I'm not using Pro hardware, so I wouldn't know. Problems with NVIDIA's discrete GPUs are completely unrelated to discussion of their chipsets and embedded GPUs.
I don't see your point. There's nothing preventing incoming calls on Skype. In fact, if you do it right, you can even have it forward your home phone calls so you only need one number where people can reach you no matter where you are. It's a lot better than having to have a separate cell phone number.
The only benefit to having a cell phone at this point is the lack of ubiquitous Wi-Fi and the lack of the ability to rapidly roam from one Wi-Fi AP to another. Both can be fairly easily solved at this point, at least within major cities.
Why keep voice? If I were in Mountain View most of the time, I'd be tempted to ditch my iPhone for an iPod Touch running Skype. Why carry a pay-per-minute cell phone at all if there's free ubiquitous Wi-Fi. The last thing AT&T is likely to do would be to add caps to iPhone data usage. As soon as they drive people to Wi-Fi, they run too great a risk of losing all their voice traffic, too.
Funny, I've been using Macs with NVIDIA chipsets for a while and haven't noticed any of those problems. Maybe it's not the chipset so much as poor BIOS and Windows support for ACPI interrupt steering, poor chipset drivers for Windows, poor Windows drivers that spend way too much time in interrupt handlers... hmm... I think I see a common theme here... Windows....:-)
Or the 50 other popups that say "Your computer is broadcasting an IP address" that everyone ignores because the supposed "virus scanners" install malware?
Normal users also don't buy routers specifically based on open source firmware so that they can tweak their firmware in strange and unusual ways. As soon as you're buying something like these, you're at least two standard deviations from the norm.:-) For normal users, the odds of bricking the router are probably very small, so the manufacturer can absorb it without trouble. For devices that they know people are going to mess around with, it makes sense to make it as easy as possible to recover when you screw it up because people *will* be screwing them up.:-D
Having to throw away expensive hardware because of a failed firmware update is asinine even if you have a thousand spares. It's not about downtime. It's about money.
Spending an extra $2 on hardware that can safely be flashed is more than worth it when the alternative is having a significant chance of having to spend several times that much to replace the hardware outright when flashing fails.
And lest you say that the statistical odds of the failure are so small that you'd likely be better off saving the $2, I would counter by noting that the cost of in-warranty replacements for failed flashes ends up built into the cost of the product, and that you are already probably paying more for everyone else's flashing failures than it would cost to make it safely flashable in the first place.
Setting up TFTP can still be a pain, but I will agree that it's better than nothing. Still nowhere near as convenient as being able to "tar -xzf" a firmware image onto a flash card and stick it in a slot, though.
The problem is that the firmware (which as you correctly point out is really just an OS) is also responsible for providing the (generally web-based) mechanism for updating the OS. So if you hose it, there's no way to get back into a state where you can upload a new OS.
It's like reinstalling a Linux distribution remotely using SSH on a computer that has no optical drive. Can you brick it? Easily. Will you? Probably.:-) It's even worse than that, though. With a desktop Linux box, you can always add an optical drive externally and boot from that. Booting from an external device typically isn't an option with a headless, keyboardless router box.
RMS correctly points out that the war was won long ago
...
If I may state the obvious, RMS is a hardliner with zero tolerance or forgiveness.
I'd go one step further. RMS is a rigid hardliner who would rather have closed source than half-open source. He claims to be anti-closed source, but his actions end up frequently being contrary to his goals by dividing the community needlessly, weakening us as a group and allowing closed source to regain lost territory.
Case in point, GPLv3 is so extreme that corporate support and *BSD support are basically gone. In five years, GCC will no longer be relevant. Many corporations that have been contributing significantly to its development are now contributing resources to other, non-GPLv3 compiler technologies like llvm/clang---resources that otherwise would likely have been spent on GCC. In a perverse way, the open source community is going to end up with better compiler technology because of GPLv3, but it won't be free software like Stallman wants. I'd be surprised if Samba didn't suffer the same fate for the same reason.
Instead of browbeating people whose license views differ and continuing to tighten the screws, Stallman should be trying to push things like end user testing, unit tests, security analysis, and other things that actually improve the quality of open source and free software. The only way open source/free/libre software will ever supplant closed source software is if it is consistently better, easier to use, and still provides the ability to adapt it. We're a long way from there, but Stallman acts as if he honestly believes that free software has won, and in so doing, ensures that the war has already been won and subsequently lost again.
