If you abolished Social Security, but continued to collect the Social Security taxes, I think we'd have a second tea party (and I'm not talking about the teabaggers).
Obviously you wouldn't call it social security tax anymore, that would be silly. But you could clearly cover the deficit by not funding social security, without raising anyone's overall effective tax rate.
The military is too large a portion of our expenses, and the welfare programs are mostly self-funding. If you eliminated all welfare programs and pork, the IRS doesn't take in enough in income tax to cover just the debt interest and military. That indicates to me that welfare isn't the problem. What does that suggest to you?
The idea that welfare programs are self-funding is accounting hocus pocus. You could just as easily pass a law (which would make no actual change in what the government does in any way) which renamed social security tax to "defense tax" and allocated the money to the military, then fund social security out of general revenues and argue that social security is what should be cut because the military is self-funding.
The government spends more money than it collects by almost a trillion dollars a year. That means some of the spending needs to go. You can argue about what it should be, but the fact is, if you have to make significant cuts, you start with the things that cost the most money. That means welfare programs and military spending.
When the reforms were pushed through, the real estate industry carved out an exception for itself, on the (correct) grounds that eliminating mortgage interest deductibility would cause a huge decrease in home prices.
The sad part is that it was still sold as a way to make home ownership more accessible to the less well off. How do you do that? By making homes cost less money. What does the mortgage interest tax credit do? Makes home cost more money. Thanks, idiots.
Oh wait. You mean asking rich people to pay taxes to maintain the country they find it so lucrative to do business in is a return to feudalism? That would be hilarious if it weren't so pathetic.
Is that what the US government has been doing? Because it doesn't look that way to me. I mean look at social security: A millionaire pays a lower effective social security tax rate than a janitor, then when he retires he gets a bigger government check. Very progressive, no?
Or the alternative minimum tax -- sold as a way to get rich people to pay taxes, all it does is increasingly skewer the middle class by not being tied to inflation while the super rich just find and buy loopholes.
The problem is that it's politically impossible to tax the rich because they own the government, so instead we get feel good bills that people say will tax the rich but just end up screwing the middle class.
Think about the ideal welfare-optimizing system: The government would tax the super rich and use the money to provide social programs for the poor while leaving the middle class alone. Instead, the government taxes the middle class and leaves the super rich alone, which flattens the welfare curve below a six figure income and makes it so that no one there can rise above their means because anyone who gets close ends up being taxed back into debt.
Try proposing a system where no one who makes below $250,000/year pays any taxes and cuts social spending by the amount that people making less than that are no longer paying in taxes. It's win-win-win: Lower taxes, less government bureaucracy, but no less money in the hands of the lower 90%. But we can't do it because it would shatter the illusion of class warfare where the poor fight the good fight against the rich, and admit that the only thing that actually happens is that the poor fight a war of attrition against the middle class where the middle class loses but the poor never win.
They have the option to opt out of society, for example by moving to Somalia.
There is an intermediate position between socialism/communism and anarchy. You can have police, fire, roads and schools without having the government consume half the economy in social spending.
I understand what you're saying, but I don't think it's that simple. The terrorists full well know that the vast, overwhelming majority of the economic damage to us caused by a terrorist attack is our self-inflicted irrational response to it, and that includes adopting policies that promote corruption and autocracy.
More importantly, it doesn't matter what the terrorists think will get them what they want, eroding liberties is the thing most likely to lead to them actually getting what they want.
You can't say "If we give up our freedoms, the terrorists win" because no terrorist organization that I am aware of specifically wants you to give your own government more power. It's not an objective they can check off on a list. They don't benefit. They might gloat, granted, but whatever possessed them to resort to mass murder in the first place isn't advanced by the erosion of civil liberties in the name of imagined security.
Sure it is. Our liberties are what make us. If we continue on this path of eroding them, it will literally destroy America, which is what the terrorists want.
