Several years ago, I wrote a simple point-point VOIP thingy.
I would not install skype on a system I want to be secure, because I do not know what it does.
The datastream format is not documented.
The code seems to have been unlikely to be reviewed by a third party for security
The encrypted aspect means that it's impossible for any admin to quickly discover if a copy of skype is operating normally, or if it's been hacked, and is reporting home to the hacker.
The widespread nature of skype, and its list of 'phone numbers' means a skype virus could spread very, very rapidly.
It's not only electricity, to do significant damage to carbon emissions, you need to go further.
From memory.
About 80% of home and buisness heating in the UK has converted to gas.
Essentially all of the UK use of gas, apart from a tiny fraction that is used as chemical feedstocks could be converted over to electricity.
The remaining long pole in the tent is transport use.
Some of this - trains - can be electrified without impossible hastle.
There are options to make cars nuclear powered - you use the heat from a nuclear power station to make hydrogen. But this does mean replacing the entire fleet of cars/...
The other point is that 'nuclear is more expensive'.
While this may be true (and I have doubts, if a significant production line was instituted), it's not the whole argument
A significant factor is that with gas/oil/..., you are paying well over half the money you spend on energy to the Sultan of Bling, meaning you need to export more stuff to pay for it.
With nuclear, you are paying the money to local people, who shop at local buisnesses, contribute taxes,...
Even 'just' substituting gas, would however mean an unprecedented (well, perhaps apart from china) level of construction, ~50 plants, and massive upgrades of the national grid, to cope with an order of magnitude increase in demand.
It's about a hundred years out, and, if it's aimed dead-on, then it needs moved about 10000Km, to be 'safe'.
80 years (say a 20 year lead time), that's about 120Km/year, or 4mm/s.
It has a mass of 3*10^11Kg, so would need an impulse of 1.2GN m/s.
Or, a thrust of about 40N (4Kg) over 1 year.
At an ISP of 10000 (typical of some ion engines), that's around 15 tons of propellant, and power use of some 40Kw.
Assuming an average radius of 2 AU, that's under 5 tons of solar panels.
This can be done with todays technology.
The alternative is not to drill into it and nuke it, which won't really work, but to detonate large nuclear weapons at some 600m from the surface.
The intense x-ray radiation from the bomb is absorbed in the top few centimeters of the body. This heats it to beyond its boiling point, and it explodes outwards, pushing the rock the other way.
(round numbers, I haven't checked very thouroughly)
If you want, you can do 'pickup only' sales on ebay.
Not to mention that I'd much rather someone cheats me out of the money, than has a friend with a weapon hanging around the pickup site, to relieve me of the item, and whatever else.
At the moment, ebay have essentially no competition.
This does not make for a healthy marketplace.
There is of course competition between sellers, but if ebay raises prices, makes it impossible to find items by completely eliminating categories, or decides that it'd rather heavily weight the market towards those who pay for featured ads, the users have no comeback, other than to not use ebay.
They can't just go to the 2nd highest auction site in many cases, as there effectively isn't one.
I have no problem with this - as long as it's sturdy enough for a ~100Kg person to sit on, or fall over, without it ending up as a Rorschach Blob, like many displays are prone to do.
Superconductors: In addition to not being available, these make the structure more complex, requiring insulation in addition to the superconductor.
At the moment, not possible, as it will make the cable significantly heavier.
Copper: (actually, sodium has best weight/conductivity) Basically, never. All the problems of superconductors, and in addition, huge resistance.
Solar power: 183w/Kg solar cells have been made, better presumably can be. Problems are that you've got to stow them for the first 20Km or so, and then unfold, you need LARGE arrays (1 ton raised 50m/s requires half a megawatt, or maybe 700 square meters), and they only work during the day (which may cause resonance problems)
Laser driven solar: much more compact arrays are possible, as it's a point source, and can be more optically intense. Optically, it's not that much of a problem, at 20000Km, a 20cm lens will project a beam about 50m wide.
You'll need lots of them, positioned in different locations so that clouds aren't a problem, but the lasers are more or less 50% efficient, and available off the shelf.
Chemical: Worse than copper, you can get up a few hundred Km, but not much more.
Nuclear: at this sort of power level, (low megawatts) there is no nice simple solution that's at all lightweight. (or that will even lift itself)
Microwave driven: Similar to laser-driven-solar, but the transmitters and recirvers have to be thousands of times larger.
To wrap up, laser-driven-solar is "least bad", and is actually not the hardest part to resolve with todays technology.
Active tags - ones with their own battery, are going to be fundamentally immune to this.
Also, in addition to tags that have a simple 'password', that they must have before they do anything - that may be trivially vulnerable to power analysis, there are tags that do more complex things - such as for example, send the reader a random token, which it then has to encrypt with a key known to both of them.
