haven't read this book, but I doubt that the world would be much better off if they learned the proper implementation of Bayes' theorem...just teach them not to use logical fallacies, PLEASE!
The book is about a lot more than just Bayes theorem. Jaynes starts out with a list of basic desiderata for a plausible reasoner:
(I) Degrees of plausibility are represented by real numbers.
(II) Qualitative correspondence with common sense.
(IIIa) If a conclusion can be reasoned out in more than one way, then every possible way must lead to the same result.
(IIIb) The [reasoner] takes into account all of the evidence it has relevant to a question.
(IIIc) The [reasoner] always represents equivalent states of knowlege by equivalent plausibility assignments.
He then shows that these desiderata lead to a unique set of quantitative rules for plausible reasoning.
But the part I find most interesting is the comment sections at the end of each chapter. For example, after listing his desiderata, his comments begin:
As politicians, advertisers, salesmen, and propagandists for various political, economic, moral, religious, psychic, envrionmental, dietary, and artistic doctrinaire positions know only too well, fallible human minds are easily tricked, by clever verbiage, into committing violations of the above desiderata. We shall try to ensure that they do not suceed with our [reasoner].
We all know that only a morally void character will flip-flop when presented with new evidence. I mean, otherwise it means they held on to the first opinion without substantial evidence.
That's the most ridiculous thing I've read all day. I think you were trying to be funny but unfortunately you were modded insightful so I feel compelled to respond if not to you then to the moderators who thought your comment was insightful.
I've recently been re-reading E. T. Jaynes' wonderful book,
Probability Theory : The Logic of Science which gives a mathematically rigorous treatment of plausible reasoning using, among other things, Bayes Theorem.
One of the things he makes perfectly clear is that new relevant evidence will always affect the decisions of a rational/perfect reasoner unless that evidence is totally redundant with respect to evidence that was already known.
The book was published posthumously in tree form but there are still.pdf and.ps available on the web. I think the world would be a much better place if everyone were to read this book. Unfortunately it has a lot of math in it that makes it un-readable for people without a technical background. But certainly anyone who uses probability theory or statistics really owes it to themselves to read this book.
The email linked to and the quote from Linus are significant since they show that Linus was advocating that someone should make a BK --> CVS gateway which is just what Tridge was working on.
Neither Tridge or anyone else is compelling Linus to drop Bitkeeper...
This whole bruhaha hit Slashdot when Larry McVoy withdrew all the free Bitkeeper licenses to all employees of the OSDL where both Linus and Tridge are working.
If Linus kept using Bitkeeper then he would have to pay for it and anyone else at the OSDL who wanted to work on the kernel would also have to pay to use Bitkeeper. These seem like pretty compelling reasons for Linus to drop Bitkeeper.
- the owner can completely cripple your use of his property at any time, and you may have to smile and be understanding to have a hope in hell of getting your own data back out of the proprietary solution and into something you can actually use, never mind actually working on whatever it is that you were trying to do with that data in the first place.
Perhaps Linus is not being a dunce in this situation. Perhaps he sees that his friend Larry is enraged and irrational. Perhaps Linus is willing to put the welfare of the kernel above his own public image.
If all of the above is true (and that is a big if), then it could be that Linus is blaming Tridge and praising Larry in order to ensure that there is a smooth transition from Bitkeeper. If Linus came out rooting for Tridge isn't it likely that Larry would yank Bitkeeper immediately and not allow a smooth transition to some other solution?
First, this isn't just theory, they've measured gamma ray bursts from other galaxies:
For around 10 seconds, intense pulses of energy are fired off, which can be detected right across the universe. All the bursts recorded by astronomers so far have come from distant galaxies and are therefore harmless to the Earth.
Second, for all those posting that a 10 second gamma ray burst won't be lethal to all of us:
Such a burst would strip the Earth of its protective ozone layer, allowing deadly ultraviolet radiation to pour down from the sun.
They don't RTFA, and they don't read all the other posts saying the same stupid thing. What do they think this is? Slashdot?
A couple of things. The reason that energy is not released when a virtual pair recombine is because that energy was already "borrowed" from the universe in order to create the pair. This is why they are called "virtual" -- they don't exist in a form that allows us to extract energy from them. The length of time a virtual pair can exist is controlled by the uncertainly principle and is thus inversely proportional to the energy of the pair:
Energy x Time = h/(2 pi)
Theoretically, if you can break the laws of physics as your device does then an infinite amount of energy would become available. The limit should be set by some physical limitation of your device. If you want to know more about background fluctuations google around for "casimir effect".
