This looks good because it is specialized for 8-bit manually created graphics. 8-bit graphics with a broad palette has little shading, so it's usually unambiguous which pixel attaches to which adjacent pixel. Vectorizing 24-bit color images, or 8-bit images from photographs, requires more guesses about how wide things are and where the edges are supposed to be.
Nice if you really want to play 80's games in higher resolution.
Really? It was my impression that even when scrammed there's enough self-reacting of a "spent" fuel rod that it takes weeks or months for the temperature to decrease to where you can remove it from a vessel, even to move to the pond to continue cooling until it's "cold".
It takes years before it reaches a state where little cooling is required. Cold shutdown, though, simply means that the reactor temperature is below the boiling point of water, the reactor vessel is at atmospheric pressure, and the fuel rods are submerged in water. In that state, a pressurized reactor can be opened at the top. The fuel rods can then be removed, one at a time, to the spent fuel pool. This is how normal refueling takes place.
Some pumps were still running after the earthquake and tsunami, and they continued to run until the backup batteries ran down. Loss of power was the real cause of the disaster. If they'd some backup power source that worked, the reactors would have reached cold shutdown in a day or two, there would have been no hydrogen explosions, and no core melting.
This is really important. A plant could lose backup power for many other reasons: fire, flood, hurricanes, terrorism, contaminated fuel, tank leakage, transformer damage, maintenance outages, or exhaustion of fuel supplies. Hospitals and data centers with backup power have at times lost power for all those reasons.
Read NUREG/CR-6890, "Reevaluation of Station Blackout Risk at Nuclear Power Plants ", from 2005. Volume 2, page 22, has the line "Risk is evaluated only for critical
operation, not for shutdown operation. External events, such as seismic, fire, or flood, are also excluded." That, as we know now, is an overoptimistic assumption. The NRC does a statistical analysis on backup power sources, assuming independent failure of separate units, and computes the odds accordingly.
Nuclear plants that need power to reach shutdown need power sources as tough as the containment vessel. That's now very clear.
Yes. About half of Switzerland's power comes from hydroelectric plants. But the good sites are already developed. This is a general problem with hydroelectric power. For large power dams, "all the good sites were gone by 1940". The ideal hydroelectric site was Hoover Dam - narrow gorge to dam, big level drop, large unpopulated desert basin area. Almost every other location is worse.
Business is not going to tolerate smartphones which are slaves of the phone provider and tell them everything. That's why Blackberry is so successful. You can have your own Blackberry server with crypto between your server and your employees' phones. Crypto for which no external provider has the keys.
The patent can be licensed for 0.575% of revenue. Apple, on the other hand, demands 30% of third-party sales revenue for a transaction which doesn't involve them.
The last major terrorism incident in Australia was in 1978, and that one is generally attributed to Australia's own security forces. In 1986, somebody tried a bombing, and blew themselves up. There have been some foreign bombings in Indonesia that killed Australians. That's it. There is no significant terrorism problem.
As for external attack, Australia is an island, has a respectable army, navy and air force, and nobody has attacked since WWII.
Vision systems for checkout are available. There's LaneHawk, for recognizing big items at the bottom of the cart, and VeggieVision, for recognizing vegetables on a scale pan. Automated checkout is getting better.
The future of retail looks more like WebVan. WebVan was a flop, but not because of customer acceptance. WebVan was popular, but the operating costs were too high. "Soap.com" (acquired by Amazon) is now doing the WebVan thing of delivering routine items. But now, with Kiva robotic order picking, it's profitable. Kiva's system is now doing about 10% of online order picking in the US. Costs are about 1/5 of human picking.
Delivery uses less fuel than driving a ton of car to the store to move a few pounds of merchandise. At $4 per gallon and up, Soap.com's shipping rates (Max of $5, free for orders over $39) look really good.
The future of retail is online ordering and delivery. Been to a record store lately? A video rental store? A bank branch? A travel agency? Look at all the vacant retail space that will never again be occupied.
Most programs should run in an environment that has far fewer privileges than the user running it. Especially games. All a game really needs to talk to are its own files, the screen, the input devices when it has focus, and its own Internet server. Those are essentially the restrictions under which a web page or Flash program runs.
Anything which needs more privileges than that should either have to be signed by somebody to indicate responsibility for the program, or the entire system has to be put in "developer mode", with additional debug facilities and logging enabled.
Microsoft has, at least,enforced that model for drivers. Doing it for applications is the next step.