I've been using them for my personal site for several months. Once you figure out how to get it set up correctly, it works just fine with Safari and FireFox. And, of course, the number of MSIE users on my personal site is so close to zero that it amounts to a rounding error.:-D
I think this brings up a crucial point. We should not be relying on any equipment in any serious network infrastructure that can be bricked by even the most colossally broken firmware update. A proper router would either:
have two sets of firmware with a physical button to force booting from the backup firmware to allow reflashing, or
have a flash card slot and a user-removable flash card for the firmware image.
The latter would be far preferable, as it would make the amount of soldered-in flash a moot point. Instead of sticking in flash chips, stick in an SD or CF card reader and a low end flash card that's just big enough to hold the stock firmware. Want to use firmware that's bigger? Copy it onto a bigger flash card and swap them out. Doesn't work? Swap back to the previous flash card.
The idea of firmware flashers makes sense for a device that is not critical and is not updated often, nor typically updated with custom firmware. Network infrastructure fails all of those tests, however.
This I can agree with, but lets not forget the investors of blu ray and DVD couldn't know before hand just how far hard drive would come. Who imagined that you would be able to buy a 2 terabytes for $200 when you were gettting between 4-20 GB in the late 90's.
Hard drive storage has seen a fairly consistent rate of growth since the mid 1990s, with areal density doubling roughly every 18 months. If they couldn't imagine that the trend would continue for another few years after it had done so for essentially a decade, they had no business being in charge of a standard, as that shows a rather remarkable lack of foresight.:-)
I can excuse it for the DVD standard; computer video was in its infancy when the DVD format came out. By the time Blu-Ray started to happen, though, it should have been obvious that this would be a problem.
That's a good point about archiving junk to DVD, though I would argue that if it isn't something you'll ever use again, that's what the trash can is for.:-)
For some types of surgery, a double blind study isn't even possible. If someone is bleeding to death, there's no way the patient or the person doing the study could not know which patient got the surgery to stop the bleeding. There's also no possible way any sane person could question whether such surgery is more effective than a placebo.
Likewise, like the GP was saying, for late stage renal failure, a double blind study is pointless. People in that state who don't get surgery die, period, with essentially 100% certainty. Generally speaking, kidneys don't get better. Therefore, the evidence for the effectiveness of that treatment is so overwhelming that you'd have to be completely clueless to question it. That's what survey studies are for---studies in which a double blind study would be unethical. Given a broad enough survey study, it's essentially as good as a double blind study.
I agree with what I think you are trying to say---that we are often far too quick to perform surgical intervention in cases where the benefits are dubious. I don't think that it's reasonable to paint surgery as generally useless, though.
The biggest problem with mass transit in the Bay Area, IMHO, is that we have a half dozen or more mass transit systems, all run by different people who don't talk to each other. The systems don't always connect, and when they do, they often connect in poorly chosen locations as you've noted.
The schedules also aren't intelligently chosen to intersect. I wrote a letter complaining about this to the VTA folks about the light rail connection in Mountain View departing once an hour, but departing ten minutes after Caltrain southbound got there and almost an hour after the previous one got there. They said they would fix the schedule, but by the time they did, it was two months later and too late for it to do any good for our conference.
And then, there was the BART/Caltrain nightmare. I took that once on a weekend. BART was running about three minutes late and the Caltrain engineer apparently decided that being on schedule was more important than the 50+ passengers who were running across the bridge trying to get down there. So there were 50+ people standing around on the platform waiting for an entire HOUR for the next Caltrain. I don't think I'll EVER take Caltrain and BART again on a weekend.
I was so pissed off that I called Caltrain while standing on the platform and chewed them out. Their exact response was "We're not responsible for synchronizing our schedule with BART's schedule." I replied, "Well, if you aren't, who is?" The customer support rep didn't have an answer for that. And therein is the real cause of our mass transit problem. Every agency blames the problems on every other agency, so nothing ever gets fixed in a timely fashion.
What we need is a single Bay Area Mass Transit Authority responsible for managing all of the transit agencies from north of Oakland all the way down to and including Santa Cruz Metro, VTA (as far south as Gilroy), etc. It's completely clear that having multiple organizations running things has been a complete disaster. The only way things will ever get better is if there is someone in charge of all of it.
Oh, yeah, and a better, more standardized payment system. I should be able to go to the VTA light rail station and buy my VTA single ride pass, a Caltrain single ride pass, and a BART card with enough money on it for the last leg of the trip, all from a single vending machine. It's insane that you have to leave enough layover to stand in a long line and wait to buy a ticket. Can you imagine if airline flights worked that way? If everybody had to leave security, go out to the lobby, wait in line, buy the second leg ticket, go back through security, and fly? Our airline system would be gridlocked and nobody would fly.
