I wouldn't say that what they did was "evil", but it does smack of trying for profit at the expense of their customers' safety. And public perception is important...this is why Toyota got into so much hot water over their throttle issue--they were perceived to be a very safe vehicle so it was a shock to discover that they were hiding safety issues.
What GM should have done was to change the part number and notify the owners of vehicles with the old ignition switch to come in to a dealer to get them swapped out for a good one.
Look at Tesla...very good safety record but they're raising the ride height a bit and adding a titanium shield just to improve the safety that much more.
It's that GM didn't actually admit that anything *had* changed (the part number was the same), so he had no reason to suspect that there had been a change in the ignition switch.
Even if you don't *think* there is a functional change, any change in the part should result in a different part number. The computer databases will all have part number equivalencies in them, so anyone looking for the old part number will get redirected to the new part.
In that way, if a problem shows up later (like in this case) it's much simpler to determine that a part was changed.
Companies did the same thing in other industries. One of the Linksys 10/100 ethernet NICs (lne100tx maybe) had about 4 different ethernet controllers over time but kept the same part number. They needed totally different drivers under Linux.
I have a 25Mbps down connection, and Netflix movies actually look pretty good after the initial ramp up in quality as it detects the available bandwidth. I very rarely get glitches.
The problem with negotiating the price is that it makes people feel like they might have gotten a worse deal then the next guy, and it eats at them.
We in North America generally don't negotiate prices for new items. You go to the store, you buy the item, you pay the same as the next guy. (Unless you're buying high volume, or are a contractor, or something.) Generally the only new things we negotiate on are cars. Even most new houses have a set price.
Most cars require a "major service" every couple of years, which can easily run into a couple thousand bucks
Huh? I've got a 9-year old Toyota Matrix. It gets an oil change twice a year at about $50 each time, and has had one transmission oil flush at a few hundred bucks, one radiator fluid flush, and new low beam bulbs. I just replaced the clock spring myself for a few hundred bucks. Soonish it'll need new brake pads and new spark plugs.
Once you accept the notion of patents in the first place (as a temporary monopoly on something in return for disclosing it instead of keeping it secret), it's a very small jump to buying and selling them. If an independent inventor comes up with an idea but doesn't have the capital to bring it to market, they can sell the patent to a bigger company.
This is fine in and of itself. Even patent trolls would be fine if the patents they were holding were actually valid and the companies they were suing had in fact copied the idea from the published patent.
The problems are twofold: 1) If someone else comes up with the same idea totally independently, it should invalidate the patent as too obvious. 2) Patents have been granted for things that should never have received them. In particular people have been allowed to patent *concepts* rather than *specific ways to implement a concept*.
It just means that the ISPs shouldn't be doing traffic shaping by type across multiple users.
So if you want to watch a movie on Netflix, and I want to rsync my hard drive across the country, and someone else wants to something else, our packets should be treated equally based on how much bandwidth we've paid for.
I have no problem if you as a subscriber want to assign relative priorities to your packets. But those priorities shouldn't affect how *your* packets are treated relative to *my* packets.
Does the consumer ISP need to pay to make sure that the peering relationship is such that all their users have the ability to stream from Netflix unfettered? Considering 1) above, is this fair to the ISP? They could do so, but to maintain their existing cost structure it'd likely mean that they may have a smaller pipe to another peer. Is it fair to users using those other peers or do they also have simply make sure ALL of their peers are able to fully pass 100% of traffic unfettered at peak times?
Given that ISPs are essentially regulated monopolies, I'd suggest that they have a moral obligation to ensure that they have sufficient bandwidth to sites that are popular with their customers.
If large numbers of people want to connect to Netflix, it means that those people are not connecting to other places and the ISPs don't need as much bandwidth to those places.
I don't see a problem with telecoms charging Netflix extra for a higher priority service
I do. It violates net neutrality. Your netflix stream shouldn't take priority over my rsync backup.
ISPs should be allowed do traffic shaping on a per-subscriber basis regardless of traffic type. They can even be allowed to traffic-shape different types of traffic *for a particular subscriber* if that subscriber asks them to. But they should absolutely NOT be allowed to prioritize packets belonging to different subscribers based on traffic type.
I thought of this as well, but I figured it would probably be unworkable since people's home connections don't really have reliable performance, while AWS should have pretty good control over their bandwidth.
