Yikes. Let's start with some basics: power (in the physics / mechanics sense) is not related to power (in the computational performance sense). They share a name, but nothing of their meanings overlaps. So in trying to make the two quantities commensurable you did not succeed. Frequency is s^-1. Power (in the computational performance sense) is tricky to define exactly, but all possible definitions would include some unit that is derived from a number of computational units of some form. Let's take MIPS as a quick example as all measures of this form are flawed in at least one way. So Power (in this context) would have dimensions of the form Is^-1, rather than s^-1.
This is the point that the GP was making. He didn't spell it out in any further detail as this is seriously well trodden ground going back to the Megahertz wars. While Osgeld touches one aspect of the problem he doesn't really cover it. This is why I left my point general: work performed per clock-cycle varies depending on the type of workload and the execution context (previous trace) *within* a single model of processor. It also varies between models of processors. This is why we don't measure performance in MIPS, and why GLOPS are a notoriously loose definition of performance. Really the AC's original point was a good one "Words mean things, use them correctly."
How can a veneer be dense, and where can I buy one? There is so much that I don't understand about marketing and may need to subcontract out to a specialist.
Why on earth would your beliefs be in any way relevant? Are you aware that the machine in front of you allows you to search for information so that you can test your beliefs. Shocking, eh?
A member of a volunteer committee organized to suppress and punish crime summarily (as when the processes of law are viewed as inadequate)
The key point of the definition is that a vigilante takes extra-legal action; a vigilante is one who acts outside of the existing legal framework. Publishing the identity of the looters is not operating outside of the law. In fact it actually supports the official effort to identify looters from video shot during the riots. Only acting on the identities to go and mete out some kind of illegal retribution would be vigilante action, and as there is no suggestion that they will do so the use of the word is loaded as the GP originally stated.
Indeed that is the real question, and you would get very different answers from a cross-section of the slashdot population. But regardless of where people stand on the issue of gun control, it is disingenuous to pretend that guns (bombs etc) don't enable the task of killing people to be carried out much more effectively. Whether or not arming a population creates a stable equilibrium where less people are killed is an entirely different (and much more interesting question). Pretending that guns don't kill people is inane.
FACT: Nobody can get shot if there are no guns. FACT: Nobody can get blown up without bombs.... CLUESTICK: While communication tech may not *cause* riots it does *enable* effective organisation of rioters. It allows them to move from location to location to avoid the police. So, no sensationalism involved. Exactly the same tech that enabled protestors in the middle east to stay ahead of security forces is enabling rioters to stay ahead of the police. As a well organised riot is much more effective at destroying property it is much more dangerous than the old fashioned unorganised kind.
Here is a simple example (leaks way more information than the real system). Let's say that the two numbers that you want are elements on a ring (or in CS terms they are numbers modulo some N). You have two numbers, x mod N and y mod N. You want me to perform the modulo addition without learning x and y. 1. You pick two random numbers, p mod N and q mod N. 2. You send me (x+p) mod N and (y+q) mod N. As long as your selections were really random this provides no information about x or y. 3. I compute (x+p) + (y+q) mod N and send you the result. This leaks nothing about the sum. 4. You then compute r - (q+p) mod N to recover the real sum.
There are two problems with this simple scheme (which is why the real scheme took many years to discover and is quite hard to implement). The first problem is that you do as much work blinding and unblinding the numbers as you would computing the real sum. The second problem is that this scheme leaks some information (can't remember what, it's been quite a while).
A Somewhat Homomorphic encryption scheme will solve both of these issues for addition (for some value of solve and some value of efficiency), while a Fully Homomorphic will also allow you to perform multiplications in the ring.
It also seems disingenuous to claim that their search engine was developed ten years ago by a few dozen people. Their search engine is very different now to what it was ten years ago, probably because of the army of CS PhDs working on it. One frequently overlooked aspect is the several orders of magnitude increase in the size of the index - not many products scale up by that amount in their lifetime. The current search engine and the original are very different beasts. Google had to develop whole new methods of building and managing clusters of datacenters to make that happen. Saying that they've only had one successful product doesn't really mean that they've rested on their laurels when that product has had to change so much to stay competitive.
