I meant in reference to the computer industry, and you knew that, but as an SI nerd you had to argue about it and alter my argument to make a point.
Uh, I'd hardly call myself an SI-nerd. I would call myself a chemist. Surprisingly enough, chemists tend to use SI units. Then again, so does just about every profession outside of the US, and quite a few inside the US. A noteworthy exception would be some people in the IT industry. But apparently not all - otherwise we wouldn't have various groups using the same abbreviation for two different things.
I'll be the first to admit that Seagate benefits from using the SI abbreviations, but that hardly makes them wrong.
However the end point should be that the expected number is on the hard drive, whether or not the characters after it are "GB" or "GiB". Power-of-10 capacities should not be given.
I want to know what my hard drive capacity is. For some strange reason I was born with ten appendages sticking out of my hands, oddly enough I prefer to use power-of-10 arithmetic. If I need to know how that translates into base-2, LBA, or whatever, I'll ask a computer (one of the many reasons I own one). Sure, I know how to do the conversion by hand as well, but while I might know how to add it doesn't stop me from using a calculator.
So, I'll take my hard drive capacities in GB please (and not GiB). Ditto for RAM, filesizes, and whatever else I need to know. And go ahead and round it all off - I really don't care if I have 107 GB or 107.374GB - 107GB is just fine. The inode table will probably waste more space than I'm not accounting for...
Oh, right, so the SI are the only people who have the right to use the letters 'K' 'M' and 'G', then?
No, but since the 18th century the prefixes have ALWAYS meant the same thing.
Suppose personally-tailored medications take off and I start a business selling them. I define the term "high quality" to mean "completely counterfeit and likely to kill you". I get sued after 45 people die. I point out that nobody owns a monopoly on the english language. Am I right?
Ignoring international conventions that are 200 years old is a recipe for confusion, which this issue completely highlights.
There wasn't an internationally standardised unit of data storage until 2000. Until then, with no official standard meaning to the letters "KB", "MB" and "GB", they were available for use for whatever the innovators who started using them wanted them to mean.
Ok, you have to pick one or the other. If no standard existed then why bash hard drive manufacturers for making up whatever they wanted to? If a standard existed then point to it.
It seems like a lot of folks around here are arguing "SI, who do they think they are?" Uh, they would be the most recognized standards body on the planet. And for the most part they're actually interested in standards and not politics (unlike so many other standards bodies that promote technology standards that are patent-encumbered). The only thing they are really concerned with is making sure that everybody's meterstick is the same length, and that people use terms that are unambiguous.
And they were first used with the meaning of 2^10, 2^20 and 2^30 bytes.
And the people who decided to do things this way laid the foundation for all the current problems. What were they thinking, anyway?
Most sciences/technologies deal in odd numbers. When they come up the usual solution isn't to just redefine SI prefixes to make the numbers nice and even. The solution is to round when appropriate, and just write out lots of digits when it isn't appropriate. Or to invent new units as convenience-units and register them with the SI.
There is no real compelling reason that the computer industry has to use base-2 to measure things. Sure, at the implementation level it comes down to base-2, but for practical use I care more about whether I have 1GB or 20GB of disk space free, and less about the possible 10% error in the ambiguous definition of a gigabyte.
The fact that SI took the side of the hard drive manufacturers should really tell you something...
I don't know - if you're doing something specialized out of your home where no enthusiast backchannel network exists I could see why somebody would be totally dependant on adwords.
Let's think of a hypothetical business model (which I know is probably not viable). Suppose you buy up used books for $0.10 each at the local library, and sell them online for $1 plus shipping. You set up adwords in conjunction with titles so that anybody searching for some 1980s best-seller gets a link to buy it for $1. You start getting sales, and since your scope is international you actually can stay in business. However, if the ads stop working then you'll never be able to attract hits to fredsusedbooks.com, and you don't have any local presence to fall back on.
There are a lot of business models that work on a national scale that wouldn't work for a local store-front. If you are in such a business and are smaller than netflix or amazon then you're going to be dependant on advertising to make it work.
Now, usually these kinds of business models target an enthusiast market of some kind. For example, there was a small business that did nothing but upgrade hard drives on Tivos - you FedEx them a Tivo they FedEx an expanded one back. Such a business could probably survive an adwords outage because the various Tivo mailing lists would keep their name in play for six months.
Adwords has enabled lots of small businesses to be viable that never would have been viable otherwise. Sure, you could call that being marginal, but it is still good for the economy. Services get provided that otherwise wouldn't, for prices cheaper than they would otherwise be, often with better customer service than your general mega-chain.
That's because the folks that originally operated the domain name system were left to do their jobs without too much politicizing.
That would be because nobody had heard of the Internet back in 1990...
Try it today and I guarantee that it will become VERY political. I'm sure that back when the Secretary of Commerce and Labor was regulating ship-to-shore transmissions and police bands it was a very non-political task. The FCC of today is a completely different story - largely because the stakes have risen.
Now lets see if he can actually herald the creation of some new tlds.
As long as they don't allow a period of time in advance for trademark and current domain holders to register corresponding domains in the new TLD. What's the point in having a new TLD if it just becomes a copy of.COM?
