Yes, but if you notice the CO2 increases tend to lag the temperature increases. One theory is that whatever causes the temperature jump causes increased evaporation from the oceans, which increases cloud cover and the reduced sunlight causes a die-off of plants which causes the CO2 levels to jump (but not by enough to offset the reduced insolation heating, so after the temperature spikes we drop into a snowball). You can see the lag clearly in the chart on the next page you reference. There's a major temperature jump starting around 12,500 years back and peaking just before 11,000 years back, but the CO2 level doesn't start to jump until around 11,500 years back (a thousand years after the temperature increase starts) and peaks about 11,000 years back (just after temperatures start to drop a bit).
Also note the differences in variance for temperatures on the various charts. We only have accurate temperature readings back about 42 years, so temperatures after that point show intra-year variation while temperatures before that are annual averages. Comparing chart 2 to charts 3 and 4, the entire increase the authors are worried about fits well within the 200-year normal variations present in the data for the last 11 millenia. I'm sorry, but there's no discernable "signal" in the temperature chart.
That said, I'd be concerned not about the CO2 levels themselves but about rising CO2 emissions coupled with the erosion of carbon-fixing plant life (eg. rain forests). Ignoring temperature data completely, that still can't possibly be good. The problem is that the majority of the CO2 emissions are in the developing world, which doesn't have the resources to do anything about it. And it's going to get worse as they go through the same progression the developed world went through. None of the "solutions" I've seen proposed do a thing about that.
Unprecedented high temperatures in recent history, perhaps. Unprecedented in terms of Earth's history? I'm afraid not. Notice the three sharp spikes occurring at roughly 130,000 year intervals. We started such a rise about 15,000 years ago, right at the expected time if the pattern repeats, but something levelled it off around present-day levels and has kept it there for the last 10,000 years. Whatever cause the levelling-out it wasn't humans, we weren't doing anything on a scale large enough to cause global effects 15,000 years back. If whatever it is stops, I'd expect global temperatures to spike by another 2-3 degrees C, then drop sharply to 4-6 degrees C below "normal".
That's another factor. WSS used to have a noscript section in their page code, harder to get right because you had to hand-populate the query string with things like the page name but it'd track browsers that didn't do JavaScript. They removed that a while back, which IMHO was a mistake.
Just counting all hits doesn't do much if you're trying to count people. However, since I'd expect people's aggregate browsing behavior is roughly similar across all browsers, percentage of hits and of people should be roughly equal so for market share as percentages raw hit data should do. Unfortunately the higher-ups at WSS don't believe that.
And as for "a visit is a visit", um... no. The problem is auto-refresh on pages and people who never close their browsers. The record is a single browser session spanning more than 96 hours, alternating between the same two pages every 5 minutes due to a meta-refresh setting (and this wasn't an isolated aberration by at least 4 orders of magnitude). There's also people who refresh the results page every 10-15 seconds during big sports or similar events. Basically, if you can think of a completely insane, utterly ridiculous, nobody-would-ever-do-that scenario that would screw up the data, chances are WSS is seeing it a hundred times a minute.
WSS has one big issue: they depend on cookies. If your browser doesn't accept their cookies, they don't track browser stats on your hits. If your browser only accepts their cookie for a session and then discards it, it skews their stats. And I'd bet that the majority of people who use FireFox have it set to not accept third-party cookies (cookies from outside the domain of the page being viewed). They're also most likely to have blocked the first-party cookies WSS uses with some customers. And of course as you noted WSS monitors mostly mainstream sites so their numbers tend to reflect the population of those sites (you wouldn't believe the traffic volume associated with Disney or ESPN).
Then again, I'd say WSS's stats are about as good as it's possible to get without some way to hook into the browser itself. With NAT and proxies and such you need some sort of persistent identifier in the browser itself to distinguish 10 different people behind a proxy each hitting one page from 1 person hitting 10 pages, and cookies are the least intrusive way of doing a persistent identifier.
Actualy I'd say Craigslist is in it for the money. They're just in it for the money year after year after year for the rest of the owner's life if possible. As Buckmaster noted, they may not be making huge bucks but, unlike a lot of Internet ventures raking in larger revenues, Craigslist has been profitable for 6-7 years now and shows no signs of losing money or going away anytime soon. They aren't making as much every year as they could be, but they'll be making it for a lot more years than they would if they starting inserting ads on their pages.