If they have an email address, they can mail a password reset to it, but simply allowing users to enter it as if it were a password is a bit much.
Of course, the problem is that if they have an email address and a password for their own system, for a large number of accounts, that password will be the password for the email system as well.
If it didn't connect to a Microsoft server to test connectivity, somebody would be screaming about the hundred million pings per day they were receiving.
Windows also connects to Microsoft's time server. There are many other time servers, but most of them aren't sized for hundreds of millions of connections per day.
What are they using to rectify the signal to convert to DC?
Unobtainum diodes. They don't actually know how to do that.
Terahertz diodes do exist. Low-cost, high-efficiency, integrated terahertz diodes, no. But as work proceeds on terahertz electronics, someone may solve that problem. Each nanoantenna needs its own nanodiode, so the diodes have to be fabricated on the substrate with the antenna, which complicates the fab problem. The enthusiasm about roll-to-roll low cost fabrication in the article is premature. We'll probably see this working first on a wafer, and it may not be cheap.
Even if it's expensive, there's an initial market for satellite power panels. The performance improvement would be worth it.
That's 10 year old technology, and it's a chemical laser. Back then it took three semitrailers for all the support equipment. Since then, electrically-powered lasers are catching up. The Navy Laser Weapons System is not as powerful, but it's a much smaller package, only needs electrical power. and can shoot down small UAVs.
Remote lighting control has been around for decades.
X10 has been available for a long time, it's inexpensive, and you can buy the gear at any Home Depot.
The next generation system after that was Echelon LONworks, which is a bidirectional power-line network for home control. That system really does give every device a unique address, set during manufacture, like Ethernet addresses. It's only 78kb/s, but that's enough for lighting control. It never caught on for home control, but it turned out to be useful for subway and railroad trains, because it has extremely good noise immunity. It's used on trains to control lights, HVAC, destination signs, doors, and other auxiliary equipment. Some office buildings use it, and it's sometimes used in semiconductor manufacturing plants.
Full Internet connectivity for each lighting device is a bit much. Do you really want to bridge that data to the outside world?
Bing keeps faithfully replicating Google's bad ideas. When Google merged "places" data from the map search engine into web search results, Bing immediately followed. Google search quality went way down as their "places" system was heavily spammed. Bing had the same problem. Google de-emphasized Places results, but that was subtle and Bing didn't pick up on it. So, right now, Bing's results are about where Google was at their low point in late 2010.
Now Google adds "like" (but not "hate" or "sucks" or "spam"), and so does Bing.
Bing could potentially do better than Google, because Bing doesn't have an AdSense revenue stream to protect. Sites with third-party ads need to be looked at more critically. (We do that with SiteTruth, and it helps.) Google gets 30% of their ad revenue from AdSense ads, and can't afford to raise their standards too high. But Bing's model seems to be to emulate Google, good or bad.
Most likely the investigators will not be listening to it like most people listen to a news broadcast. They will repeatedly be going over every second of the voice recording analyzing every pop, hiss, bang, etc.
For this flight, the flight data is the crucial issue. It's known from the maintenance telemetry that there were some system failures prior to the accident. The data from the aircraft systems is the big issue. It's non-trivial to analyze. Often, especially on newer aircraft where there's a lot of data, the data can be converted into a format that can be loaded into an aircraft simulator, allowing investigators to replay the accident. That was done with the aircraft that landed in the Hudson River.
The cockpit voice recorder is secondary for a loss-of-control accident like this.
(In contrast, the voice data is most important for a "controlled flight into terrain" accident, where the airplane was working just fine until it hit something.)
The NTSC routinely publishes transcripts of such recordings, and Flying magazine reprints some of them in their Aftermath column.
If you've never seen a full NTSB accident report, here's one.
There's so much plumbing in place at Niagara Falls that the falls can almost be turned off. There's a minimum water flow over the falls established by international agreement, but that's for aesthetics. At night, and during the tourist off season, more water is run through the hydroelectric plants.
Back in the 1980s, some boater was upstream of the falls, closer than he should have been, and lost power. He managed to run aground upstream of the falls. This was noticed at the Niagara Mohawk power plant control room, where an operator opened all feed and diversion tunnels and closed gates at the upstream weir, shutting off most of the falls until a rescue crew could fetch the boater.
A new public key is usually generated for each transaction, so this doesn't actually tell them anything.