The idea of your device only allowing one sort of matter to be created might be very unappealing to physicists because you are breaking all sorts of conservation laws that are dear to their hearts. A more appealing device might be some sort of a trap for anti-matter. Remember that current atomic bombs don't destroy atoms, they just convert neutrons and protons from one grouping to another. The energy released is the difference in binding energy of the groupings. You would get much more bang for you buck if you were able to achieve total annihilation using anti-matter. But alas, this idea seems worn and hackneyed. Although there might be some interesting ideas to explore in the trap itself that holds the anti-matter.
If you really want to harness the vacuum fluctuations then I suggest using some sort of sub-sub-atomic mirrors that harness the casimir effect. The mirrors exist for a VERY short period of time, but they are so flat and so perfectly reflecting that they slam together at high speed due to the casimir effect. The mirrors should probably be made up out of strings in some configuration that is not found in nature. If fact, you would probably want to use 'branes instead of strings. This idea is probably just as ridiculous as the first but the details can be swept under a larger and perhaps more appealing rug.
The rules for creating black holes is most simply expressed in terms of escape velocity. A black hole is achieved when the escape velocity is greater than the speed of light.
1/2 mv^2 = GMm/r
v^2 = 2GM/r
On earth the escape velocity is about 11 km/s. The speed of light is roughly 300,000 km/s. So something with the mass of the earth would need to be roughly (300,000/11)^2 times smaller than the earth to form a black hole, roughly 1 cm across.
No, I didn't get the wrong datatype. The datatype I tried to use was called "long" (as you so kindly point out). It was their 32 bit integer which is twice (or double) the size of the standard 16 bit integer. They (Microsoft) implemented their "long" 32 bit integer datatype by converting to and from double precision floats for every operation.
I realize this sounds totally insane. That is why I posted the original anecdote. **sigh**
That makes everything 16 bit integers by default. Not all real world problems can be solved with 16 bit arithmetic. I needed a 32 bit integer. They had a datatype that was officially a 32 bit integer but when I tried to use it, it turned out that the 32 bit integers were converted to and from double precision floats for every arithmetic operation. The answers it gave were correct but about 1,000 times too slow.
Next time save all of us some time by actually reading a post before replying to it. Oops, for a moment there I forgot that this is/.
Many moons ago, I got called in to put out fires in a project that had a DOS based computer in part of the feedback loop of an industrial cutting machine. If the feedback wasn't fast enough the loop would open and things would break.
The software was written in QBASIC, which had just recently come out. I needed double precision (32 bit) integers for the control loop. QBASIC had this type built-in. Problem was that when I switched to 32 bit integers the program ran about 1,000 times slower and things in the real world got broken.
I couldn't figure it out. After carefully checking and re-checking my code, I did an assembly level debug. Turns out the brainiac billionaires at Microsoft had decided to "save" about 10 minutes of programming time by using floating point double precision for all their 32 bit calculations, even though 32 bit add and subtract were either already part of the machine language instruction set or took just two or three instructions at worst case. Instead, for every math operation the 32 bit values were converted to double precision floats, the calculation was done in floating point and then the answer was converted back to 32 bits. To make matters worse, the hardware didn't have a floating point co-proccessor (because the designer knew that no floating point calculations were needed) so all the floating point stuff was done in software emulation. Of course, there wasn't a word or a warning about this in any of the manuals.
Once I figured out the problem (morons had written the 32 bit integer support) I was able to write my own 32 bit routines in QBASIC that were 100's of times faster than Microsoft's built in routines, even without dipping down into assembly and taking advantage of the carry flag.
Quick Basic indeed! If it were any quicker it would be running backwards.
When I was a young kid, very many years ago, I was introduced to computers and programming in a summer school for bright students.
We were each given (or made, I don't remember) a long strip of paper about an inch or two wide, with lines making it into a long row of squares, like a single row from a page of graph paper. Each square was like a memory location. We were also given a paper clip that acted like the instruction pointer.
There was a simple instruction set, I think they associated numbers with instructions, like move forward or backward so many squares and simple arithmetic. The whole class together worked through a simple program.
There were some downsides to this. Some of the kids were totally lost on what we were trying to do. So we had to go real slow which then made it very boring for me. But the upside was that when it was over, I understood how computers worked and that simple model has helped my throughout my life as a programmer.