We so often defend liberty without explaining why it is that we do, because it was so well established so long ago that freedom is superior to the alternative, but we do so at the risk of forgetting the why. Liberty is the right to question and challenge the government, which absolutely necessary to prevent corruption and tyranny. Privacy allows dissenters to build a movement without the knowledge of those who would suppress it. The evils these rights are designed to prevent are very real. Take them away and you open Pandora's box, and it becomes only a matter of time before those evils manifest. A despot will destroy his country in ways that a terrorist can only dream.
Running insecure servers is fine until it isn't. Especially now with companies issuing mobile devices that employees take home to infect with all variety of trojans etc., you basically have to treat the internal network as the internet. Unless you're looking forward to cleaning up the enormous clusterfuck that results when some crime syndicate compromises one of your domain controllers, best to keep things patched.
So do "developers" need a second monitor? Probably not.
No, probably not... unless they run a debugger on their code, or read documentation, or want to compare two different source files to one another, etc.
Look, monitors cost ~$200 once. Programmers cost ~$80,000/year. Just buy the second monitor.
Here's something to think about though. If everyone were able to print money, money would be worthless. It would be completely over inflated and totally worthless. Why would I take money you printer when I can print my own? I think the same holds true of bitcoin. Someone with a bunch of computers laying around can just generate bitcoin all day long and cash it in. That seems, to me, to make the currency essentially worthless.
Computing time is a scarce resource. It requires equipment, electricity, etc. What will happen is that the cost of a bitcoin will fall to the price of the resources necessary to generate one. So if it takes X cents worth of electricity to produce a bitcoin, a bitcoin will be worth about X cents. At that point it makes no sense to waste resources generating more of them because the resources cost more than the bitcoins are worth, so the price will stabilize there.
There are alternatives other than "is legal tender" and "is worthless." A thing is worth whatever people will pay for it. If some people start accepting bitcoins as payment, other people will be willing to pay money for bitcoins in order to use them as payment -- particularly if the exchange rate is favorable. Once people realize that they can thereby sell their bitcoins for money to those people, more people will be willing to accept them as payment.
This has nothing to do with its status as legal tender. Canadian dollars are not legal tender in the US, but if someone on the US side of Niagra Falls wants to operate a store and list all the prices in US and Canadian dollars and accept either one as payment, is that illegal? Even assuming it is, if they put a vending machine in the lobby where you can exchange Canadian dollars for US dollars and then spend the US dollars, does it result in any practical difference, or do anything someone couldn't do with bitcoins?
But none of it matters until they reach critical mass. People will only buy them if vendors accept them, and vendors will only accept them if people are buying them. The problem is how you get one to happen in significant numbers before the other.
My first reaction was, "Is there really so little happening in our State that you couldn't fill the entire first 10 minutes with State news?"
You are clearly not understanding the economics of news gathering. They can send one reporter to interview someone at the Army Corps of Engineers and get a five minute story which they can feed to every anchor at every affiliate all over the country. In order to cover state news, they would have to have 50 times as many reporters so that they could have 50 different stories instead of just running the same story everywhere. That wouldn't be as profitable, you understand.
I would bet big that Nintendo's new system will not include optional Linux. No one will be touching that feature in a long time.
Anyone paying attention would understand that the lesson from all this is the opposite: By offering Linux, Sony put off the day when people broke the PS3 for several years. Their biggest mistake was not giving Linux full access to the hardware, because it caused people to have to break their security in order to do that, which is what kicked off this whole mess.
The fact is that there are people who will do whatever it takes to install Linux on anything they own. If you give them what they want at the outset, they don't have to break your security model in order to get it. If you try to make it hard for them, they just see it as a challenge and when they succeed they bust everything wide open and that allows the pirates and the cheaters to get in.
It depends on the organization. If all you're doing with your network is sharing a single 20Mbps Internet connection, you can just install a 54Mbps access point for the whole building because the bottleneck is the uplink rather than the wireless.
It's just that not all businesses have such minimal requirements. Sometimes you have scientific data sets, HD video, distcc with large projects and fast machines, etc. There are situations in which the network really is the performance bottleneck, even with a wired network. In those situations you simply can't put more than one such machine in the same collision domain without materially impacting performance.
