Another is "off chain transactions".... clearing house... That's pretty much what happens in the traditional banking system.
You've now replicated the costs and headaches of the traditional banking system, added on top of the costs and headaches of Bitcoin. The whole point of Bitcoin was supposed to be to eliminate the need for centralized, trusted organizations.
This isn't a "government conspiracy" sending out bogus transactions. It's some jerk.
If you need to sell Bitcoins right now, Coinbase and Kraken are still up and running. Bitstamp is off line, and Mt. Gox is, as usual, screwed up. Mt. Gox hasn't paid out US dollars since June 2013. Whether they are incompetent, broke, or crooked is a subject of considerable speculation.
There's a technical fix in the works, but it will have the annoying side effect that when you spend Bitcoins in your own wallet, some Bitcoins you are not spending will be tied up for an hour or so. Bitcoin wallets don't really have an "account balance". What they have is a collection of items of different values. When you spend Bitcoins, the wallet software tries to put together a set of items that's over the value to be spent, with one output to the recipient and one output ("change") sent back to you.
Until now, you could can spend that "change" immediately, even though the distributed network hadn't yet confirmed it. It looks like that will be disallowed, and only confirmed items will be usable. The way this looks to the user with a wallet program is that you have a "Balance" and an "Unconfirmed" amount. Soon, when you spend, the "Unconfirmed" amount (which you can't spend) will go up for a while, then go to zero when the network catches up. Bitcoin is a distributed "consistent eventually" system. "Eventually" is about an hour. Longer during busy periods. (That's the next Bitcoin problem. The whole network has a limit of about 7 transactions per second. A few times in 2013, that limit was hit.)
Expect everyone except Mt. Gox to have this straightened out in a few days.
The article is just blithering without much useful content. They couldn't even get the right illustration. The steam engine shown is just some random engine with Corliss valve gear. This is the engine that powered much of the 1876 exhibition. It was big, impressive, and inefficient, even for that exhibition.
The "Internet of Things" may be the Next Big Thing from the industry that brought you 3D TV.
This might, marginally, make sense for serious off-roaders, which is what Renault is pitching. There's been a lot of interest in this in the military, where knowing what's over the next hill can save your life. For road driving, it's silly.
That's a perfectly good desktop PC for business. It doesn't need to be set up as a Chromebook. This thing will be powering call centers and other desktops with modest requirements.
You could probably put Windows 7 Embedded (which is simply a version of Windows that lets you make a distro with unwanted features removed) on it.
Most of this should be done with vehicle radars, not WiFi. Vehicles should be using radars to see what the other cars are doing. That works whether the other car has special equipment or not. Short range radars are cheap in volume - every automatic door has one. That can do a lot towards preventing collisions.
Radars can be arranged to talk to each other, by acting as transponders. Because the radar knows where the other party is, it's not too important who the other party is. Useful information to send is "My current turning radius is NNN, my speed is NNN, my acceleration is NNN, transmission in forward gear, turn signals off, emergency flashers off, vehicle has occupants." All of which any human observer can observe now, although not as accurately. It's useful to have an ID, but it doesn't have to be permanent. A new random ID each time the vehicle comes out of Park is good enough for safety purposes.
But no. We're going to get some kludge that reports everything to central control for marketing purposes, and might secondarily be useful for something involving vehicle control.
From memory: Pascal was a language meant for training, not for production use!
Most of the original Macintosh applications were written in a slightly modified version of Pascal. Turbo Pascal had wide usage in the DOS era. But the UNIX compilers for Pascal sucked. The Berkeley one was a hack on the C compiler, and made a subroutine call for every subscript check. So it was too slow to use.
Amusingly, having been taught too much Logic, I think everyone should be taught Statistics.
I've studied basic mathematical logic. Digital logic design and optimization. Proof of correctness. Automatic theorem proving. Constructive mathematics. (Boyer-Moore theory, which is quite elegant.) Expert systems. Even "Dr. John's Mystery Hour", John McCarthy's AI course at Stanford. And I've used all that stuff.
But not in recent years.
Statistics, though, is a win for everybody. Everybody needs enough statistics to calculate the expectations on a Lotto ticket. Programmers need enough statistics to get into machine learning.