So you're suggesting that if Netflix modified their client to send as much data back to Netflix as Netflix sends to them that the ISPs would be happier?
Somehow, I don't think so given that ISPs typically have crappy upstream bandwidth to begin with.
There's nothing stopping ISPs from doing per-subscriber traffic shaping in a traffic-agnostic way. Thus, your neighbours could all be watching netflix and it would have ZERO impact on your Vimeo stream.
Internet should be billed like my electricity bill--a certain fixed-cost amount per month to cover the costs of being a subscriber, and a reasonable cost-per-GB.
This is a closer fit to the real cost model of an ISP.
The problem is that the ISPs see metered billing as a cash cow and want to charge stupidly high amounts per GB.
Imagine if Netflix just tweaked their client software to send as much data upstream as it downloaded. Their network would then just drop the upstream packets on the floor.
At that point the traffic would be balanced, but I'm pretty sure the ISPs wouldn't be any happier...
Given that they're specifically asking *Netflix* to pay extra. That's not just a peering issue, that's a type of traffic issue, which makes it a Net Neutrality issue.
It would be a straight peering/transit issue if the Tier-1 ISPs just told Cogent "your traffic isn't balanced, pay for the imbalance" without bringing up what type of traffic it was. In that case Cogent could pay the extra and then charge Netflix more for upstream. Netflix would then pass the costs on to their customers or would find a new ISP.
This has been in the news before...many cars that aren't "high-performance" cars will cut the throttle as soon as they detect that the brake pedal has been pushed.
Given that, it's entirely reasonable to suggest that Tesla do the same.
Look at Tiffany, they have trademarked a very specific shade of blue. But anyone can use other shades.
The colours "yellow" and "dark grey" aren't very specific. When does something stop being yellow and start being orange or green? When does "dark grey" become "grey" or "charcoal"?
Slashdot Mirror
I wouldn't say that what they did was "evil", but it does smack of trying for profit at the expense of their customers' safety. And public perception is important...this is why Toyota got into so much hot water over their throttle issue--they were perceived to be a very safe vehicle so it was a shock to discover that they were hiding safety issues.
What GM should have done was to change the part number and notify the owners of vehicles with the old ignition switch to come in to a dealer to get them swapped out for a good one.
Look at Tesla...very good safety record but they're raising the ride height a bit and adding a titanium shield just to improve the safety that much more.
It's that GM didn't actually admit that anything *had* changed (the part number was the same), so he had no reason to suspect that there had been a change in the ignition switch.
Even if you don't *think* there is a functional change, any change in the part should result in a different part number. The computer databases will all have part number equivalencies in them, so anyone looking for the old part number will get redirected to the new part.
In that way, if a problem shows up later (like in this case) it's much simpler to determine that a part was changed.
Companies did the same thing in other industries. One of the Linksys 10/100 ethernet NICs (lne100tx maybe) had about 4 different ethernet controllers over time but kept the same part number. They needed totally different drivers under Linux.
If the part was changed for whatever reason, the part number should have been changed.
The screen printers still run DOS.
DOS was actually decent as an RTOS if you could dedicate it to one task. I did some control systems work for a lab-size Tokomak fusion reactor in DOS.
Copy the current machine as a VM and backup the image somewhere. When the current hardware dies, run the VM on something newer.
And keep your old working code.
With the parallel port on the host passed through to the guest.
I have a 25Mbps down connection, and Netflix movies actually look pretty good after the initial ramp up in quality as it detects the available bandwidth. I very rarely get glitches.
The problem with negotiating the price is that it makes people feel like they might have gotten a worse deal then the next guy, and it eats at them.
We in North America generally don't negotiate prices for new items. You go to the store, you buy the item, you pay the same as the next guy. (Unless you're buying high volume, or are a contractor, or something.) Generally the only new things we negotiate on are cars. Even most new houses have a set price.
Most cars require a "major service" every couple of years, which can easily run into a couple thousand bucks
Huh? I've got a 9-year old Toyota Matrix. It gets an oil change twice a year at about $50 each time, and has had one transmission oil flush at a few hundred bucks, one radiator fluid flush, and new low beam bulbs. I just replaced the clock spring myself for a few hundred bucks. Soonish it'll need new brake pads and new spark plugs.
What is this "major service" you speak of?
They are basically an "oil and gas" energy company.