I should also be able to buy a non-dated VTA single ride pass for the return trip that I can shove into a validator and get stamped in half a second. Ditto for the Caltrain single ride pass. There's no excuse for their validators only doing multi-ride passes. That's brain damaged. Using validators everywhere would not only make travel much more convenient, but would also eliminate a whole class of excuses commonly made by fare dodgers.
The only use I can see for Twitter is creating a fake public persona. For example, I think today my public persona is going to be on the beach in Maui designing a doomsday weapon to help it take over the world, all while drinking Mai Tais and petting the kangaroos. There are no kangaroos in Maui, you say? Funny, my Twitter persona would beg to differ.
Gasoline-generated electricity is some of the most expensive power out there---far more expensive than nuclear, hydro, coal, etc. If at full retail gasoline prices, I would be almost break even using a cheap gasoline generator---and that's roughly what the numbers seem to show---then I'm paying an obscene rate for my power.... Almost 38 cents per kWh for a significant portion of my bill.
I posted a similar comment about a year ago. Optical media should be a great backup medium, but because they take so long to ramp up production and push the cost of the media down, it is useless before anyone can afford it. Blu-ray media at 50 GB per disc is already useless and it still isn't even close to price parity with hard drives. To fully back up a 500 GB hard drive (the industry average size now) takes 10 discs to back up once. At 30 minutes per disc, this is five hours of continuous burning, during which time you have to have someone swapping out discs every half hour. For a terabyte HD, you're more than an entire work day. You should be doing a full backup at least every month and incremental backups weekly. Do the math, and you're spending the better part of a week every month just doing backups. The average hard drive needs to be able to be backed up on a single disc or you've already failed. Blu-ray has already failed.
As a result, recordable optical media is basically worthless except for people burning content to give to other people, which is a tiny fraction of its potential user base. If they would ramp production way up and flood the market with cheap media immediately even before the recorders are available in quantities, people would flock to them in droves. It's counterintuitive, but the only way any optical format will ever be particularly useful to the general consumer is if the industry decides to make it a loss leader for about a year. By the end of that year, you'll have so much adoption that it won't be losing money anymore, and it will be in the hands of consumers early enough to be broadly useful.
Slashdot Mirror
I'm not using Pro hardware, so I wouldn't know. Problems with NVIDIA's discrete GPUs are completely unrelated to discussion of their chipsets and embedded GPUs.
I don't see your point. There's nothing preventing incoming calls on Skype. In fact, if you do it right, you can even have it forward your home phone calls so you only need one number where people can reach you no matter where you are. It's a lot better than having to have a separate cell phone number.
The only benefit to having a cell phone at this point is the lack of ubiquitous Wi-Fi and the lack of the ability to rapidly roam from one Wi-Fi AP to another. Both can be fairly easily solved at this point, at least within major cities.
Why keep voice? If I were in Mountain View most of the time, I'd be tempted to ditch my iPhone for an iPod Touch running Skype. Why carry a pay-per-minute cell phone at all if there's free ubiquitous Wi-Fi. The last thing AT&T is likely to do would be to add caps to iPhone data usage. As soon as they drive people to Wi-Fi, they run too great a risk of losing all their voice traffic, too.
Funny, I've been using Macs with NVIDIA chipsets for a while and haven't noticed any of those problems. Maybe it's not the chipset so much as poor BIOS and Windows support for ACPI interrupt steering, poor chipset drivers for Windows, poor Windows drivers that spend way too much time in interrupt handlers... hmm... I think I see a common theme here... Windows.... :-)
<sarcasm>Ooh! I've got it! We'll use wormhole routing!</sarcasm>
Ooh, ooh.... Can it also say, "Goodbye, cruel world"?
Not sure about age or sex, but its location was along highway 1 somewhere south of Monterey, CA.
Or the 50 other popups that say "Your computer is broadcasting an IP address" that everyone ignores because the supposed "virus scanners" install malware?
My avatar is a grey squirrel, you insensitive clod.
Normal users also don't buy routers specifically based on open source firmware so that they can tweak their firmware in strange and unusual ways. As soon as you're buying something like these, you're at least two standard deviations from the norm. :-) For normal users, the odds of bricking the router are probably very small, so the manufacturer can absorb it without trouble. For devices that they know people are going to mess around with, it makes sense to make it as easy as possible to recover when you screw it up because people *will* be screwing them up. :-D
Just anyone who has ever logged in from a Windows box running a browser other than Firefox.
