It may be mathematically provable, but it lacks the "common sense" aspect that would allow the adoption of such a system. It tends to be better to use technology "under the hood" where it works as one would expect, but is resilient to attacks on the inside. I described a system I believe would work in the last story on voting machines.
Actually the price is higher because of the higher quality parts they use. Every part is thoroughly tested and burned in to be sure that it won't fail. This is critical for Enterprise hardware that going to be taking some serious punishment. Home users could care less.
I'm not sure that "Open Source" would help anything. Sure, others could audit the system, but there's still the opportunity for a unscrupulous person to insert a piece of innocuous looking code that is actually a major security breech. The only way to do this correctly (IMHO) is to contract with a third party that has no commercial interests, and will do the work and research to prove a safety level of ~95% or higher.
A good example of this situation is when DARPA contracted with Berkley to develop BSD into the primary ARPANet operating system. The code was built around solid research and engineering practices instead of a commercial interest to "ship it if it compiles". Take DieBold as an example. They were probably the lowest bidder in the commercial war for voting machines. What foundation do their machines use? Microsoft Access! And that's despite the fact that Access has practically no security features WHAT SO EVER! It's a completely wrong technology.
Now think about a design that ships out a sealed box to a voting office. This sealed box has connections for the network and nothing else. Once powered on, it will expose only an XML-RPC API over SSL for tabulating votes. It will also use the network to contact the main office and upload votes.
Data will be stored internally in an encrypted database, and all encryption keys and configuration information is permanently burned into the firmware chip. Minor actions such as uploading to the main office can be done via a "administrative voting machine" that would require a username and password. That same voting machine would be responsible for activating the individual booths. i.e. I can vote once, but after I confirm my vote, the machine will no longer accept a voting request until the operator tells it to. This allows the same physical security that is afforded by paper ballots today.
An extension of that physical control is that the voting machine and the primary box will share a secret via public key encryption. This secret code will be given to the machine once it registers over XML-RPC with the primary box. Thus any machine can be connected to the server, but only ones that the operator approves will actually get to enter a vote.
Similarly, the voting machines themselves should also be sealed boxes running out of encrypted firmware. There should be no way of changes the settings short of swapping out the physical hardware. This will ensure that the user isn't presented with their vote while the machine actually votes for someone else.
This whole concept though, still falls flat on the social engineering phase. Just like today, if the operator of the ballots is corrupt, there is very little you can do to prevent them from stuffing the ballot. (e.g. They vote multiple times themselves, or someone upstream at the voting machine provider modifies the firmware before it goes out.) In these situations, the only solutions are the same ones we have today. Make sure the number of votes and registered voters match. Do an audit of the supplier. Etc.
A good voting machine is possible. One simply has to remove parties with conflicting interests from the equation.
Dude, chill out. It's called "playing devil's advocate". And my point is that a block like this is going to cause yield problems that could impact IBM's ability to supply them. Yes, even IBM can run into supply problems.
Yes, but what are the advantages? IBM, Sun, and HP all make a business out of selling components with very high MTBF. Yet, if I have a 64 processor machine chugging along for years on end, I have a reasonably good chance of seeing a failure. (Particularly when chips come from a bad batch.)
So, IBM is taking away the ability to hot swap individual chips in exchange for... what? That's the big question. If there's some major improvement in the design, say so! Inquiring minds want to know!:-)
For another... how often do chips die? If you can afford a machine with one of those chips, you can probably afford to replace a whole brick.
Actually, that's what a support contract is for. The bigger problem is availability. Each brick requires four processors, plus the various work to mold all the interconnects into place. The yield on a process like that can't be very high. Not to mention all the custom parts that would be needed to fit a chip like this.
In other words, if my processor fails, there's a good chance I'll be waiting on a new one for anywhere from days to months. Not a good thing for a system that's making you millions for every cycle.