The second example is true, but it is not a tautology. The statement "a is a rational number or a is not a rational number" would be a tautology. The difference is that when you translate the statement into a formal logic a tautology is always true because of the syntax e.g "A v !A" where-as the example that you give depends on the behaviour of the predicate is-rational. So although "R(a) ^ R(b) -> R(ab)" is true for some predicates R, it is not true over all possible predicates R.
I disagree with you completely, because we seem to be speaking two different languages using the same words. A web of trust using pseudonyms is an oxymoron. How can you trust an unknown quality? You also seem to be using "accountability" to mean something other than what I believe to be its meaning. According to the OED "The quality of being accountable; liability to give account of, and answer for, discharge of duties or conduct; responsibility, amenableness". If you are anonymous then you cannot be liable for your actions, and you cannot be called to account for them. A pseudonym is not under any obligation to answer for their actions, they could simply ignore any questions or summons. A real identity is accountable though, specifically the person that it refers to can be found in the real world and made to answer. This is the very essence of accountability, and why it is an antonym of anonymity. You seem to be using different definitions of these words and/or trying to redefine accountability to be something compatible with a pseudonym.
The Ars Technica article on Lion has a lot of details, but basically the little lightbulb is being deprecated. The idea is that OSX applications using the new Lion interfaces should always have their state written out to disk. They can then be killed on demand to free up memory, and transparently restarted without the user seeing any difference. The idea is that applications become completely persistent and so the whole notion of running or not running becomes invisible to the user.
Anonymity (in this context) also implies being judged by nothing but the words you write. It means that you're neither nigger, wop nor kike. It means that you're not male or female. It means that you're neither younger nor older than the company you keep. It means being free to say what you really mean.
Is being (pre)judged not a form of being held to account? Prejudice is another form of tying people's views (wrongly) to information about them. Accounting for what someone says by what we think a "person like that" should think. I wasn't claiming that anonymity is a bad thing, or that accountability is a good thing: I'm not American so I don't see it such simplistic terms:)
As fyngryrz said above, well said, sir. Well said.
You have mixed up circles and non-circles in your description. If wiedzim made a public post then you would see it and could spam the comments. But if he posted it to his circles, or probably his extended circles as well, then you would not see it and could not comment on it at all. So wiedzim does have control over whether or not you can spam his comments.
Good point, I wondered if tessellation shaders had already made them obsolete. The latest nvidia demos are impressive. The unreal demo (samaritan) is well worth checking out to see how far procedural generation can go in a polygon engine. It is jaw genuinely jaw dropping and makes this video look like a cheap hack.
The idea that they've come up with a new LoD algorithm for point cloud data is reasonable. It would then allow their ridiculous claims to be (technically) true about the size of datasets. But, if everything is held procedurally then it must have a low complexity description in order to compress that vast dataset (say 20,000Gb) into something that can be processed. Low-complexity descriptions tend to exist for highly regular geometry, and if you look at their demo they appear to have very high detail objects in a very coarse, regular and repetitive mesh to the extent that when they zoom out it looks like Minecraft.
No need for it to be a hoax, I'm guessing that they can make horrific looking (regular, craply lit, static) graphics as they claim in the video with the projected datasizes they refer to. What they gloss over is that it can't just be translated onto a real level design and scaled up to the level of complexity that you see in real level design.
It would be kind of like me saying "hey, I can draw circles at an infinite level of detail, equivalent to trillions of line segments. Can't draw more complex shapes like faces yet though....."
So now you are shifting in your claims. Yes, when metadata is used it is in memory - the same is true of any data. But it is held (to use your term) on disk, where it is loaded into memory on use, changed and saved back to disk. The primary store of metadata, the one that persists between boots, is held on the disk. A small local cache is changed, as with any data. So going back to your original (erroneous) claim: traditional file-systems *do* hold their metadata on disk, even if they cache a portion of it in memory as an optimisation.
OK, so you agree that JSTOR scanning old papers does not entitle them to a copyright. So what do you think is stopping someone from redistributing the scans for our "sanctimonious, entitled convieniece"?
