Since the issue is software blocking protected speach, I'd like to see the courts put a restriction on the software that it not block any protected speach. With the current state of technology, this would effectively kill any censorware for use under this law, but would be clearly in keeping with the first amendment. Furthermore, it would open up the possibility of putting censorware companies trying to get their products in libraries at risk of lawsuits, and thus give them a harder time making censorware for home use which blocks protected speach (but which is legal for private use), since they'd have to maintain and justify a separate list of protected speach their home-use product blocks anyway.
Of course, porn sites, or even sites with explicit content of various other sorts, frequently identify themselves as such, either by actually requiring age verification or by having a click-through page saying you have to be in a place that permits viewing such things. If those sites simply sent a header to identify themselves as such, it could be enforced by browsers in places where such content is, in fact, prohibited. I haven't actually surveyed the front pages of porn sites, so I don't know how effective it would be, but this would avoid deep-linking problems and actually make those warning pages meaningful.
If you could install this, get it happy, run some script to syncronize things, and then have it such that you could just reboot and switch from Windows+LINE to Linux+WINE and have the system run exactly the same with respect to apps, it would make a very nice switchover.
In particular, you could make sure that your important apps never stop working as you transition, because you could install and test WINE while still running Windows; once everything works, you can just switch which kernel you're actually running and which you're emulating, and keep working.
If it were set up to dump its whole state to swap space, then everything would work fine for recovereing the state quickly and on an as-needed basis. Of course, the kernel would probably have to do through its boot sequence anyway, because devices generally have to be initialized, so this probably means that, with a good BIOS, you'd be 3 seconds to where you were before shutting down instead of 3 seconds to single-user-mode.
If it's got content-control and a variety of key possibilities, and can have data stored on it, can I put my secret files on this and give keys only to people I want to be able to read it?
And if a court demands a key, can I sue them under the DMCA and have search warrents and government cracking tools declared tools for piracy?
There's got to be a problem when the government is trying to keep people from enchanging information without letting other people read it and simultaneously trying to keep people from reading information while letting other people distribute it.
but somewhat surprised. It just goes to show that, for sites that don't try too hard to support broken 4.x browsers, a good, solid, reasonably standards-compliant browser will work well. Of course, the browser I'm using is netscape 3.0 without javascript, which doesn't seem to be on the lists of browsers to upgrade to.
With all the hype about standards compliance, it's nice to see them not using non-standard tricks to get rid of browsers they don't like.
Just don't support them. Use the standards and ignore the fact that the result won't look right on certain browsers. Furthermore, don't code specifically for broken new browsers. For the page to actually be standards-compliant, it can't really treat even the latest browsers specially.
If you actually follow the standards, the only people who won't have the site work are those whose browsers are not supporting the foolishly-designed standards.
NASA is going to do all future space exploration with NEAR.
NASA: Can we please have funding for another mission? We could build this great space craft and...
Congress: What's wrong with NEAR? Can't you keep using that?
NASA: We used NEAR last time. We put it on EROS and then we found there wasn't anything to do with it once it was there, so we took it back off. Can't we have one with more toys on it?
It seems like the situation of having a monolithic tree controlled by Linus with substantial portions that he doesn't have any personal use for is somewhat silly.
It leads to situations where he doesn't want to allow big patches that he can't look at to see if they are good, but the unpatched tree doesn't work at all.
Furthermore, it means the kernel source tree contains >16M of code specifically for processors that are not going to be used in a particular installation.
Why not have each arch subdirectory in a separate tarball, which follows the core kernel but doesn't go through Linus at all, going instead through a maintainer who actually works on that platform?
If the only current kernel that works on PPC is the fork, why doesn't kernel.org carry that instead of the portion of the official kernel that is specific to PPC but doesn't work?
The current status of 2.4.x is "*we* think it's stable, but massive testing might reveal something". 2.4.1 is stable for the kernel developers, but there are a lot of combinations of hardware out there that the kernel developers don't have access to, so there could be really subtle (one in a million machines, e.g.) bugs.