This shouldn't really be a problem. It only occurs if you click on a link in the e-mail. If you ignore the link in the e-mail, go to PayPal through a bookmark of your own and proceed from there, the phisher can't inject any code. End of problem. And if what the e-mail's asking for is legitimate, you'll be able to do anything you need to do directly through PayPal without needing to use any links in the e-mail.
First rule: never trust the identity of the other party if you didn't initiate the contact yourself. When someone calls you on the phone claiming to be your bank you don't trust them, you hang up and call your bank's customer-service number yourself. When someone sends you an e-mail claiming a link will take you to PayPal you don't trust that, you fire up your browser and use your own bookmark to hit PayPal.
My opinion: DRM should be banned unless it complies with all of copyright law. That includes things like the length of copyright. If copyright law says the work passes into the public domain 70 years after the creator's death, then the DRM should allow the work to pass into the public domain then or be considered itself in violation of copyright law. The problem is that current DRM doesn't honor all of copyright law, only those parts the RIAA/MPAA want to have enforced.
I think you misunderstood the character-class rules. They're speaking about how many classes of characters have to appear in the password, not how many characters of each class. Essentially a password has to be either mixed-case or mixed letters and numbers, but it does not have to have any particular number of upper-case, lower-case or digits. It can also be based on a dictionary word, it just can't be only a dictionary word (ie. "dictionary" is illegal but "dict10nary" or "dict1ionary" are acceptable. That limits the search space in some sense, but it does so by forcing the password into a much larger search space (the set of dictionary words with 1 or more digits inserted at one or more points in the word is orders of magnitude larger than the set of plain dictionary words). This takes advantage of the fact that not all search spaces are equal, some are much harder to search than others.
And as I said, most attacks won't be on passwords. The idea here isn't to make passwords highly secure, since attackers probably aren't going to bother cracking them. The idea is just to keep them from being trivial to crack, so the casual kiddie trying the obvious things won't get in, while letting them be easy enough to remember that people won't have too many problems with the change interval or with remembering passwords for multiple systems. This frees admin and user attention for more probable vulnerabilities, eg. "You know that e-mail that promises you Jada Pinkett-Smith and Lucy Liu porn if you just open that executable? Well, they're lying.".
I'd argue that. Nowadays it's pretty hard to crack the corporate firewall to be able to attack the machines you could try a password attack on, and moderately risky to get physical access to the building and the network wiring. It's dead easy, though, to e-mail a trojan or other malware masquerading as some suitably-attractive bait (new screensaver, porn, etc.) and count on at least a few people in the company getting bit by it. Note that that malware doesn't need to crack your password, it's already logged in and running as you and Windows will happily present your credentials whenever "you" try to access a machine.
IMHO passwords are far from the weakest link anymore. You want some rules to prevent the obviously-stupid passwords but basic complexity rules (eg. at least 6 characters, at least two of upper-case, lower-case and numbers, no dictionary words, can't repeat the last 2 passwords) along with a 4- to 6-month change interval should keep password cracking from being a serious problem. Aside from some special cases (eg. Windows LANMAN hashes, offline cracking of Unix passwd files), the days of password cracking are over.
The problem is that, in all the cases presented so far where the complaint is that the kid's got access to an age-inappropriate game, the kid didn't buy it. Their parent bought it for them. Almost all game stores, and all the major chains, simply won't sell M-rated games to anyone who can't show a picture ID showing them to be over 18. This doesn't stop the parent from buying the game and giving it to their kid without even looking at the rating and then complaining when it turns out the game makes "From Dusk 'til Dawn" look like a Saturday-morning cartoon. And frankly I see no way any rating system or any set of obligations or punishments aimed at the stores can do a single bloody thing about this.
Every time I hear one of the incidents of a kid getting an inappropriately-rated game, I notice that the kid didn't just go buy the game on his own. Almost always, a parent or some other adult authorized under all these proposed laws bought the game for the kid. I think the defense needed isn't more ratings. When a politician brings up the issue, someone stand up and name names and point out that the parent bought the game, then ask the politician flat-out what they're going to do about parents who buy their children these games and when are they going to start doing it. Cite their own example case back at them, and make them answer how their proposals are going to address the problem of parents doing the buying. If they try to weasel out, bring them back on point by noting that it was their example that involved the parent doing the buying, so why can't they address their own example?
