That reminds me, why does MS DHCP have such a weird format for MAC addresses? It doesn't match anything at all! Neither the Windows nor the Linux tools output MAC addresses like that. Although in Linux at least I could use sed to convert it.
I dread working with MS GUIs. Everything has 20 tabs, sometimes pops up extra windows and requires half an hour of clicking. It's not possible to figure out what the full current config is without clicking everywhere. Then there's my favourite annoyance: Fixed size windows with list views that sometimes require scrolling to see all the information in them.
Take for instance setting up a DHCP server. Task: Switch 100 hosts from static IP address to DHCP, configured to give them the same address they currently have, by MAC.
Windows: Hand falls off due to clicking for an hour, and mindlessly copy/pasting. Error prone. MS DHCP is completely braindead by the way, and seems to confuse the description and MAC fields (did anybody even test that?)
Linux: Maybe 5 minutes to write a script that: pings subnet, parses output of 'arp -a', and automatically generates a dhcpd config file. If the hosts were Linux ones I could also automatically replace the network settings on all of them. If it goes wrong, fix script then run again. Not all hosts are on? No problem, I can tweak the script to run periodically, detecting new computers and adding the config for them.
Yes, GUIs are easier for tiny tasks, but once you need something uncommon they're a huge pain in the ass. And it's not like there can't be a GUI that generates dhcpd.conf either.
It's not new either. And it does what you want it to do. However, it's a royal pain in the ass to configure, because you need to figure out what every application should be able to do. It's definitely not something for a newbie, and probably it will be long before such a thing is usable by normal people.
Also, I doubt it'll work well for Windows. For Linux sure, distributions would just have to provide the SELinux security settings for the packages. But for Windows? Who provides the list of things the application should be able to do? It can't be the author, as all the malware would just ship rules allowing them to mess with whatever they need.
Actually, this failure mode exists almost exactly on modern operating systems.
Unless you put limits to it, any application can get all your system RAM. The OS and applications get their memory from the same place, so it's still possible to run out of it. It's perfectly possible to use so much that the computer grinds to a halt and not even the mouse pointer moves.
Now, the memory space separation makes sure your application can't hose the OS' or some other program's state, but that doesn't mean it can't do something like slowing things down so much it becomes unusable.
This sort of failure mode is easy to see in these conditions: low amounts of free RAM, large amounts of available swap space, and processes that quickly allocate large amounts of RAM and actually use it. A good example program is a compiler. On some files, gcc can use 300MB RAM. Swap is useful for offloading unused stuff to disk to make room for something more useful, but if you reach the point where there really isn't enough RAM and the system needs to continuously swap stuff out and in, the system can grind to a halt.
The traffic advisory organization runs a Certificate Authority. They have their private key, which they keep well safe. Subdivisions that need to broadcast messages get their keys signed by the CA, then use them to sign messages. GPS device checks that the message is signed by a key signed by the CA.
If a key is compromised, the CA adds it to the revocation list, which for instance could be regularly broadcasted. After hearing the broadcast, GPS devices refuse to accept messages signed with that key, and whoever used the compromised key has to get a new one.
Wasn't one of the concepts in sword forging that you want to have a lot of carbon on the edge, and less of it on the back (or middle) to get a sword that's sharp but bends so that it doesn't break?
A sword made fully of diamond would be stupid as the whole sword would shatter, but one with a diamond edge should cut really good (and chip very badly afterwards).
Here's a solution for the author:
In the Fullmetal Alchemist anime, alchemy is a sort of the science of transformation of matter. An alchemist can reshape matter, but not create it. Still, an alchemist could easily and quickly make a sword out of anything that happened to be lying around. So that's a solution to the breaking sword problem, when it breaks, just reform it. Carbon is so abundant that it could be made pretty much anywhere.
Of course such a thing would be still impractical as you couldn't parry somebody else's sword without your breaking. My solution to that: Diamond throwing knives or stars. If it breaks, even better.
Sound quality, maybe. CPU load, I'm not that sure.
