That sounds great in theory. Please never drive in Chicago, Illinois or Atlanta, Georgia here in the States. On a six-lane-wide stretch of highway you must often be less than a car and a half behind the car in front of you at speed. If you're any farther back than that, some jackass will rev his engine and put his car within that two-car lengths you allowed him, giving you at best half a car length of play once he does it.
Well, for one, it's because selling "Internet access" and the tacking on extra non-neutral fees is a truth-in-advertising issue.
For another, it's a consumer protection issue. AT&T doesn't actually pay any more if you get 200 GB of data during the month from Google than if you spread it around among 200 smaller properties, 1 GB a month each. AT&T just wants to profit unfairly by charging more for routing your data based on how popular the data is. They want to create a scarcity that isn't there in digital media, and they're not the ones with the right to do that.
If Yahoo or Google or Google or Facebook want to control ho many people get copies of their content,that's fine. AT&T's job when they sell Internet access is to deliver routed traffic. If they need to prioritize VOIP over streaming video to keep their network functional, I'm okay with that. They shouldn't be hitting one video service more than another, or making video altogether unusable without some extra fee. They also shouldn't favor one VOIP carrier over another. They are selling bandwidth, not content. They need to deliver the packets.
If they want to get into the exclusive content delivery market, they need to not sell that as Internet access. They need to sell that as a content package.
My AT&T DSL is never maxed out for longer than an hour or three, because I just don't meet that sort of usage pattern. I've never had problems keeping it maxed for as long as I needed to get a few OS DVDs downloaded via straight HTTP download, FTP download, torrent, or jigdo.
Then again, I don't usually do those downloads at peak usage times -- not because I particularly avoid it, but because I keep an odd schedule and tend to do long downloads when I'm going to be away from the computer and not competing with my own downloads for other data transfers.
Perhaps it's because they are selling you Internet access packages? You can't sell someone an Internet access package and not give them access to the Internet.
You can sell them a point-to-point line or you can sell them a line without bandwidth and charge them for their actual usage (which is how many businesses still pay). You can even sell them a specific plan on which they pay for certain types of traffic to be prioritized over other types.
What you can't do includes selling them a plan that deprioritizes the data they want, then charge them for it to have equal footing. You can't sell them access to just your own network and call that Internet access, because it isn't.
What AT&T needs to do if AT&T wants to make a bunch of money from Google and Hulu is to sell them their connectivity. Short of that, they can renegotiate their peering agreements with the companies that do provide them connectivity so that the extra data being passed from them costs money above what the share/share peering agreement cancels out. Then the people providing that connectivity to the big data providers will have to adjust based on the renegotiated contracts, and maybe pass some of that cost on to the Googles of the world.
You're conflating dry ounces and liquid ounces. They are two separate measurements. One if of weight, the other of volume. A fluid ounce of water is approximately (but not exactly) one dry ounce in weight (1.043 dry ounces or 29.6 grams).
Proof is merely double the alcohol content as a percentage by volume (which differs of the old English proof of 7/4 alcohol content by volume). US law actually requires the alcohol by volume labeling, but allows the proof labeling to be placed near the ABV.
A drink of 1.5 oz is rarely mentioned in the US, as that'd be called a shot. It would be properly measured if accuracy was necessary by a jigger, of which a standard one measures out a shot. A mixed recipe may specify either by shot or by jigger. Some jiggers are larger than standard, so filling a larger jigger device would result in more than a jigger of liquor.
A cup, besides being a drinking vessel, is also a measurement of 8 fluid ounces of liquid. Two cups is a pint, two pints a quart, and four quarts to a gallon. A popular bottle size for liquor, though, is the fifth. The true fifth gallon bottles are no longer used, 750 ml being "close enough" and readily labelled for international sale. A true fifth gallon would have a whole extra 7 ml or so.
Also, we often have 16.9 oz (-ish) bottles rather than pints, because 16.9 oz is 499.8 ml.
Our sodas are often in 20-ounce bottles, which is about 591 ml, but we often buy 1 liter and 2 liter sizes. Sports drinks are often in pints, 20 oz, quarts, or half gallons, but I'm sure I've seen 16.9 oz ones of some brand or another somewhere.
I didn't say it wasn't a more usable, more stable, higher quality service with a 2-hour technician-on-site SLA. I used to be the director of technical operations for an ISP. We had over 20 T-1s for bandwidth from points to other points, and a DS-3 at 6 Mbps 95% utilization burstable to the full 45 MBps for an additional fee with four DS-1s for backup connectivity to the world. DSL that ain't, but you pay for it. I've also dealt with frame circuits and ATM. I was also the second ranked admin at a place with OC-12s running into it.