It does ask for it. But IMHO it shouldn't ever not ask, because you shouldn't be running as admin except in unusual circumstances. UAC as Microsoft's presented it isn't a solution to a problem, it's an attempt to cover up the holes the problem causes. As we've see with zone-escalation attacks and suchlike, someone somewhere will find a way to make UAC believe it's already OK'd the action and, in an environment where you're running as admin and depending on UAC to protect you, you're now screwed.
That depends. On Debian the keyring's managed so that the only way for a program to find out whether the user's got privileges or not is to run, and if the user doesn't have privileges up pops the prompt box to alert the user that something's not right. There's no way for malware to tell when or whether the user's got privileges without triggering the prompt if they don't. And as I said, the timing can be tuned. 5 minutes is a pretty long time, 1 minute or less should suffice for most purposes. Remember that once a program's got elevated privileges it doesn't need the credentials again, the idle timeout is purely to make running of several administrative tasks in quick succession less bothersome. I can configure it to save the credentials only long enough to authenticate the current program and then drop them, at the cost of having to re-enter the password for a new program (but not the one I just ran, it's got privileges already and can pass them on to it's children).
Yeah, I can figure out ways to exploit the keyring, but in practice it's a real pain to write code to get it right without tipping off the user. And it's better than Microsoft's approach, which either teaches users to just click "OK" on any UAC dialog or leaves some actions happening without user authorization to avoid excessive dialogs. It's open what the "best" tuning would be, but the tuning's still a lot simpler than what Microsoft's proposing to do.
Yes, it has to wait. The problem is, the only way for it to find out whether the user's asked for root privs is to attempt an operation that needs them. Which will cause a prompt if the user hasn't got root privs at the moment. Which gets the correct result: the user now knows something's trying to get privs it shouldn't have.
And no, Linux and OS X don't suffer from viruses just because of their small market share. Unix in general has been the dominant OS in universities for 25 years or so. That's the environment that created several generations of crackers, hackers and generally the sorts of people who break systems. It's never, despite that, been as vulnerable as Windows even when faced with more attackers in an environment where the sysadmins had to actually give those attackers access to the systems. Unix-derived systems aren't as vulnerable to viruses simply because their design never placed usability over security. They were built from the beginning with a distinction between the administrative user (root) and ordinary users, and designed so that you didn't need administrative powers to do basic things like install software. You needed administrative powers to install software system-wide (because it'd be installed into system directories that ordinary users can't write to), but to simply install it under your home directory took no special privileges. Even letting other people use the copy you'd installed under your home directory took no special privileges. This means that malware has a much harder time convincing anyone to install it as root, and if it isn't installed as root it can't hide itself from root or affect the system itself. This, not lack of marketshare, is what makes Linux and OS X less vulnerable to malware than Windows.
The point of UAC is to make sure the user has to authorize any actions that need administrative privileges. So address the authorization instead of the actions. Do what my Debian box does when programs need root privileges. When I run a program like that from my normal user account, a wrapper prompts me to enter the root password or abort the operation. If I enter the password and it's correct, root credentials are added to my keyring temporarily and the program can run as root. As long as those credentials are on my keyring, any other programs that need root access can run without prompting. If the credentials remain unused for more than a short time, they're removed from my keyring and any programs after that that need root privs will cause a prompt again. This makes sure I have to manually authorize root access, but that I don't have to keep answering repetitive prompts. It doesn't require any fancy tuning of which actions prompt and which don't, at most it only needs tuning of how long root credentials remain on the keyring which is a lot simpler.
Typical Microsoft, crafting the most complicated solution to the problem.
I believe CRLs have been superseded by OCSP (Online Certificate Status Protocol) for just the reasons you noted. OCSP doesn't require local storage of revocation lists and the synchronization issues that go along with it.
You misunderstand the telco's position. Say I subscribe to Cox HSI for my Internet connection. I pay Cox every month for my connection and bandwidth. Google connects to the Internet through Level3. Google pays Level3 every month for their connection. What the telcos want is for Google to pay Cox for the bandwidth used when I search using Google. Yes, that's in addition to what I'm paying Cox, and what Google's paying Level3. And yes, it does in fact involve charging someone who's nto a customer. The penalty for not paying would be that traffic to and from Google gets an effectively lower priority on Cox's network, making Google slow compared to sites that're paying. The telcos say they won't penalize non-paying traffic, just give priority to paying traffic. It amounts to the same thing, though. Whether a gas station charges $3.00/gallon for cash with a 10-cent surcharge for paying by credit card or $3.10/gallon with a 10-cent discount for paying with cash, the end result's exactly the same.