Some time ago, Creative intimidated John Carmack into supporting EAX in Doom 3. This is yet another reason why I don't use Creative hardware anymore. The SB Live cards causing disk corruption with VIA boards, drivers being unstable on SMP (this was before dual core) and them taking drivers off their site for some time are some others.
But anyway, why do you think Creative had to basically force John Carmack to support their EAX tech? Apparently because Doom 3, on modern hardware, can mix 5.1 audio faster than the sound card.
Seriously, mixing sounds isn't that big of a deal. I have here a book on game programming that describes the creation of a wolfenstein-3D-like game, and part of it explains how to mix sound in software. This was on a 386 CPU. Besides that, all a sound card gives you is extra reverb/etc effects, and computers don't seem to have much of a problem with doing that in real time anymore either.
with more than just outsourcing the graphics to a GPU
AMD seems to be working on putting a GPU in ther CPU
maybe a chip dedicated to memory
Memory used to be managed by a dedicated chip -- the northbridge. But AMD moved it into the CPU because it was faster that way.
or interrupt handling
The APIC? But anyway, the slow part of interrupt handling is done in the OS kernel, which runs on the CPU. So I'm not sure how much a chip would help there.
Maybe someone with more background in hardware design can enlighten me why the race for more cores and more Hertz.
I'm not an expert, but my guess is that because computers are all-purpose devices. Specialized hardware can accelerate something like encryption or audio mixing, but there doesn't seem to be all that much of that sort of thing that's still worth accelerating. Most people don't need to encrypt the huge amounts of data that would make a dedicated accelerator make much of a difference. Notice also how now almost nobody buys sound cards anymore, because you can just mix sound in hardware.
I think your sort of crash is an intentional abort.
There's a BIG difference between an application writing junk to memory and crashing somewhere in malloc because things are completely hosed, and an application deciding data makes no sense and orderly aborting the operation.
Your program seems to do the second one, which is good. It's perfectly appropiate for the program to quit if it's for example a commandline program for batch processing. Were it a GUI program you'd stop processing, produce an error, and let the user retry or whatever. You could do that because your program is still working and in a sane state.
The Word crash isn't that, the application fails completely with an access violation. If you were running it under an IDE you'd be looking at the debugger and a stack trace. MS Word couldn't continue in this situation because at this point the state is corrupt -- there's no hope of recovering from it.
That won't help, in the end it's still the software what tells the USB key to unlock itself, and if you lose it you're screwed in any case, as the attacker's computer is under their full control. Any and all obfuscation can be broken given enough time.
There's only one good way of making this safe: Encrypt the data on the drive. But that makes the "protected memory stick" idea really pointless. AFAIK, there's no standard way of sending a passsword to an USB key. Since you're going to need special drivers for the thing anyway, it means that doing encryption in hardware inside the key would be pointless as doing it in software would be just as good.
So just use TrueCrypt and be done with it. Much safer, and a lot cheaper.
I tried to use that as a base for a silent box some years ago. This didn't go well.
I got the "Nehemiah" version, which uses a 60mm fan. I also got a case with a fanless supply (brick + electronics in case) to try to make it as silent as possible.
60mm fans are noisy and hard to silence. The case was too thin and cramped to install anything larger in it and had two 60mm fans as well.
The board itself was crap. Performance was very unimpressive (but that could be expected). It turned out my board had only the RNG and no encryption acceleration (so watch for that if you want it). Onboard LAN failed to work properly, corrupting packets only when uploading. It managed to download a whole gentoo install, but ssh connections would drop every few minutes with an error from ssh (forgot which). It also had stability issues and would regularly lock up.
Board + hard disk + CD drive made the case seriously cramped, and very hot, so I wouldn't risk running it without fans.
After some months, the power supply in the case failed (bad capacitors). I'm pretty sure that wasn't the source of the problems with the board though, as I tried a different supply before with the same results.
So, based on my experience, some recommendations:
Stay away from VIA if possible. If you go with them, research it really well. I've been burned by VIA stuff several times.
