Self-modifying code
Yup, I actually write asm code.. plus he mentions "modifying the code while it's running".. if you can't do that, you shouldn't be wielding a debugger.
Code that generates code is occasionally necessary, but code that actually modifies itself locally, to "improve performance", has been obsolete for a decade.
IA-32 CPUs still support self-modifying code for backwards compatibility. (On most RISC machines, it's disallowed, and code is read-only, to simplify cache operations.) Superscalar IA-32 CPUs still support self-modifying code. But the performance is awful. Here's what self-modifying code looks like on a modern CPU:
Execution is going along, with maybe 10-20 instructions pre-fetched and a few operations running concurrently in the integer, floating point, and jump units. Alternate executions paths may be executing simultaneously, until the jump unit decides which path is being taken and cancels the speculative execution. The retirement unit looks at what's coming out of the various execution pipelines and commits the results back to memory, checking for conflicts.
Then the code stores into an instruction in the neighborhood of execution.
The retirement unit detects a memory modification at the same address as a pre-fetched instruction. This triggers an event which looks much like an interrupt and has comparable overhead. The CPU stops loading new instructions. The pipelines are allowed to finish what they're doing, but the results are discarded. The execution units all go idle. The prefetched code registers are cleared. Only then is the store into the code is allowed to take place.
Then the CPU starts up, as if returning from an interrupt. Code is re-fetched. The pipelines refill. The execution units become busy again. Normal execution resumes.
Self-modifying code hasn't been a win for performance since the Intel 286 (PC-AT era, 1985) or so.
It might not have hurt on a 386. Anything later, it's a lose.
From the FAQ: Why is someone else's User Name appearing on my User Page's Menu?
This is not a bug. This is a feature! That name is the last user page (besides your own;) that you have visited. This is useful when you want to hop around between your user info, and someone else's: to compare friends and foes for example. Your account has not been hacked, this is totally by design.
It's a badly implemented feature. You don't really have someone else's identity, it just looks that way. Maybe. It may have a bug that lets you use someone else's mod points. I just got a "Moderate" button from someone else's account.
It works for Second Life. Almost too well. In 2007, Ginko Financial, an in-game bank, went bust. Then Midas Bank went bust. This drew the attention of The Wall Street Journal. In 2008, Linden Labs introduced
bank regulation. Most of the Second Life banks were actually Ponzi schemes, with huge interest rates. It's still possible for a real-world bank to open branches in Second Life, but nobody has bothered.
The world of high-end graphics cards went away a decade ago. Evans and Sutherland, Dynamic Pictures, and Lockheed all had graphics cards for PCs in the $1000-$5000 range. Ten years ago, I had a $3000 graphics board from Dynamic Pictures. For a while I had something called a Fujitsu Sapphire graphics board on loan; Fujitsu gave up and exited the business before launching a product. And I'm ignoring SGI here.
The high-end guys were run over by the gamer card industry, which had real volume and was "good enough" for high-end animation tools. "High end" today is a few hundred dollars, not a few thousand.
The big headache for the animation community has been insufficient graphics memory. Gamer cards tended to stress fill rate over texture memory. Nobody in animation cares about frame rate once it passes 30FPS. What you need for animation is plenty of space for big textures. Game textures are shrunk to fit, but that happens late in the development pipeline. During content creation (and for movie and TV work) you need much larger texture maps. A few gigabytes of texture memory would not be too much. For most of a decade, you couldn't get that on PCs. Finally, you can.
This is a useful mechanism for search engines, which need to distinguish original content from hundreds or thousands of blogs echoing it. Imagine the Web with all the duplicate, repetitive material ignored. No wonder Yahoo is supporting this. Someone over there is thinking.
A visually annoying problem with this book is the manner in which, on far too many lines in the text, there is an inadequate amount of space between the words. Consequently, distinguishing individual words -- particularly when trying to read at a fast pace -- is made much more difficult.
Maybe they'll come out with a "large print edition".
The trouble with "software as a service" is that you're at the mercy of the service provider. Slowly, the service quality declines and the price goes up. Think of the history of cable TV, cell phone service, and post-deregulation air travel.
For those, at least, you're getting the use of an infrastructure you couldn't afford yourself. But for applications where you could have your own infrastructure for a modest cost, software as a service means unnecessarily buying into a relationship where the other side has all the power.
