Somehow I've never heard calls to encourage more males into Nursing etc. Unlike those calls for encouraging women into Engineering or Science (along with claims of discrimination).
The proposal of a failed migration of important services from one system to a different and nonbackward compatible system, or continuing to spend _company_ money on using the same thing (but faster etc) as before?
As long as it's easier and more effective to spend the company's money to help keep your job, most people will do it.
Only a few people would take the risk, and how many will pull it off successfully?
On paper, Merlin was supposed to stunt the development of Tehran's nuclear programme by sending Iran's weapons experts down the wrong technical path. The CIA believed that once the Iranians had the blueprints and studied them, they would believe the designs were usable and so would start to build an atom bomb based on the flawed designs.
The Russian studied the blueprints the CIA had given him. Within minutes of being handed the designs, he had identified a flaw. "This isn't right," he told the CIA officers gathered around the hotel room. "There is something wrong." His comments prompted stony looks, but no straight answers from the CIA men. No one in the meeting seemed surprised by the Russian's assertion that the blueprints didn't look quite right, but no one wanted to enlighten him further on the matter, either.
In fact, the CIA case officer who was the Russian's personal handler had been stunned by his statement. During a break, he took the senior CIA officer aside. "He wasn't supposed to know that," the CIA case officer told his superior. "He wasn't supposed to find a flaw."
"Don't worry," the senior CIA officer calmly replied. "It doesn't matter."
It's not rare for IBM to sell stuff which competes against products its subunits make.
You can have one IBM unit recommending/selling Cisco products which compete against more expensive IBM products by another IBM unit. You need some Sun stuff to work with some Microsoft stuff? IBM will say they'll do it.
From what I see, IBM is about providing choice, and helping customers make that choice for $$$$:).
If there isn't much choice you don't need as much "consulting" and support. For example if your choices are: reinstall, or format and reinstall, I don't think you'll want to pay a lot.
I think IBM makes more money from customers having more choices than they can cope with. Then those customers pay IBM to help them decide:).
That's why they are happy to provide the market with tons of different choices. Java,.Net, Linux, Windows, x86, RISC, Mainframes etc. Something has got to be pretty crap/loss-making for IBM to drop support for it;).
And then as you say IBM provide consulting+support services and the hardware to handle all the zillions of combinations of choices;).
Yeah, if I'm the owner of a company and not just some "Slash and Burn" CEO, I wouldn't want to have my core assets hostage to some third party _company_.
Having it in the hands of a trusted _person_ is different. If that person works for a different company, it's harder to ensure it's always that same trusted person who manages it.
Whereas if that trusted person works for you and the assets are in your company, it's a bit easier eh?
Basically the CPU manufacturers now have technology to put lots of transistors on a die (up to billions even), but they haven't figured out how to use them all to make things go faster - it's hard to think of new tricks to make a CPU _intelligently_ faster (e.g. process more instructions per cycle).
Cache is the easy way out - but it doesn't necessarily give you much better performance beyond a certain size (for typical workloads anyway).
So what do you do if you have so much transistors left, that using all of them for cache would give you an 8MB cache that makes your CPU only 2% faster than a CPU with 4MB cache for normal workloads (rather than some artificial benchmark)?
What the CPU manufacturers did was use them for another CPU (or three or more). Multicore CPUs.
The CPU manufacturers have hit limits of speeding up the serial execution of machine code even with all the superscalar, out of order execution, and other tricks used to exploit parallelism in the machine code.
In effect they have told the Software People - OK we give up, it's _your_ problem now, you make the stuff run faster, and do all that cool "parallel" stuff. Of course they didn't put it that way since it'll make them look bad.
But why should internet connectivity have to be like other services?
When a core ISP buys fibre, whether or not the bandwidth is used on the fibre it still costs them the same (yes they can choose to make huge profits, but bear with me).
Electricity and water come from somewhere, if they aren't used it doesn't cost the suppliers the same.