Don't get one of those tiny cases. They're cute, but have no airflow. Get a case where you can stick a big, silent fan if needed.
Don't get a case with a strange power supply, it'll be a pain to fix.
SL is full of people with a little business of their own. There are many money handling scripts, and objects being sold for money. There's the CopyBot, but it's unable to copy objects fully. A rogue sim would probably be perfectly capable of making a 100% perfect copy of anything, with scripts and all (which the CopyBot can't do, because scripts aren't available to clients that have no permissions to them). Additionally it would have access to scripts (since it's what runs them), and thus be able to extract information like passwords stored inside.
So, allowing that would effectively destroy nearly all the ways there are in SL of doing any sort of business, and possibly even allow stealing money from random people. The outcry resulting of that would be of really gigantic proportions.
Now, if you don't care about the current SL world, just take OpenSim and run a distributed grid with that. I doubt it such a system can be made safe, but it should be technically possible, if you remove money, content ownership and such.
There has been a discussion of texture caching on the development list some days ago, and perhaps the result of that will be that the client will be improved to make it load the world faster.
I haven't really looked at the rendering code yet, but I know what you mean. Some data must be there, so probably something can be done about it.
SL doesn't transmit polygons to the client, it transmits "primitives". So the viewer knows that there's a sphere at position x,y,z with radius r, texture t, and so on. The amount of polygons you see depends on the detail setting.
I don't think SL is ever going to luck really stunning though. While rendering can be improved, the content is still generated by normal people, and not by a company who hires artists and has a plan of how things should look like. You can find professionally looking areas (especially ones made for big companies and such), but the grid as a whole probably will always look unimpressive.
Also, the amount of primitives that can be used is limited, so it's not really possible to create all the details you'd like.
As an example, I know that the furries exist, but on the regular Internet, I can avoid them easily just by never searching for them. In Second Life, though, you can't just avoid their island or whatever they have. They are, well, everywhere.
You can't really avoid anybody on the Internet, look at my website;-)
But, furry themed areas in SL aren't that many, and quite easy to avoid. It's not like they're going to bite you if you happen to pass by, anyway.
Don't want to see X-rate content? Too bad.
Nonsense. Look at the menu bar. Right near the region's name it says "(PG)" or "(Mature)". If you don't like the porn, then just stick to PG areas. X-rated content in a PG area is a bannable offense.
Flying penises are the SL equivalent of the well known.cx site. You can't entirely avoid that sort of thing anywhere, but it's not an all that frequent occurrence, and usually dealt with quickly.
It's pretty much a copy of a small part of the build instructions from the wiki. Absolutely nothing new, and even not enough information to actually build it.
It's also out of date. The latest versions do build with GCC 4, although it doesn't seem to be fully supported yet. Once in a while they release source with a couple of lines that GCC chokes on (such as using "class::method" in headers), but compiling with GCC 4 doesn't require any changes besides fixing that.
OpenJPEG recently became very usable as well, thanks to some good work on optimizing it.
SL currently sells land that maps to server resources. SL regions have a 256x256 virtual meters size. This is the size that's simulated as a whole, that is, when something goes wrong with the simulator process, that's the size of the chunk of the world that goes down. Regions are located on servers, with a region per CPU, so a 4 core box would host 4 256x256 regions.
Land costs what it does because it maps directly to the underlying hardware. When paying for a piece of virtual land you're paying for your share of CPU usage and a number of objects you can have (15000 per region, you get a limit that depends on how much land you got).
Of course it doesn't have to be this way. They could have made it differently, by letting you have as much land as you want and charging for resource usage. So instead of paying for 1024m^2 of land, you'd be able to have a desert as large as you'd like for very cheap, and then pay extra for the objects and scripts you put in it.
That does present a technical challenge though. In that situation, it should be possible to have an area with a small size but immense processing requirements. One problem is that load moves around as people move. So the SL grid would need to be able to somehow allocate more processing to busy areas, and dedicate less to idle ones. I suppose that such a thing must be technically doable, but it doesn't sound easy.