If you want a vision of the future, imagine a boot stamping on a human face - forever. -- Orwell.
It's usually not worth it, except for situations like pulling cables through conduit where you
can't pull the connectors through.
I once had to wire Ethernet back in the coax days. We had our own cables made, pulled through underfloor ducts, and terminated. But this was an aerospace company with a strong RF capability. Someone looked at the Ethernet cable spec, made cables accordingly, tested them with RF testgear including a TDR, and they worked perfectly.
If you make it, you must have the test gear to test it.
If they can figure out how to get this thing to understand financial data, it would be quite useful. That whole area needs more theoretical work.
Machine understanding of financial data is tough. Partly because the data is willfully obfuscated. I once developed a system for turning SEC filings into XBRL (which is an XML representation for financial statements.) At one point, I had several hundred euphemisms for "Net Loss". The connection between financial reporting and reality is at times tenuous.
Accounting is fundamentally mis-designed. The problem is that some numbers are actual, some have tolerances, some are estimates whose actual value will be known at a future date, and some are estimates whose actual value will never be known. Numbers of all four categories are added, and the result is given as a number without a tolerance. That's just wrong. Accounting works that way for historical reasons; it was designed when arithmetic was expensive. Why it stays that way is more interesting, but beyond the scope of this posting. Because of these problems, machine understanding of traditional accounting data is very difficult.
(Back when I did Downside I was more into this, but when I started getting invited to accounting conferences, I realized I didn't want to get into accounting standardization as a field.)
Torrent was EXACTLY the word I was looking for. Thank you, The Economist!
If you think that happened by accident, you don't read The Economist regularly. That's exactly the sort of dry wit their writers use.
Some years ago, The Simpsons had Homer traveling by air in first class, and he says "Look at me, I'm reading The Economist. Did you know Indonesia is at a crossroads?" The Economist published an article titled "Indonesia at a Crossroads" that week.
Can Slate stop writing articles about shit it doesn't know about?
Right.
First, most of the things Slate suggests have been tried. Timing human input behavior
is in use already, and attacks already do some randomization there.
Second, despite what the Slate article quotes, the CAPTCHA for Gmail has been cracked. The success rate is only 20%, but because the cracker is embedded in a botnet, that's good enough to survive IP blacklisting. MessageLabs says Gmail spam went from 1.3 percent of all spam e-mail in January to 2.6 percent in February.
All the proposed tasks - recognizing people, cats vs dogs, etc. - can be done by computers at the 20% accuracy level or better. So that's not going to work.
ReCAPTCHA isn't very good in practice. You get two words, one of which was recognized by an OCR program and one of which wasn't. You only have to re-recognize the one which some OCR program already got to pass the CAPTCHA. If you can do that, you have a 50% chance of success.
Then there are the outsourcing services. "We are 35 seater call center located in Hyderabad, we would be interested." The going rate is US$0.001 to US$0.003 per CAPTCHA solved successfully. There are always ads on GetAFreelancer for CAPTCHA solving.
Read Black Hat World for sources.
Reading the complaint, they have a good case. Netflix and Blockbuster don't have this problem because their items are getting special handling. They want equally good treatment.
The basic problem is that they want to ship DVDs through the first class mail system, not via parcel post. As a first class item, they can't put in enough stiffeners to protect the item and stay under the 1 oz rate, or even the 2 oz rate.
If you ship a DVD in a DVD case in a fibreboard DVD mailer, it won't be damaged unless someone steps on it or runs it over. But the mailer will cost about $0.30 in quantity, and mailing it will cost $1.34. They're trying to get the first class mail rate for "1 oz envelope containing a rigid object" of $0.62, or 2oz for $0.72. The extra dollar in shipping cost to get the product delivered safely would kill their profit margin.
The NYPD might be able to do something about this. They have a sizable anti-terrorism operation, over a thousand people. David Cohen, the NYPD's Deputy Commissioner for Intelligence, used to head Clandestine Services at the CIA. Sooner or later, many of the world's conflicts spill over into New York City, and the NYPD has to deal with it. So the NYPD has more capability to deal with external threats than most departments. They're also bigger than the FBI.