I think there's an old comment that sums it up: "I don't want someone who'll work his ass off doing a job over and over and over. I want a lazy bum who'll figure out a way to not need to do the job again.". The industrious scribe invented the typewriter so the roomful of scribes could hand-copy documents faster. The lazy scribe invented the photocopier so he didn't need to hand-copy another document. The really lazy scribe invented the document-feeder and sorter/stacker attachments so he could go get lunch while the machine made 14 copies of the 357-page stack.
Note that being lazy can involve quite a bit of work. The trick is to remember that you're not trying to minimize the amount of work you're doing right this minute, you're trying to minimize the total amount of work you have to do over the long term.
Actually not. Standard and overnight delivery are in fact handled differently when it comes to how they're moved. Overnight delivery goes by more direct links that're faster but not as cost-efficient. As for the rest, I'm afraid it's wrong as well. UPS, FedEx and DHL get enough business for their overnight services from people who need delivery faster than reasonable and are willing to pay for it, they don't need to play games with their standard service. They know why they've pretty much taken the parcel business away from the US Postal Service, they're not about to make the same mistakes the USPS did.
Christopher Yoo got it, and his point is what's wrong with the telecomm's ideas. He's right, consumers should be able to pay for better delivery, just like when I order something shipped FedEx I can pay for regular delivery or I can pay more for overnight delivery depending on what I want. But that's not what the telecomms propose. That'd be like the telecomms saying "Consumer, you're using a lot of bandwidth. If you want to download streaming video you're going to have to pay for a higher-capacity link.". What the telecomms propose, though, is to not have the consumer pay for what they want but to have whoever the consumer's asking for stuff from pay. It's like my ordering something and paying for overnight shipping, and FedEx saying to the shipper "Right then. The customer's paid for standard shipping, but unless you pay us for overnight delivery we'll shove your package in the back and deliver it whenever we feel like it. Which may be never. Oh, and the extra just gets you standard delivery, real overnight will be yet more on top of that.". Of course the telecomms don't want to phrase it that way, because people understand FedEx and the extortion attempt's blatantly obvious.
What the studios are worried about isn't the average person. They're worried about the early HD adopters, the ones who have HD sets now. A lot of those aren't truly technical, but they think they're techies 'cause they've got the latest ooh-shiny gear. And they know what HD pictures look like now, and they'll be able to tell that ICT-downgraded discs don't look nearly as good as off-the-air HD. The studios are worried that, if they implement ICT now, when Joe Average asks Techie Neighbor with the fancy HD set what brand he should buy, TN will go "Don't bother. None of the discs out there do HD, so if you don't have HD cable and your set's working fine you might as well save your money.". And as long as the majority of people who've got HD sets are saying that, JA will keep on not buying HD equipment unless it's the only stuff available.
What the studios hope is that, given a few years, they'll be able to get sets with the proper support in them out there in quantity and the complaints when they turn on ICT will be a small minority. I think they've learned from DAT and SACD that the back door's the best way.
Yes, but Washington DC wasn't supposed to be a city in the conventional sense. It was supposed to be the seat of Federal government, period. The only people supposed to be living there would be Federal employees who weren't supposed to have a direct say in their own authority.
The people running Web sites, or creating software for that matter, might want to consider some of the consequences of their current crack-down on vulnerability reports. Yes, vulnerability reports are bad PR. However, if this keeps up people who find vulnerabilities will have only two feasible alternatives:
Say nothing. This leaves the site or software wide open to exploitation by the unscrupulous. The PR when this comes out will be even worse (and it will come out).
Don't report to the creators. Report only to the general public, anonymously, with full details included so nobody has to trust the reputation of the reporter to verify the validity of the report. Of course this makes it impossible for the creators to fix problems before the world gets told about them.
I'm in the habit of quoting large portions of articles, or even the entire article, for a purely practical reason: the mutability of Web pages. I've lost track of how often I've made a comment about something in an article, only to have a lot of people asking what I was talking about because the article said no such thing. On looking at the article again, the passage I was referring to had either been removed or altered to say something it hadn't said originally. The only way I have to combat this is to preserve a copy of the article as I originally read it in a place not subject to editing by the article's owner.