Even in the case where you might get five or ten machines to share a fixed amount of bandwidth with no material performance degradation, you're still defeating almost all of the advantages of using wireless in the first place without mitigating its disadvantages. If you have ~250 users, you need 25-50 access points which become that many separate points of failure, and if you misconfigure a single one you give anyone on the street access to your internal network. You still end up having to run wires all over the building for the access points and running one wire for an access point is 95% as difficult as running ten wires at once for ten computers that are all in substantially the same location. And all it takes for your entire network to fall over on a regular basis is for the tenant on the next floor to install a cordless phone.
And as to your last point, if you need more net or peak bandwidth, you just add more or better capable wireless points. It's largely the same.
What I'm saying is that the sameness is the problem. If you need almost as much bandwidth per device as you have total (either because you don't have much spectrum available or because you have high bandwidth requirements), you end up installing as many or almost as many access points as you have workstations.
Moreover, you can't use ISP contention ratios on a LAN with good results. An ISP might have a 1Gbps uplink and put 5000 10Mbps users behind it, which is fine because on average less than 1/50th of the users will be active at once and they have highly heterogeneous usage patterns. The difference with a wireless broadcast domain in a business is that if you use the same contention ratio (i.e. fifty users in a broadcast domain), you can't make the assumption that only 1/50 of the users will be active at once, because you don't have a pool of 5000 to average it out over and because all the users have similar usage patterns. The Boss says that at 2PM everyone will sit down at their desk and watch a training video on the intranet server, so at 2PM everyone sits down at their desk and suddenly if your contention ratio is not very close to 1:1 the The Boss is mad at you because the video is choppy.
There is a reason they put a gigabit uplink on almost every 100Base-T 48 port switch: It's because the 48:1 contention ratio that would result if you had a 100Mbps uplink is unsuitable for a large variety of businesses.
The iPad being replaced by a newer version is completely different from not being able to run an application on all current devices.
How is that different at all? The user who bought an iPad six months ago when it was the "current" device doesn't give one damn why it doesn't work, all they know is that not all iOS apps work on their device which means they have to start looking on the side of the box to see whether the specific app supports the specific iOS device they have. The idea that you can guarantee support of all apps by continuously upgrading to the latest and greatest as soon as it comes out is cold comfort for anyone without unlimited money to burn.
If literally everyone only buys secondhand then the RIAA member companies go out of business and are replaced by some apparatus that it is not unconscionable to fund, so people can go back to buying new music. If not everyone gets in on the boycott, it deprives the RIAA of at least some money -- which is that much less money they have to lobby against your interests -- and there is still a supply of secondhand music from the people who don't have enough awareness or conscience to join the boycott.
If you used a valid "Lab Notebook" stile, where each page is dated, notes are in ink and all non used space was boxed and 'X'ed out, so that new stuff can not be added in.
Does the law actually say that? Because it makes about zero sense. Having no blank space to add material is an utterly frivolous requirement -- anyone willing to forge a document like that could trivially just get a new lab notebook, backdate it and write whatever they want in it without leaving any blank space.
No you route between a few *different* frequencies with basically no delay. You only need a few frequencies to avoid interference, just colour the space so no two bubbles of transmission volume are the same frequency. (Having a few different frequencies is how cell phones work for much the same reasons.)
Graph coloring is computationally difficult. How do you propose to do it ad hoc with devices that can move around as they're transmitting and where no single device can see the entire graph?
You've also obviously not understood the point about contention ratios; hosts are only sending for a fraction of the time, they don't need full capacity all the time. In fact most of the time they don't need any.
So if I've got an 11 Mbit/s access point, I can give ~11Mbit/s access to (say) 5-50 users and they won't notice too much difference.