In Javascript you can add new methods to objects during runtime...
In Python you can do that from another thread. Not that you'd ever want to. But the code has to be prepared for that. In Javascript, you can tell at compile time if the code is doing that to itself.
The optimization killer in Python is that the compiler can never look at a block of code and decide "X can't happen between these two statements". Because in Python, it can. That kills most loop-related optimizations.
Where are the cyber weapons? Already deployed and awaiting activation. Undocumented errata in major CPUs which allow bypassing memory protection. Preset keys in network cards allowing remote administration. Undocumented admin passwords in network firmware. Code signing certs in the hands of intelligence agencies. That's where.
Arguably, it should have been Python, which is a better language. But Python has a problem. Python's little tin god, Guido von Rossum, is in love with his own implementation, CPython. CPython is a naive interpreter. (A "naive interpreter" is one which does the specified operations in the specified order, with little optimization across operations.) In CPython, everything is a dictionary and a lot of time is spent doing lookups. This allows everything to be dynamic. In Python, one thread can patch objects in another thread while both are running. Objects can gain or lose elements on the fly. Even running code can be "monkey-patched".
The price paid for that excessive dynamism is that a Python compiler is hard to write, and an optimizing Python compiler can't optiimize much. Google tried in-house to make Python faster, and their "Unladed Swallow" failed humililatingly. (A different group at Google then developed Go, aimed at the same problem of producing something good for server-side processing.) The PyPy crowd has tried, hard, to make an optimizing Python compiler, and with an incredible amount of complexity under the hood, has made considerable progress, but not enough that PyPy is used much in production.
Pascal went down for a similar reason. Wirth was in love with his elegant recursive-descent compiler. But it didn't optimize, couldn't handle separate compilation, and had no way to handle errors other than aborting. Python seems to be headed for similar irrelevance. It hasn't even been able to replace Perl, which ought to be as marginal as "awk" by now.
The article says "factories are static". In reality, only factories for long production runs are static. Many factories are constantly setting up production lines, running them for weeks or months, then reconfiguring for a different product. This is expensive, slow, and often requires a completely different workforce than the one used during operation.
The ability to set up a production line with robots would be useful. One reason that production lines for smartphones are so manual is that the product life cycle is so short. The production engineering and plant setup time is long for a robotic assembly line. Faster line changeover would be a big win for fast-changing product lines.
One of the better robotics ideas of the 1980s was a pair of small, cooperating forklift-type robots. A pair of these could pick up and move a couch much bigger than the robots. It's time to revive that idea.
Nadella used to run Bing. Bing had a leadership vacuum after he left (and still does), but it didn't do all that great while he was there.
Microsoft's approach to Bing upper management is very strange. Microsoft sends people there, but you never hear about them while they're there. You hear about them after they're promoted to better parts of Microsoft. Mark Penn was brought in to turn around Bing, and accomplished little there. Now he heads Microsoft Advertising. Qi Lu ran Bing for a while, and now he's head of Applications and Services. So failing to turn Bing around doesn't seem to hurt executive careers at Microsoft.
What some employers want are Stakhanovites. Most don't really want "failure is not an option" types, who want off-site backup systems, fail-soft recovery, a Q/A organzation with the authority to delay deployment, expensive testing tools, automatic code analyzers... Most don't want programmers who polish their code until it's beautiful, like MIT students are taught.
Yes, there's a point in using a lawyer when $50K is involved. You don't just use lawyers for litigation. They're also useful in negotiation. If you put a lawyer on the job, the other side will usually put their lawyer on the job. This totally bypasses the customer service level and makes things happen.
I'd been looking into this in a slightly different context. Recently, at Hacker Dojo, someone demonstrated an Android mod to me which dealt with applications that demand too many privileges. It has the usual "disable privileges" option, but for apps that won't run with privileges disabled, it sends fake info.
The demo showed generation of fake phone serial numbers and such. That's easy. Apps that improperly try to upload your address book, though, require generation of a plausible, but fake, address book. That's wasn't in the demo, but it's worth doing.
Location data should probably be sent as a random walk from some random starting point.
If enough people do this, it will garbage marketing databases.