The 2.5 series will start when there's nothing left to do on 2.4.x, essentially. Presently, people are looking for bug reports on 2.4.x stuff and fixing problems found by distribution-makers and end users who wait for stable kernels. Then the developers will get some sleep for a bit, and then figure out what changes should go into 2.5
I think a significant effect which needs to be taken into account is that PDAs can remind you of information that you would have forgotten that you once knew. So they cause there to be more information that you are aware that you've forgotten, even if they do not make your actual memory any worse.
For example, I used to know what times certain places close. Now I still remember those, but also have in my PDA the hours of more places, including openning times and more variation based on day of week. This is a whole bunch of information which I have come in contact with and forgotten due to my PDA-- but the role of the PDA was making me pay attention to it in the first place, not making me forget it afterwards.
The question boils down to what is part of the game and what is not.
The main play of a basketball game is not zero-sum, since each team makes progress. But including the win/lose descision at the end makes it a zero-sum game. But including having fun as a goal makes it not zero-sum again. Games theory applies to what makes sense to do within a framework of rules, but does not define what constitutes "the game being played". Anything which produces some variation can be made into a zero-sum game by having a cutoff to determine a winner.
What the original question should have asked, more preceisely, is what games promote strategies characteristic of non-zero-sum games. I would say that games where players tend to interact directly in ways that benefit those interacting (at the eventual expense of the non-participants) have this aspect: trading games (Res Publica, Settlers of Catan), the >2 player iterated prisoner's dilemma, and so forth.
The part that makes many of these games zero-sum is the comparison at the end. This step means that subtracting from each team's score the average of the scores doesn't change the result, since the result you care about is just the direction of the difference.
But it's not a zero-sum game when the teams want to be able to say that it was a good match afterwards. From a zero-sum perspective, it doesn't make sense to play against people who are not worse than you or to give worse players an advantage, since both of these obviously reduce the chance that you'll win.
Go is zero-sum, but go with rankings and handicaps, considering the whole process and not just the play, is not. Both players take steps at the beginning to cause the game to be difficult but possible for both of them, and they have, as a longer-term goal, to improve their skill, and, as far as I know, nobody tries to give their opponents bad habits.
It's not so much the internal aspects of the game that determine whether it is zero-sum or not, but more the higher-level goals of the players.
Two more steps are required to make it pro-cooperation:
Play a number of rounds, each scored that way.
Play with more than two people. Your score is the sum of your scored with each other person. Each pair of people plays the same number of rounds.
This is pro-cooperation because if you and your partner both compete, or if you trade off competing and cooperating, you both fall behind pairs who both cooperated.
Of course, you'd probably want some variation to make it interesting...
I think it would be great if MS released a new version of the OS and had all the bugs fixed in a week. I'd want the patch to be free and not to generate a whole bunch of new bugs, though.
Releasing a new kernel version means that it will be tried on hardware that the developers don't have, even in the case of MicroSoft, so there's always a reasonable chance that some bug will emerge that couldn't have been anticipated. These bugs will never be discovered without a general release of an official stable version, since the people with the obscure hardware won't try anything else. But they should be fixed quickly once they're found.
MicroSoft service packs tend to include not only the fix for the major bug they're dealing with but also changes to the user interface and changes that break other people's software and make you upgrade your microsoft programs. Linux 2.4.1 will probably not break anything that worked under 2.4.0; at least, that was the idea for whether patches should be included.
It's amusing to see an article asking if Linus is killing Linux, because I'm used to seeing articles asking if the $2 billion industry is killing Linux.
The article is interesting in that it entirely fails to show that the actual industry has any problem with the current model of Linux development. I suspect if they could have gotten a quote like "We wish 2.4 had been released sooner, but Linus held it up" from anyone actually in the industry, they'd have used it. Instead they say that Linux should be controlled by a consortium of companies that show no evidence of wanting to do that.
The industry depends on being able to use a steadily improving piece of software in the products that they produce. If they knew how to run the development process better, they'd have done it before. They have enough to do to make their products without being directly involved in kernel development. The industry understands this, but some analysts seem not to.
It probably does take a significant chunk of power to support, say, making this post. You need power for my desktop box, power for the routers, power for the machines at slashdot, power for the servers at salon and the other mentioned sites, and the internet to connect me to all of them. Between all of these, it's a big total.
And that's just my internet. Everybody else has their own internet, which uses up... wait...