Having to throw away expensive hardware because of a failed firmware update is asinine even if you have a thousand spares. It's not about downtime. It's about money.
Spending an extra $2 on hardware that can safely be flashed is more than worth it when the alternative is having a significant chance of having to spend several times that much to replace the hardware outright when flashing fails.
And lest you say that the statistical odds of the failure are so small that you'd likely be better off saving the $2, I would counter by noting that the cost of in-warranty replacements for failed flashes ends up built into the cost of the product, and that you are already probably paying more for everyone else's flashing failures than it would cost to make it safely flashable in the first place.
Setting up TFTP can still be a pain, but I will agree that it's better than nothing. Still nowhere near as convenient as being able to "tar -xzf" a firmware image onto a flash card and stick it in a slot, though.
The problem is that the firmware (which as you correctly point out is really just an OS) is also responsible for providing the (generally web-based) mechanism for updating the OS. So if you hose it, there's no way to get back into a state where you can upload a new OS.
It's like reinstalling a Linux distribution remotely using SSH on a computer that has no optical drive. Can you brick it? Easily. Will you? Probably. :-) It's even worse than that, though. With a desktop Linux box, you can always add an optical drive externally and boot from that. Booting from an external device typically isn't an option with a headless, keyboardless router box.
I'd go one step further. RMS is a rigid hardliner who would rather have closed source than half-open source. He claims to be anti-closed source, but his actions end up frequently being contrary to his goals by dividing the community needlessly, weakening us as a group and allowing closed source to regain lost territory.
Case in point, GPLv3 is so extreme that corporate support and *BSD support are basically gone. In five years, GCC will no longer be relevant. Many corporations that have been contributing significantly to its development are now contributing resources to other, non-GPLv3 compiler technologies like llvm/clang---resources that otherwise would likely have been spent on GCC. In a perverse way, the open source community is going to end up with better compiler technology because of GPLv3, but it won't be free software like Stallman wants. I'd be surprised if Samba didn't suffer the same fate for the same reason.
Instead of browbeating people whose license views differ and continuing to tighten the screws, Stallman should be trying to push things like end user testing, unit tests, security analysis, and other things that actually improve the quality of open source and free software. The only way open source/free/libre software will ever supplant closed source software is if it is consistently better, easier to use, and still provides the ability to adapt it. We're a long way from there, but Stallman acts as if he honestly believes that free software has won, and in so doing, ensures that the war has already been won and subsequently lost again.
And only a hardline Stallmanite would disagree with the statement.
I've been using them for my personal site for several months. Once you figure out how to get it set up correctly, it works just fine with Safari and FireFox. And, of course, the number of MSIE users on my personal site is so close to zero that it amounts to a rounding error. :-D
I think this brings up a crucial point. We should not be relying on any equipment in any serious network infrastructure that can be bricked by even the most colossally broken firmware update. A proper router would either:
The latter would be far preferable, as it would make the amount of soldered-in flash a moot point. Instead of sticking in flash chips, stick in an SD or CF card reader and a low end flash card that's just big enough to hold the stock firmware. Want to use firmware that's bigger? Copy it onto a bigger flash card and swap them out. Doesn't work? Swap back to the previous flash card.
The idea of firmware flashers makes sense for a device that is not critical and is not updated often, nor typically updated with custom firmware. Network infrastructure fails all of those tests, however.
Hard drive storage has seen a fairly consistent rate of growth since the mid 1990s, with areal density doubling roughly every 18 months. If they couldn't imagine that the trend would continue for another few years after it had done so for essentially a decade, they had no business being in charge of a standard, as that shows a rather remarkable lack of foresight. :-)
I can excuse it for the DVD standard; computer video was in its infancy when the DVD format came out. By the time Blu-Ray started to happen, though, it should have been obvious that this would be a problem.
That's a good point about archiving junk to DVD, though I would argue that if it isn't something you'll ever use again, that's what the trash can is for. :-)
For some types of surgery, a double blind study isn't even possible. If someone is bleeding to death, there's no way the patient or the person doing the study could not know which patient got the surgery to stop the bleeding. There's also no possible way any sane person could question whether such surgery is more effective than a placebo.
Likewise, like the GP was saying, for late stage renal failure, a double blind study is pointless. People in that state who don't get surgery die, period, with essentially 100% certainty. Generally speaking, kidneys don't get better. Therefore, the evidence for the effectiveness of that treatment is so overwhelming that you'd have to be completely clueless to question it. That's what survey studies are for---studies in which a double blind study would be unethical. Given a broad enough survey study, it's essentially as good as a double blind study.