He's in Pasedena, California. Twitter is in San Francisco. It's a cheap flight. He needs to get a lawyer in SF, and make an appointment with Twitter's general counsel. Bring birth certificate and passport. If Twitter then fails to return the handle, use phrases such as "complicit with extortion".
" Indeed. We don't mean "common sense" from a Marvin Minsky-like strong AI perspective. Baxter's "execution" application consists of a series of behavior-based systems. During "training," the robot detects task-relevant features and uses it to build up the behavior based system.
For example, let's say a user is training the robot for a pick and place task. During the "pick" phase, a user places the gripper above an object and closes the gripper. The force on the gripper is detected by the robot. Our "training" application detects this sequence as "the robot is grasping an object"... so during "execution", Baxter won't proceed unless it actually detects an object in the robots gripper. Thus, if the object fell out, it would stop (or do something else). This is different from how existing industrial robots work -- they'd just merrily continue the pick-and-place without the object.
Collectively, these "behavior primitives" are assigned and composed, ie. "learned", during "training" by having non-technical users directly manipulate the robot rather than programming it (which is also possible for those inclined). This gives the robot an air of common sense."
This is useful, but not that intelligent. Take a look at these PR videos to see what it can do. Basically, it can pack and unpack things, and move them from one place to another. It's not good enough to assemble much of anything. Plugging in connectors to assemble a phone? Not with this machine and software.
Mod parent up. A U.S. Senate commitee held a hearing on Bitcoin a few months ago. All the regulatory agencies were there.
FinCen: We're watching it, no big deal. DEA: the big cartels aren't using it, the street dealers aren't using it. FBI: We took down Silk Road, which was using Bitcoin, and Bitcoin didn't make it harder to do that. Secret Service: No big deal. Homeland Security: terrorists don't seem to be using it. IRS: Taxable income is taxable income; we'll deal with it. SEC: Trading Bitcoin is just like trading anything else. We busted one guy running a Ponzi scheme with Bitcoins and the judge agreed that using Bitcoins didn't make it special.
Conclusion of the Senate committee: no need for special Bitcoin legislation.
All the US law enforcement action involving Bitcoins has been for doing routine crook stuff. Now, China is cracking down on Bitcoins, but they have exchange controls; you have to get permission to swap yuan for dollars or euros. The US has an open market in foreign currencies; you can swap dollars for yen without asking anyone's permission. There are reports to make to FinCen, but they just log the info.
This is about normal for Bitcoin exchanges. All of them are flaky. Most of them don't even have a legit business address. About half of all Bitcoin "exchanges" so far have failed. The problem is that they're not just exchanges, they're also deposit-taking institutions holding customer funds. With no regulation. A sizable number of Bitcoin "exchanges" just took the money and ran.
With the arrest of the CEO, "bitinstant.com" dropped offline.
Right now, the formerly biggest Bitcoin exchange, Mt. Gox, in Tokyo, is tanking. They stopped US dollar withdrawals back in June 2013, then EUR withdrawals slowed down, then JPY withdrawals, and now Bitcoin withdrawals. There's much discussion over whether they're broke, crooked, or merely incompetent. Right now, Bitcoins on Mt. Gox are priced 25% higher than on the other exchanges, because if you sell Bitcoins on Mt. Gox, you can't get the money out, and that spread has been climbing rapidly for the last few days.
Irrevocable anonymous remote transactions are the scammer's dream. Scammers can rip people off without ever meeting the marks and with a low chance of getting caught. That's Bitcoin's big problem. It takes ten honest people to support one crook, and Bitcoin's ratio of crooks is much higher than that.
I visited Bletchley Park in 2002, when it was an all-volunteer operation with limited funding. It was great seeing the bombe rebuild, the unfinished Colossus rebuild, a Lorenz crypto machine, and a working Enigma. But I knew about all those machines and what Bletchley Park had done. For people who hadn't done the reading, it wasn't much of an experience. The guide was more into the architecture of the manor house than the crypto anyway. There were maybe 20 visitors on the grounds when I was there.
The Science Museum in London has been dumbed down. I saw it in 1985 and 2002. The big thing in 2002 was the Aston-Martin from an early James Bond movie. Some of the railroad equipment had been moved out. But they still had Maudsley's lathe, which looks amazingly like a modern lathe, but complely different than any lathe before it.