I'd note this after-the-fact rewriting tends to be most common where the original article contained egregiously and provably incorrect statements and the authors got called on the matter and now want to never have said that (as opposed to wanting to admit they mis-stated).
Slashdot Mirror
Yes, but if you notice the CO2 increases tend to lag the temperature increases. One theory is that whatever causes the temperature jump causes increased evaporation from the oceans, which increases cloud cover and the reduced sunlight causes a die-off of plants which causes the CO2 levels to jump (but not by enough to offset the reduced insolation heating, so after the temperature spikes we drop into a snowball). You can see the lag clearly in the chart on the next page you reference. There's a major temperature jump starting around 12,500 years back and peaking just before 11,000 years back, but the CO2 level doesn't start to jump until around 11,500 years back (a thousand years after the temperature increase starts) and peaks about 11,000 years back (just after temperatures start to drop a bit).
Also note the differences in variance for temperatures on the various charts. We only have accurate temperature readings back about 42 years, so temperatures after that point show intra-year variation while temperatures before that are annual averages. Comparing chart 2 to charts 3 and 4, the entire increase the authors are worried about fits well within the 200-year normal variations present in the data for the last 11 millenia. I'm sorry, but there's no discernable "signal" in the temperature chart.
That said, I'd be concerned not about the CO2 levels themselves but about rising CO2 emissions coupled with the erosion of carbon-fixing plant life (eg. rain forests). Ignoring temperature data completely, that still can't possibly be good. The problem is that the majority of the CO2 emissions are in the developing world, which doesn't have the resources to do anything about it. And it's going to get worse as they go through the same progression the developed world went through. None of the "solutions" I've seen proposed do a thing about that.
Unprecedented high temperatures in recent history, perhaps. Unprecedented in terms of Earth's history? I'm afraid not. Notice the three sharp spikes occurring at roughly 130,000 year intervals. We started such a rise about 15,000 years ago, right at the expected time if the pattern repeats, but something levelled it off around present-day levels and has kept it there for the last 10,000 years. Whatever cause the levelling-out it wasn't humans, we weren't doing anything on a scale large enough to cause global effects 15,000 years back. If whatever it is stops, I'd expect global temperatures to spike by another 2-3 degrees C, then drop sharply to 4-6 degrees C below "normal".
That's another factor. WSS used to have a noscript section in their page code, harder to get right because you had to hand-populate the query string with things like the page name but it'd track browsers that didn't do JavaScript. They removed that a while back, which IMHO was a mistake.
Just counting all hits doesn't do much if you're trying to count people. However, since I'd expect people's aggregate browsing behavior is roughly similar across all browsers, percentage of hits and of people should be roughly equal so for market share as percentages raw hit data should do. Unfortunately the higher-ups at WSS don't believe that.
And as for "a visit is a visit", um... no. The problem is auto-refresh on pages and people who never close their browsers. The record is a single browser session spanning more than 96 hours, alternating between the same two pages every 5 minutes due to a meta-refresh setting (and this wasn't an isolated aberration by at least 4 orders of magnitude). There's also people who refresh the results page every 10-15 seconds during big sports or similar events. Basically, if you can think of a completely insane, utterly ridiculous, nobody-would-ever-do-that scenario that would screw up the data, chances are WSS is seeing it a hundred times a minute.
WSS has one big issue: they depend on cookies. If your browser doesn't accept their cookies, they don't track browser stats on your hits. If your browser only accepts their cookie for a session and then discards it, it skews their stats. And I'd bet that the majority of people who use FireFox have it set to not accept third-party cookies (cookies from outside the domain of the page being viewed). They're also most likely to have blocked the first-party cookies WSS uses with some customers. And of course as you noted WSS monitors mostly mainstream sites so their numbers tend to reflect the population of those sites (you wouldn't believe the traffic volume associated with Disney or ESPN).
Then again, I'd say WSS's stats are about as good as it's possible to get without some way to hook into the browser itself. With NAT and proxies and such you need some sort of persistent identifier in the browser itself to distinguish 10 different people behind a proxy each hitting one page from 1 person hitting 10 pages, and cookies are the least intrusive way of doing a persistent identifier.
Actualy I'd say Craigslist is in it for the money. They're just in it for the money year after year after year for the rest of the owner's life if possible. As Buckmaster noted, they may not be making huge bucks but, unlike a lot of Internet ventures raking in larger revenues, Craigslist has been profitable for 6-7 years now and shows no signs of losing money or going away anytime soon. They aren't making as much every year as they could be, but they'll be making it for a lot more years than they would if they starting inserting ads on their pages.