Voice calls are typically priced by time and not by data transferred.
So why not: 1) When there is contention "premium priority" traffic always takes precedence over "normal priority" traffic. 2) Users get some "dialer" where they can switch premium priority on and off (and it allows users to turn off premium priority automatically). 3) packages are $Y a month for X hours a month of traffic at "premium priority" 4) You can purchase extra premium hours for $Z 5) If you run out, your traffic is still carried but at a lower priority.
The benefit is the ISP can still oversubscribe users - they will have a good idea of how much they can oversubscribe, even if users leave their PCs connected all the time downloading and uploading.
So users keep their P2P running, and if they really want their P2P or game to be fast they can "dial in", it doesn't matter whether the user's P2P is encrypted or not, the ISP can tell how much priority to give that user.
It might sound like a step backwards to "dial up", but the connection can still be always on, just crap most of the time when you don't care so much.
In contrast the typical charge per GB schemes don't allow an ISP to oversubscribe efficiently. say everyone signs up for premium and unlimited, if 10% _really_ want a fast download, while the other 90% users online keeping P2P running continuously etc are actually asleep and/or don't care, how does the ISP figure out who to give crap service to?
With my proposal if there are too many requests for premium service, a responsible ISP can actually say "Sorry too many premium users already" and reject the request, while that might piss off the user, the user doesn't end up paying for premium and not getting premium.
The ISP can do "peak" and "off peak" rates and all the joys of time of day/week/month based pricing. Most ISP users are humans and live in the real world, they eat, they sleep, they go to work (and misuse their company connection), so I daresay even with ubiquitous computing, most won't be needing "premium priority" all the time.
Why would you need sequential transaction numbers? I can understand the need for unique transaction numbers.
I can also understand why if you write to the same row in the same table you need to serialize. I don't see why you can't take advantage of the 16 drive controllers if you don't write to the same row.
So far to me it seems like I/O is a bigger problem. After decades hard drives still have about the same access times. Add as many cores as you want, but when the time comes to write the results permanently you have to wait for the disks, and that sure is slow.
Enough men? I think there are very many Slashdotters who would sign up to make mythical children with mythical women for a month or however long it takes.
After all many Slashdotters already have years of "hands on" experience making virtual children with virtual women.
I'm sure they'd be game to tackle the learning curves head on.
The US Gov loves to use China as a bogeyman to the citizens e.g. "Oh no the chinese will kill us all".
Sure the Chinese Gov is evil etc, but it's maybe only slightly above average in evilness[1] and I doubt they are stupid enough to attack US Gov - it will hurt China a lot too.
After all the USA buys what China makes and pays them in US Dollars. Whenever the USA runs short on dollars it "prints" more (for example by issuing bonds, which China buys;) ). Printing more USD = USD becomes worth less, the USA thus indirectly "taxes" China and the other countries that hold lots of USD in reserves for trade, or bonds or assets. It's a strange scheme but it's lasting longer than the other previous schemes!
China can NOT afford to confront the USA directly. I bet any Chinese leader that tries to launch nukes at the USA or declare war will end up in serious trouble with the Chinese military top brass _unless_ the USA strikes first or threatens to strike first. Judging from history and current regimes it is more likely that the USA will bomb China before China will bomb USA.
Even the USSR blinked in the Cuban crisis (though if you look it up you will find that the USA put missiles at the USSR's doorstep _first_, the USSR was most certainly a bad guy, but they had valid reason to want to defend themselves from the USA). So the USSR and USA fought proxy wars in Afghanistan and other countries. Instead of killing each other's civilians - they pick some unfortunate country to play their deadly game. The USSR lost.
Similarly, the most China will do is fight a proxy war with the USA.
China is gaining ground in economic and military strength, but the USA spends almost as much as the rest of the world combined. http://www.globalissues.org/Geopolitics/ArmsTrade/Spending.asp
So one could very well say that China has to keep up its military spending to discourage the US from attacking it, or bullying it.
