the money that you save by not having a HDD and the associated maintenance costs, on making the network more reliable. It is still going to be cheaper at the end of the day.
This is a great technical solution, and has been possible at least since 1991 - you could boot PCs off of the network that many years ago.
However, it has never taken off. Why ? Pretty much Microsoft. They make a lot of money from what's put on PC HDD, and they have had such influence over the industry direction that when they came out against it, the Network Computer model never really took off.
The philosophical lesson to learn from this is
the best technical solution isn't necessarily the one that is successful
If you want your best technical solution to succeed, you need to learn to, and work out how to best market it.
" You seriously cannot expect NVidia to release their IP simply so a handful of people can hack around with the Linux drivers. 99.99% of us really don't give a damn that we can't modify the drivers - only that they work."
I'd think 99.99% figure is way too high - do you have some sort of survey results ? There are a huge number of Linux users who care whether their hardware has GPL drivers or not.
I'd be of the opinion it is only the people who have started using Linux in the last 3 or so years that don't care. I'd suspect most of them are NVidia owners.
Linux has been available since at least 1992 - which was when I started using it. Subtracting 3 years from 2003, that leaves an 8 year period, where the majority, if not all Linux users have had to use, rely on, and come to believe in OSS device drivers.
Linux is the result of a philosophy - OSS. If you don't believe in the philosophy, I struggle to understand how you can be comfortable benefitting from it.
"NVidia develops a driver for Linux, that is better than 99% of the companies out there."
There are a number of companies supporting Linux drivers - and they release GPL code. Intel is great example.
"We can not demand that companies release their products open source, but we can support the companies that actually choose to take the time to develop drivers for Linux."
We can demand that hardware companies release their programming info. Whether they do or not is up to them. If they don't, we should refuse to buy.
Is your car bonnet welded shut ? Or can you open it up and play with it ? Since you own it, that is, you aren't licensing it from the car manufacturer, you should be able to do what ever you like with it, including pouring sugar in the petrol tank. If you choose to, that is up to you. The car manufacturer doesn't care what you do with their car, as it's yours - they have made their money. Set it on fire for all they care.
You own your video card, you don't license it from NVidia. You should be able to have access to the programming information, so that you can do what ever you like with it.
"I do not own an ATI card, so I'm not sure how their Linux driver support is, but from their homepage I get their impression that they link to an OSS project instead of distributing their own driver."
I'm not sure how that is a bad thing. OSS is good. The Linux kernel was developed using OSS methods. Are you somehow disagreeing with OSS methods, yet at the same time happy to benefit from OSS products such as Linux ?
What is wrong with OSS drivers ? If they failed on you, did you go looking for a solution ? Did you report the fault to the Linux Kernel Mailing List, or report it to the driver maintainer ?
"This I think hurts the OSS community, that hardware companies get the impression: "We don't have to write drivers for Linux, they can do that themselves".
Excellent. I'm happy with that. OSS drivers are usually better. And if they aren't, somebody, in fact, anybody, with access to the source code, can improve them. Everybody has access to the source code, so they are a potential candidate to improve them. Even you could if you have the right programming talent and skills.
"When I spend my money I want a complete product, not something that requires me to run Windows or unsupported 3rd party linux drivers."
I'd like you to think about what happens to the NVidia binary drivers once the NVidia hardware becomes obselete, say in 3 years time. NVidia won't continue to maintain binary drivers forever - it costs them money. Eventually they won't want to wear the continued support costs, and just stop developing them or fixing bugs. You will end up with "unsupported 3rd party linux drivers".
If they open up the hardware specs, the community can take over. If they open up the specs from day one, the community is better prepared and more willing to take over, because they know and are happy with the code base, because they wrote it.
This isn't just hypothetical, if the driver is GPL'ed, the kernel developers are able to fix problems, as they have the source code.
The traceback messages you mention are the ones that would be used to help fix the problems.
It's even better if the hardware specs are open, because they can make the driver work with the hardware correctly, rather than trying to make good guesses.