The problem is that LL went with the easy way of dividing the world in fixed chunks and letting people pay for a piece of a server. That might have worked in the beginning, but with the current situation more and more people often want to gather in the same place. So some regions stay idle, and some have a crushing load. And they can't really power down idle regions because there are scripts running on them, some of which have to run even when nobody is around. They also take a considerable time to start, which can be seen when a region restarts.
This sounds like something technically difficult to fix. A radical redesign of the underlying architecture on a system used by thousands of people can't be easy, and it's bound to mean changes that will cause large amounts of drama.
Imagine you could teleport instantly anywhere. No downsides. You could live in a cabin in the mountains, and instantly appear on a chair in the office. But even then I don't think people would just go live in a random place. If there's nothing but strangers around, you can't look of the window and say "Howdy, neighbour!". Teleporting would still take a conscious action, and suddenly appearing at somebody's house would be a disruptive way of trying to start a conversation (that's something you can do in SL, and which many people hate). Simply seeing somebody nearby is a perfect excuse to start talking, without being disruptive.
For example, people in SL often try to buy land next to their friends' land. Sure you can teleport, but teleporting isn't completely instant, and teleporting has to be an intentional action. If you have your friends near your location then you don't need to specifically visit them, you'll see them when they're there.
This is why despite teleporting SL still has large stores with products from many vendors. When you're looking for something (or nothing particular at all) it's easier to just walk around one large shop than to teleport from one to another.
I don't see what's strange about that, lots of people pay for things that don't exist physically. Insurance, for one. SL land is just like paying for space on a webserver.
Well, to each his own I guess, but I always thought that games like WoW are a pointless waste of time. Great, I get to grind for months and in the end have not gained anything of actual value for all that effort. In SL you can at least make some cash if you try a bit. In WoW you'll never be anything more than a little cog in a giant army. Even at level 60 you're just another level 60 character. On the other hand, in SL you can actually be rich and famous, and become so through your actual skill, and not just by being patient enough to keep killing rats.
Some people tell me that they play WoW as a way of socializing. In that case, SL shouldn't be hard to understand -- just like WoW, but graphics aren't as good, there's no game, you can make your own stuff, and it's not nearly as restricted in what you can do. If you want to look like a cleric, or a big demon that can be had in SL without having to grind to get good looking equipment. After playing WoW for years you'll still have nothing to show for it, while in SL you might actually acquire some valuable skills.
Slashdot Mirror
So what do you call the "free as in freedom" software in English, then?
Languages like Spanish and Russian have different words for "freedom" and "it doesn't cost money", but English seems to be lacking in that respect.
That reminds me, why does MS DHCP have such a weird format for MAC addresses? It doesn't match anything at all! Neither the Windows nor the Linux tools output MAC addresses like that. Although in Linux at least I could use sed to convert it.
I dread working with MS GUIs. Everything has 20 tabs, sometimes pops up extra windows and requires half an hour of clicking. It's not possible to figure out what the full current config is without clicking everywhere. Then there's my favourite annoyance: Fixed size windows with list views that sometimes require scrolling to see all the information in them.
Take for instance setting up a DHCP server.
Task: Switch 100 hosts from static IP address to DHCP, configured to give them the same address they currently have, by MAC.
Windows: Hand falls off due to clicking for an hour, and mindlessly copy/pasting. Error prone. MS DHCP is completely braindead by the way, and seems to confuse the description and MAC fields (did anybody even test that?)
Linux: Maybe 5 minutes to write a script that: pings subnet, parses output of 'arp -a', and automatically generates a dhcpd config file. If the hosts were Linux ones I could also automatically replace the network settings on all of them. If it goes wrong, fix script then run again. Not all hosts are on? No problem, I can tweak the script to run periodically, detecting new computers and adding the config for them.
Yes, GUIs are easier for tiny tasks, but once you need something uncommon they're a huge pain in the ass. And it's not like there can't be a GUI that generates dhcpd.conf either.