[1] Plenty of other evil countries in the world, the US Gov ranks as quite evil - just look at how well the US Gov respects democracy in other countries. In the past 100 years the US Gov has overthrown many democratically elected governments around the world. I suspect that the US Gov has overthrown more democractically elected leaders than it has overthrown dictators (it's no surprise many dictators know how to resist "hostile takeovers" since they themselves were installed with the help of the US Gov).
Nowadays Apple does about as much evil stuff as Microsoft. They just have less power, so the amount of evil they can do is more limited.
Google? As long as there isn't a better search engine I'll use Google. I used Infoseek when it came out, then Altavista, then Altavista/Hotbot, then Google. I don't trust Google in terms of security - they have not had a good track record with that.
As for Microsoft - I don't recall ever liking them, though the first programming language I learnt was Applesoft Basic.
0) backup your stuff regularly. Much of my work is checked in to a svn repository (svn isn't great but hey at least we're using something:) ).
1) If the laptop belongs to you, 1.1) modify your laptop (and similar stuff) so that the fence value goes down. Nowadays there's stuff like laser engraving printers, so pay a decent artist to engrave a custom work you'd like on your laptop (and maybe even airbrush too). They could steal it and then sell the parts, but I think it's a lot more work for them for the amount of money they'd get. If they liked doing extra work so much I don't think they'd be in the business of stealing laptops etc. And maybe you could engrave the memory modules and HDD too- "Stolen from <initials>".
1.2) Buy some insurance on the stuff. You could do 2.1 too, if the insurance company is ok with it - if the artist is really good maybe the insurance company might think the laptop's value is higher than standard;).
2) If the laptop doesn't belong to you, just do the backup stuff and let your company worry about it - if they haven't worried about it at all, you could suggest stuff (insurance, backups, spares etc), and then it's all up to them.
They have cameras in my workplace, but though they have videos of the person who took somebody's phone I don't think he ever got caught (the police here are crap - they are more interested in chasing the sheep than the wolves. e.g. catching people for speeding vs catching people for theft/robbery, and some of the cops steal stuff too). It wasn't an employee - he sneaked in when somebody opened the door (most people don't think about such stuff - and we run a 24 hour callcentre so lots of employees don't recognize fellow employees ), then walked around and then stole stuff.
"some asshat is using twenty times my bandwidth downloading some file "
This is because many bandwidth sharing methods out there are on a "per connection" basis. So if someone makes 20 connections they get 20x more than someone who makes one, everything else being the same.
The fair case would be everyone gets a fair share of bandwidth based on a per user basis. Now in _most_ cases this can be mapped to a "per IP" basis.
So even if some p2p person makes 100 connections, he only has one IP, and you have one IP, so this sharing method would mean you both download at the same speed.
This is actually technically viable, the company I work for does that in some sites we provide service too. But our business is providing expensive internet access in hotels, airports etc. Expensive internet access = a bit oxymoronic but if you want economy class prices go squeeze in the back with the rest.
You see, the problem with naive "fair share" is IF ISPs oversubscribe - sharing out bandwidth fairly becomes a problem once you have users running P2P stuff 24 hours a day. They just leave their P2P client and PCs on all the time even if they "don't really care" - so if all the users do that, it means the "20:80" or similar assumption no longer works - everyone is using the network all the time, not just a small percentage.
A fair 1000 slices of banana while fair still makes for very little banana per person, so they won't be happy with their tiny fair share.
So the "solutions" are 1) Don't oversubscribe (buy a much bigger banana) This usually makes your service rather costly, unless you bought tons of fibre/connectivity when it was cheap (e.g. the company that laid it went bust) and have good peering arrangements.
And ISPs that do get it cheap, probably still want to charge and make big profits $$$$. Why charge much less than the rest when you can get away with charging about the same prices?