The T-1 (or more properly DS-1) is still 1.544 Mbps and not 100 Mbps as AC mentioned. When you start talking HDLC lines, ATM, or SONET with short-timed SLAs, the price is big whether the bandwidth is or not.
Kind of like when Covad sold DSL on SBC lines, and SBC was required to give them "like pricing"? "Like pricing" doesn't mean "the same price for every unit". It means that the competitor gets the same price chart as the subsidiary. On that price chart, the huge volume discount was a sliding scale between SBC Internet's volume of orders from SBC Communications and anyone else's volume of orders. Lots of small ISPs were reselling Covad, but at some point SBC's DSL was cheaper to the end customer than Covad could sell it to the ISPs.
Take it from a survivor of The Great ISP Merger Rush. Giving a regulated monopoly any chance to gain an advantage in a lesser regulated market base don their strength as the incumbent in the regulated monopoly market is just balls caught in a bike chain. It's painful, messy, and does more harm to the little guys than the machine.
Not having seen the design, I don't know how it's been implemented, but it's possible to have a compatible design that implements all the original specifications without designing it the same way... It's also possible for an FPGA design to run faster than the original part - see the multiple-tens-of-MHz variants of the equally venerable 6502 (which maxed out at ~2MHz at the time) for example.
Um, did I claim that the Cray-1 was in the same class as a 600XL ? I don't believe so. I was using an example of how a chip of that vintage has been implemented with speeds approaching 50x the original.
Your assertions keep implying that what results one can get reimplementing some other stock MOS CPU from the same vintage one can get for the Cray 1. There are sure to be limiting factors based on the differences between the two fundamentally different architectures. I'm sure you didn't mean to state that. I'm just pointing out that the example you chose would tend to be misleading, even though you didn't intend for it to be.
Oh, to be managed by someone who only socializes and takes credit. That sounds so much better than actively sabotaging projects then passing the blame to everyone else, then taking credit when it finally succeeds.
Algorithms don't change much. Eratosthenes isn't getting any younger. Syntax may differ. Idioms may differ. One language's fastest approach to a particular problem may even not be the fastest approach in another language due to memory usage or other compiler vagaries. However, a good programmer knows how to program. A good software designer knows how to design software. A good usability expert knows how people use the software. Give a developer a chance to learn a language, a designer a chance to get up to speed on your doc system, and the UI guy a chance to get used to your widget layout editor. They'll be productive in short order if they're good.
I've worked as a programmer and as a systems admin. I worked in the ISP field during The Great Merger Rush. Can you guess what working with Sendmail, Postfix, Exim, IMail, Sendmail NT, qmail, tpop3d, qpopper, Eudora POP3, WU IMAP, Perdition, SpamAssassin, procmail, and more in five years' time did for me? I got really good at identifying basic patterns of mail problems and getting them sorted. I also worked with Ascend RADIUS, Livingston RADIUS, and half a dozen other RADIUS servers talking to access concentrators (and terminal servers hooked up to USR Sportster external serial modems) from Livingston, Lucent, USR, Ascend, Cisco, and Cyclades, sometimes with L2TP involved. My management at different times wanted monitoring done with Big Brother, Big Sister, What's Up Gold, or mon. I finally pressured one boss into letting me switch from the commercial What's Up Gold to Argus, and it did many more types of monitoring than he ever dreamed for us. I replaced a $6,000 Cisco 1U with Windows NT on it with a RedHat (pre-Fedora split) Pentium 133 we dug out of a closet for forcing routes onto unpaid customers to remind them to pay their bill.
Learning new tools is easy. Programming languages are just tools, albeit by their nature somewhat more complicated than most other tools. Hey, though, if you can learn m4 just to configure your mail server, then learning to put your domain knowledge in a different programming language isn't that bad.
A business analyst or user advocate is an important part of many software development teams, and that requires people skills. Technical writers are often more writers than they are technical, so a programmer who is also a good writer would be great for writing documentation (at least the docs for the development staff -- it's easy to get blinded to the user's view if you're a developer).
Software glitches are security issues. If someone can deprive me of the use of a system by crashing it or corrupting the data, it's no more useful to me than one compromised in any other way.
And stand over them and watch them "fix it" over and over until your Cayman trip is gone and your BMW is gone and your wife takes the house, because twice as many coders at half the cost working three times as long to get it right costs a lot more than half as many guys working a third as long.
Slashdot Mirror
That sounds great in theory. Please never drive in Chicago, Illinois or Atlanta, Georgia here in the States. On a six-lane-wide stretch of highway you must often be less than a car and a half behind the car in front of you at speed. If you're any farther back than that, some jackass will rev his engine and put his car within that two-car lengths you allowed him, giving you at best half a car length of play once he does it.