The moment end-to-end encryption and authentication is enabled, either via tunnels or by just encrypting the IP payload, no authority trying to assert control over VoIP will be able to identify one application verses another e.g., VoIP verses HTTP verses SMTP.
They will have to either ban encryption, or ban all applications, which is the equivalent of banning the Internet.
Deploying encryption in this manner will actually restore the Internet to its original design - an application agnostic network, whose sole job is to just make a best effort to deliver bits between the hosts at the edges. Only the hosts should know and will know what applications the Internet is being used for.
The technology already exists, albeit in early forms:
DNSSEC
Opportunistic tunnel setup within IPsec
This will also obselete firewalls, proxy servers, NAT, and any other devices that perform applications processing within the Internet. The only applications processing devices left will be those at the edges. Security, aka firewalling for example, will be deployed on each edge device.
Steve Bellovin (one of the Wily Hacker authors) wrote about distributed firewalls in 1999, here :
Distributed Firewalls
Slashdot Mirror
ment company.
Their logo is three tuning forks.
I'm a proud owner of a Yamaha YAS-62 Alto Saxophone.
and then spread it around as though it was a legitimate document.
I prefer better, even it it takes a bit longer.
"All good things come to those who wait."
Have you mentioned it on LKML ?
They need to know about your problems so they can have a go at fixing them./p.
Too treu
didn't they ?
Isn't they why they killed a lot of people ?
I very much doubt ARIA are so powerful as to be able to get cryptography banned. Maybe RIAA are, but certainly not ARIA.
It's such an evil, evil thing to do to anybody, even your worst enemy.
I was describing this idea back in 1995, and claim no originality for it. This is the NC or Network Computer model.
People talk about the TCO of a PC, such that only about 20% of the total cost of ownership is the cost of the hardware, over the life of the device.
If you break the PC down further, you end up with three functional components, without any of which the device won't operate :
If you then look at where the TCO costs break down over those three components, nearly all of them are incurred by the presence of the HDD. Specifically, it's all the futzing around with the operation system, applications and user data. Additionally, failure rates of HDDs are significantly higher than CPUs or RAM, again incurring higher costs.
Get rid of the HDD, make the PC boot off of the CD-ROM drive, or even better, off of the network, and a significant portion of the TCO disappears.
Some people will argue that you must have a HDD in case the network goes down. Well, what you do is spend some of the money that you save by not having a HDD and the associated maintenance costs, on making the network more reliable. It is still going to be cheaper at the end of the day.
This is a great technical solution, and has been possible at least since 1991 - you could boot PCs off of the network that many years ago.
However, it has never taken off. Why ? Pretty much Microsoft. They make a lot of money from what's put on PC HDD, and they have had such influence over the industry direction that when they came out against it, the Network Computer model never really took off.
The philosophical lesson to learn from this is
the best technical solution isn't necessarily the one that is successful
If you want your best technical solution to succeed, you need to learn to, and work out how to best market it.
You cannot achieve perfect security. It is impossible. You can only aim for it.
The Debian project will not only retain their credibility, but I'd suggest they'll improve it by
if the programming specs for the hardware were published ... oh that's right, they aren't available.
fortunately we've managed to send them your way :-)
I'm Australian, and damn, that's the truth ...
magnatune.com
" You seriously cannot expect NVidia to release their IP simply so a handful of people can hack around with the Linux drivers. 99.99% of us really don't give a damn that we can't modify the drivers - only that they work."
I'd think 99.99% figure is way too high - do you have some sort of survey results ? There are a huge number of Linux users who care whether their hardware has GPL drivers or not.
I'd be of the opinion it is only the people who have started using Linux in the last 3 or so years that don't care. I'd suspect most of them are NVidia owners.
Linux has been available since at least 1992 - which was when I started using it. Subtracting 3 years from 2003, that leaves an 8 year period, where the majority, if not all Linux users have had to use, rely on, and come to believe in OSS device drivers.