2) Cache the P2P stuff (get a banana, and then copy it) This is actually technically a good idea - ISP runs their own "super peers" that automatically cache and seed to their users, internal bandwidth is quite cheap. But legally it is not - the RIAA, MPAA etc will kill you.
3) Figure out which P2P stuff you can throttle (not everyone actually needs a banana "right now", they just ask for it anyway )
Problem is some users actually want their P2P stuff ASAP. And the others will still grumble anyway - since they paid for "all you can eat" and aren't getting it.
If 80% of your users run P2P in the background all the time it's going to be slow for all of them (if you don't do 1) or 2) ). You don't know which to prioritize - which users are going to take the trouble to tell you?
So maybe if you give the users an incentive to say "my whole connection or P2P stuff is now low priority", then this could work.
This could be something like the old days "dial up" hours model - but basically you get X hours of "premium priority" a month in your package, and the rest of it is at normal priority (otherwise you purchase additional premium priority time). Notice the marketspeak: low->normal and normal->premium;).
Trouble is how do you make it easy for Joe Sixpack to switch modes reliably and without forgetting and either racking up $$$ bills or running out of premium priority time?
This might seem similar to the "charge you for extra bandwidth" but it doesn't hurt you so much if you suddenly get DDoSed or get infected by a worm - as long as you don't go to premium priority mode, and even so, you don't lose internet access (you just get crap) and you don't get huge bills.
Plus the junk you spew is sent at a lower priority:).
I actually don't mind my money being taxed away to help the poor. I mind my money being taxed away to make the rich even richer or more powerful.
Sure some of the poor are stupid and/or lazy, but you know when stupid people get desperate they might do really stupid things like kill you just so they can eat or afford medical treatment or drugs.
In Indonesia you often have riots, one of the factors is the poor there really have nothing much to lose. When your odds of survival from day to day aren't that high in the first place, you stop caring so much even if the cops threaten to shoot you for rioting. "Breads and circuses" does work - if they know their stomachs will be filled every day, things will be safer for everyone.
Rich countries (or countries with extensive lines of credit like the USA;) ) can afford to keep their poor alive[1].
Apparently the USA spends USD200 million a day in Iraq. Assuming 10% of the USA are poor, keeping 30 million US poor alive with USD6/day each is difficult, but I wouldn't say it's impossible. It's more than what many people in poorer countries get for wages but the USA is more expensive to live in.
I wouldn't mind paying my equivalent (I don't live in the USA) of USD0.70/ day (90% pay for 10%) just to reduce the "desperate crime" rates significantly. Heck I'm sure the top 10% rich in the USA can afford to pay more, not that they'd want to.
[1] Excluding very expensive therapies - I think there has to be a point where people must be considered "beyond economical repair by the State" then it's up to themselves and others. Because as technology improves I forsee more and more ultra-expensive treatments appearing.
I think I actually got media player classic to play quicktime stuff last year, but somehow recently something broke and it stopped working - crashes. But nowadays there are usually alternatives in different video formats (e.g. WMV)so I download those instead. People might say WMV is more evil, but so far quicktime has annoyed me a LOT more than windows media player 6 and 9.
Slashdot Mirror
Somehow I've never heard calls to encourage more males into Nursing etc. Unlike those calls for encouraging women into Engineering or Science (along with claims of discrimination).
Which is more likely to get a Boss sacked?
The proposal of a failed migration of important services from one system to a different and nonbackward compatible system, or continuing to spend _company_ money on using the same thing (but faster etc) as before?
As long as it's easier and more effective to spend the company's money to help keep your job, most people will do it.
Only a few people would take the risk, and how many will pull it off successfully?
Makes things more interesting...
http://www.guardian.co.uk/environment/2006/jan/05/energy.g2
On paper, Merlin was supposed to stunt the development of Tehran's nuclear programme by sending Iran's weapons experts down the wrong technical path. The CIA believed that once the Iranians had the blueprints and studied them, they would believe the designs were usable and so would start to build an atom bomb based on the flawed designs.