Well, for one, it's because selling "Internet access" and the tacking on extra non-neutral fees is a truth-in-advertising issue.
For another, it's a consumer protection issue. AT&T doesn't actually pay any more if you get 200 GB of data during the month from Google than if you spread it around among 200 smaller properties, 1 GB a month each. AT&T just wants to profit unfairly by charging more for routing your data based on how popular the data is. They want to create a scarcity that isn't there in digital media, and they're not the ones with the right to do that.
If Yahoo or Google or Google or Facebook want to control ho many people get copies of their content,that's fine. AT&T's job when they sell Internet access is to deliver routed traffic. If they need to prioritize VOIP over streaming video to keep their network functional, I'm okay with that. They shouldn't be hitting one video service more than another, or making video altogether unusable without some extra fee. They also shouldn't favor one VOIP carrier over another. They are selling bandwidth, not content. They need to deliver the packets.
If they want to get into the exclusive content delivery market, they need to not sell that as Internet access. They need to sell that as a content package.
My AT&T DSL is never maxed out for longer than an hour or three, because I just don't meet that sort of usage pattern. I've never had problems keeping it maxed for as long as I needed to get a few OS DVDs downloaded via straight HTTP download, FTP download, torrent, or jigdo.
Then again, I don't usually do those downloads at peak usage times -- not because I particularly avoid it, but because I keep an odd schedule and tend to do long downloads when I'm going to be away from the computer and not competing with my own downloads for other data transfers.
Perhaps it's because they are selling you Internet access packages? You can't sell someone an Internet access package and not give them access to the Internet.
You can sell them a point-to-point line or you can sell them a line without bandwidth and charge them for their actual usage (which is how many businesses still pay). You can even sell them a specific plan on which they pay for certain types of traffic to be prioritized over other types.
What you can't do includes selling them a plan that deprioritizes the data they want, then charge them for it to have equal footing. You can't sell them access to just your own network and call that Internet access, because it isn't.
What AT&T needs to do if AT&T wants to make a bunch of money from Google and Hulu is to sell them their connectivity. Short of that, they can renegotiate their peering agreements with the companies that do provide them connectivity so that the extra data being passed from them costs money above what the share/share peering agreement cancels out. Then the people providing that connectivity to the big data providers will have to adjust based on the renegotiated contracts, and maybe pass some of that cost on to the Googles of the world.
How many furlongs per fortnight do you travel on your velocipede?
You're conflating dry ounces and liquid ounces. They are two separate measurements. One if of weight, the other of volume. A fluid ounce of water is approximately (but not exactly) one dry ounce in weight (1.043 dry ounces or 29.6 grams).
Proof is merely double the alcohol content as a percentage by volume (which differs of the old English proof of 7/4 alcohol content by volume). US law actually requires the alcohol by volume labeling, but allows the proof labeling to be placed near the ABV.
A drink of 1.5 oz is rarely mentioned in the US, as that'd be called a shot. It would be properly measured if accuracy was necessary by a jigger, of which a standard one measures out a shot. A mixed recipe may specify either by shot or by jigger. Some jiggers are larger than standard, so filling a larger jigger device would result in more than a jigger of liquor.
A cup, besides being a drinking vessel, is also a measurement of 8 fluid ounces of liquid. Two cups is a pint, two pints a quart, and four quarts to a gallon. A popular bottle size for liquor, though, is the fifth. The true fifth gallon bottles are no longer used, 750 ml being "close enough" and readily labelled for international sale. A true fifth gallon would have a whole extra 7 ml or so.
Also, we often have 16.9 oz (-ish) bottles rather than pints, because 16.9 oz is 499.8 ml.
Our sodas are often in 20-ounce bottles, which is about 591 ml, but we often buy 1 liter and 2 liter sizes. Sports drinks are often in pints, 20 oz, quarts, or half gallons, but I'm sure I've seen 16.9 oz ones of some brand or another somewhere.
I didn't say it wasn't a more usable, more stable, higher quality service with a 2-hour technician-on-site SLA. I used to be the director of technical operations for an ISP. We had over 20 T-1s for bandwidth from points to other points, and a DS-3 at 6 Mbps 95% utilization burstable to the full 45 MBps for an additional fee with four DS-1s for backup connectivity to the world. DSL that ain't, but you pay for it. I've also dealt with frame circuits and ATM. I was also the second ranked admin at a place with OC-12s running into it.