One word: SELinux
It's not new either. And it does what you want it to do. However, it's a royal pain in the ass to configure, because you need to figure out what every application should be able to do. It's definitely not something for a newbie, and probably it will be long before such a thing is usable by normal people.
Also, I doubt it'll work well for Windows. For Linux sure, distributions would just have to provide the SELinux security settings for the packages. But for Windows? Who provides the list of things the application should be able to do? It can't be the author, as all the malware would just ship rules allowing them to mess with whatever they need.
Actually, this failure mode exists almost exactly on modern operating systems.
Unless you put limits to it, any application can get all your system RAM. The OS and applications get their memory from the same place, so it's still possible to run out of it. It's perfectly possible to use so much that the computer grinds to a halt and not even the mouse pointer moves.
Now, the memory space separation makes sure your application can't hose the OS' or some other program's state, but that doesn't mean it can't do something like slowing things down so much it becomes unusable.
This sort of failure mode is easy to see in these conditions: low amounts of free RAM, large amounts of available swap space, and processes that quickly allocate large amounts of RAM and actually use it. A good example program is a compiler. On some files, gcc can use 300MB RAM. Swap is useful for offloading unused stuff to disk to make room for something more useful, but if you reach the point where there really isn't enough RAM and the system needs to continuously swap stuff out and in, the system can grind to a halt.
You do it the same way SSL certificates work.
The traffic advisory organization runs a Certificate Authority. They have their private key, which they keep well safe. Subdivisions that need to broadcast messages get their keys signed by the CA, then use them to sign messages. GPS device checks that the message is signed by a key signed by the CA.
If a key is compromised, the CA adds it to the revocation list, which for instance could be regularly broadcasted. After hearing the broadcast, GPS devices refuse to accept messages signed with that key, and whoever used the compromised key has to get a new one.
atop seems to have support for disk monitoring, but it requires a kernel patch:
http://www.atconsultancy.nl/atop/kernpatch.html
Wasn't one of the concepts in sword forging that you want to have a lot of carbon on the edge, and less of it on the back (or middle) to get a sword that's sharp but bends so that it doesn't break?
A sword made fully of diamond would be stupid as the whole sword would shatter, but one with a diamond edge should cut really good (and chip very badly afterwards).
Here's a solution for the author:
In the Fullmetal Alchemist anime, alchemy is a sort of the science of transformation of matter. An alchemist can reshape matter, but not create it. Still, an alchemist could easily and quickly make a sword out of anything that happened to be lying around. So that's a solution to the breaking sword problem, when it breaks, just reform it. Carbon is so abundant that it could be made pretty much anywhere.
Of course such a thing would be still impractical as you couldn't parry somebody else's sword without your breaking. My solution to that: Diamond throwing knives or stars. If it breaks, even better.
Sound quality, maybe. CPU load, I'm not that sure.
Some time ago, Creative intimidated John Carmack into supporting EAX in Doom 3. This is yet another reason why I don't use Creative hardware anymore. The SB Live cards causing disk corruption with VIA boards, drivers being unstable on SMP (this was before dual core) and them taking drivers off their site for some time are some others.
But anyway, why do you think Creative had to basically force John Carmack to support their EAX tech? Apparently because Doom 3, on modern hardware, can mix 5.1 audio faster than the sound card.
Seriously, mixing sounds isn't that big of a deal. I have here a book on game programming that describes the creation of a wolfenstein-3D-like game, and part of it explains how to mix sound in software. This was on a 386 CPU. Besides that, all a sound card gives you is extra reverb/etc effects, and computers don't seem to have much of a problem with doing that in real time anymore either.
AMD64 has 16 registers
AMD seems to be working on putting a GPU in ther CPU
Memory used to be managed by a dedicated chip -- the northbridge. But AMD moved it into the CPU because it was faster that way.
The APIC? But anyway, the slow part of interrupt handling is done in the OS kernel, which runs on the CPU. So I'm not sure how much a chip would help there.