The Russian studied the blueprints the CIA had given him. Within minutes of being handed the designs, he had identified a flaw. "This isn't right," he told the CIA officers gathered around the hotel room. "There is something wrong." His comments prompted stony looks, but no straight answers from the CIA men. No one in the meeting seemed surprised by the Russian's assertion that the blueprints didn't look quite right, but no one wanted to enlighten him further on the matter, either.
In fact, the CIA case officer who was the Russian's personal handler had been stunned by his statement. During a break, he took the senior CIA officer aside. "He wasn't supposed to know that," the CIA case officer told his superior. "He wasn't supposed to find a flaw."
"Don't worry," the senior CIA officer calmly replied. "It doesn't matter."
Banks are a better counter example to my OP. Companies have to trust banks with their cash and financial transactions.
So yeah I'm wrong.
It's not rare for IBM to sell stuff which competes against products its subunits make.
:).
You can have one IBM unit recommending/selling Cisco products which compete against more expensive IBM products by another IBM unit. You need some Sun stuff to work with some Microsoft stuff? IBM will say they'll do it.
From what I see, IBM is about providing choice, and helping customers make that choice for $$$$
If there isn't much choice you don't need as much "consulting" and support. For example if your choices are: reinstall, or format and reinstall, I don't think you'll want to pay a lot.
Yep.
:).
.Net, Linux, Windows, x86, RISC, Mainframes etc. Something has got to be pretty crap/loss-making for IBM to drop support for it ;).
;).
I think IBM makes more money from customers having more choices than they can cope with. Then those customers pay IBM to help them decide
That's why they are happy to provide the market with tons of different choices. Java,
And then as you say IBM provide consulting+support services and the hardware to handle all the zillions of combinations of choices
Yeah, if I'm the owner of a company and not just some "Slash and Burn" CEO, I wouldn't want to have my core assets hostage to some third party _company_.
Having it in the hands of a trusted _person_ is different. If that person works for a different company, it's harder to ensure it's always that same trusted person who manages it.
Whereas if that trusted person works for you and the assets are in your company, it's a bit easier eh?
Well it's the job of the language people to do that NFA stuff.
As long as they do it in a backward compatible way I don't care.
If you bother to read it, my proposal attempts to deal with what you say.
It is easier do "Time of use" fees with my proposal than with "bytes transferred" schemes.
Basically the CPU manufacturers now have technology to put lots of transistors on a die (up to billions even), but they haven't figured out how to use them all to make things go faster - it's hard to think of new tricks to make a CPU _intelligently_ faster (e.g. process more instructions per cycle).
;).
Cache is the easy way out - but it doesn't necessarily give you much better performance beyond a certain size (for typical workloads anyway).
So what do you do if you have so much transistors left, that using all of them for cache would give you an 8MB cache that makes your CPU only 2% faster than a CPU with 4MB cache for normal workloads (rather than some artificial benchmark)?
What the CPU manufacturers did was use them for another CPU (or three or more). Multicore CPUs.
The CPU manufacturers have hit limits of speeding up the serial execution of machine code even with all the superscalar, out of order execution, and other tricks used to exploit parallelism in the machine code.
In effect they have told the Software People - OK we give up, it's _your_ problem now, you make the stuff run faster, and do all that cool "parallel" stuff. Of course they didn't put it that way since it'll make them look bad.
Well that's my made up version of reality
"The human body is the most complicated thing on the planet.. and yet engineers consistently believe they are smarter than doctors"
:).
The first part of your sentence doesn't appear well related to the second part.
As for the engineers who think they're smarter than doctors, they should provide some good evidence if they are real engineers
But why should internet connectivity have to be like other services?
When a core ISP buys fibre, whether or not the bandwidth is used on the fibre it still costs them the same (yes they can choose to make huge profits, but bear with me).
Electricity and water come from somewhere, if they aren't used it doesn't cost the suppliers the same.