Sure, I can understand why Ford wouldn't want the owners of ihateford.com to register ford.new when it comes out. So, instead they lobby for a 30-day window or whatever so that they can grab ford.new before anybody else can. But then EVERYBODY else does the same thing, and before you know it the.new zone is essentially identical to.com, but everybody gets to pay twice as much per year for their domain registrations.
I can see the use for new TLDs when they actually mean something or have some functional purpose, but not if they're just generic domains like com/net/biz/info/whatever. The use of domains would actually need to be enforced for any new TLD scheme to work (ie if you put a standard website on a.mobi domain you lose it - or if you just redirect it to your.com site).
I appreciate that there are SI units, but they are not in common use apart from total SI/standards nerds, and their creation was misguided and based upon a core misunderstanding of the issue.
Uh, the SI prefixes are used EVERYWHERE, and with the exception of IT it always referrs to a power of 10.
The only reason that a KB of ram = 1024 bytes is because some computer designer was lazy 50 years ago. In EVERY other industry the SI terms refer to a power of 10.
The SI didn't misunderstand anything when they suggested using a different prefix for powers-of-2. Their whole point is that having one prefix mean two different things is completely contrary to the whole point of the metric system. And they're right. And the fact that a manufacturer is being sued and everybody is arguing over what "kilo" ought to mean just proves that the SI has a point...
Uh, when you factor in partitioning, filesystems, etc, they're going to be disappointed no matter what.
And they wouldn't be disappointed if the OS reported disk space in SI units and not IT-units. The problem is that for whatever reason nobody in the IT world can agree on one set of units to use for everything. Most of the rest of the world formed SI for this reason, and for whatever reason IT had to do its own thing...
And while how much you fit on a hard drive platter is arbitrary, you're reading and storing base 2 data in base 2 quantities from base 2 machines through base 2 controller chips, so the drive capacity measurements really should be in base 2.
I don't see how one follows the other.
The numbers in my spreadsheet are probably stored as some crazy combination of binary mantissa and exponent with signs and twos-complement and all that - and yet we just display them in good-old base 10.
Sure, back when drive sizes tended to fall on addressing boundaries the distinction probably mattered more (sticking a huge drive in a computer with 16-bit sector addressing would be wasteful). However, these days everybody is thinking 64- or even 128-bit - it will be a while before hard drives catch up with the addressing schemes like they used to back in the 90s.
Everything your computer does is in base-2. Even the clock counts in base-2 - just so far to the right of the decimal point that you don't notice it (sure, the ticks might come at some odd interval, but the actual time calculated after any tick is rounded to a power of 2 at some point). Software is just kind enough to store more precision than is rendered on-screen so that when you type in 12.3 you don't see 12.299999999999999964532 show up (or whatever it ends up being stored as internally).
Considering that 99% of the time use of disk space is presented to the user in rounded form, why not go ahead and just use the SI units? If a filesize of 18.2 MiB is good enough, then a size of 18.3MB should be fine too. If somebody really cares about a difference of 1% then they'll want to know that the file size is 18221048 bytes (plus x bytes of allocated empty space).
So SI can use base 24 for hours, base 60 for minutes and seconds yet can't handle base 2 for bits?
Yes, but they would be right to complain if somebody started calling minutes centihours, or hectoseconds. A month could be a kilohour.
And the Americans could be metric after all - miles could be kilofathoms.
So now kilo could mean anywhere from 500ish to 1500ish, and we'd just know what factor to use based upon the application. Isn't this metric system easy?
The whole point of the metric system is that EVERYTHING uses the standard SI prefixes - in powers of 10. For whatever reason we tolerate seconds, minutes, hours, and days, but I don't think that every new discipline that comes along should go ahead and violate the SI. Switching to metric can be a painful transition - why on earth would we invent new disciplines that AREN'T metric already?
SI's complaint wasn't that IT folks were using powers-of-2. Their complaint was that they were using the SI prefixs to mean something different than what they mean EVERYWHERE else. The SI doesn't complain about Americans measuring stuff in miles, but they would complain if they just redefined the meter to equal the yard and the kilometer to equal the mile (with the kilo prefix now being an easy-to-remember 1760).
Which, interestingly enough, they were doing long before the SI standardised them.
Uh, the SI standardized the unit prefixes in 1960 (well, at least the prefixes relevant to consumer hard drives). Before that they were standardized by just about anybody using the metric system - starting around the 18th century or so (the concept is much older, but that is when it really took off).
Lazy programmers started using the base-2 prefixes because it was easier to say 1 KB than 1.024KB. However, for any other application worldwide the kilo prefix would mean 1000.
The only thing the SI did recently is point out that everybody is misusing the prefixes that have been standardized for centuries - and suggest some new ones for lazy programmers to use.
I think the whole lawsuit is silly. Who could have possibly been confused by the claims of hard drive makers? Those who are computer illiterate would assume that 1 MB = 1*10^6 bytes - and they'd be right with respect to hard drive claims. Those who are computer literate would know that hard drive manufacturers routinely measure everything in powers of 10.
Even if both sides don't have the absolute best interests in mind, there is certainly a lesser of two evils.
And that would be the one with the registration opposite yours?
I find it ironic that the one thing that 99% of D's and R's agree on is that the whole country will go downhill if the other party's candidate gets elected.
Then again, I guess that they're both right most of the time! I suppose that is why I don't vote for either unless I actually am happy with one of them.