Linux is the result of a philosophy - OSS. If you don't believe in the philosophy, I struggle to understand how you can be comfortable benefitting from it.
"NVidia develops a driver for Linux, that is better than 99% of the companies out there."
There are a number of companies supporting Linux drivers - and they release GPL code. Intel is great example.
"We can not demand that companies release their products open source, but we can support the companies that actually choose to take the time to develop drivers for Linux."
We can demand that hardware companies release their programming info. Whether they do or not is up to them. If they don't, we should refuse to buy.
Is your car bonnet welded shut ? Or can you open it up and play with it ? Since you own it, that is, you aren't licensing it from the car manufacturer, you should be able to do what ever you like with it, including pouring sugar in the petrol tank. If you choose to, that is up to you. The car manufacturer doesn't care what you do with their car, as it's yours - they have made their money. Set it on fire for all they care.
You own your video card, you don't license it from NVidia. You should be able to have access to the programming information, so that you can do what ever you like with it.
"I do not own an ATI card, so I'm not sure how their Linux driver support is, but from their homepage I get their impression that they link to an OSS project instead of distributing their own driver."
I'm not sure how that is a bad thing. OSS is good. The Linux kernel was developed using OSS methods. Are you somehow disagreeing with OSS methods, yet at the same time happy to benefit from OSS products such as Linux ?
What is wrong with OSS drivers ? If they failed on you, did you go looking for a solution ? Did you report the fault to the Linux Kernel Mailing List, or report it to the driver maintainer ?
"This I think hurts the OSS community, that hardware companies get the impression: "We don't have to write drivers for Linux, they can do that themselves".
Excellent. I'm happy with that. OSS drivers are usually better. And if they aren't, somebody, in fact, anybody, with access to the source code, can improve them. Everybody has access to the source code, so they are a potential candidate to improve them. Even you could if you have the right programming talent and skills.
"When I spend my money I want a complete product, not something that requires me to run Windows or unsupported 3rd party linux drivers."
I'd like you to think about what happens to the NVidia binary drivers once the NVidia hardware becomes obselete, say in 3 years time. NVidia won't continue to maintain binary drivers forever - it costs them money. Eventually they won't want to wear the continued support costs, and just stop developing them or fixing bugs. You will end up with "unsupported 3rd party linux drivers".
If they open up the hardware specs, the community can take over. If they open up the specs from day one, the community is better prepared and more willing to take over, because they know and are happy with the code base, because they wrote it.
This isn't just hypothetical, if the driver is GPL'ed, the kernel developers are able to fix problems, as they have the source code.
The traceback messages you mention are the ones that would be used to help fix the problems.
It's even better if the hardware specs are open, because they can make the driver work with the hardware correctly, rather than trying to make good guesses.
The moment end-to-end encryption and authentication is enabled, either via tunnels or by just encrypting the IP payload, no authority trying to assert control over VoIP will be able to identify one application verses another e.g., VoIP verses HTTP verses SMTP.
They will have to either ban encryption, or ban all applications, which is the equivalent of banning the Internet.
Deploying encryption in this manner will actually restore the Internet to its original design - an application agnostic network, whose sole job is to just make a best effort to deliver bits between the hosts at the edges. Only the hosts should know and will know what applications the Internet is being used for.
The technology already exists, albeit in early forms :
This will also obselete firewalls, proxy servers, NAT, and any other devices that perform applications processing within the Internet. The only applications processing devices left will be those at the edges. Security, aka firewalling for example, will be deployed on each edge device.
Steve Bellovin (one of the Wily Hacker authors) wrote about distributed firewalls in 1999, here : Distributed Firewalls
My post on how to get around it
clickEven better, have they stopped you from running another email client to send a typed in copy ? I doubt it, although this _is_ Microsoft.
Much simpler than screen capping etc.
Or have Microsoft stopped you from creating a new email while you are reading one you have been sent ?
jabber
It used to be that almost everybody who read slashdot would know who Dennis is.
Now it appears that almost nobody does.
What is this world coming to ?
RIAA free music.
And I live in Australia !