See the Bitcoin paper, page 2, "Transactions": "Each owner transfers the coin to the next by digitally signing a hash of the previous transaction and the public key of the next owner and adding these to the end of the coin." Note, in the diagram, the "verify" link, where the previous owner's public key is transferred to the next owner, so the next owner can check that the previous owner owned the coin.
If a party changes their public key, their coin hashes won't match their coins, and the next transaction will fail.
This has the useful property that your "coins" aren't valuable without your private key. If someone copies your coin files, they can't spend them unless they also obtain (i.e. steal) your private key. Conversely, if you lose your private key, your coins become worthless.
Future generations will walk to the rusted ruins of Pad 39A as they walk to the Pyramids today, and wonder at the massive effort required to accomplish such a pointless but impressive task.
This has been done before. See DigiCash, from 1990.
"Clouds gather over Amsterdam as I ride into the city center after a day at the headquarters of DigiCash, a company whose mission is to change the world through the introduction of anonymous digital money technology. I have been inundated with talk of smart cards and automated toll takers and tamper-proof observer chips and virtual coinage for anonymous network ftps. I have made photocopies using a digital wallet and would have bought a soda from a DigiCash vending machine, but it was out of order. " - Wired, 1994.
See the article for what went wrong.
The soundness of Bitcoin's crypto doesn't seem to have been analyzed by third parties yet. There's nothing in Cryptologia or sci.crypt. Until there's agreement in the crypto community that it's sound, I'd be suspicious. There's also the problem that if the money resides on user PCs and smartphones, the usual attacks on those devices can steal it. Once stolen and used, there's no way to get it back.
Transactions are not very anonymous. If you spend a coin with a server, the server now knows your public key, and can associate it with any other identity information it has for you ( IP address, Facebook login, shipping address, etc.) If Amazon, eBay, Google Checkout, or Facebook accepted bitcoins, they'd be able to collect this info for a sizable fraction of the online world. Since your public key remains associated with the coin for at least the next few transactions, it's possible to follow the money.
Systems like this detect duplicate spending of the same item, but you can't tell if someone has a duplicate but unspent copy of your coins. So you don't know your money been stolen until you try to spend it.
There's also the technical problem that "new transactions are broadcast to all nodes". That won't scale.
It would be useful to have systems which automatically compare news stories on the same subject and note similarities and differences. Osama bin Laden dead? Checking... CNN. Yes. Fox News. Yes. Al-Jazeera - Yes. China Daily - Yes. Russia Today - Yes. Dawn (Pakistan) Yes. Asharq Al-Awsat - Yes. Reuters quote of statement by al-Queda - Yes. Conclusion: dead.
A fantastic summary, but I quibble with the "no evidence of any significant release of radiation" quote for Fukushima. Two months ago, I would have said it was impossible for a reactor in Japan to contaminate the drinking water in Tokyo, but that's exactly what happened. To the detriment of the industry (and I'm a nuclear engineer), there was a significant release of radiation.
Right. The number of casualties is small, but the area evacuated is large, and may be evacuated for decades.
For actuarial purposes, insurance for nuclear plants now has to be repriced. Total power reactor years worldwide is now about 14,000, with two major evacuation incidents. So an assumption of one evacuation of a 30km circle around the plant and acquisition of that real estate per 7000 reactor-years is appropriate for insurance purposes.
The insurance cost will vary with location. That's a big problem. Many US power reactors are sited near major cities. Indian Point in NY is probably the worst case for evacuation cost.
The discouraging thing about the Fukushima reactor disasters is that the real problem was loss of power. The earthquake and tsunami damage was contained. The cooling systems survived and ran until the batteries ran down. With no power, there was little cooling, resulting in hydrogen explosions and meltdowns in several units.
Plants are vulnerable to power loss for a variety of reasons - hurricanes, tornadoes, fires, floods, regional blackouts, poor maintenance, fuel shortages, and sabotage. Most of the NRC literature on loss of coolant accidents focuses on pipe breaks. That may have been the wrong emphasis.
Slashdot Mirror
That's like nominating yourself for a Nobel Peace Prize.
True. It's a bit much to be organizing a petition for this.
The Encyclopædia Britannica, in its heyday (the 11th edition), is far more worthy of monumental status.
I'd remove the search bar before removing the URL bar. Notice who's behind removing the URL bar.
This looks good because it is specialized for 8-bit manually created graphics. 8-bit graphics with a broad palette has little shading, so it's usually unambiguous which pixel attaches to which adjacent pixel. Vectorizing 24-bit color images, or 8-bit images from photographs, requires more guesses about how wide things are and where the edges are supposed to be.