First rule: if it's working, don't fix it.
This shouldn't really be a problem. It only occurs if you click on a link in the e-mail. If you ignore the link in the e-mail, go to PayPal through a bookmark of your own and proceed from there, the phisher can't inject any code. End of problem. And if what the e-mail's asking for is legitimate, you'll be able to do anything you need to do directly through PayPal without needing to use any links in the e-mail.
First rule: never trust the identity of the other party if you didn't initiate the contact yourself. When someone calls you on the phone claiming to be your bank you don't trust them, you hang up and call your bank's customer-service number yourself. When someone sends you an e-mail claiming a link will take you to PayPal you don't trust that, you fire up your browser and use your own bookmark to hit PayPal.
My opinion: DRM should be banned unless it complies with all of copyright law. That includes things like the length of copyright. If copyright law says the work passes into the public domain 70 years after the creator's death, then the DRM should allow the work to pass into the public domain then or be considered itself in violation of copyright law. The problem is that current DRM doesn't honor all of copyright law, only those parts the RIAA/MPAA want to have enforced.
I think you misunderstood the character-class rules. They're speaking about how many classes of characters have to appear in the password, not how many characters of each class. Essentially a password has to be either mixed-case or mixed letters and numbers, but it does not have to have any particular number of upper-case, lower-case or digits. It can also be based on a dictionary word, it just can't be only a dictionary word (ie. "dictionary" is illegal but "dict10nary" or "dict1ionary" are acceptable. That limits the search space in some sense, but it does so by forcing the password into a much larger search space (the set of dictionary words with 1 or more digits inserted at one or more points in the word is orders of magnitude larger than the set of plain dictionary words). This takes advantage of the fact that not all search spaces are equal, some are much harder to search than others.
And as I said, most attacks won't be on passwords. The idea here isn't to make passwords highly secure, since attackers probably aren't going to bother cracking them. The idea is just to keep them from being trivial to crack, so the casual kiddie trying the obvious things won't get in, while letting them be easy enough to remember that people won't have too many problems with the change interval or with remembering passwords for multiple systems. This frees admin and user attention for more probable vulnerabilities, eg. "You know that e-mail that promises you Jada Pinkett-Smith and Lucy Liu porn if you just open that executable? Well, they're lying. ".
I'd argue that. Nowadays it's pretty hard to crack the corporate firewall to be able to attack the machines you could try a password attack on, and moderately risky to get physical access to the building and the network wiring. It's dead easy, though, to e-mail a trojan or other malware masquerading as some suitably-attractive bait (new screensaver, porn, etc.) and count on at least a few people in the company getting bit by it. Note that that malware doesn't need to crack your password, it's already logged in and running as you and Windows will happily present your credentials whenever "you" try to access a machine.
IMHO passwords are far from the weakest link anymore. You want some rules to prevent the obviously-stupid passwords but basic complexity rules (eg. at least 6 characters, at least two of upper-case, lower-case and numbers, no dictionary words, can't repeat the last 2 passwords) along with a 4- to 6-month change interval should keep password cracking from being a serious problem. Aside from some special cases (eg. Windows LANMAN hashes, offline cracking of Unix passwd files), the days of password cracking are over.
The problem is that, in all the cases presented so far where the complaint is that the kid's got access to an age-inappropriate game, the kid didn't buy it. Their parent bought it for them. Almost all game stores, and all the major chains, simply won't sell M-rated games to anyone who can't show a picture ID showing them to be over 18. This doesn't stop the parent from buying the game and giving it to their kid without even looking at the rating and then complaining when it turns out the game makes "From Dusk 'til Dawn" look like a Saturday-morning cartoon. And frankly I see no way any rating system or any set of obligations or punishments aimed at the stores can do a single bloody thing about this.
Every time I hear one of the incidents of a kid getting an inappropriately-rated game, I notice that the kid didn't just go buy the game on his own. Almost always, a parent or some other adult authorized under all these proposed laws bought the game for the kid. I think the defense needed isn't more ratings. When a politician brings up the issue, someone stand up and name names and point out that the parent bought the game, then ask the politician flat-out what they're going to do about parents who buy their children these games and when are they going to start doing it. Cite their own example case back at them, and make them answer how their proposals are going to address the problem of parents doing the buying. If they try to weasel out, bring them back on point by noting that it was their example that involved the parent doing the buying, so why can't they address their own example?