Ah, the "markets will fix everything" (but didn't read the linked article) types.
Wholesale electric prices did rise, to about $950/MWH, for about half an hour around 5 AM EST this morning. That didn't last long. It's now around $150/MWH. The price goes up and down by a factor of 3 or so in a normal 24 hour cycle.
There's hedging going on in power, natural gas, and weather. But it doesn't affect the amount of generating capacity online on an hour by hour basis.
Read PJM 101 to understand how this works. Electric power in the PJM region is normally driven entirely by markets. However, PJM grid control in Valley Forge, PA can order "non-market actions" to keep power on, and generating companies (which are not all utilities) are obligated by their contracts with PJM to obey those instructions or pay huge penalties. PJM doesn't do this often. Yesterday and the day before, though, were bad days. Both days, there were Max Emergency Generation alerts . The longest was from 19:19 EST on Thursday to 08:45 Friday. That's because some generating capacity was down, and peaking plants had to be used to make up capacity. That's part of what peaking plants are for.
Wind power didn't help. Wind power was at a low when power was most needed. Even with wind farms spread over many states, wind power in the PJM area goes up and down over a 4x range.
(Sometimes power is really cheap. The price can even go negative. Load varies over about a 3x range during a normal day, and around 2-5 AM, it's at minimum. All the plants that burn fuel shut down first. Much of PJM's power comes from Ontario Hydro, and when they have too much water in their reservoirs, they have to let some out through their generators. So they continue to produce power even if the price they're being paid briefly goes below 0. Adjusting the output of nuclear plants is slow, and they'll also sometimes generate even if it costs them. The wind farms usually prefer to shut down rather than pay, and so, late at night, sometimes the giant wind turbines feather their props and slow to a stop.)
Slashdot Mirror
Another is "off chain transactions". ... clearing house ... That's pretty much what happens in the traditional banking system.
You've now replicated the costs and headaches of the traditional banking system, added on top of the costs and headaches of Bitcoin. The whole point of Bitcoin was supposed to be to eliminate the need for centralized, trusted organizations.
This isn't a "government conspiracy" sending out bogus transactions. It's some jerk.
If you need to sell Bitcoins right now, Coinbase and Kraken are still up and running. Bitstamp is off line, and Mt. Gox is, as usual, screwed up. Mt. Gox hasn't paid out US dollars since June 2013. Whether they are incompetent, broke, or crooked is a subject of considerable speculation.
There's a technical fix in the works, but it will have the annoying side effect that when you spend Bitcoins in your own wallet, some Bitcoins you are not spending will be tied up for an hour or so. Bitcoin wallets don't really have an "account balance". What they have is a collection of items of different values. When you spend Bitcoins, the wallet software tries to put together a set of items that's over the value to be spent, with one output to the recipient and one output ("change") sent back to you.
Until now, you could can spend that "change" immediately, even though the distributed network hadn't yet confirmed it. It looks like that will be disallowed, and only confirmed items will be usable. The way this looks to the user with a wallet program is that you have a "Balance" and an "Unconfirmed" amount. Soon, when you spend, the "Unconfirmed" amount (which you can't spend) will go up for a while, then go to zero when the network catches up. Bitcoin is a distributed "consistent eventually" system. "Eventually" is about an hour. Longer during busy periods. (That's the next Bitcoin problem. The whole network has a limit of about 7 transactions per second. A few times in 2013, that limit was hit.)
Expect everyone except Mt. Gox to have this straightened out in a few days.
The article is just blithering without much useful content. They couldn't even get the right illustration. The steam engine shown is just some random engine with Corliss valve gear. This is the engine that powered much of the 1876 exhibition. It was big, impressive, and inefficient, even for that exhibition.
The "Internet of Things" may be the Next Big Thing from the industry that brought you 3D TV.
This might, marginally, make sense for serious off-roaders, which is what Renault is pitching. There's been a lot of interest in this in the military, where knowing what's over the next hill can save your life. For road driving, it's silly.
The machine at Bletchley Park is a working replica, not the original.
