The big problem impacting speed is dispersion. The light takes different paths down the cable with the net result that some of it travels more slowly than the rest. At the other end, its as if you received a blurry picture: you can't tell what the signal was supposed to be.
If that doesn't make sense, let me explain it this way: light doesn't travel straight down a fiber optic cable. Instead, it bounces back and forth down the cable, first hitting the cladding at one side and then hitting the cladding at the other. That's why the light can go around curves; its not traveling straight, its bouncing back and forth off the walls. The index of refraction for the cladding material is much higher than the index of refraction for the fiber, so the light obeys a principle called "total internal reflection" instead of the cladding absorbing it.
Some photons go pretty straight, rarely hitting the walls. Others bounce off the walls a lot. That changes distance they travel, which changes the time it takes them to reach the other end. With a thick plastic cable, the ones that bounce a lot will travel a much longer distance thus you have to space the changes in the signal further apart for them to be detectable at the other end of the cable.
You'd have to equal their political power before you could abolish copyright
Who's throwing up a straw man now? I'm pretty sure I didn't say anything about abolishing copyright. I'm pretty sure I didn't even imply it. I don't even consider it a worthwhile goal.
What I want to see is corrections to IP law. To wit:
1. No more lifetime-plus copyrights. You get a couple decades and that's it. If it doesn't become popular until 40 years after the fact, well that's just rough luck.
2. Full disclosure! If I can't build your invention from the patent application or create a copy of the work from your copyright filing then it is insufficient to garner protection. Patents should be engineering documents, not legalese.
3. No post-sale right to control. An owner of a copy can do as he pleases with that one copy. Convert the format. Edit the scenes. Split it in half and sell the halves separately. Anything except make new copies. Its his property now, not the authors.
4. No right to control distribution. You can choose the cost per copy but after than anyone has a right to make copies so long as they pay you for them.
5. No exclusive right to make derivative works. If I want to write a story in your world setting, I can do it and sell it as much as I want, as long as it accompanies a legally acquired copy of your work as well.
6. Discourage DRM. The rights holder should have a choice: DRM or statutory damages. One of the other but not both.
And sadly, you're right. I'm not the King of America and I haven't the political power.:)
The rights of the minority must be protected, even if the minority is currently unpopular.
By that logic, murder must be acceptable because a minority (namely murderers) want their right to kill protected.
Your basic premise is correct: a majority must not be able to abuse a minority merely because they have the votes. However, the minority should rarely have the right to compel the behavior of the majority and then only with excruciatingly careful attention to fairness. Modern intellectual property law got its start that way those many decades ago, but the path is long lost.
I do believe that the right to get paid for your labor is a fundamental right.
As a software developer, I'm inclined to agree. But being paid for your work is entirely different from being able to indefinitely control the entire idea-space which branches from the notions embodied in your work. Disney doesn't just assert ownership of a drawing; they legally own of a piece of the American psyche: those little parts of your and my minds which are indelibly stamped with a certain red-pantsed rodent.
Nor is it fair for you to do work once and then be paid for it over and over again. Do that with any physical work-product and such a thing is criminal fraud. That privilege should come with a price. Current IP law exacts no toll to balance work-once sell-many. When meaningful additional work was needed to make a copy, there was no need.
Technology has changed this aspect of intellectual property and folks have ever seen it as righteous to disregard an unfair law. Only 2 of the 500 thought downloading was wrong, but surely 498 of the 500 would have agreed stealing a physical CD was.
I'm sorry, but its simply time for free domain tasting to go. It costs something like $6 at the back end to register a domain for one year and its a hardship on no legitimate use if they have to pay another $6 to correct a typo.
Dan Heller poses a great question. After describing a situation where only 2 of 500 polled students thought downloading a copyrighted work was wrong, he asks:
What do we do about a society that is already predisposed to ignoring copyright in the first place?
The answer is simply this: where the law fails to reflect the will of the people it is the law, not the people, which is in error.
That's what I wrote: if you go from 2^n addresses to (2^n)^2, execution time doubles: log(2^n^2) / log(2^n) = 2.