I'm not an expert, but my guess is that because computers are all-purpose devices. Specialized hardware can accelerate something like encryption or audio mixing, but there doesn't seem to be all that much of that sort of thing that's still worth accelerating. Most people don't need to encrypt the huge amounts of data that would make a dedicated accelerator make much of a difference. Notice also how now almost nobody buys sound cards anymore, because you can just mix sound in hardware.
I think your sort of crash is an intentional abort.
There's a BIG difference between an application writing junk to memory and crashing somewhere in malloc because things are completely hosed, and an application deciding data makes no sense and orderly aborting the operation.
Your program seems to do the second one, which is good. It's perfectly appropiate for the program to quit if it's for example a commandline program for batch processing. Were it a GUI program you'd stop processing, produce an error, and let the user retry or whatever. You could do that because your program is still working and in a sane state.
The Word crash isn't that, the application fails completely with an access violation. If you were running it under an IDE you'd be looking at the debugger and a stack trace. MS Word couldn't continue in this situation because at this point the state is corrupt -- there's no hope of recovering from it.
That won't help, in the end it's still the software what tells the USB key to unlock itself, and if you lose it you're screwed in any case, as the attacker's computer is under their full control. Any and all obfuscation can be broken given enough time.
There's only one good way of making this safe: Encrypt the data on the drive. But that makes the "protected memory stick" idea really pointless. AFAIK, there's no standard way of sending a passsword to an USB key. Since you're going to need special drivers for the thing anyway, it means that doing encryption in hardware inside the key would be pointless as doing it in software would be just as good.
So just use TrueCrypt and be done with it. Much safer, and a lot cheaper.
You can't be sued for libel if it's true.
So don't post the text itself, post links to an archive, preferrably controlled by some completely independent party (say, Google for newsgroups).
I got the "Nehemiah" version, which uses a 60mm fan. I also got a case with a fanless supply (brick + electronics in case) to try to make it as silent as possible.
So, based on my experience, some recommendations:
It's got nothing to do with LL's business model.
SL is full of people with a little business of their own. There are many money handling scripts, and objects being sold for money. There's the CopyBot, but it's unable to copy objects fully. A rogue sim would probably be perfectly capable of making a 100% perfect copy of anything, with scripts and all (which the CopyBot can't do, because scripts aren't available to clients that have no permissions to them). Additionally it would have access to scripts (since it's what runs them), and thus be able to extract information like passwords stored inside.
So, allowing that would effectively destroy nearly all the ways there are in SL of doing any sort of business, and possibly even allow stealing money from random people. The outcry resulting of that would be of really gigantic proportions.
Now, if you don't care about the current SL world, just take OpenSim and run a distributed grid with that. I doubt it such a system can be made safe, but it should be technically possible, if you remove money, content ownership and such.
There has been a discussion of texture caching on the development list some days ago, and perhaps the result of that will be that the client will be improved to make it load the world faster.
I haven't really looked at the rendering code yet, but I know what you mean. Some data must be there, so probably something can be done about it.
People are already working on the OpenSim project, which is exactly that.
But, I doubt it'll ever connect into the main grid. There are huge security issues there.
SL doesn't transmit polygons to the client, it transmits "primitives". So the viewer knows that there's a sphere at position x,y,z with radius r, texture t, and so on. The amount of polygons you see depends on the detail setting.
I don't think SL is ever going to luck really stunning though. While rendering can be improved, the content is still generated by normal people, and not by a company who hires artists and has a plan of how things should look like. You can find professionally looking areas (especially ones made for big companies and such), but the grid as a whole probably will always look unimpressive.
Also, the amount of primitives that can be used is limited, so it's not really possible to create all the details you'd like.
You can't really avoid anybody on the Internet, look at my website
But, furry themed areas in SL aren't that many, and quite easy to avoid. It's not like they're going to bite you if you happen to pass by, anyway.
Nonsense. Look at the menu bar. Right near the region's name it says "(PG)" or "(Mature)". If you don't like the porn, then just stick to PG areas. X-rated content in a PG area is a bannable offense.
