It's not arbitrary. There's actually a good reason for minimizing (y-yobs)^2, assuming that your observations have a Gaussian distribution. The resulting estimators provide a maximum likelihood estimator of the parameters of the distribution, if and only if it really was Gaussian. Thus, of course, if it isn't Gaussian (outliers of various sorts, et.c), the x^2 may not be the best bet. There is an entire field of 'robust estimators' of quantities, which are more resistant to outliers than least squares. There are also cases in which the underlying distribution is pathologically different from Gaussian; it could be Lorentzian (Cauchy), in which case it is so completely unlike a Gaussian, it doesn't even have a defined standard deviation (it is infinite). There are weighted methods which can fix this too.
So, in short, least squares is the right answer (in the sense that it yields results which provable have the maximum likelihood describing the data at hand) if you have a perfect Gaussian variate; otherwise, it may well not be.
You missed the good part of the T-Mobile PAYGO plan: once you have paid the $100 once (or accumulated it via smaller payments), all your future added minutes last a year. So if you haven't used $100, you can top off with $50, or even $10, at the end of the first year, and it roll the leftover minutes forward to the next year and add the new minutes. If you buy fewer minutes, you are paying a bit higher price per minute, but you never lose the old minutes as long as you top off. This is why the original purchase of $100 is worth it, since you get the best rate ($0.10/minute).
Also, T-Mobile has great deals on refurb phones often. I bought a Dart (not a great phone, but it works), For $90, which came with a $50 card for minutes. Thus, a useable Android phone for $40. It does support WiFi data tethering, so I can pay the $2 or $3 rate for a day when I need data, and have that for my other devices, too.
There are a number of issues with the calculations here.
First, you estimate that it takes about 1 gallon of gasoline equivalent per pound of aluminum (1800 gallons for 2000 pounds). This then seems to say that the energy cost for aluminum should hover around $3.00/pound (the price of the gasoline), or maybe half that since electricity is cheaper. Maybe $1.50/pound for the energy. It currently sells for $0.75/pound on the spot market, so the energy input is somehow wrong.
Also, if a truck gets 20 miles/gallon, it burns 5000 gallons in 100k miles. If it gets 25 MPG, it burns 4000 gallons in 100k miles, so the difference seems to be about a 1000 gallon improvement in consumption, not a 200 gallon improvement. If your original arithmetic is sound, you would still end up in the hole if you burned the truck at the end, but aluminum is highly recyclable with only a tiny fraction of the original energy, so in the long run I think it comes out ahead even if the original numbers are right. However, I think that with some mis-estimate of the original energy budget, it might come out ahead even on the first pass, without considering recycling.
The Union of Concerned Scientists includes some scientists, but is an anti-nuclear political organization. This headline is like saying "Teenagers have unhealthy fantasies playing D&D, say mothers" amd omitting from the headline that "mothers" really refers to "Mothers Against Dungeons and Dragons".
This is not even close to correct about the policies of UCS. See:
http://www.ucsusa.org/nuclear_power/ and http://www.ucsusa.org/nuclear_power/nuclear-power-and-our-energy-choices/nuclear-power-and-global-warming/house-testimony-on-nuclear.html
They are very strongly looking at nuclear safety issues, but specifically are neither pro-or-con on nuclear power itself. The organization does a great deal of research into all matters related to energy and safety and sustainability issues. They are well aware of the carbon-free nature of nuclear power, and that if it would be managed safely, it could be highly beneficial.
I know they are different things, but it was the camera access that got my attention. Disabling Control Panel access, I think, as I mentioned in the original post, avoids the issue. As far as I can tell, there is no way to get to anything on my iPad without unlocking.
The ad hominem about my lacking imagination and/or sense was not needed or polite.
As soon as I did the iOS7 update, I noticed that you could access the camera from the lock screen, and I didn't want someone taking inappropriate pictures on my iPad if they stole it. There is an option in the settings which controls what features are available from the lock screen. If you turn off the Control Panel access from the lock screen, and everything else, this goes away.
So, it's annoying but not fatal as a security issue. I can't imagine anyone wanting to have the device open for the camera when it is locked. I do wish the options were flexible enough that one could still adjust audio settings with it locked.
Any app that provides programmability is not allowed....
Well, that would be true if you couldn't get Python 2.7 for iOS. There are, in fact, two different full python implementations of python on the App store. I have used it to run my vxi-11 stack to talk to oscilloscopes and other data acquisition stuff. Works fine. The only annoyance is that you have to cut & paste large programs from email (for now) to get them in. You can edit code in the editor, though, so small programs can be done right in place.