Why can't you just admit that you were incorrect to include the signal path length on only one side of your comparison of algorithm efficiency between a TCAM and a radix tree? Does it hurt so bad to be wrong once in a while?
It implies our current universe is not computable/predictable by our mathematics.
It doesn't particularly damage probability and statistics. The results take on a different meaning but the equations don't change. Then too, a nondeterministic equation is frequently identical to a longer deterministic equation and the reverse, so only some parts of math suffer for it.
the signal path increases by a factor 2 and thus execution time.
That's a specious argument.
First, you can show the same property with any real hardware. By that logic, all algorithms would have to have their running times multiplied log n to account for their operation on a computer with real, finite memory. In which case we're back to square one: a TCAM takes the minimum possible running time of any real computer algorithm for any purpose.
And second, the signal path length for adding memory capacity to a device (any device) doesn't increase by n, it increases by log n. It only appears to be a straight bus. You feed the signal to x regenerators which each feed it to x more regenerators which each feed it to x cells. The difference between a TCAM and a standard memory is that in a TCAM the signal is always fed to every cell while in a standard memory one of the lines selects or deselects each regenerator and the deselected regenerators don't propagate the signal. At the tail end of a standard memory call, the full path for only one row of cells has been selected. This difference impacts several things, but the algorithmic runtime isn't one of them.
TCAMs have a number of problems including heat output and the O(n) insertion time. Failure to complete a lookup in O(1) is not one of those problems.
However, that's the very definition of NP
At the quantum physics level, things like electron location appear nondeterministic... So much so that chemists describe it as the shell of a sphere surrounding the atom. If quanta are the building blocks of our universe and they display nondeterministic properties even though everything larger appears thoroughly deterministic... Does that imply anything about the hypothetical ur-universe?
Range voting reduces the ability of minority parties to influence the political system.
Right now, major candidates have a strong incentive to prevent serious spoilers by subsuming those spoilers' key ideas into their own campaigns. The Republican candidate will preach small government because if he doesn't the libertarian candidate will pull away enough voters for the democrat to beat him.
In a range system, why bother? Folks who oppose his rival will rank him high anyway to assure that his rival loses. If the third party candidate can't spoil your race, why bother paying any attention to his supporters' desires at all?
Truth is, our government stays pretty centrist (even in times of crisis like 9/11) and the reason it does is that whenever a candidate strays too far, a spoiler comes in and wipes him out. With range voting, nothing prevents large unstable swings in governance.
The only common hardware is the AC power and the modem that has a/28 assigned to it.
Which means that if a customer in his room hacks the modem, he has access to your admin network.
No doubt this is the problem with Boeing's system. The radio is on the safety network but there's a gateway attached to both the safety and passenger networks that rebroadcasts the radio traffic so the passengers can listen in. If devices on the passenger network can send packets to that gateway then it is a potential point of breach.
Yeah, and there are speed of light delays propagating the bits down the wire nor does the charged match line bleed to ground instantaneously. Similar design challenges are why we have gigahertz processors instead of petahertz processors. None of it speaks to algorithmic efficiency. If it finishes in a constant number of clocks regardless of the value n then its O(1). Especially if the number of clocks is exactly one.
we still don't know if NP problems can be solved in polynomial time or not.
And you don't think that working out the basic rules of physics of a hypothetical ur-universe in which there is a a polynomial-time solution would offer additional insight in to the problem? In your studied opinion, that new vantage would definitely not offer another step towards either finding a polynomial-time algorithm that works in our universe, nor offer a step towards a proof that no such algorithm exists?
you can circumvent this by looking at another property, X, for which there is a proof.
Sure, anyone can design a bad experiment where the desired result is defined to exist. When is that not the case? The challenge is to define a basic property of the hypothetical ur-universe which leads to an algorithm that solves the equation in polynomial time.
In the tcam/tree example, the basic property is the data-processor tie. In the computer, data sits unchanged and unchanging until a processor comes around to perform an operation on it. It's not possible to construct a simulated device where every piece of data changes in parallel with the rest. Even if the inhabitants of that digital simulation could not perceive the difference between a simple operation affecting little of the data and a complex operation affecting a lot of it, they could detect the apparent choppiness in the inputs that occurs as a result.