Nice if you really want to play 80's games in higher resolution.
Really? It was my impression that even when scrammed there's enough self-reacting of a "spent" fuel rod that it takes weeks or months for the temperature to decrease to where you can remove it from a vessel, even to move to the pond to continue cooling until it's "cold".
It takes years before it reaches a state where little cooling is required. Cold shutdown, though, simply means that the reactor temperature is below the boiling point of water, the reactor vessel is at atmospheric pressure, and the fuel rods are submerged in water. In that state, a pressurized reactor can be opened at the top. The fuel rods can then be removed, one at a time, to the spent fuel pool. This is how normal refueling takes place.
Mod parent up.
Some pumps were still running after the earthquake and tsunami, and they continued to run until the backup batteries ran down. Loss of power was the real cause of the disaster. If they'd some backup power source that worked, the reactors would have reached cold shutdown in a day or two, there would have been no hydrogen explosions, and no core melting.
This is really important. A plant could lose backup power for many other reasons: fire, flood, hurricanes, terrorism, contaminated fuel, tank leakage, transformer damage, maintenance outages, or exhaustion of fuel supplies. Hospitals and data centers with backup power have at times lost power for all those reasons.
Read NUREG/CR-6890, "Reevaluation of Station Blackout Risk at Nuclear Power Plants ", from 2005. Volume 2, page 22, has the line "Risk is evaluated only for critical operation, not for shutdown operation. External events, such as seismic, fire, or flood, are also excluded." That, as we know now, is an overoptimistic assumption. The NRC does a statistical analysis on backup power sources, assuming independent failure of separate units, and computes the odds accordingly.
Nuclear plants that need power to reach shutdown need power sources as tough as the containment vessel. That's now very clear.
The sciences and engineering programs tend to have intro classes that weed out the weaker students.
That's a property of low-end schools. If you can get into MIT, you have over a 90% chance of graduating. (Cal Poly engineering, 45% - 50%)
Switzerland has good solar irradiation potential.
Er, no.
It also has mountains.
Yes. About half of Switzerland's power comes from hydroelectric plants. But the good sites are already developed. This is a general problem with hydroelectric power. For large power dams, "all the good sites were gone by 1940". The ideal hydroelectric site was Hoover Dam - narrow gorge to dam, big level drop, large unpopulated desert basin area. Almost every other location is worse.
Business is not going to tolerate smartphones which are slaves of the phone provider and tell them everything. That's why Blackberry is so successful. You can have your own Blackberry server with crypto between your server and your employees' phones. Crypto for which no external provider has the keys.
The patent can be licensed for 0.575% of revenue. Apple, on the other hand, demands 30% of third-party sales revenue for a transaction which doesn't involve them.
The last major terrorism incident in Australia was in 1978, and that one is generally attributed to Australia's own security forces. In 1986, somebody tried a bombing, and blew themselves up. There have been some foreign bombings in Indonesia that killed Australians. That's it. There is no significant terrorism problem.
As for external attack, Australia is an island, has a respectable army, navy and air force, and nobody has attacked since WWII.
What, exactly, justifies stringent security measures?
Whatever happened to the concept that you'd just push your cart through an RFID portal, everything in your cart would be interrogated, and you'd get an immediate bill? Wal-Mart was behind that. NCR demonstrated it in 2004. That was a more promising idea.
Vision systems for checkout are available. There's LaneHawk, for recognizing big items at the bottom of the cart, and VeggieVision, for recognizing vegetables on a scale pan. Automated checkout is getting better.
The future of retail looks more like WebVan. WebVan was a flop, but not because of customer acceptance. WebVan was popular, but the operating costs were too high. "Soap.com" (acquired by Amazon) is now doing the WebVan thing of delivering routine items. But now, with Kiva robotic order picking, it's profitable. Kiva's system is now doing about 10% of online order picking in the US. Costs are about 1/5 of human picking.
Delivery uses less fuel than driving a ton of car to the store to move a few pounds of merchandise. At $4 per gallon and up, Soap.com's shipping rates (Max of $5, free for orders over $39) look really good.
The future of retail is online ordering and delivery. Been to a record store lately? A video rental store? A bank branch? A travel agency? Look at all the vacant retail space that will never again be occupied.
Most programs should run in an environment that has far fewer privileges than the user running it. Especially games. All a game really needs to talk to are its own files, the screen, the input devices when it has focus, and its own Internet server. Those are essentially the restrictions under which a web page or Flash program runs.