This is formally a perfectly acceptable way to present the discovery. Two events (points) in Minkowski 4-space which are connected by a ray of light have an invariant time separation tau = delta_t - delta_x/c = 0. To us, it did just happen.
I am a t-mobile prepaid customer, with one of their LG Android Optimus-T handsets, which I got for $130 refurb. I can get legal tethering from them by setting up the phone as a mobile hotspot for $1.50 a day. If you use it every day, it is quite expensive, but if you just use it a few days a month for travel or whatever, it is great. They give you 30 MB of very high speed (I think I have seen 1 MB/sec peaks), and unlimited 2G for the rest of the day. If you are just checking email, or doing bandwidth-restricted video chat back home for a while, it is plenty.
I juts got back from a conference trip, where the hotel charged $10/day for internet and you couldn't even get it in the main conference area, and they didn't even have free internet in the lobby. I spent my $1.50/day, and was able to edit code via ssh, talk to my wife via video chat for a while, and look at the web. I had to make sure I did the video chat at the beginning of the session, while I still was under the 30 MB throttling limit.
You do know that the Union of Concerned Scientists has been around since 1969, not since Bush Jr., don't you? Did you read their history on their web site?
They (we, in this case) certainly have been more vocal during the most egregiously anti-scientific administrations, but Bush Jr. wasn't the first.
There is a truly excellent book, "Hitler's Uranium Club" which documents what the Germans themselves said about their efforts. It is edited by Jeremy Bernstein. It is a collection of transcriptions of conversations among the leading German scientists (Heisenberg, Laue, etc., not all of whom were actually doing nuclear physics), who were captured lat in the war and transferred to Farm Hall in England. They were recorded secretly, so what is said is very candid.
Anyone interested in this history should definitely read the book. The conversations run the gamut from very technical, to various fights over social issues.
OK, I'm in the process of debugging somebody else's 43,000 lines of FORTRAN code. (I hate FORTRAN...). What I see in here would require a number of offsets which would cost approximately the entire US GDP to buy. This is not the first time I have seen code like this, either.
If I could get my hands on say an ounce of Pu 238 I could build a RTG that would power my home, all my vehicles, and enable me to quit my job and live of the check my local electricity provider would have to pay me for the excess power I would generate. It would generate full power for ~ 87 years and not only wold I be using the greenest power available I would be providing a community service of disposing of a radioactive material.
OK, you must have a very efficient house. The specific heat output of Pu238 is about 0.5 Watts/gram, so an ounce puts out about 14 Watts. Given 10% RTG efficiency (which is much better than normal), I give you about 1.5 watts from this. Can you run your house on that?
I think your sample is badly biased by ending on the 1,000,000th prime, since essentially all your primes are then those between 10,000,000 and 15,485,863 which is the last prime in the first million file.
It took me a while to notice this, too, since any sampling which doesn't sample exact decades is badly biased. By the time you get all the primes less than 1e9, the distribution is very flat. Here are the stats:
OK, this is an interesting new marketing strategy for a company as a way to remove a product from their line. You don't ever have to stop selling it. You just keep halving its size until no one is sure whether they have bought one or not.
With some good access to the RDF, everyone will continue to hear music, whether or not there was actually a device in the box.
I still own a first generation Shuffle. I think it weighs 50 grams. Really, that's just to much to bear, carrying it in my briefcase. I know that if my briefcase only had a 10.7 gram Shuffle in it, it would be MUCH easier on my walk to work.
Interestingly, a few months ago, my financial services company (Merrill Lynch) changed the way their online login works to make this attack very hard. They required me to select an image from a large catalog, and a phrase I made up to go with it. Now, when I log in, I am presented the image and the phrase. Since these images come from a huge catalog, and the phrase is entirely up to the user, the probability that a hijacked page would have the same information is very small. In effect, the site is presenting _me_ with a pasword, before I present it with a password. (Cue, on 3, In Soviet Russia, sites log onto you)
I think this makes these pages fairly secure against the various DNS and other redirect attacks people have come up with. Someone would have to get very deep access to the main server, to figure out the image everyone chose, to successfully hijack a site.
Actually 'found no evidence' is the strongest possible scientifically justifiable statement.
You cannot ever prove that the machines do not, and have never, emitted such fumes, unless you have monitored each and every computer continuously with a perfectly sensitive detector (which is not possible to build...).
I was never sure if they named the register this way just so they could use this error message, or otherwise. It was indeed on a UniBus adapter. And,as the SYS$F prefix indicates, it was _very_ fatal.
On the efficiency of the human body as an engine, the number you quote is about right. However, assessing that as a terribly inefficient heat engine is a bit odd.