In the universe as we know it, elements and changes to their activity appear to be tied 1:1. Each element that makes up the earth accelerates under the influence of gravity in parallel. Because of this, its possible to construct a device, the TCAM, where a function is applied to all data elements in as close to the same instant as is possible with relativity. Consequently, we have an algorithm in our reality that completes in O(1) even though the best comparable algorithm inside the simulation completes in O(x) where x=the maximum length of any stored data elements.
So the challenge is: if we allow ourselves to consider basic properties of a universe that are different that what we know them to be in our universe, is it possible to design an algorithm in that universe which, for example, factors the products of primes quickly? And how many other NP problems fall to the same altered universe?
Now, you have to admit that if we could design an alternate universe by changing several basic properties of our own in which we could prove that a whole slew of NP math problems have easy solutions, that would be a remarkably interesting result.
How about "The fact that the universe could be a VR simulation doesn't mean that we should suspect it is"?
Of course we shouldn't expect that it is. The notion is absurd. But if we don't put unproven possibilities to the test then there is no science and if we discard testable possibilities merely because they are strange and unlikely then we don't achieve the great discoveries like Relativity.
Whitworth has challenged us with a possibility that is testable, at least to a point. At a minimum we will learn more about the character of our reality in the process of designing and implementing those tests. And we won't need bazillion-dollar supercolliders to gain that knowledge.
A TCAM is O(log(N)) too, it just has a really small constant term.
No, its not. See http://www.pagiamtzis.com/cam/camintro.html for a good intro on how TCAMs work. All cells are activated in parallel and the cells which don't match pull down the match line for their row. Also in parallel. One gigantic operation. O(1). Generates a crapload of heat, so I supposed you might have to slow the cycles to keep the heat under control, but it's still O(1).
For one thing, the fact that there could conceivably be an Operator standing by to adjust the results doesn't mean that we should suspect there is. That was one of the wisecrack points made in the Flying Spaghetti Monster letter to the Kansas Board of Education.
For another, you can't name a single real-time simulation in which an operator stands by making adjustments whenever one of the simulated components bumps up against the edge of the system. Its just not a useful way to run a simulation. There's no reason to suspect that a posited ur-universe would treat its simulations any differently than we treat our own. However, if there was an operator fudging results, that would make the borders of the VR easy to detect. Just look for the objective problem that produces inconsistent answers each time its evaluated.
This is the failure of reconciling the metaphysical with the physical.
No, no! He's on to something. Consider this example:
When routing TCP/IP packets, the best available software algorithms are tree-based. You step down the branches of the tree until you find the most specific route known for the destination address. Its O(log n).
However, if you step out of the software universe running on a general-purpose computer, you can design a hardware device called a "TCAM." A TCAM is a special kind of static ram where a request is processed across all cells in the same cycle in order to produce the best match. Not only does it return a routing decision in O(1), it returns that decision in exactly one clock cycle.
Now, we could describe how a TCAM works within software and we could even simulate it but the simulation would run in O(n) because the simulation would have to activate each cell in sequence instead of activating all cells at once the way a real TCAM does.
So the challenge for detecting whether we're in a virtual reality is this: find a mathematical problem which is conceptually simple (e.g. factoring the product of large primes) but which we know to be hard ( O(x^n) ) and then construct a simulation of a finite ur-universe in which the problem is easy. The simulation itself won't run any faster than the best known factoring algorithms but it would be able to prove that given the physical rules of the ur-universe the factoring would have completed in O(1).
Successfully constructing such a simulation wouldn't prove that we're actually in a virtual reality, but proving that such a simulation can't be constructed would prove that we're not. Thus the theory is falsifiable. Thus it is science, not philosophy.
I have to share my office space with all the network equipment. Just 4 standard racks
And let me guess, by "4 standard racks" you mean four of the two-post relay racks, not 2'x3' four-post cabinets. You're a noob. You wouldn't consider such a setup if you weren't a greenhorn and you certainly wouldn't describe it as your "dream office." If the noise doesn't drive you nuts, the temperature issues will.