The NYPD is well-connected with infrastructure organizations. They're generally thought of as better organized than U.S. Homeland Security in that area. Homeland Security tends to be political. The NYPD, for better or for worse, is just cops. They also have a large number of people with good connections outside the US and good foreign language skills. The NYPD has liaison officers with key police departments around the the world, and they're willing to put somebody on a plane to go somewhere when necessary.
Most computer intrusions, once the attacker has been localized, yield to ordinary cop work. Now the attackers have the full attention of the NYPD.
The trouble with arming merchant ships is that few of the major cargo ports will let an armed ship dock. US ports certainly wouldn't.
The big fire nozzle is a good idea. Especially if it's radar-directed well enough to hit a speedboat, as Raytheon is proposing. Raytheon has a long history of making stuff that hits targets. Water cannons have been used successfully as military weapons in the past. Egypt used them in the 1973 war against Israel to slice gaps through Israeli fortifications along the Suez Canal. Pirates may try attacking using larger boats, but then it's harder for them
to hide from warships.
Pirates are easier to deal with than terrorists. Pirates not only don't want to die, they have to make a profit.
All I need is a PDF reader that will render correctly, won't create security problems, and will run on Win32. What's current thinking on this? The alternatives listed:
MuPDF Seems to be mostly a demo for a new graphics library.
Okular Does that even run on Windows? The table says yes, but the site says no.
Sumatra PDF Do I want to trust something that comes from "blog.kowalczyk.info"?
Yap Just a front end for GhostScript, which does a mediocre job on PostScript.
The reality: Skyy Vodka is a marketing company. Manufacturing is outsourced. They buy bulk ethanol from a MGP Ingredients (formerly Midwest Solvents Company) plant in Pekin, IL. MGP makes ethanol for beverage and industrial purposes. They used to sell ethanol for fuel, too, but that ended in February 2009 due to financial losses; their production costs were too high for fuel use.
The ethanol is pumped into tank cars and shipped by rail to Frank-Lin Distillers Products in San Jose, CA., which has their own railroad sidings. Frank-Lin bottles, along with Skyy Vodka, most of the low-end booze on the West Coast. They make everything from brandy to whiskey, by mixing ethanol, water, and flavoring. They make over a thousand different "brands", although they only have about a hundred different recipes.
Frank-Lin is very automated. They have automated bottling lines that can change from one bottle and product to another without human intervention, and equally flexible packaging systems. So they can create the illusion of thousands of products, all coming from one plant.
The cut location in San Carlos was reported as being at Bing St and Old County Road. That's
actually alongside the rail line that runs up the SF Peninsula. There are many fibre optic cables along that right of way. It used to be a Southern Pacific Railroad line, and "Sprint" was originally Southern Pacific Communications.
There aren't that many long haul fibre optic cable routes. Many of them run along rail lines, because the railroad owns the right of way and doesn't need anyone else's permission to run cables. Often you can run cable for miles without crossing a street, which makes installation much simpler.
He's previously proposed battery-swapping stations in two places where they actually make sense - Hawaii and Israel. In both places, you can't drive very far. So the number of battery swap stations needed is small.
But deployment isn't actually happening in those locations. Instead, he's going for more PR, not an initial product rollout. Hype.
Battery swapping was actually used around 1900. As with this new scheme, the batteries were in a big rectangular box on the bottom of the vehicle, and there was a mechanical arrangement to swap the packs quickly. But the competition was horse-drawn cabs.
A similar scheme was tried in London in 1896. Interesting, their problem was tires, not the batteries.
In 1898, New York got electric cabs with battery-charging stations. The battery change time was 75 seconds. Tires were still a problem; the battery weight overloaded existing tire technology.
The New York service was modestly successful and ran until 1907.
The problems with this idea are 1) it requires standardization of cars, 2) it's liable to be overtaken by improvements in fast battery charging technology, and 3) it takes a huge infrastructure investment before it's useful at all.
When this guy gets Honolulu or Tel Aviv running on his battery packs, it's time to listen. Until then, forget it.
1980s style: computers would reduce the use of paper in offices
fact: paper use in offices has gone right on up,
Actually, paper consumption peaked about two years ago. Printing is starting to decline. It took a while, but now that everything is networked, everything is searchable, and good, cheap displays are everywhere, it's happening.