And yes, I'd love to see a more proportionate system of government like 99% of the rest of the democratic world...
Shouldn't the states be in charge of when the Primaries are held?
Uh, no. His point was that states shouldn't have anything to do with primaries. Political parties might choose to have internal primary elections. They might choose to organize these along state lines. They're welcome to do them however they'd like to. If they want to burn the ballots before they're counted and appoint whoever the party leadership likes that shouldn't be a matter of state concern except to the degree that it violates the same laws that deal with the administration of any other non-profit's bylaws. It would probably be a civil matter for the members of the party to bring forth.
If two parties want to coordinate their elections on the same day that is fine as well. However, they shouldn't be officially regulated, use official resources (voting machiens, etc), and they shouldn't have any kind of official regulation unless the parties want to hire local cops much as a neighborhood might do when having a block party.
Parties should not be part of the official governance of the US. They shouldn't be banned or anything like that, but government shouldn't be deciding who does or doesn't get official support as a party - it amounts to preferential treatment for particular non-governmental organizations that do not answer to the people (beyond their membership).
That is almost trivial to accomplish. Give each article a "Notable" flag. If it is set it shows up by default in all searches. If it is not set it does not. The flag can be changed at any time as appropriate. There would be a search option to search un-notable articles, and this could be set as a default on a per-user basis.
But then we couldn't argue about whether topics are worthy of inclusion all day...
Also I would postulate that even if as a webcomic writer you obviously know a lot more webcomics than an average schmuck like me, that is only a justification why _not_ to include them in an encyclopedia.
Yes, but who cares about Propylene Oxide outside of a few specific industries? Would the average wikipedia reader even know what it is? How about the Blue-Gray mouse?
If we delete every article that isn't common knowledge then what is the point of having an encyclopedia in the first place? If a topic is of general interest to some segment of society (beyond just a few individuals) why not allow it to be given an article?
If there is a webcomic that just about ANY webcomic author would know about, then it is noteworthy. I don't know anything about milking cows, but that doesn't make the most common method used to perform this task non-noteworthy, even if it hasn't been published in the New York Times...
No, VMs can provide services to one another through the hypervisor. I'd expect the TCP stack will in the VM considered to be "The OS". I'd expect nearly all applications to run in this OS VM as well, some of them using security services from other VMs.
Doesn't being able to communicate through the hypervisor turn the hypervisor into essentially a microkernel? The whole point of a hypervisor is that it completely isolates the OSes that run under it. When various components under the hypervisor talk to each other but are generally protected from each other we call those components applications, and the "hypervisor" is called an OS. How is your proposed hypervisor any different from a microkernel-based OS, such as Windows? And if an application has to rely on various components to do its job, how is that different from what we have today - just as many opportunities for security holes...
Vista will never function in the sort of environment I'm talking about. Doing this is going to require a radical restructuring of how an operating system works. I expect FLOSS to get there first.
I think the idea is that it is more-or-less already there - Vista IS the "hypervisor" you describe. It just isn't a hypervisor, but neither is what you propose. It will eventually support remote attestation of applications running under it, and it is designed to provide security to the individual applications running on top of it. Of course, we can question whether this security is airtight, but that is true of ANY particular software implementation - particularly a complex one. A very thin but impenetrable hypervisor can in theory be made fairly secure, but you're not proposing a very thin hypervisor but essentially a fairly complex and porous one.
The bottom line is that I don't want the companies I do business with dictating what software I use to do business with them. If they want to recommend or support a given set of software that is of course fine, but they shouldn't be able to enforce my usage of that software...
Not to mention that for what we could be saving by transitioning to something other than the shuttle we could probably send up a new Hubble every year or two (sure, the cost of the Hubble was huge - but that was for ONE - the second has to be a lot cheaper). It is like buying non-refundable tickets - the 1/100 flights you cancel doesn't cost nearly as much as what you'd pay every time for the right to cancel at the last minute.
The brokerage could provide its own secure VM (no OS, just the core functionality required, so it's very small) that lives on top of a standard hypervisor. Any operating system could be running -- the brokerage would have no reason to care. As long as the hypervisor and special-purpose VM hash to the correct values, you're golden.
Sure, and that would be fine, but:
1. I have to trust the hypervisor - I don't get any choice there. In theory if it is minimal and FOSS that isn't a problem. Of course, it sounds like it might be GPL-3 incompatible due to the anti-tivoization clauses (it probably would depend on implementation, and it could lead to a GPL-4 to intentionally break compatibility depending on how this goes).
2. MS has to decide to promote the hypervisor. Otherwise this is all a pipe dream.
3. Instead of apps running on an OS we now have dozens of applications BEING their own OS. Everybody has their own TCP stack, GUI, etc. If this stuff moves to the hypervisor then it is no longer merely a hypervisor - it is an OS.
4. Everybody has to create these runs-on-bare-hypervisor apps instead of just using Palladium, which is what will most likely happen in actuality.
This idea sounds nice and in a hypothetical sense addresses my concerns, but it isn't likely to become reality. What is likely to become reality is that Vista becomes the hypervisor, and at best I might be able to run linux in a VM on vista - not the other way around.