You're assuming ISP-like contention ratios for LAN traffic. There are a wide variety of usage patterns for which that is not reasonable. Suppose my devices include HD TV set top boxes to which I want to stream HD content at ~30Mbps per device. There is no way that you can have 50 users in the same broadcast domain under that usage pattern with any sane amount of spectrum. If you're using e.g. 802.11g with 54Mbps available, having even two is unacceptable because it means they can't both watch TV at the same time. Moreover, the way ISPs do it is by having a much faster uplink than any single user can consume. The problem with wireless is that you want any single user to be able to consume all of the available bandwidth if no one else is using it, but if that is possible then any concurrent use whatsoever results in immediate reduced performance in direct proportion to the amount of actual concurrent use. Granted there are a lot of situations in which this doesn't matter -- sometimes you have 100Mbps of spectrum and each user only needs 1Mbps of bandwidth -- but that doesn't mean it isn't a performance disadvantage of the medium, only that there are situations in which you can live with the disadvantage in exchange for the convenience.
Essentially ALL internet access is (in reality) a contended service. You can get uncontended, it's just much more expensive.
That's the point: If you need uncontended service (or nearly so) with a wired network, you pay more money and lay more fiber and you get it. If you need uncontended service or anything like it with a wireless network, you have to build a wired network instead.
You don't need one per desk, you need one per n users, where n is the designed contention ratio.
So the total bandwidth winds out to be per volume, but you can design in how much bandwidth you want to support.
Having less than one access point per device requires you to have more spectrum to achieve the same amount of bandwidth per device because the available spectrum has to be shared between n devices.
Which brings up the other problem with wireless: Sometimes you just don't have enough spectrum, full stop. It doesn't matter how many access points you build, if you have 600Mbps worth of spectrum total and you have something like live video processing machines that require 1Gbps or 10Gbps connections, there is simply nothing you can do to fit 10 pounds of video into a 5 pound bag.
The thing about latency is that there's virtually no minimum, some routing techniques give 1 or just a few bits of delay at whatever speed you're using (say 16 bits at 100 Mbit/s, which is well under a microsecond per hop); you don't necessarily have to wait for receiving the whole packet before routing it on.
Processing delay is not the only source of latency. If node 2 is routing through node 1 and both are transmitting at the same time, somebody's packets are getting buffered or dropped. The more hops you have, the more opportunities for that sort of collision. All it takes is for one node in the chain to be transferring a large file and you max out your bandwidth and start filling up the buffers on every hop. The only way to avoid that is to try to implement some kind of QoS that prioritizes smaller data transfers, but that requires extra processing and increases latency in its own right, and incorrect classifications can really screw up latency sensitive high bandwidth applications.
I don't necessarily agree about the latency thing, individual nodes can have access to two different frequencies and can be listening and transmitting at the same time, with very low delays indeed.
I'm not talking about half duplex. In order to have what you suggest, i.e. increasing bandwidth by reducing transmission power levels, you either have to have endpoint devices route through endpoint devices and create a hop for every five feet you are away from the access point (and each hop adds latency), or you have to have to install a wired-in access point every five feet.
And the problem about having too many nodes trying to use the same internet access point, all you're saying is that you don't have enough access points. But then so what? No network, not even wired networks, can survive oversubscription of any one node.
What you're missing is that with wired networks, over-subscription is resolved by adding more bandwidth. You can run another fiber pair or upgrade terminating equipment. There is no equivalent with wireless -- there is 1Gbps of spectrum allocated for unlicensed use, that's what you have. You can't just pay a service tech $500 to install more wireless spectrum in your building.
What you can do is disperse the access points so that instead of having one 1Gbps access point for the whole building, you have an access point on every user's desk which only covers that desk, so that each user can get 1Gbps. But that's totally useless -- if have to run a cable to every desk then you can just plug the cable into the device and have better security, no interference-related network problems and a non-frivolous cost savings in not needing to buy a huge pile of wireless access points.
That requires each device to become a router, which at scale creates a huge amount of latency. Moreover, you still end up with shared bandwidth -- if you have to route through my device then you're necessarily sharing its available bandwidth to the uplink. Your model only works if there are lots of nodes that each talk to their neighbors as endpoints. When you have lots of nodes that all want to talk to the same node (like the closest uplink to the rest of the internet), you haven't really gained much if anything.
Slashdot Mirror
If you abolished Social Security, but continued to collect the Social Security taxes, I think we'd have a second tea party (and I'm not talking about the teabaggers).