Yes, the AP-1000 generates about 1.1GW of electricity. Yes, it's a traditional pressurized water reactor. That's because all the alternatives to pressurized water reactors have in practice turned out to be worse. Gas-cooled reactors have a troublesome record. Pebble-bed reactors tend to have jams in pebble handing.
It's an ongoing frustration with nuclear plant design that the operating temperatures are low, which makes for low efficiency and too much waste heat. But to date, nobody has ever built a gigawatt-sized plant with any of the more exotic technologies. So it makes sense to go forward with the classic approach.
The AP-1000 isn't a new technology reactor. That's the whole point. It's a conventional pressurized-water reactor. It's built mostly from existing Westinghouse components which Westinghouse had type-approved by the US Nuclear Regulatory Commission, so that multiple identical units could be built without going through a full design review for each one. So far, nobody has ordered one. Until now.
Most US reactors are unique designs, which is a headache. France has 34 reactors of the same design, which has cost and maintenance advantages, although there's been at least one common design flaw found.
Westinghouse is no longer a US company. It's owned by Toshiba.
I went to a "Green Expo" last year, and saw one booth with a small windmill, with about a 2 meter diameter 3-blade carbon-fiber blade assembly. The blades were fixed, and there was no overspeed feathering/furling capability. This was $10,000. Using their numbers, payback time was a century.
Take a look at the Logic Controls 8600. That's a server for fast-food restaurants and similar harsh environments. 1.6GHZ, 2GB, 40GB hard drive. Will run Linux. Fanless and ventless. Temp range 5C to 40C. Relative humidity 8 to 80%, non-condensing.
What do you need 4 terabytes of storage for? Unless you're running a movie piracy service?
But they are exactly like the TV shows (Spooks/MI-5)
The woman who actually runs MI-5 watches the show. She has commented that the two big errors are the assumption that everything is eventually knowable and that five people can do it all.
Slashdot Mirror
Self-modifying code
Yup, I actually write asm code.. plus he mentions "modifying the code while it's running".. if you can't do that, you shouldn't be wielding a debugger.
Code that generates code is occasionally necessary, but code that actually modifies itself locally, to "improve performance", has been obsolete for a decade.
IA-32 CPUs still support self-modifying code for backwards compatibility. (On most RISC machines, it's disallowed, and code is read-only, to simplify cache operations.) Superscalar IA-32 CPUs still support self-modifying code. But the performance is awful. Here's what self-modifying code looks like on a modern CPU:
Execution is going along, with maybe 10-20 instructions pre-fetched and a few operations running concurrently in the integer, floating point, and jump units. Alternate executions paths may be executing simultaneously, until the jump unit decides which path is being taken and cancels the speculative execution. The retirement unit looks at what's coming out of the various execution pipelines and commits the results back to memory, checking for conflicts.
Then the code stores into an instruction in the neighborhood of execution. The retirement unit detects a memory modification at the same address as a pre-fetched instruction. This triggers an event which looks much like an interrupt and has comparable overhead. The CPU stops loading new instructions. The pipelines are allowed to finish what they're doing, but the results are discarded. The execution units all go idle. The prefetched code registers are cleared. Only then is the store into the code is allowed to take place.
Then the CPU starts up, as if returning from an interrupt. Code is re-fetched. The pipelines refill. The execution units become busy again. Normal execution resumes.
Self-modifying code hasn't been a win for performance since the Intel 286 (PC-AT era, 1985) or so. It might not have hurt on a 386. Anything later, it's a lose.
From the FAQ: Why is someone else's User Name appearing on my User Page's Menu? ;) that you have visited. This is useful when you want to hop around between your user info, and someone else's: to compare friends and foes for example. Your account has not been hacked, this is totally by design.
This is not a bug. This is a feature! That name is the last user page (besides your own
It's a badly implemented feature. You don't really have someone else's identity, it just looks that way. Maybe. It may have a bug that lets you use someone else's mod points. I just got a "Moderate" button from someone else's account.