Anything which needs more privileges than that should either have to be signed by somebody to indicate responsibility for the program, or the entire system has to be put in "developer mode", with additional debug facilities and logging enabled.
Microsoft has, at least,enforced that model for drivers. Doing it for applications is the next step.
If they have an email address, they can mail a password reset to it, but simply allowing users to enter it as if it were a password is a bit much.
Of course, the problem is that if they have an email address and a password for their own system, for a large number of accounts, that password will be the password for the email system as well.
If it didn't connect to a Microsoft server to test connectivity, somebody would be screaming about the hundred million pings per day they were receiving.
Windows also connects to Microsoft's time server. There are many other time servers, but most of them aren't sized for hundreds of millions of connections per day.
What are they using to rectify the signal to convert to DC?
Unobtainum diodes. They don't actually know how to do that.
Terahertz diodes do exist. Low-cost, high-efficiency, integrated terahertz diodes, no. But as work proceeds on terahertz electronics, someone may solve that problem. Each nanoantenna needs its own nanodiode, so the diodes have to be fabricated on the substrate with the antenna, which complicates the fab problem. The enthusiasm about roll-to-roll low cost fabrication in the article is premature. We'll probably see this working first on a wafer, and it may not be cheap.
Even if it's expensive, there's an initial market for satellite power panels. The performance improvement would be worth it.
The real deal: MTHEL, from Northrop Grumman.
That's 10 year old technology, and it's a chemical laser. Back then it took three semitrailers for all the support equipment. Since then, electrically-powered lasers are catching up. The Navy Laser Weapons System is not as powerful, but it's a much smaller package, only needs electrical power. and can shoot down small UAVs.
Remote lighting control has been around for decades. X10 has been available for a long time, it's inexpensive, and you can buy the gear at any Home Depot.
The next generation system after that was Echelon LONworks, which is a bidirectional power-line network for home control. That system really does give every device a unique address, set during manufacture, like Ethernet addresses. It's only 78kb/s, but that's enough for lighting control. It never caught on for home control, but it turned out to be useful for subway and railroad trains, because it has extremely good noise immunity. It's used on trains to control lights, HVAC, destination signs, doors, and other auxiliary equipment. Some office buildings use it, and it's sometimes used in semiconductor manufacturing plants.
Full Internet connectivity for each lighting device is a bit much. Do you really want to bridge that data to the outside world?
Bing keeps faithfully replicating Google's bad ideas. When Google merged "places" data from the map search engine into web search results, Bing immediately followed. Google search quality went way down as their "places" system was heavily spammed. Bing had the same problem. Google de-emphasized Places results, but that was subtle and Bing didn't pick up on it. So, right now, Bing's results are about where Google was at their low point in late 2010.
Now Google adds "like" (but not "hate" or "sucks" or "spam"), and so does Bing.
Bing could potentially do better than Google, because Bing doesn't have an AdSense revenue stream to protect. Sites with third-party ads need to be looked at more critically. (We do that with SiteTruth, and it helps.) Google gets 30% of their ad revenue from AdSense ads, and can't afford to raise their standards too high. But Bing's model seems to be to emulate Google, good or bad.
Most likely the investigators will not be listening to it like most people listen to a news broadcast. They will repeatedly be going over every second of the voice recording analyzing every pop, hiss, bang, etc.
For this flight, the flight data is the crucial issue. It's known from the maintenance telemetry that there were some system failures prior to the accident. The data from the aircraft systems is the big issue. It's non-trivial to analyze. Often, especially on newer aircraft where there's a lot of data, the data can be converted into a format that can be loaded into an aircraft simulator, allowing investigators to replay the accident. That was done with the aircraft that landed in the Hudson River.
The cockpit voice recorder is secondary for a loss-of-control accident like this. (In contrast, the voice data is most important for a "controlled flight into terrain" accident, where the airplane was working just fine until it hit something.) The NTSC routinely publishes transcripts of such recordings, and Flying magazine reprints some of them in their Aftermath column.
If you've never seen a full NTSB accident report, here's one.
It's a feed tunnel for the Sir Adam Beck power plant below the falls. It's the third tunnel built for that purpose, and adds 194MW of generating capacity.
There's so much plumbing in place at Niagara Falls that the falls can almost be turned off. There's a minimum water flow over the falls established by international agreement, but that's for aesthetics. At night, and during the tourist off season, more water is run through the hydroelectric plants.