They seem to be failing to realize that all of the desktop boxes use a single overlapping internet, which means that, while a substantial amount of power is necessary to maintain my machine, that same power goes into maintaining hundreds of other machines.
Having a Pilot with internet use takes up close to the same amount of power that a desktop box takes up. Having both takes up about the same amount of power as just one.
The reason that servers and server farms exist is that it's more efficient for a group of people to share resources than for everyone to have their own. It only looks more when you charge the cost of the whole thing to each person.
I'm pretty sure the Playstation 2 has a hard drive in it. It's got these loading screens where you can hear the DVD drive spinning and the sound of a hard drive in use.
MS is certainly doing this the open way, at least for now. Of course, it's no more open than PC gaming, which is where they're coming from. The PS2 definitely suffers from insufficient testing on the games right now-- there are bugs in almost all of the ones I've seen which could not have been missed if anyone actually played the games through before the release.
I expect to see things turn more closed later. Perhaps the DVDs are encoded in some way that makes them inpossible to write without a license, so, while you know exactly how the box works inside, you can't get code into the actual hardware. Developer kits would be less strict, so you could write games, but you could only use them on consumer models if MS has approved.
This is essentially a product of using the same processor features you need for other performance gains (register renaming, out-of-order execution) better, with a small amount of extra state.
The first thing you do is get the OS out of the way by making the processor able to handle multiple threads at once without involving OS code in context switching. This is a definite win, because context switches done by the OS kill the pipeline and often kill the cache. If the processor is in charge of all of this, it can do better, because it is closer to what is going on.
After that, the only innovation is to use the out-of-order-execution support on the whole set of things the processor is doing, instead of just one thread.
Considering what they're already doing, this isn't a lot of new complexity, but it should help performance significantly. The main problem with out-of-order stuff has generally been that there isn't really all that much implicit parallelism within a thread. Adding other threads which do not, in general, have any dependancies gives the instruction scheduler much more to work with.
When everyone moves up a notch, that means that nobody in the department is doing what they were hired to do, or even what they've been getting good at. You'll have to train everyone, and the people who know the jobs are either gone or trying to learn other jobs themselves.
The best solution is to promote someone who wants the job, has the right skills for it, and is easier to replace than the original person. Failing that, it's better to high a new person at the top level and train them to work with the department than to mess with the existing structure of the department. Of course, that requires finding someone who can be new at the top level without destroying the department.
The existing community does a good job of writing code. IBM could provide additional developers with unusual experience, but this is on more of an individual level than a corporate level. One thing I think the corporation could do usefully is provide testing platforms. Linux developers generally can't tell how their code will do on massive SMP machines with huge memory spaces, huge disk arrays, and so forth, because they don't have the machines to try stuff. IBM ought to do testing for people on some of the really big boxes, so that ultra-high-end performance can be tuned.
For some reason, Silverman doesn't respond to the idea that, when connecting to a server for the first time, you cannot detect a man-in-the-middle attack.
This boils down to the identity problem: there doesn't have to be an actual server involved at all; you may be actually trying to log into the wrong machine, whether by mischief or by accident, and you'd be none the wiser because you don't know the difference between the different servers. You need to have some way of distinguishing the totally unfamiliar, but correct, server from the totally unfamiliar, but wrong, one.
Now there is a problem with most SSH clients: you might actually know the server you want from experience, but be using a computer which does not have access to this knowledge. You are expecting the client not to recognize the server, but you have more information than the client. In these cases, it would be helpful if the client would get the user involved in the authentication process. OpenSSH actually does this, by putting up for approval the server's key fingerprint, which a user can tell has changed even if the client has no information.
One of the main things that secure protocols are designed to stop is that man-in-the-middle attack.
The basic idea is to exchange session keys and verification data before a random public key. That way, the attacker can't substitute other public keys, because the attacker would have to be able to generate data based on information which has not yet been sent.
The article isn't actually talking about the described attack at all, but about spoofing servers. Identification of servers is a substantially different problem, and is a human-engineering one rather than a protocol one: what makes a server the one you are trying to connect to?
The attack is where the attacker wants to convince you that you're talking to the server you think you are by having you actually talk to that server, except that you are not talking directly. Most encrypted protocols do a good job of insuring that you are connected directly, over an encrypted channel to whatever it is you seem to be talking to. Whether the server you're talking directly to is the one you want to talk to is up to the user and other parts of the system.