It will also eliminate the ability of, for example, your stock brokerage to remotely check the security of your machine before allowing large transactions. There are lots of legitimate uses for remote attestation.
Ie, the ability for your stock brokerage to turn away customers using anything other than a few flavors of windows with their preferred web browser. For remote attestation to be of any use companies are forced to pick particular software vendors that they want to prefer. Within an organization that is fine - if I deploy 1000 machines I know what I imaged them with. However, once you leave the organization boundary it isn't appropriate for outsiders to dictate what software you use.
It still drives me nuts that I have to boot up vmware to run half the windows software I can't find a viable linux replacement for. If remote attestation takes off I won't even be able to do that - I'd need windows running natively on blessed hardware. I dual-boot loader might even potentially mess up the chain of trust - we're talking about dedicated hardware, or using boot CDs to access linux.
Computers shouldn't be tattling on their owners. Sure, I can defeat this on my own computer, but if 99% of the population doesn't then suddenly my ISP thinks I have a virus, my bank won't do business with me, and I can't watch videos on youtube (hmm - I guess no major loss there). Suddenly you're effectively punished by not running the exact same software as everybody else...
Assuming no one has taken control of their machine -- which is the whole point of attestation; having a way to verify that no one has tampered with your box, and to be able to do so remotely.
How is a PAPER copy of the machine's private key in the box the machine came in going to help a hacker who has taken control of the machine defeat attestation? The sysadmin would lock it up in a safe - without the piece of paper the mechanism could not be defeated. The goal is to allow a computer owner to defeat the mechanism if they feel the need, while still allowing the full utility of attestation TO THE MACHINE OWNER.
I still assert that machines not be distributed with certified keys. If a sysadmin wants to certify the validity of TPM modules they can do so on their own - this won't really deter any legitimate use of TPM - just the DRM aspects.
If it were possible to falsify an attestation message, then sysadmins wouldn't be able to remotely verify the software their systems are running, which is the goal of the TC committee members like IBM and Intel (who don't care about DRM but do care about security
Note that I said any SYSTEM OWNER - not anybody who happens to log into the computer.
Sysadmins can rest assured their systems are running just fine. They'd be the only ones able to falsify attestation messages on their own PCs.
It's much better to be able to generate a new random key than to get a dump of the existing key. The latter voids all semblance of security.
I'm not convinced of this. Key escrow and management is generally considered an important part of any encryption solution. Suppose something goes wrong and you want to be able to recover a hard drive - perhaps the motherboard was smashed and you need to get at data on the hard drive of a PC that uses TPM.
Also - PCs using TPM should be issued with random keys and without any certificates (ie their trust state should be a clean slate) - not with certificates issued by whoever made them. Otherwise anybody who does "take ownership" of their PC will suddenly find themselves locked out of just about any kind of multimedia in the future - and who knows what else. If this becomes the norm then it will be a feature that nobody will be able to use - hence eliminating its value...
You're missing the point. Of course that is what those pushing TC WANT the system to do. It shouldn't be allowed to take off.
There is a lot in TC that is good. You could potentially eliminate viruses, for example, with a hardware-backed chain of trust. The issue is that the chain of trust should lead back to the computer owner - not the computer manufacturer.
This one can't happen. No one is policing *your* computer, the worst that could theoretically happen is someone *else* will deny service to you if you're not running the right software. If you have the word file, no one can stop you from reading it.
Have you read about Palladium? Software will be able to tell the OS to keep a file in "protected storage" where other programs can't read it. Sure, you have the file, but it is encrypted. Most of the OS partition will also be encrypted. The decryption key will be stored in a chip with instructions to only yield the key to a program that has a given hash. The whole system is designed to make it VERY difficult to bypass - even with physical access to the machine. You'd need to tear apart the TCPM chip to get the key out. Fileservers hosting documents will only allow machiens with secure OSes to download them - directly to protected storage.
The design isn't actually that difficult. Sure, like all DRM it is vulnerable to physical-level attacks, but those can be made very expensive. And if a given model of TCPM chip turns out to be vulnerable it will be removed from the chain of trust - no new multimedia for anything running on those systems.
The solution is simple - require anybody selling a computer to give a paper copy of any keys embedded in the hardware. Most people won't use them, but if TC gets out of control then owners can download programs and type in their keys and bypass the whole thing (or keep only the parts they want).
TC is coming whether you like it or not. You can't stop a technology. Your best bet is to prepare. That is, if you're truly afraid of this scenario happening, vote for politicians that are willing to regulate monopolies. You can't stop people from researching this stuff.
Nobody wants to stop technology. We just want to make sure it is properly used. The simplest solution is to require that every computer include a printed copy of any and all keys embedded in it, and ideally any private keys associated with any public-keys installed on it. That will keep everybody honest. If TC gets out of control people will write software that lets an average owner just type in their machine key and then any onerous component of TC can be bypassed. You'd still be able to protect from viruses since the average computer owner won't go typing in a 200-character hex key to run some cute program a friend emailed them.
I'd love having TC on my computer if I COULD TRUST IT. A hardware security mechanism that works FOR me is great - one that works AGAINST me is not.
Slashdot Mirror
I meant in reference to the computer industry, and you knew that, but as an SI nerd you had to argue about it and alter my argument to make a point.