Alright, 4 two-way chips. But does it actually improve anything over individual processors? If I have to yank a board on an UltraSparc, I'm not going to throw away the entire board and all its processors! I'm simply going to replace the bad one and slap the board right back in the system. With IBM's design, I have to throw the whole thing away and get a new block of cement^W^W^W processor chip for my machine.
Geez! That thing looks like you could club someone over the head with it! Does putting 8 processors into a block on cement really improve things that much over multiple processors?
I think it would have been best to have an article devoted to the TransMeta chip, and split the Power5/UltraSparc discussion out into its own article. That way he could have given a great deal more attention to the powerhouse chips and how they're going to change the future. TransMeta's chips are on the level of ARM, not UltraSparc.
> Microsoft thinks it can use the very generic word > 'windows' exclusively across all computer markets
The amusing part is that Bill Gates' own book states that they were unable to register "Windows" because it was too generic. As a result, they were forced to register names like "Microsoft Windows", "Microsoft Windows 95", "Windows NT", etc. Technically, that means you could build an OS and call it "Bob's Windows".
However, you might still get into trouble for intentionally trying to cause brand confusion. Microsoft would try to convince a judge that you're using the word "Windows" in your name to trick people into thinking that the OS is from Microsoft. You would probably have to build a case around the concept that "Windows" is a generic term for GUI based operating systems. Unfortunately, the only help you'd get there is "X Windows" which is not actually a brand name.
IANAL, and this is not legal advice, blah, blah, blah.
fedora is a "cute" inside joke that most people won't even get. [how many PHB's or joe-sixpacks or even college students do you really expect know what the hell a Fedora is anyway]
Plenty. I actually think that geeks would have a harder time knowing what a Fedora is than Joe Sixpack. Joe Sixpack watched all the Chicago Gangster movies growing up. Geeks watched Sci-Fi.
Case and point: My wife (who's originally from Russia!) knew what a Fedora was when I asked her. I had to Google to figure it out.
> I don't see why they didn't just keep the name "RedHat Linux."
Because they didn't want to cause brand confusion. RedHat is basically saying that RedHat Enterprise and RedHat Desktop Linux are two completely separate products, the later of which they no longer officially support. As a result, they've renamed the later to "Fedora Linux" in an attempt to keep two products separate.
Unfortunately, the really GOOD applicants are helpfully filtered out by HR departments. There's no question in my mind that I could have passed your tests easily. Yet, when I was unemployed, I rarely even got interviews to the jobs I was applying for. Apparently, my experience and ability didn't count as much to HR depts as that pretty piece of paper that says "BS Degree" (the real meaning of BS is left to the reader). Unfortunately, as long as technology is a "hot market" where lots of money can be made, real programmers will be hard to find.
The primary reason for mining the moon is that it makes a convenient base for doing things like mining asteroids. Imaging the expense of launching tons of mining gear into orbit. Imagine that it has to be enough stuff and people to efficiently mine before the asteroid passes out of range. Now imagine if we could build that gear on the moon, plus build and use nuclear powered engines to catch those asteroids.
The slashdot blurb makes it sound like private key encryption, but a quick look at the article and site suggest that it is actually public key encryption. Which makes sense in my opinion. Why would I only ever want to call one other phone (with the exception of a RED phone, heh)? I simply want to send my public key, receive his, and then chat on a secure line with whomever I choose.
Slashdot Mirror
It may be mathematically provable, but it lacks the "common sense" aspect that would allow the adoption of such a system. It tends to be better to use technology "under the hood" where it works as one would expect, but is resilient to attacks on the inside. I described a system I believe would work in the last story on voting machines.
Actually the price is higher because of the higher quality parts they use. Every part is thoroughly tested and burned in to be sure that it won't fail. This is critical for Enterprise hardware that going to be taking some serious punishment. Home users could care less.
...is to refer to it as "Solaris X" or "Solaris OS X". That way it can join the ranks of:
Mac OS X
JBuilder X
MegaMan X
And others!