This can be immune to power analysis - in the simplest case, as it does not check each bit as recieved, but only at the end of a computation.
And, the fact that getting the first bit correct of a hash with a given key does not help you to guess the rest.
You do not need a license for "live or nearly live".
You need a license for recieving "television programs".
Looking further up the chain, into the broadcasting legislation, which defines "television programme", it's that broadcast by a "television programme service". http://www.opsi.gov.uk/acts/acts1996/1996055.htm
This is not a remote PC, sending you data, whether or not that data is sourced off-air.
The transcoder would, as I understand it though require a license.
There are dumb tags, that just spit out their serial number on being powered.
Then, you can spend the extra $2/tag or so, and get ones with internal writable storage, and SHA or other hashing algorithms in.
Personally - I'd be quite happy with an implanted tag - as long as I was able to turn off and on the contained authorisation tokens (those with the correct keys would be able to delete, but not enable tokens), and could add new ones of my own, for car, laptop, phone,...
Unfortunately, the standards for this are not there yet.
Not to mention I'd want full medical cover, and life insurance of say $10M, for any complications due to the tag.
That's the problem of course, you're limited at best to the bandwidth of the average upstream, minus a bit.
TCP acks are not a problem - you roll your own protocol.
You're probably going to want to do forward error correction, dedicationg 5% or so of the stream to error correction, rather than wasting the bandwidth the other way with acks.
OTOH, in some cases, such as radio, 128K is just fine, and is a lot better than 44K that some stations webcast in noe.
I've found three seperate devices of this sort, that are significantly inaccirate on power factor corrected switch mode supplies.
Compared with measuring the actual power on a scope, it was overestimating the power 100%.
This was with a server in a cupboard.
I tested that the scope measurement was real by turning the server off, leaving the fans on, and comparing the temperature of the outgoing air with 150W of ordinary light bulbs.
It was reading about 150W, with one drive, one duron 1800 processor.
I think this is as they sample too slowly, they are accurate for non power-factor-corrected power supplies (linear and SMPS) as well as motors, heaters,...
Unfortunately not. This is not that much more than Apollo. A few people on the moon, for a bit longer than apollo.
No manufacturing things from lunar resources, or long-term bases at all. More up to date sensors on the experiments of course, which is nice. No significant development of anything that will make spaceflight cheaper. It's been $10K/lb (approximately) since the end of apollo, and it still is.
There are 20 million journeys in progress at any time, at an average speed of 30mph.
Assuming all cars pass cameras every 5 miles (which is obviously incorrect, but probably not horribly so), and 20 million are driving at once
A car will pass a camera every 600 seconds.
Or all cars will pass some camera every 30us.
Let's assume the database of images problem is seperate - we're working on 8 char text strings (16 bit timestamp, 16 bit camera, remainder for plate
30000*8 = 240K/second.
Now, assume that there are 40 million vehicles in total, driven on average 100 miles, for a total of 4 billion miles, or a little under a billion camera events.
So, we're looking at around 10Gb/day, so a day is storable in RAM on a high end system.
Ok, so a number comes in, and we've got a camera number, plate ID pair. So, we now need to look up where the plate was last, and generate an exception if it's been going >77MPH (to reduce false hits, but once you've got most of them out, a few fast moving vehicles are not hard to prune).
We 'only' for each vehicle need to store the last position, let's say to the nearest.1 mile, which is maybe 20 bits, and a 32 bit timestamp, so call a record 16 bytes, in a hash table with half occupancy, for a total amount of information oif only 1.2Gb.
Now, we need to look up the old position in a hash table, compute the speed, throw an exception if it's too fast, and rewrite, all in 30us.
30us is a long, long time for modern computers
So, it looks like a single cheap PC can do the work - and that's without trivially obvious tricks such as splitting the job by first/last letter.
Of course, you'd want to split it up for other reasons, but realtime isn't a problem.
Then you'd want a seperate system that spat out pairs of images at operators, to see if it was a real or imagined match.
Slashdot Mirror
I would not install skype on a system I want to be secure, because I do not know what it does.
The datastream format is not documented.
The code seems to have been unlikely to be reviewed by a third party for security
The encrypted aspect means that it's impossible for any admin to quickly discover if a copy of skype is operating normally, or if it's been hacked, and is reporting home to the hacker.
The widespread nature of skype, and its list of 'phone numbers' means a skype virus could spread very, very rapidly.
More like 3.7 pounds (486/75 (DEC hinote ultra CT475), with 3 hour + battery life.
Laptops at the ultraportable end have not got appreciably lighter, or with more battery life.
The screens have gotten a little bigger, and more stuff has gotten built in, and they've gotten cheaper - the above laptop was $5000.