A really well-tuned automobile engine, running on pre-refined fuel, might get 40% thermal efficiency or so. The human body, of course, starts with rather unrefined fuel (food, to the non-techie:-) ), runs all the necessary chemical conversion machinery, and produces its output. It also expends a lot of energy in self-repair and maintenance, which for your car is separate. And if your car had to provide enough power for all the computational work we are doing, too, it would further increase the overall energy budget.
A full-cycle efficiency of the 33% or so (assuming your 2:1 ratio of waste heat to output) seems very good. Almost all the fuel conversion techniques we have (oil refining, fermentation to convert the stuff we eat to ethanol, coal gasification, etc.) lose more than this, I suspect.
BRAVO! you are the first person to correctly describe the source of the problem. I have no mod points, so I will comment.
The parasitic capacitance in the high-frequency transformer used in switching supplies allows a tiny, high frequency current to trickle through. As one previous comment mis-stated, even switching supplies have the transformer, it's just much smaller (roughly by the ratio of the switching frequency to the mains frequency). However, at high frequency, the capacitive couple is increase, so unless the transformer is extremely well balanced and shielded, you get feed-through.
If the case of the computer is earthed, of course, this is not an issue. If it is not, even if the computer user is not grounded, he/she can feel the tingle, since the stray current is driven through your capacitance to ground.
The limits on the current are quite small, but it can be distracting.
A few years ago, I had some kids shine a laser pointer in my eyes while I was driving a car on a dark road, with my family aboard. It caused me to drive off onto the shoulder of the road. If I had lost slightly more control, it would have cost my family their lives.
Doing this to someone is no joke. It is not an innocent crime, as it is far more dangerous than it sounds. In my case, it amounted to something approaching attempted murder (since I'm sure that the people doing it knew it had the potential to be dangerous, although they may not have considered how dangerous).
Doing it to a helicopter, where it could cause the pilot to lose control, and not only kill the pilot but occupants of dwellings below, is even more serious.
Slashdot Mirror
It's not arbitrary. There's actually a good reason for minimizing (y-yobs)^2, assuming that your observations have a Gaussian distribution. The resulting estimators provide a maximum likelihood estimator of the parameters of the distribution, if and only if it really was Gaussian. Thus, of course, if it isn't Gaussian (outliers of various sorts, et.c), the x^2 may not be the best bet. There is an entire field of 'robust estimators' of quantities, which are more resistant to outliers than least squares. There are also cases in which the underlying distribution is pathologically different from Gaussian; it could be Lorentzian (Cauchy), in which case it is so completely unlike a Gaussian, it doesn't even have a defined standard deviation (it is infinite). There are weighted methods which can fix this too.
So, in short, least squares is the right answer (in the sense that it yields results which provable have the maximum likelihood describing the data at hand) if you have a perfect Gaussian variate; otherwise, it may well not be.
You missed the good part of the T-Mobile PAYGO plan: once you have paid the $100 once (or accumulated it via smaller payments), all your future added minutes last a year. So if you haven't used $100, you can top off with $50, or even $10, at the end of the first year, and it roll the leftover minutes forward to the next year and add the new minutes. If you buy fewer minutes, you are paying a bit higher price per minute, but you never lose the old minutes as long as you top off. This is why the original purchase of $100 is worth it, since you get the best rate ($0.10/minute).
Also, T-Mobile has great deals on refurb phones often. I bought a Dart (not a great phone, but it works), For $90, which came with a $50 card for minutes. Thus, a useable Android phone for $40. It does support WiFi data tethering, so I can pay the $2 or $3 rate for a day when I need data, and have that for my other devices, too.
There are a number of issues with the calculations here.
First, you estimate that it takes about 1 gallon of gasoline equivalent per pound of aluminum (1800 gallons for 2000 pounds). This then seems to say that the energy cost for aluminum should hover around $3.00/pound (the price of the gasoline), or maybe half that since electricity is cheaper. Maybe $1.50/pound for the energy. It currently sells for $0.75/pound on the spot market, so the energy input is somehow wrong.
Also, if a truck gets 20 miles/gallon, it burns 5000 gallons in 100k miles. If it gets 25 MPG, it burns 4000 gallons in 100k miles, so the difference seems to be about a 1000 gallon improvement in consumption, not a 200 gallon improvement. If your original arithmetic is sound, you would still end up in the hole if you burned the truck at the end, but aluminum is highly recyclable with only a tiny fraction of the original energy, so in the long run I think it comes out ahead even if the original numbers are right. However, I think that with some mis-estimate of the original energy budget, it might come out ahead even on the first pass, without considering recycling.