I'm not sure that "Open Source" would help anything. Sure, others could audit the system, but there's still the opportunity for a unscrupulous person to insert a piece of innocuous looking code that is actually a major security breech. The only way to do this correctly (IMHO) is to contract with a third party that has no commercial interests, and will do the work and research to prove a safety level of ~95% or higher.
A good example of this situation is when DARPA contracted with Berkley to develop BSD into the primary ARPANet operating system. The code was built around solid research and engineering practices instead of a commercial interest to "ship it if it compiles". Take DieBold as an example. They were probably the lowest bidder in the commercial war for voting machines. What foundation do their machines use? Microsoft Access! And that's despite the fact that Access has practically no security features WHAT SO EVER! It's a completely wrong technology.
Now think about a design that ships out a sealed box to a voting office. This sealed box has connections for the network and nothing else. Once powered on, it will expose only an XML-RPC API over SSL for tabulating votes. It will also use the network to contact the main office and upload votes.
Data will be stored internally in an encrypted database, and all encryption keys and configuration information is permanently burned into the firmware chip. Minor actions such as uploading to the main office can be done via a "administrative voting machine" that would require a username and password. That same voting machine would be responsible for activating the individual booths. i.e. I can vote once, but after I confirm my vote, the machine will no longer accept a voting request until the operator tells it to. This allows the same physical security that is afforded by paper ballots today.
An extension of that physical control is that the voting machine and the primary box will share a secret via public key encryption. This secret code will be given to the machine once it registers over XML-RPC with the primary box. Thus any machine can be connected to the server, but only ones that the operator approves will actually get to enter a vote.
Similarly, the voting machines themselves should also be sealed boxes running out of encrypted firmware. There should be no way of changes the settings short of swapping out the physical hardware. This will ensure that the user isn't presented with their vote while the machine actually votes for someone else.
This whole concept though, still falls flat on the social engineering phase. Just like today, if the operator of the ballots is corrupt, there is very little you can do to prevent them from stuffing the ballot. (e.g. They vote multiple times themselves, or someone upstream at the voting machine provider modifies the firmware before it goes out.) In these situations, the only solutions are the same ones we have today. Make sure the number of votes and registered voters match. Do an audit of the supplier. Etc.
A good voting machine is possible. One simply has to remove parties with conflicting interests from the equation.
I believe I said that. Please pay more attention.
> so the name "Lindows" is a trademark infringement, but
> the name "Red Hat Windows" would not be?
AKAIK, no court of law has yet declared "Lindows" to be infringing. If they had, Lindows.com wouldn't work right now.
Technically, neither of your examples are infringing. That doesn't stop Microsoft from trying to claim they infringe.
Dude, chill out. It's called "playing devil's advocate". And my point is that a block like this is going to cause yield problems that could impact IBM's ability to supply them. Yes, even IBM can run into supply problems.
Thanks to you and the AC. That's all I wanted to know. :-)
Now if you'll excuse me, I need to see how useful these new chips are as boat anchors...
Do better than that! Make it look like an array of PDP-11s! That'll keep 'em guessing. :-)
Yes, but what are the advantages? IBM, Sun, and HP all make a business out of selling components with very high MTBF. Yet, if I have a 64 processor machine chugging along for years on end, I have a reasonably good chance of seeing a failure. (Particularly when chips come from a bad batch.)
:-)
So, IBM is taking away the ability to hot swap individual chips in exchange for... what? That's the big question. If there's some major improvement in the design, say so! Inquiring minds want to know!
For another... how often do chips die? If you can afford a machine with one of those chips, you can probably afford to replace a whole brick.
Actually, that's what a support contract is for. The bigger problem is availability. Each brick requires four processors, plus the various work to mold all the interconnects into place. The yield on a process like that can't be very high. Not to mention all the custom parts that would be needed to fit a chip like this.
In other words, if my processor fails, there's a good chance I'll be waiting on a new one for anywhere from days to months. Not a good thing for a system that's making you millions for every cycle.