That's a perfectly good desktop PC for business. It doesn't need to be set up as a Chromebook. This thing will be powering call centers and other desktops with modest requirements.
You could probably put Windows 7 Embedded (which is simply a version of Windows that lets you make a distro with unwanted features removed) on it.
Most of this should be done with vehicle radars, not WiFi. Vehicles should be using radars to see what the other cars are doing. That works whether the other car has special equipment or not. Short range radars are cheap in volume - every automatic door has one. That can do a lot towards preventing collisions.
Radars can be arranged to talk to each other, by acting as transponders. Because the radar knows where the other party is, it's not too important who the other party is. Useful information to send is "My current turning radius is NNN, my speed is NNN, my acceleration is NNN, transmission in forward gear, turn signals off, emergency flashers off, vehicle has occupants." All of which any human observer can observe now, although not as accurately. It's useful to have an ID, but it doesn't have to be permanent. A new random ID each time the vehicle comes out of Park is good enough for safety purposes.
But no. We're going to get some kludge that reports everything to central control for marketing purposes, and might secondarily be useful for something involving vehicle control.
From memory: Pascal was a language meant for training, not for production use!
Most of the original Macintosh applications were written in a slightly modified version of Pascal. Turbo Pascal had wide usage in the DOS era. But the UNIX compilers for Pascal sucked. The Berkeley one was a hack on the C compiler, and made a subroutine call for every subscript check. So it was too slow to use.
Soon, there will be other heads-up displays. This is one of the more useful applications for them. I'm looking forward to seeing how well it works.
Everyone should be taught Logic.
Amusingly, having been taught too much Logic, I think everyone should be taught Statistics.
I've studied basic mathematical logic. Digital logic design and optimization. Proof of correctness. Automatic theorem proving. Constructive mathematics. (Boyer-Moore theory, which is quite elegant.) Expert systems. Even "Dr. John's Mystery Hour", John McCarthy's AI course at Stanford. And I've used all that stuff. But not in recent years.
Statistics, though, is a win for everybody. Everybody needs enough statistics to calculate the expectations on a Lotto ticket. Programmers need enough statistics to get into machine learning.
Are things different now?
No, VOIP still sucks. Cellular sucks. Cellular plus VOIP really sucks. Lags as high as 1 second.
Telephony has gone from "You can hear a pin drop" to "Can you hear me now?".
In Javascript you can add new methods to objects during runtime...
In Python you can do that from another thread. Not that you'd ever want to. But the code has to be prepared for that. In Javascript, you can tell at compile time if the code is doing that to itself.
The optimization killer in Python is that the compiler can never look at a block of code and decide "X can't happen between these two statements". Because in Python, it can. That kills most loop-related optimizations.
Where are the cyber weapons? Already deployed and awaiting activation. Undocumented errata in major CPUs which allow bypassing memory protection. Preset keys in network cards allowing remote administration. Undocumented admin passwords in network firmware. Code signing certs in the hands of intelligence agencies. That's where.
Arguably, it should have been Python, which is a better language. But Python has a problem. Python's little tin god, Guido von Rossum, is in love with his own implementation, CPython. CPython is a naive interpreter. (A "naive interpreter" is one which does the specified operations in the specified order, with little optimization across operations.) In CPython, everything is a dictionary and a lot of time is spent doing lookups. This allows everything to be dynamic. In Python, one thread can patch objects in another thread while both are running. Objects can gain or lose elements on the fly. Even running code can be "monkey-patched".
The price paid for that excessive dynamism is that a Python compiler is hard to write, and an optimizing Python compiler can't optiimize much. Google tried in-house to make Python faster, and their "Unladed Swallow" failed humililatingly. (A different group at Google then developed Go, aimed at the same problem of producing something good for server-side processing.) The PyPy crowd has tried, hard, to make an optimizing Python compiler, and with an incredible amount of complexity under the hood, has made considerable progress, but not enough that PyPy is used much in production.
Pascal went down for a similar reason. Wirth was in love with his elegant recursive-descent compiler. But it didn't optimize, couldn't handle separate compilation, and had no way to handle errors other than aborting. Python seems to be headed for similar irrelevance. It hasn't even been able to replace Perl, which ought to be as marginal as "awk" by now.