It works for Second Life. Almost too well. In 2007, Ginko Financial, an in-game bank, went bust. Then Midas Bank went bust. This drew the attention of The Wall Street Journal. In 2008, Linden Labs introduced bank regulation. Most of the Second Life banks were actually Ponzi schemes, with huge interest rates. It's still possible for a real-world bank to open branches in Second Life, but nobody has bothered.
The world of high-end graphics cards went away a decade ago. Evans and Sutherland, Dynamic Pictures, and Lockheed all had graphics cards for PCs in the $1000-$5000 range. Ten years ago, I had a $3000 graphics board from Dynamic Pictures. For a while I had something called a Fujitsu Sapphire graphics board on loan; Fujitsu gave up and exited the business before launching a product. And I'm ignoring SGI here.
The high-end guys were run over by the gamer card industry, which had real volume and was "good enough" for high-end animation tools. "High end" today is a few hundred dollars, not a few thousand.
The big headache for the animation community has been insufficient graphics memory. Gamer cards tended to stress fill rate over texture memory. Nobody in animation cares about frame rate once it passes 30FPS. What you need for animation is plenty of space for big textures. Game textures are shrunk to fit, but that happens late in the development pipeline. During content creation (and for movie and TV work) you need much larger texture maps. A few gigabytes of texture memory would not be too much. For most of a decade, you couldn't get that on PCs. Finally, you can.
This is a useful mechanism for search engines, which need to distinguish original content from hundreds or thousands of blogs echoing it. Imagine the Web with all the duplicate, repetitive material ignored. No wonder Yahoo is supporting this. Someone over there is thinking.
A visually annoying problem with this book is the manner in which, on far too many lines in the text, there is an inadequate amount of space between the words. Consequently, distinguishing individual words -- particularly when trying to read at a fast pace -- is made much more difficult.
Maybe they'll come out with a "large print edition".
The trouble with "software as a service" is that you're at the mercy of the service provider. Slowly, the service quality declines and the price goes up. Think of the history of cable TV, cell phone service, and post-deregulation air travel.
For those, at least, you're getting the use of an infrastructure you couldn't afford yourself. But for applications where you could have your own infrastructure for a modest cost, software as a service means unnecessarily buying into a relationship where the other side has all the power.
If you want a vision of the future, imagine a boot stamping on a human face - forever. -- Orwell.
It's usually not worth it, except for situations like pulling cables through conduit where you can't pull the connectors through.
I once had to wire Ethernet back in the coax days. We had our own cables made, pulled through underfloor ducts, and terminated. But this was an aerospace company with a strong RF capability. Someone looked at the Ethernet cable spec, made cables accordingly, tested them with RF testgear including a TDR, and they worked perfectly.
If you make it, you must have the test gear to test it.
If they can figure out how to get this thing to understand financial data, it would be quite useful. That whole area needs more theoretical work.
Machine understanding of financial data is tough. Partly because the data is willfully obfuscated. I once developed a system for turning SEC filings into XBRL (which is an XML representation for financial statements.) At one point, I had several hundred euphemisms for "Net Loss". The connection between financial reporting and reality is at times tenuous.
Accounting is fundamentally mis-designed. The problem is that some numbers are actual, some have tolerances, some are estimates whose actual value will be known at a future date, and some are estimates whose actual value will never be known. Numbers of all four categories are added, and the result is given as a number without a tolerance. That's just wrong. Accounting works that way for historical reasons; it was designed when arithmetic was expensive. Why it stays that way is more interesting, but beyond the scope of this posting. Because of these problems, machine understanding of traditional accounting data is very difficult.
(Back when I did Downside I was more into this, but when I started getting invited to accounting conferences, I realized I didn't want to get into accounting standardization as a field.)
Torrent was EXACTLY the word I was looking for. Thank you, The Economist!
If you think that happened by accident, you don't read The Economist regularly. That's exactly the sort of dry wit their writers use.
Some years ago, The Simpsons had Homer traveling by air in first class, and he says "Look at me, I'm reading The Economist. Did you know Indonesia is at a crossroads?" The Economist published an article titled "Indonesia at a Crossroads" that week.
Can Slate stop writing articles about shit it doesn't know about?
Right.
First, most of the things Slate suggests have been tried. Timing human input behavior is in use already, and attacks already do some randomization there.