Obviously you wouldn't call it social security tax anymore, that would be silly. But you could clearly cover the deficit by not funding social security, without raising anyone's overall effective tax rate.
The military is too large a portion of our expenses, and the welfare programs are mostly self-funding. If you eliminated all welfare programs and pork, the IRS doesn't take in enough in income tax to cover just the debt interest and military. That indicates to me that welfare isn't the problem. What does that suggest to you?
The idea that welfare programs are self-funding is accounting hocus pocus. You could just as easily pass a law (which would make no actual change in what the government does in any way) which renamed social security tax to "defense tax" and allocated the money to the military, then fund social security out of general revenues and argue that social security is what should be cut because the military is self-funding.
The government spends more money than it collects by almost a trillion dollars a year. That means some of the spending needs to go. You can argue about what it should be, but the fact is, if you have to make significant cuts, you start with the things that cost the most money. That means welfare programs and military spending.
When the reforms were pushed through, the real estate industry carved out an exception for itself, on the (correct) grounds that eliminating mortgage interest deductibility would cause a huge decrease in home prices.
The sad part is that it was still sold as a way to make home ownership more accessible to the less well off. How do you do that? By making homes cost less money. What does the mortgage interest tax credit do? Makes home cost more money. Thanks, idiots.
Oh wait. You mean asking rich people to pay taxes to maintain the country they find it so lucrative to do business in is a return to feudalism? That would be hilarious if it weren't so pathetic.
Is that what the US government has been doing? Because it doesn't look that way to me. I mean look at social security: A millionaire pays a lower effective social security tax rate than a janitor, then when he retires he gets a bigger government check. Very progressive, no?
Or the alternative minimum tax -- sold as a way to get rich people to pay taxes, all it does is increasingly skewer the middle class by not being tied to inflation while the super rich just find and buy loopholes.
The problem is that it's politically impossible to tax the rich because they own the government, so instead we get feel good bills that people say will tax the rich but just end up screwing the middle class.
Think about the ideal welfare-optimizing system: The government would tax the super rich and use the money to provide social programs for the poor while leaving the middle class alone. Instead, the government taxes the middle class and leaves the super rich alone, which flattens the welfare curve below a six figure income and makes it so that no one there can rise above their means because anyone who gets close ends up being taxed back into debt.
Try proposing a system where no one who makes below $250,000/year pays any taxes and cuts social spending by the amount that people making less than that are no longer paying in taxes. It's win-win-win: Lower taxes, less government bureaucracy, but no less money in the hands of the lower 90%. But we can't do it because it would shatter the illusion of class warfare where the poor fight the good fight against the rich, and admit that the only thing that actually happens is that the poor fight a war of attrition against the middle class where the middle class loses but the poor never win.
They have the option to opt out of society, for example by moving to Somalia.
There is an intermediate position between socialism/communism and anarchy. You can have police, fire, roads and schools without having the government consume half the economy in social spending.
If we eliminated the services but not the taxes we would.
But privacy is key to allowing people to live their own lives and to form and talk about their own ideas.
That's what I said. Read it again.
I understand what you're saying, but I don't think it's that simple. The terrorists full well know that the vast, overwhelming majority of the economic damage to us caused by a terrorist attack is our self-inflicted irrational response to it, and that includes adopting policies that promote corruption and autocracy.
More importantly, it doesn't matter what the terrorists think will get them what they want, eroding liberties is the thing most likely to lead to them actually getting what they want.
If the government didn't spend money on social services they would be able to barely balance the budget
FTFY.
You can't say "If we give up our freedoms, the terrorists win" because no terrorist organization that I am aware of specifically wants you to give your own government more power. It's not an objective they can check off on a list. They don't benefit. They might gloat, granted, but whatever possessed them to resort to mass murder in the first place isn't advanced by the erosion of civil liberties in the name of imagined security.
Sure it is. Our liberties are what make us. If we continue on this path of eroding them, it will literally destroy America, which is what the terrorists want.