The article says "factories are static". In reality, only factories for long production runs are static. Many factories are constantly setting up production lines, running them for weeks or months, then reconfiguring for a different product. This is expensive, slow, and often requires a completely different workforce than the one used during operation.
The ability to set up a production line with robots would be useful. One reason that production lines for smartphones are so manual is that the product life cycle is so short. The production engineering and plant setup time is long for a robotic assembly line. Faster line changeover would be a big win for fast-changing product lines.
One of the better robotics ideas of the 1980s was a pair of small, cooperating forklift-type robots. A pair of these could pick up and move a couch much bigger than the robots. It's time to revive that idea.
Nadella used to run Bing. Bing had a leadership vacuum after he left (and still does), but it didn't do all that great while he was there.
Microsoft's approach to Bing upper management is very strange. Microsoft sends people there, but you never hear about them while they're there. You hear about them after they're promoted to better parts of Microsoft. Mark Penn was brought in to turn around Bing, and accomplished little there. Now he heads Microsoft Advertising. Qi Lu ran Bing for a while, and now he's head of Applications and Services. So failing to turn Bing around doesn't seem to hurt executive careers at Microsoft.
What some employers want are Stakhanovites. Most don't really want "failure is not an option" types, who want off-site backup systems, fail-soft recovery, a Q/A organzation with the authority to delay deployment, expensive testing tools, automatic code analyzers... Most don't want programmers who polish their code until it's beautiful, like MIT students are taught.
Yes, there's a point in using a lawyer when $50K is involved. You don't just use lawyers for litigation. They're also useful in negotiation. If you put a lawyer on the job, the other side will usually put their lawyer on the job. This totally bypasses the customer service level and makes things happen.
I'd been looking into this in a slightly different context. Recently, at Hacker Dojo, someone demonstrated an Android mod to me which dealt with applications that demand too many privileges. It has the usual "disable privileges" option, but for apps that won't run with privileges disabled, it sends fake info.
The demo showed generation of fake phone serial numbers and such. That's easy. Apps that improperly try to upload your address book, though, require generation of a plausible, but fake, address book. That's wasn't in the demo, but it's worth doing. Location data should probably be sent as a random walk from some random starting point.
If enough people do this, it will garbage marketing databases.
He's in Pasedena, California. Twitter is in San Francisco. It's a cheap flight. He needs to get a lawyer in SF, and make an appointment with Twitter's general counsel. Bring birth certificate and passport. If Twitter then fails to return the handle, use phrases such as "complicit with extortion".
The 2012 interview was more informative:
" Indeed. We don't mean "common sense" from a Marvin Minsky-like strong AI perspective. Baxter's "execution" application consists of a series of behavior-based systems. During "training," the robot detects task-relevant features and uses it to build up the behavior based system.
For example, let's say a user is training the robot for a pick and place task. During the "pick" phase, a user places the gripper above an object and closes the gripper. The force on the gripper is detected by the robot. Our "training" application detects this sequence as "the robot is grasping an object"... so during "execution", Baxter won't proceed unless it actually detects an object in the robots gripper. Thus, if the object fell out, it would stop (or do something else). This is different from how existing industrial robots work -- they'd just merrily continue the pick-and-place without the object.
Collectively, these "behavior primitives" are assigned and composed, ie. "learned", during "training" by having non-technical users directly manipulate the robot rather than programming it (which is also possible for those inclined). This gives the robot an air of common sense."
This is useful, but not that intelligent. Take a look at these PR videos to see what it can do. Basically, it can pack and unpack things, and move them from one place to another. It's not good enough to assemble much of anything. Plugging in connectors to assemble a phone? Not with this machine and software.
Mod parent up. A U.S. Senate commitee held a hearing on Bitcoin a few months ago. All the regulatory agencies were there.
FinCen: We're watching it, no big deal.
DEA: the big cartels aren't using it, the street dealers aren't using it.
FBI: We took down Silk Road, which was using Bitcoin, and Bitcoin didn't make it harder to do that.
Secret Service: No big deal.
Homeland Security: terrorists don't seem to be using it.
IRS: Taxable income is taxable income; we'll deal with it.