From memory.
About 80% of home and buisness heating in the UK has converted to gas.
Essentially all of the UK use of gas, apart from a tiny fraction that is used as chemical feedstocks could be converted over to electricity.
The remaining long pole in the tent is transport use.
Some of this - trains - can be electrified without impossible hastle.
There are options to make cars nuclear powered - you use the heat from a nuclear power station to make hydrogen. But this does mean replacing the entire fleet of cars/...
The other point is that 'nuclear is more expensive'.
While this may be true (and I have doubts, if a significant production line was instituted), it's not the whole argument
A significant factor is that with gas/oil/..., you are paying well over half the money you spend on energy to the Sultan of Bling, meaning you need to export more stuff to pay for it.
With nuclear, you are paying the money to local people, who shop at local buisnesses, contribute taxes, ...
Even 'just' substituting gas, would however mean an unprecedented (well, perhaps apart from china) level of construction, ~50 plants, and massive upgrades of the national grid, to cope with an order of magnitude increase in demand.
It's about a hundred years out, and, if it's aimed dead-on, then it needs moved about 10000Km, to be 'safe'.
80 years (say a 20 year lead time), that's about 120Km/year, or 4mm/s.
It has a mass of 3*10^11Kg, so would need an impulse of 1.2GN m/s.
Or, a thrust of about 40N (4Kg) over 1 year.
At an ISP of 10000 (typical of some ion engines), that's around 15 tons of propellant, and power use of some 40Kw.
Assuming an average radius of 2 AU, that's under 5 tons of solar panels.
This can be done with todays technology.
The alternative is not to drill into it and nuke it, which won't really work, but to detonate large nuclear weapons at some 600m from the surface.
The intense x-ray radiation from the bomb is absorbed in the top few centimeters of the body. This heats it to beyond its boiling point, and it explodes outwards, pushing the rock the other way.
(round numbers, I haven't checked very thouroughly)
http://www.bose.com/controller?event=VIEW_STATIC_P AGE_EVENT&url=/learning/project_sound/bose_suspens ion.jsp&ck=0 Is a similar idea from bose, but improved suspensions, with lower road noise, rather than military apps.
Tablet PCs, ubid - 9 auctions from 1 seller, ebay - 549 items, yahoo - 15 items from 4 sellers
Craigslist has 114 items for sale.
Ebay has 142203.
If you want, you can do 'pickup only' sales on ebay.
Not to mention that I'd much rather someone cheats me out of the money, than has a friend with a weapon hanging around the pickup site, to relieve me of the item, and whatever else.
This does not make for a healthy marketplace.
There is of course competition between sellers, but if ebay raises prices, makes it impossible to find items by completely eliminating categories, or decides that it'd rather heavily weight the market towards those who pay for featured ads, the users have no comeback, other than to not use ebay.
They can't just go to the 2nd highest auction site in many cases, as there effectively isn't one.
Competition would greatly help users.
I have no problem with this - as long as it's sturdy enough for a ~100Kg person to sit on, or fall over, without it ending up as a Rorschach Blob, like many displays are prone to do.
Superconductors: In addition to not being available, these make the structure more complex, requiring insulation in addition to the superconductor. At the moment, not possible, as it will make the cable significantly heavier.
Copper: (actually, sodium has best weight/conductivity) Basically, never. All the problems of superconductors, and in addition, huge resistance.
Solar power: 183w/Kg solar cells have been made, better presumably can be. Problems are that you've got to stow them for the first 20Km or so, and then unfold, you need LARGE arrays (1 ton raised 50m/s requires half a megawatt, or maybe 700 square meters), and they only work during the day (which may cause resonance problems)
Laser driven solar: much more compact arrays are possible, as it's a point source, and can be more optically intense. Optically, it's not that much of a problem, at 20000Km, a 20cm lens will project a beam about 50m wide. You'll need lots of them, positioned in different locations so that clouds aren't a problem, but the lasers are more or less 50% efficient, and available off the shelf.
Chemical: Worse than copper, you can get up a few hundred Km, but not much more.
Nuclear: at this sort of power level, (low megawatts) there is no nice simple solution that's at all lightweight. (or that will even lift itself)
Microwave driven: Similar to laser-driven-solar, but the transmitters and recirvers have to be thousands of times larger.
To wrap up, laser-driven-solar is "least bad", and is actually not the hardest part to resolve with todays technology.
Also, in addition to tags that have a simple 'password', that they must have before they do anything - that may be trivially vulnerable to power analysis, there are tags that do more complex things - such as for example, send the reader a random token, which it then has to encrypt with a key known to both of them.
This can be immune to power analysis - in the simplest case, as it does not check each bit as recieved, but only at the end of a computation.