Uh, I'd hardly call myself an SI-nerd. I would call myself a chemist. Surprisingly enough, chemists tend to use SI units. Then again, so does just about every profession outside of the US, and quite a few inside the US. A noteworthy exception would be some people in the IT industry. But apparently not all - otherwise we wouldn't have various groups using the same abbreviation for two different things.
I'll be the first to admit that Seagate benefits from using the SI abbreviations, but that hardly makes them wrong.
However the end point should be that the expected number is on the hard drive, whether or not the characters after it are "GB" or "GiB". Power-of-10 capacities should not be given.
I want to know what my hard drive capacity is. For some strange reason I was born with ten appendages sticking out of my hands, oddly enough I prefer to use power-of-10 arithmetic. If I need to know how that translates into base-2, LBA, or whatever, I'll ask a computer (one of the many reasons I own one). Sure, I know how to do the conversion by hand as well, but while I might know how to add it doesn't stop me from using a calculator.
So, I'll take my hard drive capacities in GB please (and not GiB). Ditto for RAM, filesizes, and whatever else I need to know. And go ahead and round it all off - I really don't care if I have 107 GB or 107.374GB - 107GB is just fine. The inode table will probably waste more space than I'm not accounting for...
Just to be clear: It shouldn't be Watt hours or Watts per hour. It should just be Watts.
:)
Well, I guess you could state it in terms of Wh/h or even kWh/h - but I'm not quite sure why you'd want to do that...
Oh, right, so the SI are the only people who have the right to use the letters 'K' 'M' and 'G', then?
No, but since the 18th century the prefixes have ALWAYS meant the same thing.
Suppose personally-tailored medications take off and I start a business selling them. I define the term "high quality" to mean "completely counterfeit and likely to kill you". I get sued after 45 people die. I point out that nobody owns a monopoly on the english language. Am I right?
Ignoring international conventions that are 200 years old is a recipe for confusion, which this issue completely highlights.
There wasn't an internationally standardised unit of data storage until 2000. Until then, with no official standard meaning to the letters "KB", "MB" and "GB", they were available for use for whatever the innovators who started using them wanted them to mean.
Ok, you have to pick one or the other. If no standard existed then why bash hard drive manufacturers for making up whatever they wanted to? If a standard existed then point to it.
It seems like a lot of folks around here are arguing "SI, who do they think they are?" Uh, they would be the most recognized standards body on the planet. And for the most part they're actually interested in standards and not politics (unlike so many other standards bodies that promote technology standards that are patent-encumbered). The only thing they are really concerned with is making sure that everybody's meterstick is the same length, and that people use terms that are unambiguous.
And they were first used with the meaning of 2^10, 2^20 and 2^30 bytes.
And the people who decided to do things this way laid the foundation for all the current problems. What were they thinking, anyway?
Most sciences/technologies deal in odd numbers. When they come up the usual solution isn't to just redefine SI prefixes to make the numbers nice and even. The solution is to round when appropriate, and just write out lots of digits when it isn't appropriate. Or to invent new units as convenience-units and register them with the SI.
There is no real compelling reason that the computer industry has to use base-2 to measure things. Sure, at the implementation level it comes down to base-2, but for practical use I care more about whether I have 1GB or 20GB of disk space free, and less about the possible 10% error in the ambiguous definition of a gigabyte.
The fact that SI took the side of the hard drive manufacturers should really tell you something...
I don't know - if you're doing something specialized out of your home where no enthusiast backchannel network exists I could see why somebody would be totally dependant on adwords.
Let's think of a hypothetical business model (which I know is probably not viable). Suppose you buy up used books for $0.10 each at the local library, and sell them online for $1 plus shipping. You set up adwords in conjunction with titles so that anybody searching for some 1980s best-seller gets a link to buy it for $1. You start getting sales, and since your scope is international you actually can stay in business. However, if the ads stop working then you'll never be able to attract hits to fredsusedbooks.com, and you don't have any local presence to fall back on.
There are a lot of business models that work on a national scale that wouldn't work for a local store-front. If you are in such a business and are smaller than netflix or amazon then you're going to be dependant on advertising to make it work.
Now, usually these kinds of business models target an enthusiast market of some kind. For example, there was a small business that did nothing but upgrade hard drives on Tivos - you FedEx them a Tivo they FedEx an expanded one back. Such a business could probably survive an adwords outage because the various Tivo mailing lists would keep their name in play for six months.
Adwords has enabled lots of small businesses to be viable that never would have been viable otherwise. Sure, you could call that being marginal, but it is still good for the economy. Services get provided that otherwise wouldn't, for prices cheaper than they would otherwise be, often with better customer service than your general mega-chain.
That's because the folks that originally operated the domain name system were left to do their jobs without too much politicizing.
That would be because nobody had heard of the Internet back in 1990...
Try it today and I guarantee that it will become VERY political. I'm sure that back when the Secretary of Commerce and Labor was regulating ship-to-shore transmissions and police bands it was a very non-political task. The FCC of today is a completely different story - largely because the stakes have risen.
Now lets see if he can actually herald the creation of some new tlds.
.COM?
.new zone is essentially identical to .com, but everybody gets to pay twice as much per year for their domain registrations.
.mobi domain you lose it - or if you just redirect it to your .com site).