What I expect, in the situation referred to by the article, is:
A MS Linux distribution which costs a bunch of money and comes with a bunch of Microsoft programs that don't work without their closed-source libraries. All of the other people's stuff they ship comes with original source. The latest version of Office for Linux + MS libraries, later possibly only for MS Linux.
If they have a way to make sure that users have to buy MS products, and they can require other MS products to have those work, there's no reason for them to care if Windows dies. I mean, if they suddenly dumped Windows and released Office for Linux (requiring the MS library set, of course), not only would people still be tied to MS products, they'd have to buy a new version of the OS yet again. Plus, MS doesn't have to deal with the Windows codebase any more; they don't have to pay anything to distribute the basic stuff, so they lose absolutely no profit from this switch.
They don't need the support of the community at all; they can go for the users who are using Linux because it's better; they give the impression that their distribution has all the power of Linux that you can get from other sources, plus you can use the authentic MS products which read the latest formats, etc.
Is there a good reason to continue to do Linux R&D? Sure; if Linux-based products progressively beats MS products in more and more fields, the ability to use MS products, whether on Windows or on Linux, becomes less of a big deal.
Consider this possibility: Intuit discovers that 8% of their customers won't buy the next version of Quick Books if they use any Whistler features that Wine doesn't support. They could follow the latest API, but then they automatically lose market share.
On the other hand, they could stick with the old API and have their product run, not only on the latest windows boxes, but also on machines that have been around a little while and on linux boxes.
If a significant portion of the windows boxes in use are not following the newest API, the only company for whom it makes sense to use that API is Microsoft, since anyone who wants to use a product that uses the API will have to pay Microsoft for the OS upgrade.
So the critical mass needed to make Wine beneficial to linux and a thorn in MS's side is the point at which the portion of customers using the windows API who aren't going to switch to the latest Microsoft version is significant; a much smaller mass than one might think, especially with the problems MS has had at making upgrading safe and affordable.
Slashdot Mirror
Since the issue is software blocking protected speach, I'd like to see the courts put a restriction on the software that it not block any protected speach. With the current state of technology, this would effectively kill any censorware for use under this law, but would be clearly in keeping with the first amendment. Furthermore, it would open up the possibility of putting censorware companies trying to get their products in libraries at risk of lawsuits, and thus give them a harder time making censorware for home use which blocks protected speach (but which is legal for private use), since they'd have to maintain and justify a separate list of protected speach their home-use product blocks anyway.
Of course, porn sites, or even sites with explicit content of various other sorts, frequently identify themselves as such, either by actually requiring age verification or by having a click-through page saying you have to be in a place that permits viewing such things. If those sites simply sent a header to identify themselves as such, it could be enforced by browsers in places where such content is, in fact, prohibited. I haven't actually surveyed the front pages of porn sites, so I don't know how effective it would be, but this would avoid deep-linking problems and actually make those warning pages meaningful.
If you could install this, get it happy, run some script to syncronize things, and then have it such that you could just reboot and switch from Windows+LINE to Linux+WINE and have the system run exactly the same with respect to apps, it would make a very nice switchover.
In particular, you could make sure that your important apps never stop working as you transition, because you could install and test WINE while still running Windows; once everything works, you can just switch which kernel you're actually running and which you're emulating, and keep working.
If it were set up to dump its whole state to swap space, then everything would work fine for recovereing the state quickly and on an as-needed basis. Of course, the kernel would probably have to do through its boot sequence anyway, because devices generally have to be initialized, so this probably means that, with a good BIOS, you'd be 3 seconds to where you were before shutting down instead of 3 seconds to single-user-mode.
If it's got content-control and a variety of key possibilities, and can have data stored on it, can I put my secret files on this and give keys only to people I want to be able to read it?
And if a court demands a key, can I sue them under the DMCA and have search warrents and government cracking tools declared tools for piracy?
There's got to be a problem when the government is trying to keep people from enchanging information without letting other people read it and simultaneously trying to keep people from reading information while letting other people distribute it.
but somewhat surprised. It just goes to show that, for sites that don't try too hard to support broken 4.x browsers, a good, solid, reasonably standards-compliant browser will work well. Of course, the browser I'm using is netscape 3.0 without javascript, which doesn't seem to be on the lists of browsers to upgrade to.