To me, a true "energy company" should be producing energy from multiple sources: oil, gas, wind, solar, coal, nuclear, tidal, hydrogen, alcohol, etc.
Once you accept the notion of patents in the first place (as a temporary monopoly on something in return for disclosing it instead of keeping it secret), it's a very small jump to buying and selling them. If an independent inventor comes up with an idea but doesn't have the capital to bring it to market, they can sell the patent to a bigger company.
This is fine in and of itself. Even patent trolls would be fine if the patents they were holding were actually valid and the companies they were suing had in fact copied the idea from the published patent.
The problems are twofold:
1) If someone else comes up with the same idea totally independently, it should invalidate the patent as too obvious.
2) Patents have been granted for things that should never have received them. In particular people have been allowed to patent *concepts* rather than *specific ways to implement a concept*.
It just means that the ISPs shouldn't be doing traffic shaping by type across multiple users.
So if you want to watch a movie on Netflix, and I want to rsync my hard drive across the country, and someone else wants to something else, our packets should be treated equally based on how much bandwidth we've paid for.
I have no problem if you as a subscriber want to assign relative priorities to your packets. But those priorities shouldn't affect how *your* packets are treated relative to *my* packets.
In the end, I pay both Netflix and my ISP. Both of them (and any transit in the middle) are making a profit from that already.
Does the consumer ISP need to pay to make sure that the peering relationship is such that all their users have the ability to stream from Netflix unfettered? Considering 1) above, is this fair to the ISP? They could do so, but to maintain their existing cost structure it'd likely mean that they may have a smaller pipe to another peer. Is it fair to users using those other peers or do they also have simply make sure ALL of their peers are able to fully pass 100% of traffic unfettered at peak times?
Given that ISPs are essentially regulated monopolies, I'd suggest that they have a moral obligation to ensure that they have sufficient bandwidth to sites that are popular with their customers.
If large numbers of people want to connect to Netflix, it means that those people are not connecting to other places and the ISPs don't need as much bandwidth to those places.
I don't see a problem with telecoms charging Netflix extra for a higher priority service
I do. It violates net neutrality. Your netflix stream shouldn't take priority over my rsync backup.
ISPs should be allowed do traffic shaping on a per-subscriber basis regardless of traffic type. They can even be allowed to traffic-shape different types of traffic *for a particular subscriber* if that subscriber asks them to. But they should absolutely NOT be allowed to prioritize packets belonging to different subscribers based on traffic type.
I thought of this as well, but I figured it would probably be unworkable since people's home connections don't really have reliable performance, while AWS should have pretty good control over their bandwidth.
So you're suggesting that if Netflix modified their client to send as much data back to Netflix as Netflix sends to them that the ISPs would be happier?
Somehow, I don't think so given that ISPs typically have crappy upstream bandwidth to begin with.
There's nothing stopping ISPs from doing per-subscriber traffic shaping in a traffic-agnostic way. Thus, your neighbours could all be watching netflix and it would have ZERO impact on your Vimeo stream.
Internet should be billed like my electricity bill--a certain fixed-cost amount per month to cover the costs of being a subscriber, and a reasonable cost-per-GB.
This is a closer fit to the real cost model of an ISP.
The problem is that the ISPs see metered billing as a cash cow and want to charge stupidly high amounts per GB.
Imagine if Netflix just tweaked their client software to send as much data upstream as it downloaded. Their network would then just drop the upstream packets on the floor.
At that point the traffic would be balanced, but I'm pretty sure the ISPs wouldn't be any happier...
Given that they're specifically asking *Netflix* to pay extra. That's not just a peering issue, that's a type of traffic issue, which makes it a Net Neutrality issue.
It would be a straight peering/transit issue if the Tier-1 ISPs just told Cogent "your traffic isn't balanced, pay for the imbalance" without bringing up what type of traffic it was. In that case Cogent could pay the extra and then charge Netflix more for upstream. Netflix would then pass the costs on to their customers or would find a new ISP.
This has been in the news before...many cars that aren't "high-performance" cars will cut the throttle as soon as they detect that the brake pedal has been pushed.
Given that, it's entirely reasonable to suggest that Tesla do the same.
Look at Tiffany, they have trademarked a very specific shade of blue. But anyone can use other shades.
The colours "yellow" and "dark grey" aren't very specific. When does something stop being yellow and start being orange or green? When does "dark grey" become "grey" or "charcoal"?