As long as they don't allow a period of time in advance for trademark and current domain holders to register corresponding domains in the new TLD. What's the point in having a new TLD if it just becomes a copy of
Sure, I can understand why Ford wouldn't want the owners of ihateford.com to register ford.new when it comes out. So, instead they lobby for a 30-day window or whatever so that they can grab ford.new before anybody else can. But then EVERYBODY else does the same thing, and before you know it the
I can see the use for new TLDs when they actually mean something or have some functional purpose, but not if they're just generic domains like com/net/biz/info/whatever. The use of domains would actually need to be enforced for any new TLD scheme to work (ie if you put a standard website on a
I appreciate that there are SI units, but they are not in common use apart from total SI/standards nerds, and their creation was misguided and based upon a core misunderstanding of the issue.
Uh, the SI prefixes are used EVERYWHERE, and with the exception of IT it always referrs to a power of 10.
The only reason that a KB of ram = 1024 bytes is because some computer designer was lazy 50 years ago. In EVERY other industry the SI terms refer to a power of 10.
The SI didn't misunderstand anything when they suggested using a different prefix for powers-of-2. Their whole point is that having one prefix mean two different things is completely contrary to the whole point of the metric system. And they're right. And the fact that a manufacturer is being sued and everybody is arguing over what "kilo" ought to mean just proves that the SI has a point...
Uh, when you factor in partitioning, filesystems, etc, they're going to be disappointed no matter what.
And they wouldn't be disappointed if the OS reported disk space in SI units and not IT-units. The problem is that for whatever reason nobody in the IT world can agree on one set of units to use for everything. Most of the rest of the world formed SI for this reason, and for whatever reason IT had to do its own thing...
And while how much you fit on a hard drive platter is arbitrary, you're reading and storing base 2 data in base 2 quantities from base 2 machines through base 2 controller chips, so the drive capacity measurements really should be in base 2.
I don't see how one follows the other.
The numbers in my spreadsheet are probably stored as some crazy combination of binary mantissa and exponent with signs and twos-complement and all that - and yet we just display them in good-old base 10.
Sure, back when drive sizes tended to fall on addressing boundaries the distinction probably mattered more (sticking a huge drive in a computer with 16-bit sector addressing would be wasteful). However, these days everybody is thinking 64- or even 128-bit - it will be a while before hard drives catch up with the addressing schemes like they used to back in the 90s.
Everything your computer does is in base-2. Even the clock counts in base-2 - just so far to the right of the decimal point that you don't notice it (sure, the ticks might come at some odd interval, but the actual time calculated after any tick is rounded to a power of 2 at some point). Software is just kind enough to store more precision than is rendered on-screen so that when you type in 12.3 you don't see 12.299999999999999964532 show up (or whatever it ends up being stored as internally).
Considering that 99% of the time use of disk space is presented to the user in rounded form, why not go ahead and just use the SI units? If a filesize of 18.2 MiB is good enough, then a size of 18.3MB should be fine too. If somebody really cares about a difference of 1% then they'll want to know that the file size is 18221048 bytes (plus x bytes of allocated empty space).
So SI can use base 24 for hours, base 60 for minutes and seconds yet can't handle base 2 for bits?
Yes, but they would be right to complain if somebody started calling minutes centihours, or hectoseconds. A month could be a kilohour.
And the Americans could be metric after all - miles could be kilofathoms.
So now kilo could mean anywhere from 500ish to 1500ish, and we'd just know what factor to use based upon the application. Isn't this metric system easy?
The whole point of the metric system is that EVERYTHING uses the standard SI prefixes - in powers of 10. For whatever reason we tolerate seconds, minutes, hours, and days, but I don't think that every new discipline that comes along should go ahead and violate the SI. Switching to metric can be a painful transition - why on earth would we invent new disciplines that AREN'T metric already?
SI's complaint wasn't that IT folks were using powers-of-2. Their complaint was that they were using the SI prefixs to mean something different than what they mean EVERYWHERE else. The SI doesn't complain about Americans measuring stuff in miles, but they would complain if they just redefined the meter to equal the yard and the kilometer to equal the mile (with the kilo prefix now being an easy-to-remember 1760).
Which, interestingly enough, they were doing long before the SI standardised them.
Uh, the SI standardized the unit prefixes in 1960 (well, at least the prefixes relevant to consumer hard drives). Before that they were standardized by just about anybody using the metric system - starting around the 18th century or so (the concept is much older, but that is when it really took off).
Lazy programmers started using the base-2 prefixes because it was easier to say 1 KB than 1.024KB. However, for any other application worldwide the kilo prefix would mean 1000.
The only thing the SI did recently is point out that everybody is misusing the prefixes that have been standardized for centuries - and suggest some new ones for lazy programmers to use.
I think the whole lawsuit is silly. Who could have possibly been confused by the claims of hard drive makers? Those who are computer illiterate would assume that 1 MB = 1*10^6 bytes - and they'd be right with respect to hard drive claims. Those who are computer literate would know that hard drive manufacturers routinely measure everything in powers of 10.
Even if both sides don't have the absolute best interests in mind, there is certainly a lesser of two evils.
And that would be the one with the registration opposite yours?
I find it ironic that the one thing that 99% of D's and R's agree on is that the whole country will go downhill if the other party's candidate gets elected.