Second, despite what the Slate article quotes, the CAPTCHA for Gmail has been cracked. The success rate is only 20%, but because the cracker is embedded in a botnet, that's good enough to survive IP blacklisting. MessageLabs says Gmail spam went from 1.3 percent of all spam e-mail in January to 2.6 percent in February.
All the proposed tasks - recognizing people, cats vs dogs, etc. - can be done by computers at the 20% accuracy level or better. So that's not going to work.
ReCAPTCHA isn't very good in practice. You get two words, one of which was recognized by an OCR program and one of which wasn't. You only have to re-recognize the one which some OCR program already got to pass the CAPTCHA. If you can do that, you have a 50% chance of success.
Then there are the outsourcing services. "We are 35 seater call center located in Hyderabad, we would be interested." The going rate is US$0.001 to US$0.003 per CAPTCHA solved successfully. There are always ads on GetAFreelancer for CAPTCHA solving. Read Black Hat World for sources.
Reading the complaint, they have a good case. Netflix and Blockbuster don't have this problem because their items are getting special handling. They want equally good treatment.
The basic problem is that they want to ship DVDs through the first class mail system, not via parcel post. As a first class item, they can't put in enough stiffeners to protect the item and stay under the 1 oz rate, or even the 2 oz rate.
If you ship a DVD in a DVD case in a fibreboard DVD mailer, it won't be damaged unless someone steps on it or runs it over. But the mailer will cost about $0.30 in quantity, and mailing it will cost $1.34. They're trying to get the first class mail rate for "1 oz envelope containing a rigid object" of $0.62, or 2oz for $0.72. The extra dollar in shipping cost to get the product delivered safely would kill their profit margin.
That was expected; it was predicted in Robotics Business Review last month. The price point was far too high.
WowWee's RoboReptile is almost as advanced, and has a price point around $90.
WowWee is a company to watch. They have a broad line of reasonably good robotic toys at modest price points. They even sell a fembot.
The NYPD might be able to do something about this. They have a sizable anti-terrorism operation, over a thousand people. David Cohen, the NYPD's Deputy Commissioner for Intelligence, used to head Clandestine Services at the CIA. Sooner or later, many of the world's conflicts spill over into New York City, and the NYPD has to deal with it. So the NYPD has more capability to deal with external threats than most departments. They're also bigger than the FBI.
The NYPD is well-connected with infrastructure organizations. They're generally thought of as better organized than U.S. Homeland Security in that area. Homeland Security tends to be political. The NYPD, for better or for worse, is just cops. They also have a large number of people with good connections outside the US and good foreign language skills. The NYPD has liaison officers with key police departments around the the world, and they're willing to put somebody on a plane to go somewhere when necessary.
Most computer intrusions, once the attacker has been localized, yield to ordinary cop work. Now the attackers have the full attention of the NYPD.
The trouble with arming merchant ships is that few of the major cargo ports will let an armed ship dock. US ports certainly wouldn't.
The big fire nozzle is a good idea. Especially if it's radar-directed well enough to hit a speedboat, as Raytheon is proposing. Raytheon has a long history of making stuff that hits targets. Water cannons have been used successfully as military weapons in the past. Egypt used them in the 1973 war against Israel to slice gaps through Israeli fortifications along the Suez Canal. Pirates may try attacking using larger boats, but then it's harder for them to hide from warships.
Pirates are easier to deal with than terrorists. Pirates not only don't want to die, they have to make a profit.
All I need is a PDF reader that will render correctly, won't create security problems, and will run on Win32. What's current thinking on this? The alternatives listed:
The hype: Skyy Vodka
The reality: Skyy Vodka is a marketing company. Manufacturing is outsourced. They buy bulk ethanol from a MGP Ingredients (formerly Midwest Solvents Company) plant in Pekin, IL. MGP makes ethanol for beverage and industrial purposes. They used to sell ethanol for fuel, too, but that ended in February 2009 due to financial losses; their production costs were too high for fuel use.
The ethanol is pumped into tank cars and shipped by rail to Frank-Lin Distillers Products in San Jose, CA., which has their own railroad sidings. Frank-Lin bottles, along with Skyy Vodka, most of the low-end booze on the West Coast. They make everything from brandy to whiskey, by mixing ethanol, water, and flavoring. They make over a thousand different "brands", although they only have about a hundred different recipes.