We so often defend liberty without explaining why it is that we do, because it was so well established so long ago that freedom is superior to the alternative, but we do so at the risk of forgetting the why. Liberty is the right to question and challenge the government, which absolutely necessary to prevent corruption and tyranny. Privacy allows dissenters to build a movement without the knowledge of those who would suppress it. The evils these rights are designed to prevent are very real. Take them away and you open Pandora's box, and it becomes only a matter of time before those evils manifest. A despot will destroy his country in ways that a terrorist can only dream.
Running insecure servers is fine until it isn't. Especially now with companies issuing mobile devices that employees take home to infect with all variety of trojans etc., you basically have to treat the internal network as the internet. Unless you're looking forward to cleaning up the enormous clusterfuck that results when some crime syndicate compromises one of your domain controllers, best to keep things patched.
So do "developers" need a second monitor? Probably not.
No, probably not... unless they run a debugger on their code, or read documentation, or want to compare two different source files to one another, etc.
Look, monitors cost ~$200 once. Programmers cost ~$80,000/year. Just buy the second monitor.
Here's something to think about though. If everyone were able to print money, money would be worthless. It would be completely over inflated and totally worthless. Why would I take money you printer when I can print my own? I think the same holds true of bitcoin. Someone with a bunch of computers laying around can just generate bitcoin all day long and cash it in. That seems, to me, to make the currency essentially worthless.
Computing time is a scarce resource. It requires equipment, electricity, etc. What will happen is that the cost of a bitcoin will fall to the price of the resources necessary to generate one. So if it takes X cents worth of electricity to produce a bitcoin, a bitcoin will be worth about X cents. At that point it makes no sense to waste resources generating more of them because the resources cost more than the bitcoins are worth, so the price will stabilize there.
There are alternatives other than "is legal tender" and "is worthless." A thing is worth whatever people will pay for it. If some people start accepting bitcoins as payment, other people will be willing to pay money for bitcoins in order to use them as payment -- particularly if the exchange rate is favorable. Once people realize that they can thereby sell their bitcoins for money to those people, more people will be willing to accept them as payment.
This has nothing to do with its status as legal tender. Canadian dollars are not legal tender in the US, but if someone on the US side of Niagra Falls wants to operate a store and list all the prices in US and Canadian dollars and accept either one as payment, is that illegal? Even assuming it is, if they put a vending machine in the lobby where you can exchange Canadian dollars for US dollars and then spend the US dollars, does it result in any practical difference, or do anything someone couldn't do with bitcoins?
But none of it matters until they reach critical mass. People will only buy them if vendors accept them, and vendors will only accept them if people are buying them. The problem is how you get one to happen in significant numbers before the other.
My first reaction was, "Is there really so little happening in our State that you couldn't fill the entire first 10 minutes with State news?"
You are clearly not understanding the economics of news gathering. They can send one reporter to interview someone at the Army Corps of Engineers and get a five minute story which they can feed to every anchor at every affiliate all over the country. In order to cover state news, they would have to have 50 times as many reporters so that they could have 50 different stories instead of just running the same story everywhere. That wouldn't be as profitable, you understand.
I would bet big that Nintendo's new system will not include optional Linux. No one will be touching that feature in a long time.
Anyone paying attention would understand that the lesson from all this is the opposite: By offering Linux, Sony put off the day when people broke the PS3 for several years. Their biggest mistake was not giving Linux full access to the hardware, because it caused people to have to break their security in order to do that, which is what kicked off this whole mess.
The fact is that there are people who will do whatever it takes to install Linux on anything they own. If you give them what they want at the outset, they don't have to break your security model in order to get it. If you try to make it hard for them, they just see it as a challenge and when they succeed they bust everything wide open and that allows the pirates and the cheaters to get in.
It depends on the organization. If all you're doing with your network is sharing a single 20Mbps Internet connection, you can just install a 54Mbps access point for the whole building because the bottleneck is the uplink rather than the wireless.
It's just that not all businesses have such minimal requirements. Sometimes you have scientific data sets, HD video, distcc with large projects and fast machines, etc. There are situations in which the network really is the performance bottleneck, even with a wired network. In those situations you simply can't put more than one such machine in the same collision domain without materially impacting performance.