It does ask for it. But IMHO it shouldn't ever not ask, because you shouldn't be running as admin except in unusual circumstances. UAC as Microsoft's presented it isn't a solution to a problem, it's an attempt to cover up the holes the problem causes. As we've see with zone-escalation attacks and suchlike, someone somewhere will find a way to make UAC believe it's already OK'd the action and, in an environment where you're running as admin and depending on UAC to protect you, you're now screwed.
That depends. On Debian the keyring's managed so that the only way for a program to find out whether the user's got privileges or not is to run, and if the user doesn't have privileges up pops the prompt box to alert the user that something's not right. There's no way for malware to tell when or whether the user's got privileges without triggering the prompt if they don't. And as I said, the timing can be tuned. 5 minutes is a pretty long time, 1 minute or less should suffice for most purposes. Remember that once a program's got elevated privileges it doesn't need the credentials again, the idle timeout is purely to make running of several administrative tasks in quick succession less bothersome. I can configure it to save the credentials only long enough to authenticate the current program and then drop them, at the cost of having to re-enter the password for a new program (but not the one I just ran, it's got privileges already and can pass them on to it's children).
Yeah, I can figure out ways to exploit the keyring, but in practice it's a real pain to write code to get it right without tipping off the user. And it's better than Microsoft's approach, which either teaches users to just click "OK" on any UAC dialog or leaves some actions happening without user authorization to avoid excessive dialogs. It's open what the "best" tuning would be, but the tuning's still a lot simpler than what Microsoft's proposing to do.
Yes, it has to wait. The problem is, the only way for it to find out whether the user's asked for root privs is to attempt an operation that needs them. Which will cause a prompt if the user hasn't got root privs at the moment. Which gets the correct result: the user now knows something's trying to get privs it shouldn't have.
And no, Linux and OS X don't suffer from viruses just because of their small market share. Unix in general has been the dominant OS in universities for 25 years or so. That's the environment that created several generations of crackers, hackers and generally the sorts of people who break systems. It's never, despite that, been as vulnerable as Windows even when faced with more attackers in an environment where the sysadmins had to actually give those attackers access to the systems. Unix-derived systems aren't as vulnerable to viruses simply because their design never placed usability over security. They were built from the beginning with a distinction between the administrative user (root) and ordinary users, and designed so that you didn't need administrative powers to do basic things like install software. You needed administrative powers to install software system-wide (because it'd be installed into system directories that ordinary users can't write to), but to simply install it under your home directory took no special privileges. Even letting other people use the copy you'd installed under your home directory took no special privileges. This means that malware has a much harder time convincing anyone to install it as root, and if it isn't installed as root it can't hide itself from root or affect the system itself. This, not lack of marketshare, is what makes Linux and OS X less vulnerable to malware than Windows.
The point of UAC is to make sure the user has to authorize any actions that need administrative privileges. So address the authorization instead of the actions. Do what my Debian box does when programs need root privileges. When I run a program like that from my normal user account, a wrapper prompts me to enter the root password or abort the operation. If I enter the password and it's correct, root credentials are added to my keyring temporarily and the program can run as root. As long as those credentials are on my keyring, any other programs that need root access can run without prompting. If the credentials remain unused for more than a short time, they're removed from my keyring and any programs after that that need root privs will cause a prompt again. This makes sure I have to manually authorize root access, but that I don't have to keep answering repetitive prompts. It doesn't require any fancy tuning of which actions prompt and which don't, at most it only needs tuning of how long root credentials remain on the keyring which is a lot simpler.
Typical Microsoft, crafting the most complicated solution to the problem.
I believe CRLs have been superseded by OCSP (Online Certificate Status Protocol) for just the reasons you noted. OCSP doesn't require local storage of revocation lists and the synchronization issues that go along with it.
You misunderstand the telco's position. Say I subscribe to Cox HSI for my Internet connection. I pay Cox every month for my connection and bandwidth. Google connects to the Internet through Level3. Google pays Level3 every month for their connection. What the telcos want is for Google to pay Cox for the bandwidth used when I search using Google. Yes, that's in addition to what I'm paying Cox, and what Google's paying Level3. And yes, it does in fact involve charging someone who's nto a customer. The penalty for not paying would be that traffic to and from Google gets an effectively lower priority on Cox's network, making Google slow compared to sites that're paying. The telcos say they won't penalize non-paying traffic, just give priority to paying traffic. It amounts to the same thing, though. Whether a gas station charges $3.00/gallon for cash with a 10-cent surcharge for paying by credit card or $3.10/gallon with a 10-cent discount for paying with cash, the end result's exactly the same.