And, the fact that getting the first bit correct of a hash with a given key does not help you to guess the rest.
I first read that as
I'd imagine room temperature superconductor plates on the fingertips of gloves.
These are repelled by an array of electromagnets in the screen, to give tactility.
Not tomorrow.
Card contains biometric information.
The only time a central database is checked is when issuing cards, to ensure you get only one per individual.
The card contains a means of taking external biometric information, and comparing to stored model
You can if you wish add other tokens to the card, in addition to the ID function, for example credit card, or door access.
There is no compulsion to carry, or show, and verifying the card does not require a hit to a central database.
You do not need a license for "live or nearly live".
You need a license for recieving "television programs".
Looking further up the chain, into the broadcasting legislation, which defines "television programme", it's that broadcast by a "television programme service". http://www.opsi.gov.uk/acts/acts1996/1996055.htm
This is not a remote PC, sending you data, whether or not that data is sourced off-air.
The transcoder would, as I understand it though require a license.
Then, you can spend the extra $2/tag or so, and get ones with internal writable storage, and SHA or other hashing algorithms in.
Personally - I'd be quite happy with an implanted tag - as long as I was able to turn off and on the contained authorisation tokens (those with the correct keys would be able to delete, but not enable tokens), and could add new ones of my own, for car, laptop, phone, ...
Unfortunately, the standards for this are not there yet.
Not to mention I'd want full medical cover, and life insurance of say $10M, for any complications due to the tag.
You're probably going to want to do forward error correction, dedicationg 5% or so of the stream to error correction, rather than wasting the bandwidth the other way with acks.
OTOH, in some cases, such as radio, 128K is just fine, and is a lot better than 44K that some stations webcast in noe.
They both accurately reported normal loads, including ones with non unity power factors.
Compared with measuring the actual power on a scope, it was overestimating the power 100%.
This was with a server in a cupboard.
I tested that the scope measurement was real by turning the server off, leaving the fans on, and comparing the temperature of the outgoing air with 150W of ordinary light bulbs.
It was reading about 150W, with one drive, one duron 1800 processor.
I think this is as they sample too slowly, they are accurate for non power-factor-corrected power supplies (linear and SMPS) as well as motors, heaters, ...
Unfortunately not.
This is not that much more than Apollo.
A few people on the moon, for a bit longer than apollo.
No manufacturing things from lunar resources, or long-term bases at all.
More up to date sensors on the experiments of course, which is nice.
No significant development of anything that will make spaceflight cheaper.
It's been $10K/lb (approximately) since the end of apollo, and it still is.
Ballpark time.
There are 20 million journeys in progress at any time, at an average speed of 30mph.
Assuming all cars pass cameras every 5 miles (which is obviously incorrect, but probably not horribly so), and 20 million are driving at once
A car will pass a camera every 600 seconds.
Or all cars will pass some camera every 30us.
Let's assume the database of images problem is seperate - we're working on 8 char text strings (16 bit timestamp, 16 bit camera, remainder for plate
30000*8 = 240K/second.
Now, assume that there are 40 million vehicles in total, driven on average 100 miles, for a total of 4 billion miles, or a little under a billion camera events.
So, we're looking at around 10Gb/day, so a day is storable in RAM on a high end system.
Ok, so a number comes in, and we've got a camera number, plate ID pair. So, we now need to look up where the plate was last, and generate an exception if it's been going >77MPH (to reduce false hits, but once you've got most of them out, a few fast moving vehicles are not hard to prune).
We 'only' for each vehicle need to store the last position, let's say to the nearest
which is maybe 20 bits, and a 32 bit timestamp, so call a record 16 bytes, in a hash table with half occupancy, for a total amount of information oif only 1.2Gb.
Now, we need to look up the old position in a hash table, compute the speed, throw an
exception if it's too fast, and rewrite, all in 30us.
30us is a long, long time for modern computers
So, it looks like a single cheap PC can do the work - and that's without trivially obvious
tricks such as splitting the job by first/last letter.
Of course, you'd want to split it up for other reasons, but realtime isn't a problem.
Then you'd want a seperate system that spat out pairs of images at operators, to see if it was a real or imagined match.
My WRT54G (linux based wireless router) currently has an ebay sniping tool on it.
If I used a conventional PC to do this job, I'd be using around 30W more electricity.
24h*365*30W = 232Kwh.
At local prices, that's about 1/2 the price of the router.
Couple of weeks in, and you've got enough to do "Halt, or I shoot".
Hmm. Combine this with a little badge, indicating knowledge of "30/100/300" words, and local distribution of word cards...
The possibilities are endless.
Flash began expensive, and though the cost per byte has fallen, it's still a couple of orders of magnitude more expensive than disk drives.
IIRC, that's about always been the price differential.