I'll make a couple recommendations. You're a noob, so you wont take them but maybe someone else reading them will.
1. No humans in the equipment room. Aside from the discomfort, it leads to accidents with soda and tripping over wires.
2. Use 4-post cabinets unless you know exactly what you want the 2-post rack for, and get ones that a three-feet deep with square holes. You can successfully rack just about anything in one of those. Not so with a relay rack or a cabinet with threaded or round holes.
3. 3 feet front and back. Especially back.
4. Don't put non-rack equipment on the racks. You think you're going to set all your old tower servers on shelves in the rack? Well, you can do that but you'll find its inconvenient and wasteful. Get yourself a single cabinet and get a food service shelf a la https://www.precisiontools.com/shop/product_118.html . 2 feet deep by 4 feet wide by 6 feet tall is just about right. Lots of tie points for a clean cabling job, it'll efficiently fit all your tower servers and it won't collect dust.
Unfortunately, the mathematicians weren't smart enough to follow through to understand the cause of heavy braking: following too closely, less than 2 seconds behind the car in front of you so that when you brake lightly the next guy has to brake hard.
Nor did they follow that back to its root cause: too many cars on a section of road so that they pack too tightly. Nor did they notice that in light traffic flows fine regardless of braking because the large gaps consume the time lost so that more than a couple cars behind you no braking is needed.
If you want to understand road traffic, you need only understand data traffic on CSMA/CD half-duplex Ethernet. It works poorly after about 60% of theoretical capacity and has a cascade failure approaching 100%. Actually, that's badly phrased because half-duplex ethernet never approaches 100% throughput. The wire can be consumed 100% of the time, but when it is, total throughput is close to zero. Most transmission attempts are retried due to collisions and most packets that do get through end up in a higher-level retransmission because a timeout has been hit or the packet is out of sequence with another packet that was dropped from the overflowing transmission buffer.
Ethernet collisions are analogous to someone tapping the brakes lightly while a packet lost due to a buffer overrun is analogous to someone hitting the brakes hard. As the probability of each event increases, the throughput on the approaches zero.
My preferred e-book reader is my blackberry. It pulls books in html format from my server at need. It's conveniently located on my belt whenever I feel like reading. And of course it does less important things too: like handle email and phone calls.
Perhaps I oversimplified. Let me put it to you this way:
If I open Bill's Diner and register the trademark, no one else can open a Bill's Diner anywhere in the US that there isn't already a Bill's Diner. If they do, I can sue and win.
If I open Bill's Diner and place a "TM" after it, no one else can open a Bill's Diner nearby. By placing the TM after it, I have claimed the trademark even though its a relatively generic name.
If I open Bill's Diner and do nothing, someone else can open a Bill's Diner a block away and there's very little I can do about it. Its pretty generic and there's no reason the other Bill's Diner should have suspected that I was claiming a mark.
You have a legal right to slap TM on anything that you could reasonably claim as a trademark. Doing so carries the specific legal meaning that you are claiming the mark. And trademark law is very slippery when it comes to how close you have to be to violate. If you want to claim a mark and don't at least put TM beside it, you're not adequately protected.
Nevertheless, you'll find that they respond to insult the same as any real human being. If you want to convince one of them to take a particular action, tossing an insult his way is not an effective strategy.
Believe it or not, those people are human beings too. They can be insulted and almost anyone can afford to walk away from $100 because the person offering insulted them.
You're thinking copyrights. Copyright vests as soon as you create the work.
For a trademark to be valid, you have to declare that its a trademark. You generally do this by placing the letters "TM" after the mark or R in a circle if you've registered the mark. If you just use the name and don't include that notification then it's not a trademark and you have no protection under trademark law.
Slashdot Mirror
The big problem impacting speed is dispersion. The light takes different paths down the cable with the net result that some of it travels more slowly than the rest. At the other end, its as if you received a blurry picture: you can't tell what the signal was supposed to be.