SEC: Trading Bitcoin is just like trading anything else. We busted one guy running a Ponzi scheme with Bitcoins and the judge agreed that using Bitcoins didn't make it special.
Conclusion of the Senate committee: no need for special Bitcoin legislation.
All the US law enforcement action involving Bitcoins has been for doing routine crook stuff. Now, China is cracking down on Bitcoins, but they have exchange controls; you have to get permission to swap yuan for dollars or euros. The US has an open market in foreign currencies; you can swap dollars for yen without asking anyone's permission. There are reports to make to FinCen, but they just log the info.
This is about normal for Bitcoin exchanges. All of them are flaky. Most of them don't even have a legit business address. About half of all Bitcoin "exchanges" so far have failed. The problem is that they're not just exchanges, they're also deposit-taking institutions holding customer funds. With no regulation. A sizable number of Bitcoin "exchanges" just took the money and ran. With the arrest of the CEO, "bitinstant.com" dropped offline.
Right now, the formerly biggest Bitcoin exchange, Mt. Gox, in Tokyo, is tanking. They stopped US dollar withdrawals back in June 2013, then EUR withdrawals slowed down, then JPY withdrawals, and now Bitcoin withdrawals. There's much discussion over whether they're broke, crooked, or merely incompetent. Right now, Bitcoins on Mt. Gox are priced 25% higher than on the other exchanges, because if you sell Bitcoins on Mt. Gox, you can't get the money out, and that spread has been climbing rapidly for the last few days.
Irrevocable anonymous remote transactions are the scammer's dream. Scammers can rip people off without ever meeting the marks and with a low chance of getting caught. That's Bitcoin's big problem. It takes ten honest people to support one crook, and Bitcoin's ratio of crooks is much higher than that.
I visited Bletchley Park in 2002, when it was an all-volunteer operation with limited funding. It was great seeing the bombe rebuild, the unfinished Colossus rebuild, a Lorenz crypto machine, and a working Enigma. But I knew about all those machines and what Bletchley Park had done. For people who hadn't done the reading, it wasn't much of an experience. The guide was more into the architecture of the manor house than the crypto anyway. There were maybe 20 visitors on the grounds when I was there.
The Science Museum in London has been dumbed down. I saw it in 1985 and 2002. The big thing in 2002 was the Aston-Martin from an early James Bond movie. Some of the railroad equipment had been moved out. But they still had Maudsley's lathe, which looks amazingly like a modern lathe, but complely different than any lathe before it.
Ah, the "markets will fix everything" (but didn't read the linked article) types.
Wholesale electric prices did rise, to about $950/MWH, for about half an hour around 5 AM EST this morning. That didn't last long. It's now around $150/MWH. The price goes up and down by a factor of 3 or so in a normal 24 hour cycle.
There's hedging going on in power, natural gas, and weather. But it doesn't affect the amount of generating capacity online on an hour by hour basis.
Read PJM 101 to understand how this works. Electric power in the PJM region is normally driven entirely by markets. However, PJM grid control in Valley Forge, PA can order "non-market actions" to keep power on, and generating companies (which are not all utilities) are obligated by their contracts with PJM to obey those instructions or pay huge penalties. PJM doesn't do this often. Yesterday and the day before, though, were bad days. Both days, there were Max Emergency Generation alerts . The longest was from 19:19 EST on Thursday to 08:45 Friday. That's because some generating capacity was down, and peaking plants had to be used to make up capacity. That's part of what peaking plants are for.
Wind power didn't help. Wind power was at a low when power was most needed. Even with wind farms spread over many states, wind power in the PJM area goes up and down over a 4x range.
(Sometimes power is really cheap. The price can even go negative. Load varies over about a 3x range during a normal day, and around 2-5 AM, it's at minimum. All the plants that burn fuel shut down first. Much of PJM's power comes from Ontario Hydro, and when they have too much water in their reservoirs, they have to let some out through their generators. So they continue to produce power even if the price they're being paid briefly goes below 0. Adjusting the output of nuclear plants is slow, and they'll also sometimes generate even if it costs them. The wind farms usually prefer to shut down rather than pay, and so, late at night, sometimes the giant wind turbines feather their props and slow to a stop.)