Slashdot Mirror
Yikes. Let's start with some basics: power (in the physics / mechanics sense) is not related to power (in the computational performance sense). They share a name, but nothing of their meanings overlaps. So in trying to make the two quantities commensurable you did not succeed. Frequency is s^-1. Power (in the computational performance sense) is tricky to define exactly, but all possible definitions would include some unit that is derived from a number of computational units of some form. Let's take MIPS as a quick example as all measures of this form are flawed in at least one way. So Power (in this context) would have dimensions of the form Is^-1, rather than s^-1.
This is the point that the GP was making. He didn't spell it out in any further detail as this is seriously well trodden ground going back to the Megahertz wars. While Osgeld touches one aspect of the problem he doesn't really cover it. This is why I left my point general: work performed per clock-cycle varies depending on the type of workload and the execution context (previous trace) *within* a single model of processor. It also varies between models of processors. This is why we don't measure performance in MIPS, and why GLOPS are a notoriously loose definition of performance. Really the AC's original point was a good one "Words mean things, use them correctly."
How can a veneer be dense, and where can I buy one? There is so much that I don't understand about marketing and may need to subcontract out to a specialist.
His objection is almost certainly that work per cycle is not constant. Why assume something that is clearly false?
Why on earth would your beliefs be in any way relevant? Are you aware that the machine in front of you allows you to search for information so that you can test your beliefs. Shocking, eh?
The key point of the definition is that a vigilante takes extra-legal action; a vigilante is one who acts outside of the existing legal framework. Publishing the identity of the looters is not operating outside of the law. In fact it actually supports the official effort to identify looters from video shot during the riots. Only acting on the identities to go and mete out some kind of illegal retribution would be vigilante action, and as there is no suggestion that they will do so the use of the word is loaded as the GP originally stated.
Indeed that is the real question, and you would get very different answers from a cross-section of the slashdot population. But regardless of where people stand on the issue of gun control, it is disingenuous to pretend that guns (bombs etc) don't enable the task of killing people to be carried out much more effectively. Whether or not arming a population creates a stable equilibrium where less people are killed is an entirely different (and much more interesting question). Pretending that guns don't kill people is inane.
You mean like the protesters torching down-town Athens?
Which doesn't change that fact that bombs and guns do the job a lot more effectively.
FACT: Nobody can get shot if there are no guns. ...
FACT: Nobody can get blown up without bombs.
CLUESTICK: While communication tech may not *cause* riots it does *enable* effective organisation of rioters. It allows them to move from location to location to avoid the police. So, no sensationalism involved. Exactly the same tech that enabled protestors in the middle east to stay ahead of security forces is enabling rioters to stay ahead of the police. As a well organised riot is much more effective at destroying property it is much more dangerous than the old fashioned unorganised kind.
Here is a simple example (leaks way more information than the real system). Let's say that the two numbers that you want are elements on a ring (or in CS terms they are numbers modulo some N). You have two numbers, x mod N and y mod N. You want me to perform the modulo addition without learning x and y.
1. You pick two random numbers, p mod N and q mod N.
2. You send me (x+p) mod N and (y+q) mod N. As long as your selections were really random this provides no information about x or y.
3. I compute (x+p) + (y+q) mod N and send you the result. This leaks nothing about the sum.
4. You then compute r - (q+p) mod N to recover the real sum.
There are two problems with this simple scheme (which is why the real scheme took many years to discover and is quite hard to implement). The first problem is that you do as much work blinding and unblinding the numbers as you would computing the real sum. The second problem is that this scheme leaks some information (can't remember what, it's been quite a while).
A Somewhat Homomorphic encryption scheme will solve both of these issues for addition (for some value of solve and some value of efficiency), while a Fully Homomorphic will also allow you to perform multiplications in the ring.
Gmail?
Maps?
It also seems disingenuous to claim that their search engine was developed ten years ago by a few dozen people. Their search engine is very different now to what it was ten years ago, probably because of the army of CS PhDs working on it. One frequently overlooked aspect is the several orders of magnitude increase in the size of the index - not many products scale up by that amount in their lifetime. The current search engine and the original are very different beasts. Google had to develop whole new methods of building and managing clusters of datacenters to make that happen. Saying that they've only had one successful product doesn't really mean that they've rested on their laurels when that product has had to change so much to stay competitive.
I've wanted a message board that threaded comments as a DAG rather than a tree for many years, but this is the first real need for it :)
Damn. I thought I was the most pedantic person on this thread. Yes, it does depend on the law of the excluded middle. Or not. Sometimes.