If that doesn't make sense, let me explain it this way: light doesn't travel straight down a fiber optic cable. Instead, it bounces back and forth down the cable, first hitting the cladding at one side and then hitting the cladding at the other. That's why the light can go around curves; its not traveling straight, its bouncing back and forth off the walls. The index of refraction for the cladding material is much higher than the index of refraction for the fiber, so the light obeys a principle called "total internal reflection" instead of the cladding absorbing it.
Some photons go pretty straight, rarely hitting the walls. Others bounce off the walls a lot. That changes distance they travel, which changes the time it takes them to reach the other end. With a thick plastic cable, the ones that bounce a lot will travel a much longer distance thus you have to space the changes in the signal further apart for them to be detectable at the other end of the cable.
Make more sense now?
You'd have to equal their political power before you could abolish copyright
:)
Who's throwing up a straw man now? I'm pretty sure I didn't say anything about abolishing copyright. I'm pretty sure I didn't even imply it. I don't even consider it a worthwhile goal.
What I want to see is corrections to IP law. To wit:
1. No more lifetime-plus copyrights. You get a couple decades and that's it. If it doesn't become popular until 40 years after the fact, well that's just rough luck.
2. Full disclosure! If I can't build your invention from the patent application or create a copy of the work from your copyright filing then it is insufficient to garner protection. Patents should be engineering documents, not legalese.
3. No post-sale right to control. An owner of a copy can do as he pleases with that one copy. Convert the format. Edit the scenes. Split it in half and sell the halves separately. Anything except make new copies. Its his property now, not the authors.
4. No right to control distribution. You can choose the cost per copy but after than anyone has a right to make copies so long as they pay you for them.
5. No exclusive right to make derivative works. If I want to write a story in your world setting, I can do it and sell it as much as I want, as long as it accompanies a legally acquired copy of your work as well.
6. Discourage DRM. The rights holder should have a choice: DRM or statutory damages. One of the other but not both.
And sadly, you're right. I'm not the King of America and I haven't the political power.
The rights of the minority must be protected, even if the minority is currently unpopular.
By that logic, murder must be acceptable because a minority (namely murderers) want their right to kill protected.
Your basic premise is correct: a majority must not be able to abuse a minority merely because they have the votes. However, the minority should rarely have the right to compel the behavior of the majority and then only with excruciatingly careful attention to fairness. Modern intellectual property law got its start that way those many decades ago, but the path is long lost.
I do believe that the right to get paid for your labor is a fundamental right.
As a software developer, I'm inclined to agree. But being paid for your work is entirely different from being able to indefinitely control the entire idea-space which branches from the notions embodied in your work. Disney doesn't just assert ownership of a drawing; they legally own of a piece of the American psyche: those little parts of your and my minds which are indelibly stamped with a certain red-pantsed rodent.
Nor is it fair for you to do work once and then be paid for it over and over again. Do that with any physical work-product and such a thing is criminal fraud. That privilege should come with a price. Current IP law exacts no toll to balance work-once sell-many. When meaningful additional work was needed to make a copy, there was no need.
Technology has changed this aspect of intellectual property and folks have ever seen it as righteous to disregard an unfair law. Only 2 of the 500 thought downloading was wrong, but surely 498 of the 500 would have agreed stealing a physical CD was.
I'm sorry, but its simply time for free domain tasting to go. It costs something like $6 at the back end to register a domain for one year and its a hardship on no legitimate use if they have to pay another $6 to correct a typo.
Dan Heller poses a great question. After describing a situation where only 2 of 500 polled students thought downloading a copyrighted work was wrong, he asks:
What do we do about a society that is already predisposed to ignoring copyright in the first place?
The answer is simply this: where the law fails to reflect the will of the people it is the law, not the people, which is in error.
And this is better than classic port-knocking how?
That's what I wrote: if you go from 2^n addresses to (2^n)^2, execution time doubles: log(2^n^2) / log(2^n) = 2.
Why can't you just admit that you were incorrect to include the signal path length on only one side of your comparison of algorithm efficiency between a TCAM and a radix tree? Does it hurt so bad to be wrong once in a while?