Frank-Lin is very automated. They have automated bottling lines that can change from one bottle and product to another without human intervention, and equally flexible packaging systems. So they can create the illusion of thousands of products, all coming from one plant.
It's all just flavored ethanol. Deal with it.
The cut location in San Carlos was reported as being at Bing St and Old County Road. That's actually alongside the rail line that runs up the SF Peninsula. There are many fibre optic cables along that right of way. It used to be a Southern Pacific Railroad line, and "Sprint" was originally Southern Pacific Communications.
There aren't that many long haul fibre optic cable routes. Many of them run along rail lines, because the railroad owns the right of way and doesn't need anyone else's permission to run cables. Often you can run cable for miles without crossing a street, which makes installation much simpler.
I think this Shai Agassi guy is mostly hype.
He's previously proposed battery-swapping stations in two places where they actually make sense - Hawaii and Israel. In both places, you can't drive very far. So the number of battery swap stations needed is small.
But deployment isn't actually happening in those locations. Instead, he's going for more PR, not an initial product rollout. Hype.
Battery swapping was actually used around 1900. As with this new scheme, the batteries were in a big rectangular box on the bottom of the vehicle, and there was a mechanical arrangement to swap the packs quickly. But the competition was horse-drawn cabs. A similar scheme was tried in London in 1896. Interesting, their problem was tires, not the batteries. In 1898, New York got electric cabs with battery-charging stations. The battery change time was 75 seconds. Tires were still a problem; the battery weight overloaded existing tire technology. The New York service was modestly successful and ran until 1907.
The problems with this idea are 1) it requires standardization of cars, 2) it's liable to be overtaken by improvements in fast battery charging technology, and 3) it takes a huge infrastructure investment before it's useful at all.
When this guy gets Honolulu or Tel Aviv running on his battery packs, it's time to listen. Until then, forget it.
1980s style: computers would reduce the use of paper in offices
fact: paper use in offices has gone right on up,
Actually, paper consumption peaked about two years ago. Printing is starting to decline. It took a while, but now that everything is networked, everything is searchable, and good, cheap displays are everywhere, it's happening.
Mod parent up.
Yes, the AP-1000 generates about 1.1GW of electricity. Yes, it's a traditional pressurized water reactor. That's because all the alternatives to pressurized water reactors have in practice turned out to be worse. Gas-cooled reactors have a troublesome record. Pebble-bed reactors tend to have jams in pebble handing.
It's an ongoing frustration with nuclear plant design that the operating temperatures are low, which makes for low efficiency and too much waste heat. But to date, nobody has ever built a gigawatt-sized plant with any of the more exotic technologies. So it makes sense to go forward with the classic approach.
The AP-1000 isn't a new technology reactor. That's the whole point. It's a conventional pressurized-water reactor. It's built mostly from existing Westinghouse components which Westinghouse had type-approved by the US Nuclear Regulatory Commission, so that multiple identical units could be built without going through a full design review for each one. So far, nobody has ordered one. Until now.
Most US reactors are unique designs, which is a headache. France has 34 reactors of the same design, which has cost and maintenance advantages, although there's been at least one common design flaw found.
Westinghouse is no longer a US company. It's owned by Toshiba.
I went to a "Green Expo" last year, and saw one booth with a small windmill, with about a 2 meter diameter 3-blade carbon-fiber blade assembly. The blades were fixed, and there was no overspeed feathering/furling capability. This was $10,000. Using their numbers, payback time was a century.
The going rate for a 2m turbine is about $1000. So I asked the sales rep why their unit was so expensive. He said "this is a status symbol, like a Mercedes". Right.
That's a big ruggedized server.
Take a look at the Logic Controls 8600. That's a server for fast-food restaurants and similar harsh environments. 1.6GHZ, 2GB, 40GB hard drive. Will run Linux. Fanless and ventless. Temp range 5C to 40C. Relative humidity 8 to 80%, non-condensing.
What do you need 4 terabytes of storage for? Unless you're running a movie piracy service?
But they are exactly like the TV shows (Spooks/MI-5)
The woman who actually runs MI-5 watches the show. She has commented that the two big errors are the assumption that everything is eventually knowable and that five people can do it all.