Most of the time when I've seen Slashdot link to a simple page of text, the site hasn't been slashdotted. The sites that go down nearly instantly are the hardware case mod sites where someone has taken a bunch of pictures (even a 50k picture will be slower to download than a.torrent file) of her work and hosted it on her server in the basement.
It's difficult because Akamai has to go through a bunch of stupid effort to get the content to where it's going. It's also difficult because you have to sign up with Akamai to arrange to be able to push your content through them in the first place.
If BitTorrent were natively supported in Mozilla, you could host entire sites using it too. In fact, I rather would. Most of those picture heavy sites that get Slashdotted within seconds of their appearing on the front page would be just fine if they were published through BitTorrent.
IMHO, HTTP is a broken protocol, and the fact that sites can be Slashdotted proves it.
Maybe it is a poor man's Akamai, but IMHO, it works a lot better than Akamai does. For example, it works well even if your ISP doesn't have a hub running BitTorrent. And the publishing step is much simpler than Akamai's. So, perhaps, Akamai replacement is a better term.
You might very well be right. The bad behavior with regards to net splits and 'rehubbing' of many IRC networks indicates you probably are. I'm sort of surprised and a bit apalled.
The definition of peer 2 peer is that it is self-organizing, as well. That is the more current definition anyway. It also must not rely upon centralization nor upon "constant routes" (like IRC does)
I did not realize that IRC could not auto-reconfigure its spanning trees. The algorithms for doing so aren't that hard. The Ethernet bridging spanning tree algorithm points the way. For maximum network efficiency, they should have a per-channel spanning tree that only encompasses nodes who have users on the channel.
In my opinion, it's actually a meta-level mixing problem. So much of computer science is devoted to creating systems with well understood properties. Most people who program are of the fundamental opinion that modularity and some sort of hierarchical design are the keys to making good computer programs.
I think this is the problem. Someone pointed out puns to me as something that would trip up a program trying to pass the Turing test. Puns are meta-level humor. They require the associating of two different meanings of a word on the basis of their acoustic similarity.
This kind of problem requires a near unthinkable mixing of information from different abstraction levels. All of our attemts to apply reductionist principles to the building of tractable large systems will work against the creation of a human level machine intelligence.
The biggest problem I see here is that the law doesn't match the expectations and beliefs of most of the people who's behavior it governs. This makes it really hard to classify any given 'black hat' or 'white hat' as bad. Because the classification 'white hat' or 'black hat' has everything to do with the law, and very little to do with what behaviors most security experts would consider good or bad.
Personally, I don't have that much of a problem with this. I think they're being really silly, but if they really want to spend that much time and money on this, they can go right ahead.
If it negatively affects picture quality though, I'll be pretty annoyed.
IPv6 could provide almost as much protection as a NAT.
Every single network gets at least a full/64 in IPv6. 64 bits is a lot of bits. Your devices IPs wouldn't be guessable. Script kiddies would have to run a very noticeable address scan, and even that would not be likely to find a randomly numbered device in a reasonable amount of time.
Actually, one extremely vexing problem is that we're getting de-facto censorship by the people at large rather than any specific government body. That's pretty scary, and a sign that we're in extremely bad shape as a democracy.
People are finding all kinds of laws to use to convince law enforcement to remove people who are saying things they don't like. Laws that are being twisted to fit the circumstances. The problem is what they're saying, not the law they're breaking. But, where you have many laws and a lot of selective enforcement, you end up with the laws being applied to political opponents, to people who are saying things you don't like.
Now, Michael Moore's statement was possibly inappropriate at the Oscars. But, everybody at the Oscars being pressured to wear red ribbons to symbolize their concern about AIDS isn't very appropriate either, but nobody bothered to edit it out.
Actually, that reductionist view isn't completely correct either. Governments can and do do things that many of the individuals who make them up may actually find apalling.
IMHO, the Soviet Union, for example, was an evil empire, despite the fact that most people in the Soviet Union were not evil.
Or, to put it in a different way, people have motivations, despite the fact that atoms don't.
Did it ever occur to you that perhaps Jamie actually really likes BitTorrent and thinks it's a fantastic idea and is trying to get the word out?
I guess you could still call it an advertisement then, but I think most people would call it advocacy. As is often pointed out, Slashdot editors have never been shy about advocacy.:-)
That was my point. The license is, for the most part, the GPL. The GPL explicitly permits redistribution.
So, am I supposed to pay attention to the terms of the license, or the original poster who wants to invent some societal norms and brand behavior permitted by the GPL as 'unethical' to insure that RedHat makes a profit?