Flying penises are the SL equivalent of the well known
It's pretty much a copy of a small part of the build instructions from the wiki. Absolutely nothing new, and even not enough information to actually build it.
It's also out of date. The latest versions do build with GCC 4, although it doesn't seem to be fully supported yet. Once in a while they release source with a couple of lines that GCC chokes on (such as using "class::method" in headers), but compiling with GCC 4 doesn't require any changes besides fixing that.
OpenJPEG recently became very usable as well, thanks to some good work on optimizing it.
SL currently sells land that maps to server resources. SL regions have a 256x256 virtual meters size. This is the size that's simulated as a whole, that is, when something goes wrong with the simulator process, that's the size of the chunk of the world that goes down. Regions are located on servers, with a region per CPU, so a 4 core box would host 4 256x256 regions.
Land costs what it does because it maps directly to the underlying hardware. When paying for a piece of virtual land you're paying for your share of CPU usage and a number of objects you can have (15000 per region, you get a limit that depends on how much land you got).
Of course it doesn't have to be this way. They could have made it differently, by letting you have as much land as you want and charging for resource usage. So instead of paying for 1024m^2 of land, you'd be able to have a desert as large as you'd like for very cheap, and then pay extra for the objects and scripts you put in it.
That does present a technical challenge though. In that situation, it should be possible to have an area with a small size but immense processing requirements. One problem is that load moves around as people move. So the SL grid would need to be able to somehow allocate more processing to busy areas, and dedicate less to idle ones. I suppose that such a thing must be technically doable, but it doesn't sound easy.
The problem is that LL went with the easy way of dividing the world in fixed chunks and letting people pay for a piece of a server. That might have worked in the beginning, but with the current situation more and more people often want to gather in the same place. So some regions stay idle, and some have a crushing load. And they can't really power down idle regions because there are scripts running on them, some of which have to run even when nobody is around. They also take a considerable time to start, which can be seen when a region restarts.
This sounds like something technically difficult to fix. A radical redesign of the underlying architecture on a system used by thousands of people can't be easy, and it's bound to mean changes that will cause large amounts of drama.
Well, it's something that will probably stay.
Imagine you could teleport instantly anywhere. No downsides. You could live in a cabin in the mountains, and instantly appear on a chair in the office. But even then I don't think people would just go live in a random place. If there's nothing but strangers around, you can't look of the window and say "Howdy, neighbour!". Teleporting would still take a conscious action, and suddenly appearing at somebody's house would be a disruptive way of trying to start a conversation (that's something you can do in SL, and which many people hate). Simply seeing somebody nearby is a perfect excuse to start talking, without being disruptive.
Convenience.
For example, people in SL often try to buy land next to their friends' land. Sure you can teleport, but teleporting isn't completely instant, and teleporting has to be an intentional action. If you have your friends near your location then you don't need to specifically visit them, you'll see them when they're there.
This is why despite teleporting SL still has large stores with products from many vendors. When you're looking for something (or nothing particular at all) it's easier to just walk around one large shop than to teleport from one to another.
I don't see what's strange about that, lots of people pay for things that don't exist physically. Insurance, for one. SL land is just like paying for space on a webserver.
Well, to each his own I guess, but I always thought that games like WoW are a pointless waste of time. Great, I get to grind for months and in the end have not gained anything of actual value for all that effort. In SL you can at least make some cash if you try a bit. In WoW you'll never be anything more than a little cog in a giant army. Even at level 60 you're just another level 60 character. On the other hand, in SL you can actually be rich and famous, and become so through your actual skill, and not just by being patient enough to keep killing rats.
Some people tell me that they play WoW as a way of socializing. In that case, SL shouldn't be hard to understand -- just like WoW, but graphics aren't as good, there's no game, you can make your own stuff, and it's not nearly as restricted in what you can do. If you want to look like a cleric, or a big demon that can be had in SL without having to grind to get good looking equipment. After playing WoW for years you'll still have nothing to show for it, while in SL you might actually acquire some valuable skills.