The T-1 (or more properly DS-1) is still 1.544 Mbps and not 100 Mbps as AC mentioned. When you start talking HDLC lines, ATM, or SONET with short-timed SLAs, the price is big whether the bandwidth is or not.
It's the example you used.
I know... periodic surprise audits by a Federal agency, with employees and lobbyists cycled in and out between the agency and those it monitors!
(hey, it's the plan most regulated industries in the US are regulated under)
That sounds like a family-friendly deal. The Gambino family, for example.
Well, there is multihoming, but in the common case you're dead on right.
Kind of like when Covad sold DSL on SBC lines, and SBC was required to give them "like pricing"? "Like pricing" doesn't mean "the same price for every unit". It means that the competitor gets the same price chart as the subsidiary. On that price chart, the huge volume discount was a sliding scale between SBC Internet's volume of orders from SBC Communications and anyone else's volume of orders. Lots of small ISPs were reselling Covad, but at some point SBC's DSL was cheaper to the end customer than Covad could sell it to the ISPs.
Take it from a survivor of The Great ISP Merger Rush. Giving a regulated monopoly any chance to gain an advantage in a lesser regulated market base don their strength as the incumbent in the regulated monopoly market is just balls caught in a bike chain. It's painful, messy, and does more harm to the little guys than the machine.
AOL and MSN walled their own gardens. AT&T didn't wall them off for AT&T customers.
A T1 is 1.5Mbps.
Your assertions keep implying that what results one can get reimplementing some other stock MOS CPU from the same vintage one can get for the Cray 1. There are sure to be limiting factors based on the differences between the two fundamentally different architectures. I'm sure you didn't mean to state that. I'm just pointing out that the example you chose would tend to be misleading, even though you didn't intend for it to be.
Sure it is, if you're an admissions counselor at a top-10 business school.
Oh, to be managed by someone who only socializes and takes credit. That sounds so much better than actively sabotaging projects then passing the blame to everyone else, then taking credit when it finally succeeds.
Algorithms don't change much. Eratosthenes isn't getting any younger. Syntax may differ. Idioms may differ. One language's fastest approach to a particular problem may even not be the fastest approach in another language due to memory usage or other compiler vagaries. However, a good programmer knows how to program. A good software designer knows how to design software. A good usability expert knows how people use the software. Give a developer a chance to learn a language, a designer a chance to get up to speed on your doc system, and the UI guy a chance to get used to your widget layout editor. They'll be productive in short order if they're good.
I've worked as a programmer and as a systems admin. I worked in the ISP field during The Great Merger Rush. Can you guess what working with Sendmail, Postfix, Exim, IMail, Sendmail NT, qmail, tpop3d, qpopper, Eudora POP3, WU IMAP, Perdition, SpamAssassin, procmail, and more in five years' time did for me? I got really good at identifying basic patterns of mail problems and getting them sorted. I also worked with Ascend RADIUS, Livingston RADIUS, and half a dozen other RADIUS servers talking to access concentrators (and terminal servers hooked up to USR Sportster external serial modems) from Livingston, Lucent, USR, Ascend, Cisco, and Cyclades, sometimes with L2TP involved. My management at different times wanted monitoring done with Big Brother, Big Sister, What's Up Gold, or mon. I finally pressured one boss into letting me switch from the commercial What's Up Gold to Argus, and it did many more types of monitoring than he ever dreamed for us. I replaced a $6,000 Cisco 1U with Windows NT on it with a RedHat (pre-Fedora split) Pentium 133 we dug out of a closet for forcing routes onto unpaid customers to remind them to pay their bill.
Learning new tools is easy. Programming languages are just tools, albeit by their nature somewhat more complicated than most other tools. Hey, though, if you can learn m4 just to configure your mail server, then learning to put your domain knowledge in a different programming language isn't that bad.
A business analyst or user advocate is an important part of many software development teams, and that requires people skills. Technical writers are often more writers than they are technical, so a programmer who is also a good writer would be great for writing documentation (at least the docs for the development staff -- it's easy to get blinded to the user's view if you're a developer).
One part of it is that plumbers, electricians, carpenters, auto workers, and city trash collectors tend to be organized labor.
Newbie. I grew up in the 2-color (green and dark green) era. The background wasn't actually even black. just not as green as the lit pixels.
... which is a remake of Buck Rogers from the 1939 film serials...
Discrimination based on familial status is illegal.
Software glitches are security issues. If someone can deprive me of the use of a system by crashing it or corrupting the data, it's no more useful to me than one compromised in any other way.
And stand over them and watch them "fix it" over and over until your Cayman trip is gone and your BMW is gone and your wife takes the house, because twice as many coders at half the cost working three times as long to get it right costs a lot more than half as many guys working a third as long.