*chuckle* You must be at the end of a T1 and all of your upstreams are on asymmetric DSL lines, like me. Hopefully more T1 people will join and things will even out for you.
I've paid for RH in the past from stores. I'll probably join RHN when I get a job. I don't think it's unethical for BitTorrent to do what it's doing. The license explicitly permits it.
Are you telling me that I should ignore the text of the license and listen to a set of societal norms you're attempting to invent? Where would we be if everybody did that with the first ammendment?
I hate it when science discoveries are reported in that uber-hyped style. It so obscures what the real finding actually is. It looks like they have something here, but in between the whole 'transcend the laws of nature' garbage and the 'this is so fantastic and revolutionary it will change absolutely everything' garbage, it's hard to see what they actually have.
I agree that the Kerberos vulnerability is a painful and difficult one to fix, being a vulnerability in the protocol itself. Perhaps, in that sort of a case, CERT's policy might be sort of OK, though I'm still suspicious of it. If you're going to do anything less than full disclosure, the set release date is an absolute necessity.
The whole system, from the withholding reports for weeks and weeks until the vendor 'has time' to fix it to the for pay ability to get reports early, it's all designed to protect the images of large corporations. They've sat on those reports for months or years sometimes. Knowing that CERT knows is no incentive at all to fix it.
Also, why should the glibc author (who may not have any money) have to pay a ton of money to a government backed agency in order to get a security vulnerability report that would enable h(im/er) to fix the bug?
The whole CERT system is designed with the notion that if you're big, or at least incorporated, or have a bunch of money, you must be a 'good' guy. It's 1950s era thinking. The establishment is always right and must be protected at all costs.
That vulnerability is a simple buffer overflow. RedHat had a patch out for it in less than a day. This whole 'wait for the vendor to fix it' thing just results in lazy vendors.
And, as the army breakin shows, the 'bad' guys often have the information whether or not the 'good' guys even know it. There are many script kiddies out there, but there are a few really intelligent people who can do their own research, and won't bother telling CERT before they go and exploit the vulnerability.
Slashdot Mirror
Most of the time when I've seen Slashdot link to a simple page of text, the site hasn't been slashdotted. The sites that go down nearly instantly are the hardware case mod sites where someone has taken a bunch of pictures (even a 50k picture will be slower to download than a .torrent file) of her work and hosted it on her server in the basement.
It's difficult because Akamai has to go through a bunch of stupid effort to get the content to where it's going. It's also difficult because you have to sign up with Akamai to arrange to be able to push your content through them in the first place.
If BitTorrent were natively supported in Mozilla, you could host entire sites using it too. In fact, I rather would. Most of those picture heavy sites that get Slashdotted within seconds of their appearing on the front page would be just fine if they were published through BitTorrent.
IMHO, HTTP is a broken protocol, and the fact that sites can be Slashdotted proves it.
Maybe it is a poor man's Akamai, but IMHO, it works a lot better than Akamai does. For example, it works well even if your ISP doesn't have a hub running BitTorrent. And the publishing step is much simpler than Akamai's. So, perhaps, Akamai replacement is a better term.
I have no college degree, but I have a large vocabulary and usually flawless spelling.
But, I agree with your general point. That part just rankled me.
He has no taken into account Benford's Law in determining transaction amounts, and is falsely assuming a flat distribution.
I was writing a b-tree based database library in C. :-)
I didn't know much about networking then, though I was also writing a distributed mandlebrot set generator in C++. :-)
You might very well be right. The bad behavior with regards to net splits and 'rehubbing' of many IRC networks indicates you probably are. I'm sort of surprised and a bit apalled.
I did not realize that IRC could not auto-reconfigure its spanning trees. The algorithms for doing so aren't that hard. The Ethernet bridging spanning tree algorithm points the way. For maximum network efficiency, they should have a per-channel spanning tree that only encompasses nodes who have users on the channel.
Yes, you've found another idiot. But, IRC is P2P in a certain sense. The IRC servers form a P2P network among themselves. And, DCC is definitely P2P.
In my opinion, it's actually a meta-level mixing problem. So much of computer science is devoted to creating systems with well understood properties. Most people who program are of the fundamental opinion that modularity and some sort of hierarchical design are the keys to making good computer programs.
I think this is the problem. Someone pointed out puns to me as something that would trip up a program trying to pass the Turing test. Puns are meta-level humor. They require the associating of two different meanings of a word on the basis of their acoustic similarity.
This kind of problem requires a near unthinkable mixing of information from different abstraction levels. All of our attemts to apply reductionist principles to the building of tractable large systems will work against the creation of a human level machine intelligence.