The second example is true, but it is not a tautology. The statement "a is a rational number or a is not a rational number" would be a tautology. The difference is that when you translate the statement into a formal logic a tautology is always true because of the syntax e.g "A v !A" where-as the example that you give depends on the behaviour of the predicate is-rational. So although "R(a) ^ R(b) -> R(ab)" is true for some predicates R, it is not true over all possible predicates R.
I disagree with you completely, because we seem to be speaking two different languages using the same words. A web of trust using pseudonyms is an oxymoron. How can you trust an unknown quality? You also seem to be using "accountability" to mean something other than what I believe to be its meaning. According to the OED "The quality of being accountable; liability to give account of, and answer for, discharge of duties or conduct; responsibility, amenableness". If you are anonymous then you cannot be liable for your actions, and you cannot be called to account for them. A pseudonym is not under any obligation to answer for their actions, they could simply ignore any questions or summons. A real identity is accountable though, specifically the person that it refers to can be found in the real world and made to answer. This is the very essence of accountability, and why it is an antonym of anonymity. You seem to be using different definitions of these words and/or trying to redefine accountability to be something compatible with a pseudonym.
The Ars Technica article on Lion has a lot of details, but basically the little lightbulb is being deprecated. The idea is that OSX applications using the new Lion interfaces should always have their state written out to disk. They can then be killed on demand to free up memory, and transparently restarted without the user seeing any difference. The idea is that applications become completely persistent and so the whole notion of running or not running becomes invisible to the user.
Is being (pre)judged not a form of being held to account? Prejudice is another form of tying people's views (wrongly) to information about them. Accounting for what someone says by what we think a "person like that" should think. I wasn't claiming that anonymity is a bad thing, or that accountability is a good thing: I'm not American so I don't see it such simplistic terms :)
As fyngryrz said above, well said, sir. Well said.
Because they are antonyms of one another.
Accountability implies no anonymity.
Anonymity implies no accountability.
It would be hard to discuss one without mentioning the other.
You have mixed up circles and non-circles in your description. If wiedzim made a public post then you would see it and could spam the comments. But if he posted it to his circles, or probably his extended circles as well, then you would not see it and could not comment on it at all. So wiedzim does have control over whether or not you can spam his comments.
Good point, I wondered if tessellation shaders had already made them obsolete. The latest nvidia demos are impressive. The unreal demo (samaritan) is well worth checking out to see how far procedural generation can go in a polygon engine. It is jaw genuinely jaw dropping and makes this video look like a cheap hack.
The idea that they've come up with a new LoD algorithm for point cloud data is reasonable. It would then allow their ridiculous claims to be (technically) true about the size of datasets. But, if everything is held procedurally then it must have a low complexity description in order to compress that vast dataset (say 20,000Gb) into something that can be processed. Low-complexity descriptions tend to exist for highly regular geometry, and if you look at their demo they appear to have very high detail objects in a very coarse, regular and repetitive mesh to the extent that when they zoom out it looks like Minecraft.
No need for it to be a hoax, I'm guessing that they can make horrific looking (regular, craply lit, static) graphics as they claim in the video with the projected datasizes they refer to. What they gloss over is that it can't just be translated onto a real level design and scaled up to the level of complexity that you see in real level design.
It would be kind of like me saying "hey, I can draw circles at an infinite level of detail, equivalent to trillions of line segments. Can't draw more complex shapes like faces yet though....."
Thousands of milliseconds, eh? Round my way we be calling them seconds...
Bad spellers, eh?
So now you are shifting in your claims. Yes, when metadata is used it is in memory - the same is true of any data. But it is held (to use your term) on disk, where it is loaded into memory on use, changed and saved back to disk. The primary store of metadata, the one that persists between boots, is held on the disk. A small local cache is changed, as with any data. So going back to your original (erroneous) claim: traditional file-systems *do* hold their metadata on disk, even if they cache a portion of it in memory as an optimisation.
It doesn't sound like you do. Sync is used to flush the cache of metadata back out to the disk. The metadata is actually stored on disk.
OK, so you agree that JSTOR scanning old papers does not entitle them to a copyright. So what do you think is stopping someone from redistributing the scans for our "sanctimonious, entitled convieniece"?