Voice calls are typically priced by time and not by data transferred.
So why not:
1) When there is contention "premium priority" traffic always takes precedence over "normal priority" traffic.
2) Users get some "dialer" where they can switch premium priority on and off (and it allows users to turn off premium priority automatically).
3) packages are $Y a month for X hours a month of traffic at "premium priority"
4) You can purchase extra premium hours for $Z
5) If you run out, your traffic is still carried but at a lower priority.
The benefit is the ISP can still oversubscribe users - they will have a good idea of how much they can oversubscribe, even if users leave their PCs connected all the time downloading and uploading.
So users keep their P2P running, and if they really want their P2P or game to be fast they can "dial in", it doesn't matter whether the user's P2P is encrypted or not, the ISP can tell how much priority to give that user.
It might sound like a step backwards to "dial up", but the connection can still be always on, just crap most of the time when you don't care so much.
In contrast the typical charge per GB schemes don't allow an ISP to oversubscribe efficiently. say everyone signs up for premium and unlimited, if 10% _really_ want a fast download, while the other 90% users online keeping P2P running continuously etc are actually asleep and/or don't care, how does the ISP figure out who to give crap service to?
With my proposal if there are too many requests for premium service, a responsible ISP can actually say "Sorry too many premium users already" and reject the request, while that might piss off the user, the user doesn't end up paying for premium and not getting premium.
The ISP can do "peak" and "off peak" rates and all the joys of time of day/week/month based pricing. Most ISP users are humans and live in the real world, they eat, they sleep, they go to work (and misuse their company connection), so I daresay even with ubiquitous computing, most won't be needing "premium priority" all the time.
Why would you need sequential transaction numbers? I can understand the need for unique transaction numbers.
I can also understand why if you write to the same row in the same table you need to serialize. I don't see why you can't take advantage of the 16 drive controllers if you don't write to the same row.
So far to me it seems like I/O is a bigger problem. After decades hard drives still have about the same access times. Add as many cores as you want, but when the time comes to write the results permanently you have to wait for the disks, and that sure is slow.
Enough men? I think there are very many Slashdotters who would sign up to make mythical children with mythical women for a month or however long it takes.
After all many Slashdotters already have years of "hands on" experience making virtual children with virtual women.
I'm sure they'd be game to tackle the learning curves head on.
They missed the part where you are also forced to watch ads in order to generate the "approved discontent" ;).
In a lot of engineering if you do your job really well, I think it is supposed to end up quite boring.
;).
;).
No crashes. No screaming. No panic. No long hours spent to try to deal with your mistakes.
In my engineering course I think the female:male ratio was less than 1:10, so to me that's another minus that guy should consider
I think it is a common saying among females in tech courses that even if the odds are good (for them) the goods are odd
The US Gov loves to use China as a bogeyman to the citizens e.g. "Oh no the chinese will kill us all".
;) ). Printing more USD = USD becomes worth less, the USA thus indirectly "taxes" China and the other countries that hold lots of USD in reserves for trade, or bonds or assets. It's a strange scheme but it's lasting longer than the other previous schemes!
Sure the Chinese Gov is evil etc, but it's maybe only slightly above average in evilness[1] and I doubt they are stupid enough to attack US Gov - it will hurt China a lot too.
After all the USA buys what China makes and pays them in US Dollars. Whenever the USA runs short on dollars it "prints" more (for example by issuing bonds, which China buys
China can NOT afford to confront the USA directly. I bet any Chinese leader that tries to launch nukes at the USA or declare war will end up in serious trouble with the Chinese military top brass _unless_ the USA strikes first or threatens to strike first. Judging from history and current regimes it is more likely that the USA will bomb China before China will bomb USA.
Even the USSR blinked in the Cuban crisis (though if you look it up you will find that the USA put missiles at the USSR's doorstep _first_, the USSR was most certainly a bad guy, but they had valid reason to want to defend themselves from the USA). So the USSR and USA fought proxy wars in Afghanistan and other countries. Instead of killing each other's civilians - they pick some unfortunate country to play their deadly game. The USSR lost.