The Union of Concerned Scientists includes some scientists, but is an anti-nuclear political organization. This headline is like saying "Teenagers have unhealthy fantasies playing D&D, say mothers" amd omitting from the headline that "mothers" really refers to "Mothers Against Dungeons and Dragons".
This is not even close to correct about the policies of UCS. See:
http://www.ucsusa.org/nuclear_power/
and
http://www.ucsusa.org/nuclear_power/nuclear-power-and-our-energy-choices/nuclear-power-and-global-warming/house-testimony-on-nuclear.html
They are very strongly looking at nuclear safety issues, but specifically are neither pro-or-con on nuclear power itself. The organization does a great deal of research into all matters related to energy and safety and sustainability issues. They are well aware of the carbon-free nature of nuclear power, and that if it would be managed safely, it could be highly beneficial.
I know they are different things, but it was the camera access that got my attention. Disabling Control Panel access, I think, as I mentioned in the original post, avoids the issue. As far as I can tell, there is no way to get to anything on my iPad without unlocking.
The ad hominem about my lacking imagination and/or sense was not needed or polite.
As soon as I did the iOS7 update, I noticed that you could access the camera from the lock screen, and I didn't want someone taking inappropriate pictures on my iPad if they stole it. There is an option in the settings which controls what features are available from the lock screen. If you turn off the Control Panel access from the lock screen, and everything else, this goes away.
So, it's annoying but not fatal as a security issue. I can't imagine anyone wanting to have the device open for the camera when it is locked. I do wish the options were flexible enough that one could still adjust audio settings with it locked.
Any app that provides programmability is not allowed....
Well, that would be true if you couldn't get Python 2.7 for iOS. There are, in fact, two different full python implementations of python on the App store. I have used it to run my vxi-11 stack to talk to oscilloscopes and other data acquisition stuff. Works fine. The only annoyance is that you have to cut & paste large programs from email (for now) to get them in. You can edit code in the editor, though, so small programs can be done right in place.
Algol-60. RT-11. TECO. Hypercard (count this one twice!).
This is formally a perfectly acceptable way to present the discovery. Two events (points) in Minkowski 4-space which are connected by a ray of light have an invariant time separation tau = delta_t - delta_x /c = 0. To us, it did just happen.
I am a t-mobile prepaid customer, with one of their LG Android Optimus-T handsets, which I got for $130 refurb.
I can get legal tethering from them by setting up the phone as a mobile hotspot for $1.50 a day. If
you use it every day, it is quite expensive, but if you just use it a few days a month for travel or
whatever, it is great. They give you 30 MB of very high speed (I think I have seen 1 MB/sec peaks),
and unlimited 2G for the rest of the day. If you are just checking email, or doing bandwidth-restricted
video chat back home for a while, it is plenty.
I juts got back from a conference trip, where the hotel charged $10/day for internet and you
couldn't even get it in the main conference area, and they didn't even have free internet in the
lobby. I spent my $1.50/day, and was able to edit code via ssh, talk to my wife via video chat
for a while, and look at the web. I had to make sure I did the video chat at the beginning of the
session, while I still was under the 30 MB throttling limit.
You do know that the Union of Concerned Scientists has been around since 1969, not since Bush Jr., don't you? Did you read their history on their web site?
They (we, in this case) certainly have been more vocal during the most egregiously anti-scientific administrations, but Bush Jr. wasn't the first.
There is a truly excellent book, "Hitler's Uranium Club" which documents what the Germans themselves said about their efforts. It is edited by Jeremy Bernstein. It is a collection of transcriptions of conversations among the leading German scientists (Heisenberg, Laue, etc., not all of whom were actually doing nuclear physics), who were captured lat in the war and transferred to Farm Hall in England. They were recorded secretly, so what is said is very candid.
Anyone interested in this history should definitely read the book. The conversations run the gamut from very technical, to various fights over social issues.
OK, I'm in the process of debugging somebody else's 43,000 lines of FORTRAN code. (I hate FORTRAN...). What I see in here would require a number of offsets which would cost approximately the entire US GDP to buy. This is not the first time I have seen code like this, either.
OK, you must have a very efficient house. The specific heat output of Pu238 is about 0.5 Watts/gram, so an ounce puts out about 14 Watts. Given 10% RTG efficiency (which is much better than normal), I give you about 1.5 watts from this. Can you run your house on that?
I think your sample is badly biased by ending on the 1,000,000th prime, since essentially all your primes are then those between 10,000,000 and 15,485,863 which is the last prime in the first million file.