It implies our current universe is not computable/predictable by our mathematics.
It doesn't particularly damage probability and statistics. The results take on a different meaning but the equations don't change. Then too, a nondeterministic equation is frequently identical to a longer deterministic equation and the reverse, so only some parts of math suffer for it.
the signal path increases by a factor 2 and thus execution time.
That's a specious argument.
First, you can show the same property with any real hardware. By that logic, all algorithms would have to have their running times multiplied log n to account for their operation on a computer with real, finite memory. In which case we're back to square one: a TCAM takes the minimum possible running time of any real computer algorithm for any purpose.
And second, the signal path length for adding memory capacity to a device (any device) doesn't increase by n, it increases by log n. It only appears to be a straight bus. You feed the signal to x regenerators which each feed it to x more regenerators which each feed it to x cells. The difference between a TCAM and a standard memory is that in a TCAM the signal is always fed to every cell while in a standard memory one of the lines selects or deselects each regenerator and the deselected regenerators don't propagate the signal. At the tail end of a standard memory call, the full path for only one row of cells has been selected. This difference impacts several things, but the algorithmic runtime isn't one of them.
TCAMs have a number of problems including heat output and the O(n) insertion time. Failure to complete a lookup in O(1) is not one of those problems.
However, that's the very definition of NP
At the quantum physics level, things like electron location appear nondeterministic... So much so that chemists describe it as the shell of a sphere surrounding the atom. If quanta are the building blocks of our universe and they display nondeterministic properties even though everything larger appears thoroughly deterministic... Does that imply anything about the hypothetical ur-universe?
Range voting reduces the ability of minority parties to influence the political system.
Right now, major candidates have a strong incentive to prevent serious spoilers by subsuming those spoilers' key ideas into their own campaigns. The Republican candidate will preach small government because if he doesn't the libertarian candidate will pull away enough voters for the democrat to beat him.
In a range system, why bother? Folks who oppose his rival will rank him high anyway to assure that his rival loses. If the third party candidate can't spoil your race, why bother paying any attention to his supporters' desires at all?
Truth is, our government stays pretty centrist (even in times of crisis like 9/11) and the reason it does is that whenever a candidate strays too far, a spoiler comes in and wipes him out. With range voting, nothing prevents large unstable swings in governance.
The only common hardware is the AC power and the modem that has a /28 assigned to it.
Which means that if a customer in his room hacks the modem, he has access to your admin network.
No doubt this is the problem with Boeing's system. The radio is on the safety network but there's a gateway attached to both the safety and passenger networks that rebroadcasts the radio traffic so the passengers can listen in. If devices on the passenger network can send packets to that gateway then it is a potential point of breach.
Yeah, and there are speed of light delays propagating the bits down the wire nor does the charged match line bleed to ground instantaneously. Similar design challenges are why we have gigahertz processors instead of petahertz processors. None of it speaks to algorithmic efficiency. If it finishes in a constant number of clocks regardless of the value n then its O(1). Especially if the number of clocks is exactly one.
we still don't know if NP problems can be solved in polynomial time or not.
And you don't think that working out the basic rules of physics of a hypothetical ur-universe in which there is a a polynomial-time solution would offer additional insight in to the problem? In your studied opinion, that new vantage would definitely not offer another step towards either finding a polynomial-time algorithm that works in our universe, nor offer a step towards a proof that no such algorithm exists?
you can circumvent this by looking at another property, X, for which there is a proof.
Sure, anyone can design a bad experiment where the desired result is defined to exist. When is that not the case? The challenge is to define a basic property of the hypothetical ur-universe which leads to an algorithm that solves the equation in polynomial time.
In the tcam/tree example, the basic property is the data-processor tie. In the computer, data sits unchanged and unchanging until a processor comes around to perform an operation on it. It's not possible to construct a simulated device where every piece of data changes in parallel with the rest. Even if the inhabitants of that digital simulation could not perceive the difference between a simple operation affecting little of the data and a complex operation affecting a lot of it, they could detect the apparent choppiness in the inputs that occurs as a result.