With all the hype about standards compliance, it's nice to see them not using non-standard tricks to get rid of browsers they don't like.
Just don't support them. Use the standards and ignore the fact that the result won't look right on certain browsers. Furthermore, don't code specifically for broken new browsers. For the page to actually be standards-compliant, it can't really treat even the latest browsers specially.
If you actually follow the standards, the only people who won't have the site work are those whose browsers are not supporting the foolishly-designed standards.
NASA is going to do all future space exploration with NEAR.
NASA: Can we please have funding for another mission? We could build this great space craft and...
Congress: What's wrong with NEAR? Can't you keep using that?
NASA: We used NEAR last time. We put it on EROS and then we found there wasn't anything to do with it once it was there, so we took it back off. Can't we have one with more toys on it?
Congress: No, go away. Can't you see we're busy?
It seems like the situation of having a monolithic tree controlled by Linus with substantial portions that he doesn't have any personal use for is somewhat silly.
It leads to situations where he doesn't want to allow big patches that he can't look at to see if they are good, but the unpatched tree doesn't work at all.
Furthermore, it means the kernel source tree contains >16M of code specifically for processors that are not going to be used in a particular installation.
Why not have each arch subdirectory in a separate tarball, which follows the core kernel but doesn't go through Linus at all, going instead through a maintainer who actually works on that platform?
If the only current kernel that works on PPC is the fork, why doesn't kernel.org carry that instead of the portion of the official kernel that is specific to PPC but doesn't work?
The current status of 2.4.x is "*we* think it's stable, but massive testing might reveal something". 2.4.1 is stable for the kernel developers, but there are a lot of combinations of hardware out there that the kernel developers don't have access to, so there could be really subtle (one in a million machines, e.g.) bugs.
The 2.5 series will start when there's nothing left to do on 2.4.x, essentially. Presently, people are looking for bug reports on 2.4.x stuff and fixing problems found by distribution-makers and end users who wait for stable kernels. Then the developers will get some sleep for a bit, and then figure out what changes should go into 2.5
I think a significant effect which needs to be taken into account is that PDAs can remind you of information that you would have forgotten that you once knew. So they cause there to be more information that you are aware that you've forgotten, even if they do not make your actual memory any worse.
For example, I used to know what times certain places close. Now I still remember those, but also have in my PDA the hours of more places, including openning times and more variation based on day of week. This is a whole bunch of information which I have come in contact with and forgotten due to my PDA-- but the role of the PDA was making me pay attention to it in the first place, not making me forget it afterwards.
The question boils down to what is part of the game and what is not.
The main play of a basketball game is not zero-sum, since each team makes progress. But including the win/lose descision at the end makes it a zero-sum game. But including having fun as a goal makes it not zero-sum again. Games theory applies to what makes sense to do within a framework of rules, but does not define what constitutes "the game being played". Anything which produces some variation can be made into a zero-sum game by having a cutoff to determine a winner.
What the original question should have asked, more preceisely, is what games promote strategies characteristic of non-zero-sum games. I would say that games where players tend to interact directly in ways that benefit those interacting (at the eventual expense of the non-participants) have this aspect: trading games (Res Publica, Settlers of Catan), the >2 player iterated prisoner's dilemma, and so forth.
The part that makes many of these games zero-sum is the comparison at the end. This step means that subtracting from each team's score the average of the scores doesn't change the result, since the result you care about is just the direction of the difference.
But it's not a zero-sum game when the teams want to be able to say that it was a good match afterwards. From a zero-sum perspective, it doesn't make sense to play against people who are not worse than you or to give worse players an advantage, since both of these obviously reduce the chance that you'll win.
Go is zero-sum, but go with rankings and handicaps, considering the whole process and not just the play, is not. Both players take steps at the beginning to cause the game to be difficult but possible for both of them, and they have, as a longer-term goal, to improve their skill, and, as far as I know, nobody tries to give their opponents bad habits.
It's not so much the internal aspects of the game that determine whether it is zero-sum or not, but more the higher-level goals of the players.
Two more steps are required to make it pro-cooperation:
Play a number of rounds, each scored that way.