Back in the 1980s, some boater was upstream of the falls, closer than he should have been, and lost power. He managed to run aground upstream of the falls. This was noticed at the Niagara Mohawk power plant control room, where an operator opened all feed and diversion tunnels and closed gates at the upstream weir, shutting off most of the falls until a rescue crew could fetch the boater.
A new public key is usually generated for each transaction, so this doesn't actually tell them anything.
See the Bitcoin paper, page 2, "Transactions": "Each owner transfers the coin to the next by digitally signing a hash of the previous transaction and the public key of the next owner and adding these to the end of the coin." Note, in the diagram, the "verify" link, where the previous owner's public key is transferred to the next owner, so the next owner can check that the previous owner owned the coin. If a party changes their public key, their coin hashes won't match their coins, and the next transaction will fail.
This has the useful property that your "coins" aren't valuable without your private key. If someone copies your coin files, they can't spend them unless they also obtain (i.e. steal) your private key. Conversely, if you lose your private key, your coins become worthless.
Future generations will walk to the rusted ruins of Pad 39A as they walk to the Pyramids today, and wonder at the massive effort required to accomplish such a pointless but impressive task.
This has been done before. See DigiCash, from 1990. "Clouds gather over Amsterdam as I ride into the city center after a day at the headquarters of DigiCash, a company whose mission is to change the world through the introduction of anonymous digital money technology. I have been inundated with talk of smart cards and automated toll takers and tamper-proof observer chips and virtual coinage for anonymous network ftps. I have made photocopies using a digital wallet and would have bought a soda from a DigiCash vending machine, but it was out of order. " - Wired, 1994. See the article for what went wrong.
The soundness of Bitcoin's crypto doesn't seem to have been analyzed by third parties yet. There's nothing in Cryptologia or sci.crypt. Until there's agreement in the crypto community that it's sound, I'd be suspicious. There's also the problem that if the money resides on user PCs and smartphones, the usual attacks on those devices can steal it. Once stolen and used, there's no way to get it back.
Transactions are not very anonymous. If you spend a coin with a server, the server now knows your public key, and can associate it with any other identity information it has for you ( IP address, Facebook login, shipping address, etc.) If Amazon, eBay, Google Checkout, or Facebook accepted bitcoins, they'd be able to collect this info for a sizable fraction of the online world. Since your public key remains associated with the coin for at least the next few transactions, it's possible to follow the money.
Systems like this detect duplicate spending of the same item, but you can't tell if someone has a duplicate but unspent copy of your coins. So you don't know your money been stolen until you try to spend it.
There's also the technical problem that "new transactions are broadcast to all nodes". That won't scale.
It would be useful to have systems which automatically compare news stories on the same subject and note similarities and differences. Osama bin Laden dead? Checking... CNN. Yes. Fox News. Yes. Al-Jazeera - Yes. China Daily - Yes. Russia Today - Yes. Dawn (Pakistan) Yes. Asharq Al-Awsat - Yes. Reuters quote of statement by al-Queda - Yes. Conclusion: dead.
A fantastic summary, but I quibble with the "no evidence of any significant release of radiation" quote for Fukushima. Two months ago, I would have said it was impossible for a reactor in Japan to contaminate the drinking water in Tokyo, but that's exactly what happened. To the detriment of the industry (and I'm a nuclear engineer), there was a significant release of radiation.
Right. The number of casualties is small, but the area evacuated is large, and may be evacuated for decades.
For actuarial purposes, insurance for nuclear plants now has to be repriced. Total power reactor years worldwide is now about 14,000, with two major evacuation incidents. So an assumption of one evacuation of a 30km circle around the plant and acquisition of that real estate per 7000 reactor-years is appropriate for insurance purposes.
The insurance cost will vary with location. That's a big problem. Many US power reactors are sited near major cities. Indian Point in NY is probably the worst case for evacuation cost.
The discouraging thing about the Fukushima reactor disasters is that the real problem was loss of power. The earthquake and tsunami damage was contained. The cooling systems survived and ran until the batteries ran down. With no power, there was little cooling, resulting in hydrogen explosions and meltdowns in several units.
Plants are vulnerable to power loss for a variety of reasons - hurricanes, tornadoes, fires, floods, regional blackouts, poor maintenance, fuel shortages, and sabotage. Most of the NRC literature on loss of coolant accidents focuses on pipe breaks. That may have been the wrong emphasis.