In the universe as we know it, elements and changes to their activity appear to be tied 1:1. Each element that makes up the earth accelerates under the influence of gravity in parallel. Because of this, its possible to construct a device, the TCAM, where a function is applied to all data elements in as close to the same instant as is possible with relativity. Consequently, we have an algorithm in our reality that completes in O(1) even though the best comparable algorithm inside the simulation completes in O(x) where x=the maximum length of any stored data elements.
So the challenge is: if we allow ourselves to consider basic properties of a universe that are different that what we know them to be in our universe, is it possible to design an algorithm in that universe which, for example, factors the products of primes quickly? And how many other NP problems fall to the same altered universe?
Now, you have to admit that if we could design an alternate universe by changing several basic properties of our own in which we could prove that a whole slew of NP math problems have easy solutions, that would be a remarkably interesting result.
How about "The fact that the universe could be a VR simulation doesn't mean that we should suspect it is"?
Of course we shouldn't expect that it is. The notion is absurd. But if we don't put unproven possibilities to the test then there is no science and if we discard testable possibilities merely because they are strange and unlikely then we don't achieve the great discoveries like Relativity.
Whitworth has challenged us with a possibility that is testable, at least to a point. At a minimum we will learn more about the character of our reality in the process of designing and implementing those tests. And we won't need bazillion-dollar supercolliders to gain that knowledge.
A TCAM is O(log(N)) too, it just has a really small constant term.
No, its not. See http://www.pagiamtzis.com/cam/camintro.html for a good intro on how TCAMs work. All cells are activated in parallel and the cells which don't match pull down the match line for their row. Also in parallel. One gigantic operation. O(1). Generates a crapload of heat, so I supposed you might have to slow the cycles to keep the heat under control, but it's still O(1).
That's a specious argument.
For one thing, the fact that there could conceivably be an Operator standing by to adjust the results doesn't mean that we should suspect there is. That was one of the wisecrack points made in the Flying Spaghetti Monster letter to the Kansas Board of Education.
For another, you can't name a single real-time simulation in which an operator stands by making adjustments whenever one of the simulated components bumps up against the edge of the system. Its just not a useful way to run a simulation. There's no reason to suspect that a posited ur-universe would treat its simulations any differently than we treat our own. However, if there was an operator fudging results, that would make the borders of the VR easy to detect. Just look for the objective problem that produces inconsistent answers each time its evaluated.
This is the failure of reconciling the metaphysical with the physical.
No, no! He's on to something. Consider this example:
When routing TCP/IP packets, the best available software algorithms are tree-based. You step down the branches of the tree until you find the most specific route known for the destination address. Its O(log n).
However, if you step out of the software universe running on a general-purpose computer, you can design a hardware device called a "TCAM." A TCAM is a special kind of static ram where a request is processed across all cells in the same cycle in order to produce the best match. Not only does it return a routing decision in O(1), it returns that decision in exactly one clock cycle.
Now, we could describe how a TCAM works within software and we could even simulate it but the simulation would run in O(n) because the simulation would have to activate each cell in sequence instead of activating all cells at once the way a real TCAM does.
So the challenge for detecting whether we're in a virtual reality is this: find a mathematical problem which is conceptually simple (e.g. factoring the product of large primes) but which we know to be hard ( O(x^n) ) and then construct a simulation of a finite ur-universe in which the problem is easy. The simulation itself won't run any faster than the best known factoring algorithms but it would be able to prove that given the physical rules of the ur-universe the factoring would have completed in O(1).
Successfully constructing such a simulation wouldn't prove that we're actually in a virtual reality, but proving that such a simulation can't be constructed would prove that we're not. Thus the theory is falsifiable. Thus it is science, not philosophy.
I have to share my office space with all the network equipment. Just 4 standard racks
And let me guess, by "4 standard racks" you mean four of the two-post relay racks, not 2'x3' four-post cabinets. You're a noob. You wouldn't consider such a setup if you weren't a greenhorn and you certainly wouldn't describe it as your "dream office." If the noise doesn't drive you nuts, the temperature issues will.