Play with more than two people. Your score is the sum of your scored with each other person. Each pair of people plays the same number of rounds.
This is pro-cooperation because if you and your partner both compete, or if you trade off competing and cooperating, you both fall behind pairs who both cooperated.
Of course, you'd probably want some variation to make it interesting...
The next devel series will be 2.5.x; 3.1.x will be after 3.0.x. 1.3.x led to 2.0, remember.
I think it would be great if MS released a new version of the OS and had all the bugs fixed in a week. I'd want the patch to be free and not to generate a whole bunch of new bugs, though.
Releasing a new kernel version means that it will be tried on hardware that the developers don't have, even in the case of MicroSoft, so there's always a reasonable chance that some bug will emerge that couldn't have been anticipated. These bugs will never be discovered without a general release of an official stable version, since the people with the obscure hardware won't try anything else. But they should be fixed quickly once they're found.
MicroSoft service packs tend to include not only the fix for the major bug they're dealing with but also changes to the user interface and changes that break other people's software and make you upgrade your microsoft programs. Linux 2.4.1 will probably not break anything that worked under 2.4.0; at least, that was the idea for whether patches should be included.
It's amusing to see an article asking if Linus is killing Linux, because I'm used to seeing articles asking if the $2 billion industry is killing Linux.
The article is interesting in that it entirely fails to show that the actual industry has any problem with the current model of Linux development. I suspect if they could have gotten a quote like "We wish 2.4 had been released sooner, but Linus held it up" from anyone actually in the industry, they'd have used it. Instead they say that Linux should be controlled by a consortium of companies that show no evidence of wanting to do that.
The industry depends on being able to use a steadily improving piece of software in the products that they produce. If they knew how to run the development process better, they'd have done it before. They have enough to do to make their products without being directly involved in kernel development. The industry understands this, but some analysts seem not to.
It probably does take a significant chunk of power to support, say, making this post. You need power for my desktop box, power for the routers, power for the machines at slashdot, power for the servers at salon and the other mentioned sites, and the internet to connect me to all of them. Between all of these, it's a big total.
And that's just my internet. Everybody else has their own internet, which uses up... wait...
They seem to be failing to realize that all of the desktop boxes use a single overlapping internet, which means that, while a substantial amount of power is necessary to maintain my machine, that same power goes into maintaining hundreds of other machines.
Having a Pilot with internet use takes up close to the same amount of power that a desktop box takes up. Having both takes up about the same amount of power as just one.
The reason that servers and server farms exist is that it's more efficient for a group of people to share resources than for everyone to have their own. It only looks more when you charge the cost of the whole thing to each person.
I'm pretty sure the Playstation 2 has a hard drive in it. It's got these loading screens where you can hear the DVD drive spinning and the sound of a hard drive in use.
MS is certainly doing this the open way, at least for now. Of course, it's no more open than PC gaming, which is where they're coming from. The PS2 definitely suffers from insufficient testing on the games right now-- there are bugs in almost all of the ones I've seen which could not have been missed if anyone actually played the games through before the release.
I expect to see things turn more closed later. Perhaps the DVDs are encoded in some way that makes them inpossible to write without a license, so, while you know exactly how the box works inside, you can't get code into the actual hardware. Developer kits would be less strict, so you could write games, but you could only use them on consumer models if MS has approved.
This is essentially a product of using the same processor features you need for other performance gains (register renaming, out-of-order execution) better, with a small amount of extra state.
The first thing you do is get the OS out of the way by making the processor able to handle multiple threads at once without involving OS code in context switching. This is a definite win, because context switches done by the OS kill the pipeline and often kill the cache. If the processor is in charge of all of this, it can do better, because it is closer to what is going on.
After that, the only innovation is to use the out-of-order-execution support on the whole set of things the processor is doing, instead of just one thread.
Considering what they're already doing, this isn't a lot of new complexity, but it should help performance significantly. The main problem with out-of-order stuff has generally been that there isn't really all that much implicit parallelism within a thread. Adding other threads which do not, in general, have any dependancies gives the instruction scheduler much more to work with.
When everyone moves up a notch, that means that nobody in the department is doing what they were hired to do, or even what they've been getting good at. You'll have to train everyone, and the people who know the jobs are either gone or trying to learn other jobs themselves.