Similarly, the most China will do is fight a proxy war with the USA.
China is gaining ground in economic and military strength, but the USA spends almost as much as the rest of the world combined. http://www.globalissues.org/Geopolitics/ArmsTrade/Spending.asp
So one could very well say that China has to keep up its military spending to discourage the US from attacking it, or bullying it.
[1] Plenty of other evil countries in the world, the US Gov ranks as quite evil - just look at how well the US Gov respects democracy in other countries. In the past 100 years the US Gov has overthrown many democratically elected governments around the world. I suspect that the US Gov has overthrown more democractically elected leaders than it has overthrown dictators (it's no surprise many dictators know how to resist "hostile takeovers" since they themselves were installed with the help of the US Gov).
Nowadays Apple does about as much evil stuff as Microsoft. They just have less power, so the amount of evil they can do is more limited.
Google? As long as there isn't a better search engine I'll use Google. I used Infoseek when it came out, then Altavista, then Altavista/Hotbot, then Google. I don't trust Google in terms of security - they have not had a good track record with that.
As for Microsoft - I don't recall ever liking them, though the first programming language I learnt was Applesoft Basic.
0) backup your stuff regularly. Much of my work is checked in to a svn repository (svn isn't great but hey at least we're using something :) ).
;).
1) If the laptop belongs to you,
1.1) modify your laptop (and similar stuff) so that the fence value goes down. Nowadays there's stuff like laser engraving printers, so pay a decent artist to engrave a custom work you'd like on your laptop (and maybe even airbrush too). They could steal it and then sell the parts, but I think it's a lot more work for them for the amount of money they'd get. If they liked doing extra work so much I don't think they'd be in the business of stealing laptops etc. And maybe you could engrave the memory modules and HDD too- "Stolen from <initials>".
1.2) Buy some insurance on the stuff. You could do 2.1 too, if the insurance company is ok with it - if the artist is really good maybe the insurance company might think the laptop's value is higher than standard
2) If the laptop doesn't belong to you, just do the backup stuff and let your company worry about it - if they haven't worried about it at all, you could suggest stuff (insurance, backups, spares etc), and then it's all up to them.
They have cameras in my workplace, but though they have videos of the person who took somebody's phone I don't think he ever got caught (the police here are crap - they are more interested in chasing the sheep than the wolves. e.g. catching people for speeding vs catching people for theft/robbery, and some of the cops steal stuff too). It wasn't an employee - he sneaked in when somebody opened the door (most people don't think about such stuff - and we run a 24 hour callcentre so lots of employees don't recognize fellow employees ), then walked around and then stole stuff.
"some asshat is using twenty times my bandwidth downloading some file "
;).
:).
This is because many bandwidth sharing methods out there are on a "per connection" basis. So if someone makes 20 connections they get 20x more than someone who makes one, everything else being the same.
The fair case would be everyone gets a fair share of bandwidth based on a per user basis. Now in _most_ cases this can be mapped to a "per IP" basis.
So even if some p2p person makes 100 connections, he only has one IP, and you have one IP, so this sharing method would mean you both download at the same speed.
This is actually technically viable, the company I work for does that in some sites we provide service too. But our business is providing expensive internet access in hotels, airports etc. Expensive internet access = a bit oxymoronic but if you want economy class prices go squeeze in the back with the rest.
You see, the problem with naive "fair share" is IF ISPs oversubscribe - sharing out bandwidth fairly becomes a problem once you have users running P2P stuff 24 hours a day. They just leave their P2P client and PCs on all the time even if they "don't really care" - so if all the users do that, it means the "20:80" or similar assumption no longer works - everyone is using the network all the time, not just a small percentage.
A fair 1000 slices of banana while fair still makes for very little banana per person, so they won't be happy with their tiny fair share.