I'll make a couple recommendations. You're a noob, so you wont take them but maybe someone else reading them will.
1. No humans in the equipment room. Aside from the discomfort, it leads to accidents with soda and tripping over wires.
2. Use 4-post cabinets unless you know exactly what you want the 2-post rack for, and get ones that a three-feet deep with square holes. You can successfully rack just about anything in one of those. Not so with a relay rack or a cabinet with threaded or round holes.
3. 3 feet front and back. Especially back.
4. Don't put non-rack equipment on the racks. You think you're going to set all your old tower servers on shelves in the rack? Well, you can do that but you'll find its inconvenient and wasteful. Get yourself a single cabinet and get a food service shelf a la https://www.precisiontools.com/shop/product_118.html . 2 feet deep by 4 feet wide by 6 feet tall is just about right. Lots of tie points for a clean cabling job, it'll efficiently fit all your tower servers and it won't collect dust.
Unfortunately, the mathematicians weren't smart enough to follow through to understand the cause of heavy braking: following too closely, less than 2 seconds behind the car in front of you so that when you brake lightly the next guy has to brake hard.
Nor did they follow that back to its root cause: too many cars on a section of road so that they pack too tightly. Nor did they notice that in light traffic flows fine regardless of braking because the large gaps consume the time lost so that more than a couple cars behind you no braking is needed.
If you want to understand road traffic, you need only understand data traffic on CSMA/CD half-duplex Ethernet. It works poorly after about 60% of theoretical capacity and has a cascade failure approaching 100%. Actually, that's badly phrased because half-duplex ethernet never approaches 100% throughput. The wire can be consumed 100% of the time, but when it is, total throughput is close to zero. Most transmission attempts are retried due to collisions and most packets that do get through end up in a higher-level retransmission because a timeout has been hit or the packet is out of sequence with another packet that was dropped from the overflowing transmission buffer.
Ethernet collisions are analogous to someone tapping the brakes lightly while a packet lost due to a buffer overrun is analogous to someone hitting the brakes hard. As the probability of each event increases, the throughput on the approaches zero.
AFRL's initial aim is to work out how to make a MAV flying at 74 kilometres per hour latch onto a power line without destroying itself or the line.
Yeah, that would be the hard part. 'Till you've figured that out, there's nothing to see here.
My preferred e-book reader is my blackberry. It pulls books in html format from my server at need. It's conveniently located on my belt whenever I feel like reading. And of course it does less important things too: like handle email and phone calls.
Truth in advertising? How is it that people busted for sex crimes manage to have names like [Sodom]sk[y]?
Perhaps I oversimplified. Let me put it to you this way:
If I open Bill's Diner and register the trademark, no one else can open a Bill's Diner anywhere in the US that there isn't already a Bill's Diner. If they do, I can sue and win.
If I open Bill's Diner and place a "TM" after it, no one else can open a Bill's Diner nearby. By placing the TM after it, I have claimed the trademark even though its a relatively generic name.
If I open Bill's Diner and do nothing, someone else can open a Bill's Diner a block away and there's very little I can do about it. Its pretty generic and there's no reason the other Bill's Diner should have suspected that I was claiming a mark.
You have a legal right to slap TM on anything that you could reasonably claim as a trademark. Doing so carries the specific legal meaning that you are claiming the mark. And trademark law is very slippery when it comes to how close you have to be to violate. If you want to claim a mark and don't at least put TM beside it, you're not adequately protected.
Nevertheless, you'll find that they respond to insult the same as any real human being. If you want to convince one of them to take a particular action, tossing an insult his way is not an effective strategy.
Believe it or not, those people are human beings too. They can be insulted and almost anyone can afford to walk away from $100 because the person offering insulted them.
You're thinking copyrights. Copyright vests as soon as you create the work.
For a trademark to be valid, you have to declare that its a trademark. You generally do this by placing the letters "TM" after the mark or R in a circle if you've registered the mark. If you just use the name and don't include that notification then it's not a trademark and you have no protection under trademark law.