I think there's an old comment that sums it up: "I don't want someone who'll work his ass off doing a job over and over and over. I want a lazy bum who'll figure out a way to not need to do the job again.". The industrious scribe invented the typewriter so the roomful of scribes could hand-copy documents faster. The lazy scribe invented the photocopier so he didn't need to hand-copy another document. The really lazy scribe invented the document-feeder and sorter/stacker attachments so he could go get lunch while the machine made 14 copies of the 357-page stack.
Note that being lazy can involve quite a bit of work. The trick is to remember that you're not trying to minimize the amount of work you're doing right this minute, you're trying to minimize the total amount of work you have to do over the long term.
Actually not. Standard and overnight delivery are in fact handled differently when it comes to how they're moved. Overnight delivery goes by more direct links that're faster but not as cost-efficient. As for the rest, I'm afraid it's wrong as well. UPS, FedEx and DHL get enough business for their overnight services from people who need delivery faster than reasonable and are willing to pay for it, they don't need to play games with their standard service. They know why they've pretty much taken the parcel business away from the US Postal Service, they're not about to make the same mistakes the USPS did.
Christopher Yoo got it, and his point is what's wrong with the telecomm's ideas. He's right, consumers should be able to pay for better delivery, just like when I order something shipped FedEx I can pay for regular delivery or I can pay more for overnight delivery depending on what I want. But that's not what the telecomms propose. That'd be like the telecomms saying "Consumer, you're using a lot of bandwidth. If you want to download streaming video you're going to have to pay for a higher-capacity link.". What the telecomms propose, though, is to not have the consumer pay for what they want but to have whoever the consumer's asking for stuff from pay. It's like my ordering something and paying for overnight shipping, and FedEx saying to the shipper "Right then. The customer's paid for standard shipping, but unless you pay us for overnight delivery we'll shove your package in the back and deliver it whenever we feel like it. Which may be never. Oh, and the extra just gets you standard delivery, real overnight will be yet more on top of that.". Of course the telecomms don't want to phrase it that way, because people understand FedEx and the extortion attempt's blatantly obvious.
What the studios are worried about isn't the average person. They're worried about the early HD adopters, the ones who have HD sets now. A lot of those aren't truly technical, but they think they're techies 'cause they've got the latest ooh-shiny gear. And they know what HD pictures look like now, and they'll be able to tell that ICT-downgraded discs don't look nearly as good as off-the-air HD. The studios are worried that, if they implement ICT now, when Joe Average asks Techie Neighbor with the fancy HD set what brand he should buy, TN will go "Don't bother. None of the discs out there do HD, so if you don't have HD cable and your set's working fine you might as well save your money.". And as long as the majority of people who've got HD sets are saying that, JA will keep on not buying HD equipment unless it's the only stuff available.
What the studios hope is that, given a few years, they'll be able to get sets with the proper support in them out there in quantity and the complaints when they turn on ICT will be a small minority. I think they've learned from DAT and SACD that the back door's the best way.
Yes, but Washington DC wasn't supposed to be a city in the conventional sense. It was supposed to be the seat of Federal government, period. The only people supposed to be living there would be Federal employees who weren't supposed to have a direct say in their own authority.
The people running Web sites, or creating software for that matter, might want to consider some of the consequences of their current crack-down on vulnerability reports. Yes, vulnerability reports are bad PR. However, if this keeps up people who find vulnerabilities will have only two feasible alternatives:
I'm in the habit of quoting large portions of articles, or even the entire article, for a purely practical reason: the mutability of Web pages. I've lost track of how often I've made a comment about something in an article, only to have a lot of people asking what I was talking about because the article said no such thing. On looking at the article again, the passage I was referring to had either been removed or altered to say something it hadn't said originally. The only way I have to combat this is to preserve a copy of the article as I originally read it in a place not subject to editing by the article's owner.
I'd note this after-the-fact rewriting tends to be most common where the original article contained egregiously and provably incorrect statements and the authors got called on the matter and now want to never have said that (as opposed to wanting to admit they mis-stated).