Then again, I guess that they're both right most of the time! I suppose that is why I don't vote for either unless I actually am happy with one of them.
And yes, I'd love to see a more proportionate system of government like 99% of the rest of the democratic world...
Shouldn't the states be in charge of when the Primaries are held?
Uh, no. His point was that states shouldn't have anything to do with primaries. Political parties might choose to have internal primary elections. They might choose to organize these along state lines. They're welcome to do them however they'd like to. If they want to burn the ballots before they're counted and appoint whoever the party leadership likes that shouldn't be a matter of state concern except to the degree that it violates the same laws that deal with the administration of any other non-profit's bylaws. It would probably be a civil matter for the members of the party to bring forth.
If two parties want to coordinate their elections on the same day that is fine as well. However, they shouldn't be officially regulated, use official resources (voting machiens, etc), and they shouldn't have any kind of official regulation unless the parties want to hire local cops much as a neighborhood might do when having a block party.
Parties should not be part of the official governance of the US. They shouldn't be banned or anything like that, but government shouldn't be deciding who does or doesn't get official support as a party - it amounts to preferential treatment for particular non-governmental organizations that do not answer to the people (beyond their membership).
That is almost trivial to accomplish. Give each article a "Notable" flag. If it is set it shows up by default in all searches. If it is not set it does not. The flag can be changed at any time as appropriate. There would be a search option to search un-notable articles, and this could be set as a default on a per-user basis.
But then we couldn't argue about whether topics are worthy of inclusion all day...
Also I would postulate that even if as a webcomic writer you obviously know a lot more webcomics than an average schmuck like me, that is only a justification why _not_ to include them in an encyclopedia.
Yes, but who cares about Propylene Oxide outside of a few specific industries? Would the average wikipedia reader even know what it is? How about the Blue-Gray mouse?
If we delete every article that isn't common knowledge then what is the point of having an encyclopedia in the first place? If a topic is of general interest to some segment of society (beyond just a few individuals) why not allow it to be given an article?
If there is a webcomic that just about ANY webcomic author would know about, then it is noteworthy. I don't know anything about milking cows, but that doesn't make the most common method used to perform this task non-noteworthy, even if it hasn't been published in the New York Times...
No, VMs can provide services to one another through the hypervisor. I'd expect the TCP stack will in the VM considered to be "The OS". I'd expect nearly all applications to run in this OS VM as well, some of them using security services from other VMs.
Doesn't being able to communicate through the hypervisor turn the hypervisor into essentially a microkernel? The whole point of a hypervisor is that it completely isolates the OSes that run under it. When various components under the hypervisor talk to each other but are generally protected from each other we call those components applications, and the "hypervisor" is called an OS. How is your proposed hypervisor any different from a microkernel-based OS, such as Windows? And if an application has to rely on various components to do its job, how is that different from what we have today - just as many opportunities for security holes...
Vista will never function in the sort of environment I'm talking about. Doing this is going to require a radical restructuring of how an operating system works. I expect FLOSS to get there first.
I think the idea is that it is more-or-less already there - Vista IS the "hypervisor" you describe. It just isn't a hypervisor, but neither is what you propose. It will eventually support remote attestation of applications running under it, and it is designed to provide security to the individual applications running on top of it. Of course, we can question whether this security is airtight, but that is true of ANY particular software implementation - particularly a complex one. A very thin but impenetrable hypervisor can in theory be made fairly secure, but you're not proposing a very thin hypervisor but essentially a fairly complex and porous one.
The bottom line is that I don't want the companies I do business with dictating what software I use to do business with them. If they want to recommend or support a given set of software that is of course fine, but they shouldn't be able to enforce my usage of that software...
Not to mention that for what we could be saving by transitioning to something other than the shuttle we could probably send up a new Hubble every year or two (sure, the cost of the Hubble was huge - but that was for ONE - the second has to be a lot cheaper). It is like buying non-refundable tickets - the 1/100 flights you cancel doesn't cost nearly as much as what you'd pay every time for the right to cancel at the last minute.
The brokerage could provide its own secure VM (no OS, just the core functionality required, so it's very small) that lives on top of a standard hypervisor. Any operating system could be running -- the brokerage would have no reason to care. As long as the hypervisor and special-purpose VM hash to the correct values, you're golden.
Sure, and that would be fine, but:
1. I have to trust the hypervisor - I don't get any choice there. In theory if it is minimal and FOSS that isn't a problem. Of course, it sounds like it might be GPL-3 incompatible due to the anti-tivoization clauses (it probably would depend on implementation, and it could lead to a GPL-4 to intentionally break compatibility depending on how this goes).
2. MS has to decide to promote the hypervisor. Otherwise this is all a pipe dream.
3. Instead of apps running on an OS we now have dozens of applications BEING their own OS. Everybody has their own TCP stack, GUI, etc. If this stuff moves to the hypervisor then it is no longer merely a hypervisor - it is an OS.
4. Everybody has to create these runs-on-bare-hypervisor apps instead of just using Palladium, which is what will most likely happen in actuality.
This idea sounds nice and in a hypothetical sense addresses my concerns, but it isn't likely to become reality. What is likely to become reality is that Vista becomes the hypervisor, and at best I might be able to run linux in a VM on vista - not the other way around.