It took me a while to notice this, too, since any sampling which doesn't sample exact decades is badly biased. By the time you get all the primes less than 1e9, the distribution is very flat. Here are the stats:
first digit histogram [6003531, 5837665, 5735086, 5661135, 5602768, 5556434, 5516130, 5481646, 5453140]
fractions:
0.118069263338
0.114807236968
0.112789853038
0.111335485585
0.110187602998
0.109276369051
0.108483724924
0.107805540623
0.107244923476
I am running through all the primes less that 1e11 right now, and will post that later.
OK, this is an interesting new marketing strategy for a company as a way to remove a product from their line. You don't ever have to stop selling it. You just keep halving its size until no one is sure whether they have bought one or not.
With some good access to the RDF, everyone will continue to hear music, whether or not there was actually a device in the box.
I still own a first generation Shuffle. I think it weighs 50 grams. Really, that's just to much to bear, carrying it in my briefcase. I know that if my briefcase only had a 10.7 gram Shuffle in it, it would be MUCH easier on my walk to work.
Do you really think the song 'Fight Fiehcely, Hahvahd' came from an MIT professor?
Interestingly, a few months ago, my financial services company (Merrill Lynch) changed the way their online login works to make this attack very hard. They required me to select an image from a large catalog, and a phrase I made up to go with it. Now, when I log in, I am presented the image and the phrase. Since these images come from a huge catalog, and the phrase is entirely up to the user, the probability that a hijacked page would have the same information is very small. In effect, the site is presenting _me_ with a pasword, before I present it with a password. (Cue, on 3, In Soviet Russia, sites log onto you)
I think this makes these pages fairly secure against the various DNS and other redirect attacks people have come up with. Someone would have to get very deep access to the main server, to figure out the image everyone chose, to successfully hijack a site.
Actually 'found no evidence' is the strongest possible scientifically justifiable statement.
You cannot ever prove that the machines do not, and have never, emitted such fumes, unless you have monitored each and every computer continuously with a perfectly sensitive detector (which is not possible to build...).
I always liked:
SYS$F_FUBAR -- Failed UniBus Adapter Register
I was never sure if they named the register this way just so they could use this error message, or otherwise. It was indeed on a UniBus adapter. And,as the SYS$F prefix indicates, it was _very_ fatal.
On the efficiency of the human body as an engine, the number you quote is about right. However, assessing that as a terribly inefficient heat engine is a bit odd.
:-) ), runs all the necessary chemical conversion machinery, and produces its output. It also expends a lot of energy in self-repair and maintenance, which for your car is separate. And if your car had to provide enough power for all the computational work we are doing, too, it would further increase the overall energy budget.
A really well-tuned automobile engine, running on pre-refined fuel, might get 40% thermal efficiency or so. The human body, of course, starts with rather unrefined fuel (food, to the non-techie
A full-cycle efficiency of the 33% or so (assuming your 2:1 ratio of waste heat to output) seems very good. Almost all the fuel conversion techniques we have (oil refining, fermentation to convert the stuff we eat to ethanol, coal gasification, etc.) lose more than this, I suspect.
0.1, base 3?
BRAVO! you are the first person to correctly describe the source of the problem. I have no mod points, so I will comment.
The parasitic capacitance in the high-frequency transformer used in switching supplies allows a tiny, high frequency current to trickle through. As one previous comment mis-stated, even switching supplies have the transformer, it's just much smaller (roughly by the ratio of the switching frequency to the mains frequency). However, at high frequency, the capacitive couple is increase, so unless the transformer is extremely well balanced and shielded, you get feed-through.
If the case of the computer is earthed, of course, this is not an issue. If it is not, even if the computer user is not grounded, he/she can feel the tingle, since the stray current is driven through your capacitance to ground.
The limits on the current are quite small, but it can be distracting.
A few years ago, I had some kids shine a laser pointer in my eyes while I was driving a car on a dark road, with my family aboard. It caused me to drive off onto the shoulder of the road. If I had lost slightly more control, it would have cost my family their lives.
Doing this to someone is no joke. It is not an innocent crime, as it is far more dangerous than it sounds. In my case, it amounted to something approaching attempted murder (since I'm sure that the people doing it knew it had the potential to be dangerous, although they may not have considered how dangerous).
Doing it to a helicopter, where it could cause the pilot to lose control, and not only kill the pilot but occupants of dwellings below, is even more serious.
I was referring to the wording:
"Internet users will create 161 exabytes of new data this year"
I don't doubt that we will _transfer_ 161 exabytes around, but _create_? most of the data is the same stuff, transferred to millions of users.