So the "solutions" are
1) Don't oversubscribe (buy a much bigger banana)
This usually makes your service rather costly, unless you bought tons of fibre/connectivity when it was cheap (e.g. the company that laid it went bust) and have good peering arrangements.
And ISPs that do get it cheap, probably still want to charge and make big profits $$$$. Why charge much less than the rest when you can get away with charging about the same prices?
2) Cache the P2P stuff (get a banana, and then copy it)
This is actually technically a good idea - ISP runs their own "super peers" that automatically cache and seed to their users, internal bandwidth is quite cheap. But legally it is not - the RIAA, MPAA etc will kill you.
3) Figure out which P2P stuff you can throttle (not everyone actually needs a banana "right now", they just ask for it anyway )
Problem is some users actually want their P2P stuff ASAP. And the others will still grumble anyway - since they paid for "all you can eat" and aren't getting it.
If 80% of your users run P2P in the background all the time it's going to be slow for all of them (if you don't do 1) or 2) ). You don't know which to prioritize - which users are going to take the trouble to tell you?
So maybe if you give the users an incentive to say "my whole connection or P2P stuff is now low priority", then this could work.
This could be something like the old days "dial up" hours model - but basically you get X hours of "premium priority" a month in your package, and the rest of it is at normal priority (otherwise you purchase additional premium priority time). Notice the marketspeak: low->normal and normal->premium
Trouble is how do you make it easy for Joe Sixpack to switch modes reliably and without forgetting and either racking up $$$ bills or running out of premium priority time?
This might seem similar to the "charge you for extra bandwidth" but it doesn't hurt you so much if you suddenly get DDoSed or get infected by a worm - as long as you don't go to premium priority mode, and even so, you don't lose internet access (you just get crap) and you don't get huge bills.
Plus the junk you spew is sent at a lower priority
I actually don't mind my money being taxed away to help the poor. I mind my money being taxed away to make the rich even richer or more powerful.
;) ) can afford to keep their poor alive[1].
Sure some of the poor are stupid and/or lazy, but you know when stupid people get desperate they might do really stupid things like kill you just so they can eat or afford medical treatment or drugs.
In Indonesia you often have riots, one of the factors is the poor there really have nothing much to lose. When your odds of survival from day to day aren't that high in the first place, you stop caring so much even if the cops threaten to shoot you for rioting. "Breads and circuses" does work - if they know their stomachs will be filled every day, things will be safer for everyone.
Rich countries (or countries with extensive lines of credit like the USA
Apparently the USA spends USD200 million a day in Iraq. Assuming 10% of the USA are poor, keeping 30 million US poor alive with USD6/day each is difficult, but I wouldn't say it's impossible. It's more than what many people in poorer countries get for wages but the USA is more expensive to live in.
I wouldn't mind paying my equivalent (I don't live in the USA) of USD0.70/ day (90% pay for 10%) just to reduce the "desperate crime" rates significantly. Heck I'm sure the top 10% rich in the USA can afford to pay more, not that they'd want to.
[1] Excluding very expensive therapies - I think there has to be a point where people must be considered "beyond economical repair by the State" then it's up to themselves and others. Because as technology improves I forsee more and more ultra-expensive treatments appearing.
Would be nice if someone came up with cheap AND really good night vision goggles - wide field of view :).
That server probably hosts multiple domains. I think it's Sam's hosting provider who got pwn3d. So linking this to Sam is probably premature.
try whois 80.65.228.130
Slashdot's lameness filter prevents me from posting the results.
I think I actually got media player classic to play quicktime stuff last year, but somehow recently something broke and it stopped working - crashes. But nowadays there are usually alternatives in different video formats (e.g. WMV)so I download those instead. People might say WMV is more evil, but so far quicktime has annoyed me a LOT more than windows media player 6 and 9.
There's also Tron, which is apparently quite popular in Japan and used in lots of embedded stuff.
http://en.wikipedia.org/wiki/TRON_Project