It will also eliminate the ability of, for example, your stock brokerage to remotely check the security of your machine before allowing large transactions. There are lots of legitimate uses for remote attestation.
Ie, the ability for your stock brokerage to turn away customers using anything other than a few flavors of windows with their preferred web browser. For remote attestation to be of any use companies are forced to pick particular software vendors that they want to prefer. Within an organization that is fine - if I deploy 1000 machines I know what I imaged them with. However, once you leave the organization boundary it isn't appropriate for outsiders to dictate what software you use.
It still drives me nuts that I have to boot up vmware to run half the windows software I can't find a viable linux replacement for. If remote attestation takes off I won't even be able to do that - I'd need windows running natively on blessed hardware. I dual-boot loader might even potentially mess up the chain of trust - we're talking about dedicated hardware, or using boot CDs to access linux.
Computers shouldn't be tattling on their owners. Sure, I can defeat this on my own computer, but if 99% of the population doesn't then suddenly my ISP thinks I have a virus, my bank won't do business with me, and I can't watch videos on youtube (hmm - I guess no major loss there). Suddenly you're effectively punished by not running the exact same software as everybody else...
Assuming no one has taken control of their machine -- which is the whole point of attestation; having a way to verify that no one has tampered with your box, and to be able to do so remotely.
How is a PAPER copy of the machine's private key in the box the machine came in going to help a hacker who has taken control of the machine defeat attestation? The sysadmin would lock it up in a safe - without the piece of paper the mechanism could not be defeated. The goal is to allow a computer owner to defeat the mechanism if they feel the need, while still allowing the full utility of attestation TO THE MACHINE OWNER.
I still assert that machines not be distributed with certified keys. If a sysadmin wants to certify the validity of TPM modules they can do so on their own - this won't really deter any legitimate use of TPM - just the DRM aspects.
Well, yes, until somebody lets some dry in a sealed glass vessel.
I'd like to see more "real" chemistry in high school, but NI3 would not be among the experiments I'd encourage...
If it were possible to falsify an attestation message, then sysadmins wouldn't be able to remotely verify the software their systems are running, which is the goal of the TC committee members like IBM and Intel (who don't care about DRM but do care about security
Note that I said any SYSTEM OWNER - not anybody who happens to log into the computer.
Sysadmins can rest assured their systems are running just fine. They'd be the only ones able to falsify attestation messages on their own PCs.
It's much better to be able to generate a new random key than to get a dump of the existing key. The latter voids all semblance of security.
I'm not convinced of this. Key escrow and management is generally considered an important part of any encryption solution. Suppose something goes wrong and you want to be able to recover a hard drive - perhaps the motherboard was smashed and you need to get at data on the hard drive of a PC that uses TPM.
Also - PCs using TPM should be issued with random keys and without any certificates (ie their trust state should be a clean slate) - not with certificates issued by whoever made them. Otherwise anybody who does "take ownership" of their PC will suddenly find themselves locked out of just about any kind of multimedia in the future - and who knows what else. If this becomes the norm then it will be a feature that nobody will be able to use - hence eliminating its value...
You're missing the point. Of course that is what those pushing TC WANT the system to do. It shouldn't be allowed to take off.
There is a lot in TC that is good. You could potentially eliminate viruses, for example, with a hardware-backed chain of trust. The issue is that the chain of trust should lead back to the computer owner - not the computer manufacturer.
This one can't happen. No one is policing *your* computer, the worst that could theoretically happen is someone *else* will deny service to you if you're not running the right software. If you have the word file, no one can stop you from reading it.
Have you read about Palladium? Software will be able to tell the OS to keep a file in "protected storage" where other programs can't read it. Sure, you have the file, but it is encrypted. Most of the OS partition will also be encrypted. The decryption key will be stored in a chip with instructions to only yield the key to a program that has a given hash. The whole system is designed to make it VERY difficult to bypass - even with physical access to the machine. You'd need to tear apart the TCPM chip to get the key out. Fileservers hosting documents will only allow machiens with secure OSes to download them - directly to protected storage.
The design isn't actually that difficult. Sure, like all DRM it is vulnerable to physical-level attacks, but those can be made very expensive. And if a given model of TCPM chip turns out to be vulnerable it will be removed from the chain of trust - no new multimedia for anything running on those systems.
The solution is simple - require anybody selling a computer to give a paper copy of any keys embedded in the hardware. Most people won't use them, but if TC gets out of control then owners can download programs and type in their keys and bypass the whole thing (or keep only the parts they want).
TC is coming whether you like it or not. You can't stop a technology. Your best bet is to prepare. That is, if you're truly afraid of this scenario happening, vote for politicians that are willing to regulate monopolies. You can't stop people from researching this stuff.
Nobody wants to stop technology. We just want to make sure it is properly used. The simplest solution is to require that every computer include a printed copy of any and all keys embedded in it, and ideally any private keys associated with any public-keys installed on it. That will keep everybody honest. If TC gets out of control people will write software that lets an average owner just type in their machine key and then any onerous component of TC can be bypassed. You'd still be able to protect from viruses since the average computer owner won't go typing in a 200-character hex key to run some cute program a friend emailed them.
I'd love having TC on my computer if I COULD TRUST IT. A hardware security mechanism that works FOR me is great - one that works AGAINST me is not.