I agree with what I think you are trying to say---that we are often far too quick to perform surgical intervention in cases where the benefits are dubious. I don't think that it's reasonable to paint surgery as generally useless, though.
The biggest problem with mass transit in the Bay Area, IMHO, is that we have a half dozen or more mass transit systems, all run by different people who don't talk to each other. The systems don't always connect, and when they do, they often connect in poorly chosen locations as you've noted.
The schedules also aren't intelligently chosen to intersect. I wrote a letter complaining about this to the VTA folks about the light rail connection in Mountain View departing once an hour, but departing ten minutes after Caltrain southbound got there and almost an hour after the previous one got there. They said they would fix the schedule, but by the time they did, it was two months later and too late for it to do any good for our conference.
And then, there was the BART/Caltrain nightmare. I took that once on a weekend. BART was running about three minutes late and the Caltrain engineer apparently decided that being on schedule was more important than the 50+ passengers who were running across the bridge trying to get down there. So there were 50+ people standing around on the platform waiting for an entire HOUR for the next Caltrain. I don't think I'll EVER take Caltrain and BART again on a weekend.
I was so pissed off that I called Caltrain while standing on the platform and chewed them out. Their exact response was "We're not responsible for synchronizing our schedule with BART's schedule." I replied, "Well, if you aren't, who is?" The customer support rep didn't have an answer for that. And therein is the real cause of our mass transit problem. Every agency blames the problems on every other agency, so nothing ever gets fixed in a timely fashion.
What we need is a single Bay Area Mass Transit Authority responsible for managing all of the transit agencies from north of Oakland all the way down to and including Santa Cruz Metro, VTA (as far south as Gilroy), etc. It's completely clear that having multiple organizations running things has been a complete disaster. The only way things will ever get better is if there is someone in charge of all of it.
Oh, yeah, and a better, more standardized payment system. I should be able to go to the VTA light rail station and buy my VTA single ride pass, a Caltrain single ride pass, and a BART card with enough money on it for the last leg of the trip, all from a single vending machine. It's insane that you have to leave enough layover to stand in a long line and wait to buy a ticket. Can you imagine if airline flights worked that way? If everybody had to leave security, go out to the lobby, wait in line, buy the second leg ticket, go back through security, and fly? Our airline system would be gridlocked and nobody would fly.
I should also be able to buy a non-dated VTA single ride pass for the return trip that I can shove into a validator and get stamped in half a second. Ditto for the Caltrain single ride pass. There's no excuse for their validators only doing multi-ride passes. That's brain damaged. Using validators everywhere would not only make travel much more convenient, but would also eliminate a whole class of excuses commonly made by fare dodgers.
The only use I can see for Twitter is creating a fake public persona. For example, I think today my public persona is going to be on the beach in Maui designing a doomsday weapon to help it take over the world, all while drinking Mai Tais and petting the kangaroos. There are no kangaroos in Maui, you say? Funny, my Twitter persona would beg to differ.
Gasoline-generated electricity is some of the most expensive power out there---far more expensive than nuclear, hydro, coal, etc. If at full retail gasoline prices, I would be almost break even using a cheap gasoline generator---and that's roughly what the numbers seem to show---then I'm paying an obscene rate for my power.... Almost 38 cents per kWh for a significant portion of my bill.
My go-kart had one drive wheel. It didn't go off into a ditch every ten seconds.... Well, it did, but only because I steered it through ditches.
I posted a similar comment about a year ago. Optical media should be a great backup medium, but because they take so long to ramp up production and push the cost of the media down, it is useless before anyone can afford it. Blu-ray media at 50 GB per disc is already useless and it still isn't even close to price parity with hard drives. To fully back up a 500 GB hard drive (the industry average size now) takes 10 discs to back up once. At 30 minutes per disc, this is five hours of continuous burning, during which time you have to have someone swapping out discs every half hour. For a terabyte HD, you're more than an entire work day. You should be doing a full backup at least every month and incremental backups weekly. Do the math, and you're spending the better part of a week every month just doing backups. The average hard drive needs to be able to be backed up on a single disc or you've already failed. Blu-ray has already failed.
As a result, recordable optical media is basically worthless except for people burning content to give to other people, which is a tiny fraction of its potential user base. If they would ramp production way up and flood the market with cheap media immediately even before the recorders are available in quantities, people would flock to them in droves. It's counterintuitive, but the only way any optical format will ever be particularly useful to the general consumer is if the industry decides to make it a loss leader for about a year. By the end of that year, you'll have so much adoption that it won't be losing money anymore, and it will be in the hands of consumers early enough to be broadly useful.