Even in the case where you might get five or ten machines to share a fixed amount of bandwidth with no material performance degradation, you're still defeating almost all of the advantages of using wireless in the first place without mitigating its disadvantages. If you have ~250 users, you need 25-50 access points which become that many separate points of failure, and if you misconfigure a single one you give anyone on the street access to your internal network. You still end up having to run wires all over the building for the access points and running one wire for an access point is 95% as difficult as running ten wires at once for ten computers that are all in substantially the same location. And all it takes for your entire network to fall over on a regular basis is for the tenant on the next floor to install a cordless phone.
And as to your last point, if you need more net or peak bandwidth, you just add more or better capable wireless points. It's largely the same.
What I'm saying is that the sameness is the problem. If you need almost as much bandwidth per device as you have total (either because you don't have much spectrum available or because you have high bandwidth requirements), you end up installing as many or almost as many access points as you have workstations.
Moreover, you can't use ISP contention ratios on a LAN with good results. An ISP might have a 1Gbps uplink and put 5000 10Mbps users behind it, which is fine because on average less than 1/50th of the users will be active at once and they have highly heterogeneous usage patterns. The difference with a wireless broadcast domain in a business is that if you use the same contention ratio (i.e. fifty users in a broadcast domain), you can't make the assumption that only 1/50 of the users will be active at once, because you don't have a pool of 5000 to average it out over and because all the users have similar usage patterns. The Boss says that at 2PM everyone will sit down at their desk and watch a training video on the intranet server, so at 2PM everyone sits down at their desk and suddenly if your contention ratio is not very close to 1:1 the The Boss is mad at you because the video is choppy.
There is a reason they put a gigabit uplink on almost every 100Base-T 48 port switch: It's because the 48:1 contention ratio that would result if you had a 100Mbps uplink is unsuitable for a large variety of businesses.
The iPad being replaced by a newer version is completely different from not being able to run an application on all current devices.
How is that different at all? The user who bought an iPad six months ago when it was the "current" device doesn't give one damn why it doesn't work, all they know is that not all iOS apps work on their device which means they have to start looking on the side of the box to see whether the specific app supports the specific iOS device they have. The idea that you can guarantee support of all apps by continuously upgrading to the latest and greatest as soon as it comes out is cold comfort for anyone without unlimited money to burn.
OK, then the damages for Jammie Thomas should be $0.95/song, right?
If literally everyone only buys secondhand then the RIAA member companies go out of business and are replaced by some apparatus that it is not unconscionable to fund, so people can go back to buying new music. If not everyone gets in on the boycott, it deprives the RIAA of at least some money -- which is that much less money they have to lobby against your interests -- and there is still a supply of secondhand music from the people who don't have enough awareness or conscience to join the boycott.
If you used a valid "Lab Notebook" stile, where each page is dated, notes are in ink and all non used space was boxed and 'X'ed out, so that new stuff can not be added in.
Does the law actually say that? Because it makes about zero sense. Having no blank space to add material is an utterly frivolous requirement -- anyone willing to forge a document like that could trivially just get a new lab notebook, backdate it and write whatever they want in it without leaving any blank space.
No you route between a few *different* frequencies with basically no delay. You only need a few frequencies to avoid interference, just colour the space so no two bubbles of transmission volume are the same frequency. (Having a few different frequencies is how cell phones work for much the same reasons.)
Graph coloring is computationally difficult. How do you propose to do it ad hoc with devices that can move around as they're transmitting and where no single device can see the entire graph?
You've also obviously not understood the point about contention ratios; hosts are only sending for a fraction of the time, they don't need full capacity all the time. In fact most of the time they don't need any.
So if I've got an 11 Mbit/s access point, I can give ~11Mbit/s access to (say) 5-50 users and they won't notice too much difference.