Alright, 4 two-way chips. But does it actually improve anything over individual processors? If I have to yank a board on an UltraSparc, I'm not going to throw away the entire board and all its processors! I'm simply going to replace the bad one and slap the board right back in the system. With IBM's design, I have to throw the whole thing away and get a new block of cement^W^W^W processor chip for my machine.
Geez! That thing looks like you could club someone over the head with it! Does putting 8 processors into a block on cement really improve things that much over multiple processors?
I think it would have been best to have an article devoted to the TransMeta chip, and split the Power5/UltraSparc discussion out into its own article. That way he could have given a great deal more attention to the powerhouse chips and how they're going to change the future. TransMeta's chips are on the level of ARM, not UltraSparc.
Snagglepuss IIRC. His other favorite line was "Exiiittt, Stage Left evvveeenn!" :-)
> And, of course, it's actually the X Window System.
:-)
Or X11.
It was called "A Piece of the Action". A whole planet had reorganized itself around a book on 1920's gangsters and Kirk had to set it straight.
Yes, I remembered that off the top of my head. So sue me.
I think they should call it "Derby Linux".
....
What-da-mean, "derbies aren't as cool"?!
> Microsoft thinks it can use the very generic word
> 'windows' exclusively across all computer markets
The amusing part is that Bill Gates' own book states that they were unable to register "Windows" because it was too generic. As a result, they were forced to register names like "Microsoft Windows", "Microsoft Windows 95", "Windows NT", etc. Technically, that means you could build an OS and call it "Bob's Windows".
However, you might still get into trouble for intentionally trying to cause brand confusion. Microsoft would try to convince a judge that you're using the word "Windows" in your name to trick people into thinking that the OS is from Microsoft. You would probably have to build a case around the concept that "Windows" is a generic term for GUI based operating systems. Unfortunately, the only help you'd get there is "X Windows" which is not actually a brand name.
IANAL, and this is not legal advice, blah, blah, blah.
fedora is a "cute" inside joke that most people won't even get. [how many PHB's or joe-sixpacks or even college students do you really expect know what the hell a Fedora is anyway]
Plenty. I actually think that geeks would have a harder time knowing what a Fedora is than Joe Sixpack. Joe Sixpack watched all the Chicago Gangster movies growing up. Geeks watched Sci-Fi.
Case and point: My wife (who's originally from Russia!) knew what a Fedora was when I asked her. I had to Google to figure it out.
> I don't see why they didn't just keep the name "RedHat Linux."
Because they didn't want to cause brand confusion. RedHat is basically saying that RedHat Enterprise and RedHat Desktop Linux are two completely separate products, the later of which they no longer officially support. As a result, they've renamed the later to "Fedora Linux" in an attempt to keep two products separate.
...here goes the FireBird scandal all over again...
Unfortunately, the really GOOD applicants are helpfully filtered out by HR departments. There's no question in my mind that I could have passed your tests easily. Yet, when I was unemployed, I rarely even got interviews to the jobs I was applying for. Apparently, my experience and ability didn't count as much to HR depts as that pretty piece of paper that says "BS Degree" (the real meaning of BS is left to the reader). Unfortunately, as long as technology is a "hot market" where lots of money can be made, real programmers will be hard to find.
The primary reason for mining the moon is that it makes a convenient base for doing things like mining asteroids. Imaging the expense of launching tons of mining gear into orbit. Imagine that it has to be enough stuff and people to efficiently mine before the asteroid passes out of range. Now imagine if we could build that gear on the moon, plus build and use nuclear powered engines to catch those asteroids.
The slashdot blurb makes it sound like private key encryption, but a quick look at the article and site suggest that it is actually public key encryption. Which makes sense in my opinion. Why would I only ever want to call one other phone (with the exception of a RED phone, heh)? I simply want to send my public key, receive his, and then chat on a secure line with whomever I choose.