The best solution is to promote someone who wants the job, has the right skills for it, and is easier to replace than the original person. Failing that, it's better to high a new person at the top level and train them to work with the department than to mess with the existing structure of the department. Of course, that requires finding someone who can be new at the top level without destroying the department.
The existing community does a good job of writing code. IBM could provide additional developers with unusual experience, but this is on more of an individual level than a corporate level. One thing I think the corporation could do usefully is provide testing platforms. Linux developers generally can't tell how their code will do on massive SMP machines with huge memory spaces, huge disk arrays, and so forth, because they don't have the machines to try stuff. IBM ought to do testing for people on some of the really big boxes, so that ultra-high-end performance can be tuned.
For some reason, Silverman doesn't respond to the idea that, when connecting to a server for the first time, you cannot detect a man-in-the-middle attack.
This boils down to the identity problem: there doesn't have to be an actual server involved at all; you may be actually trying to log into the wrong machine, whether by mischief or by accident, and you'd be none the wiser because you don't know the difference between the different servers. You need to have some way of distinguishing the totally unfamiliar, but correct, server from the totally unfamiliar, but wrong, one.
Now there is a problem with most SSH clients: you might actually know the server you want from experience, but be using a computer which does not have access to this knowledge. You are expecting the client not to recognize the server, but you have more information than the client. In these cases, it would be helpful if the client would get the user involved in the authentication process. OpenSSH actually does this, by putting up for approval the server's key fingerprint, which a user can tell has changed even if the client has no information.
One of the main things that secure protocols are designed to stop is that man-in-the-middle attack.
The basic idea is to exchange session keys and verification data before a random public key. That way, the attacker can't substitute other public keys, because the attacker would have to be able to generate data based on information which has not yet been sent.
The article isn't actually talking about the described attack at all, but about spoofing servers. Identification of servers is a substantially different problem, and is a human-engineering one rather than a protocol one: what makes a server the one you are trying to connect to?
The attack is where the attacker wants to convince you that you're talking to the server you think you are by having you actually talk to that server, except that you are not talking directly. Most encrypted protocols do a good job of insuring that you are connected directly, over an encrypted channel to whatever it is you seem to be talking to. Whether the server you're talking directly to is the one you want to talk to is up to the user and other parts of the system.
What I expect, in the situation referred to by the article, is:
A MS Linux distribution which costs a bunch of money and comes with a bunch of Microsoft programs that don't work without their closed-source libraries. All of the other people's stuff they ship comes with original source. The latest version of Office for Linux + MS libraries, later possibly only for MS Linux.
If they have a way to make sure that users have to buy MS products, and they can require other MS products to have those work, there's no reason for them to care if Windows dies. I mean, if they suddenly dumped Windows and released Office for Linux (requiring the MS library set, of course), not only would people still be tied to MS products, they'd have to buy a new version of the OS yet again. Plus, MS doesn't have to deal with the Windows codebase any more; they don't have to pay anything to distribute the basic stuff, so they lose absolutely no profit from this switch.
They don't need the support of the community at all; they can go for the users who are using Linux because it's better; they give the impression that their distribution has all the power of Linux that you can get from other sources, plus you can use the authentic MS products which read the latest formats, etc.
Is there a good reason to continue to do Linux R&D? Sure; if Linux-based products progressively beats MS products in more and more fields, the ability to use MS products, whether on Windows or on Linux, becomes less of a big deal.
Consider this possibility: Intuit discovers that 8% of their customers won't buy the next version of Quick Books if they use any Whistler features that Wine doesn't support. They could follow the latest API, but then they automatically lose market share.
On the other hand, they could stick with the old API and have their product run, not only on the latest windows boxes, but also on machines that have been around a little while and on linux boxes.
If a significant portion of the windows boxes in use are not following the newest API, the only company for whom it makes sense to use that API is Microsoft, since anyone who wants to use a product that uses the API will have to pay Microsoft for the OS upgrade.
So the critical mass needed to make Wine beneficial to linux and a thorn in MS's side is the point at which the portion of customers using the windows API who aren't going to switch to the latest Microsoft version is significant; a much smaller mass than one might think, especially with the problems MS has had at making upgrading safe and affordable.