You're assuming ISP-like contention ratios for LAN traffic. There are a wide variety of usage patterns for which that is not reasonable. Suppose my devices include HD TV set top boxes to which I want to stream HD content at ~30Mbps per device. There is no way that you can have 50 users in the same broadcast domain under that usage pattern with any sane amount of spectrum. If you're using e.g. 802.11g with 54Mbps available, having even two is unacceptable because it means they can't both watch TV at the same time. Moreover, the way ISPs do it is by having a much faster uplink than any single user can consume. The problem with wireless is that you want any single user to be able to consume all of the available bandwidth if no one else is using it, but if that is possible then any concurrent use whatsoever results in immediate reduced performance in direct proportion to the amount of actual concurrent use. Granted there are a lot of situations in which this doesn't matter -- sometimes you have 100Mbps of spectrum and each user only needs 1Mbps of bandwidth -- but that doesn't mean it isn't a performance disadvantage of the medium, only that there are situations in which you can live with the disadvantage in exchange for the convenience.
Essentially ALL internet access is (in reality) a contended service. You can get uncontended, it's just much more expensive.
That's the point: If you need uncontended service (or nearly so) with a wired network, you pay more money and lay more fiber and you get it. If you need uncontended service or anything like it with a wireless network, you have to build a wired network instead.
You don't need one per desk, you need one per n users, where n is the designed contention ratio.
So the total bandwidth winds out to be per volume, but you can design in how much bandwidth you want to support.
Having less than one access point per device requires you to have more spectrum to achieve the same amount of bandwidth per device because the available spectrum has to be shared between n devices.
Which brings up the other problem with wireless: Sometimes you just don't have enough spectrum, full stop. It doesn't matter how many access points you build, if you have 600Mbps worth of spectrum total and you have something like live video processing machines that require 1Gbps or 10Gbps connections, there is simply nothing you can do to fit 10 pounds of video into a 5 pound bag.
The thing about latency is that there's virtually no minimum, some routing techniques give 1 or just a few bits of delay at whatever speed you're using (say 16 bits at 100 Mbit/s, which is well under a microsecond per hop); you don't necessarily have to wait for receiving the whole packet before routing it on.
Processing delay is not the only source of latency. If node 2 is routing through node 1 and both are transmitting at the same time, somebody's packets are getting buffered or dropped. The more hops you have, the more opportunities for that sort of collision. All it takes is for one node in the chain to be transferring a large file and you max out your bandwidth and start filling up the buffers on every hop. The only way to avoid that is to try to implement some kind of QoS that prioritizes smaller data transfers, but that requires extra processing and increases latency in its own right, and incorrect classifications can really screw up latency sensitive high bandwidth applications.
I don't necessarily agree about the latency thing, individual nodes can have access to two different frequencies and can be listening and transmitting at the same time, with very low delays indeed.
I'm not talking about half duplex. In order to have what you suggest, i.e. increasing bandwidth by reducing transmission power levels, you either have to have endpoint devices route through endpoint devices and create a hop for every five feet you are away from the access point (and each hop adds latency), or you have to have to install a wired-in access point every five feet.
And the problem about having too many nodes trying to use the same internet access point, all you're saying is that you don't have enough access points. But then so what? No network, not even wired networks, can survive oversubscription of any one node.
What you're missing is that with wired networks, over-subscription is resolved by adding more bandwidth. You can run another fiber pair or upgrade terminating equipment. There is no equivalent with wireless -- there is 1Gbps of spectrum allocated for unlicensed use, that's what you have. You can't just pay a service tech $500 to install more wireless spectrum in your building.
What you can do is disperse the access points so that instead of having one 1Gbps access point for the whole building, you have an access point on every user's desk which only covers that desk, so that each user can get 1Gbps. But that's totally useless -- if have to run a cable to every desk then you can just plug the cable into the device and have better security, no interference-related network problems and a non-frivolous cost savings in not needing to buy a huge pile of wireless access points.
That requires each device to become a router, which at scale creates a huge amount of latency. Moreover, you still end up with shared bandwidth -- if you have to route through my device then you're necessarily sharing its available bandwidth to the uplink. Your model only works if there are lots of nodes that each talk to their neighbors as endpoints. When you have lots of nodes that all want to talk to the same node (like the closest uplink to the rest of the internet), you haven't really gained much if anything.