The biggest problem I see here is that the law doesn't match the expectations and beliefs of most of the people who's behavior it governs. This makes it really hard to classify any given 'black hat' or 'white hat' as bad. Because the classification 'white hat' or 'black hat' has everything to do with the law, and very little to do with what behaviors most security experts would consider good or bad.
Personally, I don't have that much of a problem with this. I think they're being really silly, but if they really want to spend that much time and money on this, they can go right ahead.
If it negatively affects picture quality though, I'll be pretty annoyed.
I didn't think they even had sailing vessels.
It wasn't a service they had wanted to offer in the first place. It was only added to caller id because of immense pressure from privacy groups.
IPv6 could provide almost as much protection as a NAT.
Every single network gets at least a full /64 in IPv6. 64 bits is a lot of bits. Your devices IPs wouldn't be guessable. Script kiddies would have to run a very noticeable address scan, and even that would not be likely to find a randomly numbered device in a reasonable amount of time.
Actually, one extremely vexing problem is that we're getting de-facto censorship by the people at large rather than any specific government body. That's pretty scary, and a sign that we're in extremely bad shape as a democracy.
People are finding all kinds of laws to use to convince law enforcement to remove people who are saying things they don't like. Laws that are being twisted to fit the circumstances. The problem is what they're saying, not the law they're breaking. But, where you have many laws and a lot of selective enforcement, you end up with the laws being applied to political opponents, to people who are saying things you don't like.
Now, Michael Moore's statement was possibly inappropriate at the Oscars. But, everybody at the Oscars being pressured to wear red ribbons to symbolize their concern about AIDS isn't very appropriate either, but nobody bothered to edit it out.
Actually, that reductionist view isn't completely correct either. Governments can and do do things that many of the individuals who make them up may actually find apalling.
IMHO, the Soviet Union, for example, was an evil empire, despite the fact that most people in the Soviet Union were not evil.
Or, to put it in a different way, people have motivations, despite the fact that atoms don't.
Did it ever occur to you that perhaps Jamie actually really likes BitTorrent and thinks it's a fantastic idea and is trying to get the word out?
I guess you could still call it an advertisement then, but I think most people would call it advocacy. As is often pointed out, Slashdot editors have never been shy about advocacy. :-)
That was my point. The license is, for the most part, the GPL. The GPL explicitly permits redistribution.
So, am I supposed to pay attention to the terms of the license, or the original poster who wants to invent some societal norms and brand behavior permitted by the GPL as 'unethical' to insure that RedHat makes a profit?
*chuckle* You must be at the end of a T1 and all of your upstreams are on asymmetric DSL lines, like me. Hopefully more T1 people will join and things will even out for you.
I've paid for RH in the past from stores. I'll probably join RHN when I get a job. I don't think it's unethical for BitTorrent to do what it's doing. The license explicitly permits it.
Are you telling me that I should ignore the text of the license and listen to a set of societal norms you're attempting to invent? Where would we be if everybody did that with the first ammendment?
I hate it when science discoveries are reported in that uber-hyped style. It so obscures what the real finding actually is. It looks like they have something here, but in between the whole 'transcend the laws of nature' garbage and the 'this is so fantastic and revolutionary it will change absolutely everything' garbage, it's hard to see what they actually have.
I agree that the Kerberos vulnerability is a painful and difficult one to fix, being a vulnerability in the protocol itself. Perhaps, in that sort of a case, CERT's policy might be sort of OK, though I'm still suspicious of it. If you're going to do anything less than full disclosure, the set release date is an absolute necessity.
The whole system, from the withholding reports for weeks and weeks until the vendor 'has time' to fix it to the for pay ability to get reports early, it's all designed to protect the images of large corporations. They've sat on those reports for months or years sometimes. Knowing that CERT knows is no incentive at all to fix it.
Also, why should the glibc author (who may not have any money) have to pay a ton of money to a government backed agency in order to get a security vulnerability report that would enable h(im/er) to fix the bug?
The whole CERT system is designed with the notion that if you're big, or at least incorporated, or have a bunch of money, you must be a 'good' guy. It's 1950s era thinking. The establishment is always right and must be protected at all costs.
That vulnerability is a simple buffer overflow. RedHat had a patch out for it in less than a day. This whole 'wait for the vendor to fix it' thing just results in lazy vendors.
And, as the army breakin shows, the 'bad' guys often have the information whether or not the 'good' guys even know it. There are many script kiddies out there, but there are a few really intelligent people who can do their own research, and won't bother telling CERT before they go and exploit the vulnerability.