The key is to be really prepared and work it all out ahead of time. Make sure you have the instruction set and the simple programs all worked out ahead of time. You might want to practice it a bit. Keep it simple. Treat it like a game with rules. Forget about being true to Turing, instead focus on giving the kids something they can do and understand.
Not all of the kids will "get it". But those that do will really understand what programming is all about. Especially when you combine the simple Turing machine with gee-whiz-bang stuff on your laptop.
Somebody hired by HBS screws up and makes information that should have been kept private accessible on a public web server.
Instead of firing the people who made the boo-boo, the powers that be at HBS decide to punish anyone they can find who looked at their own admission letter.
First of all, it is not at all clear to me that it is ethically wrong to look at your own admission letter when it is posted on a public web site where *many* other people can already see it. For example, if I had heard about something like this I would probably try it just to see if it was really true. I would trust that HBS was not so bone-headed as to allow such a thing to happen.
Second, even if it were established that it was ethically wrong or questionable to peek, that is one heck of a temptation to put in front of someone since so much of their future plans depend upon what is in that letter.
Finally, I don't see that any harm is done by someone just peeking at the letter. If they act upon that information then that is another matter, for example by starting apartment hunting a month early. But just looking doesn't hurt anyone. According to my own ethics, if I am not hurting someone then I am not doing something bad.
I hope some of those people who got rejected band together and sue the pants off of HBS.
Slashdot Mirror
Fox couldn't stand having themselves portrayed as the baddies.
# Here are the lines for your /etc/hosts file
# to boycott sys-con sites
127.0.0.1 coldfusion.sys-con.com
127.0.0.1 dotnet.sys-con.com
127.0.0.1 eclipse.sys-con.com
127.0.0.1 issj.sys-con.com
127.0.0.1 itsolutions.sys-con.com
127.0.0.1 jdj.sys-con.com
127.0.0.1 linux.sys-con.com
127.0.0.1 linuxbusinessweek.sys-con.com
127.0.0.1 mxdj.sys-con.com
127.0.0.1 pbdj.sys-con.com
127.0.0.1 symbian.sys-con.com
127.0.0.1 weblogic.sys-con.com
127.0.0.1 webservices.sys-con.com
127.0.0.1 websphere.sys-con.com
127.0.0.1 wireless.sys-con.com
127.0.0.1 www.sys-con.tv
127.0.0.1 xml.sys-con.com
127.0.0.1 www.linuxworld.com
127.0.0.1 www.sys-con.com
echo "127.0.0.1 www.linuxworld.com" >> /etc/hosts
echo "127.0.0.1 coldfusion.sys-con.com /etc/hosts
127.0.0.1 dotnet.sys-con.com
127.0.0.1 eclipse.sys-con.com
127.0.0.1 issj.sys-con.com
127.0.0.1 itsolutions.sys-con.com
127.0.0.1 jdj.sys-con.com
127.0.0.1 linux.sys-con.com
127.0.0.1 linuxbusinessweek.sys-con.com
127.0.0.1 mxdj.sys-con.com
127.0.0.1 pbdj.sys-con.com
127.0.0.1 symbian.sys-con.com
127.0.0.1 weblogic.sys-con.com
127.0.0.1 webservices.sys-con.com
127.0.0.1 websphere.sys-con.com
127.0.0.1 wireless.sys-con.com
127.0.0.1 www.sys-con.tv
127.0.0.1 xml.sys-con.com
127.0.0.1 www.sys-con.com" >>
Free Porn!
The book is about a lot more than just Bayes theorem. Jaynes starts out with a list of basic desiderata for a plausible reasoner:
- (I) Degrees of plausibility are represented by real numbers.
- (II) Qualitative correspondence with common sense.
- (IIIa) If a conclusion can be reasoned out in more than one way, then every possible way must lead to the same result.
- (IIIb) The [reasoner] takes into account all of the evidence it has relevant to a question.
- (IIIc) The [reasoner] always represents equivalent states of knowlege by equivalent plausibility assignments.
He then shows that these desiderata lead to a unique set of quantitative rules for plausible reasoning.But the part I find most interesting is the comment sections at the end of each chapter. For example, after listing his desiderata, his comments begin:
What a great story to start off National TV-Turnoff Week.
That's the most ridiculous thing I've read all day. I think you were trying to be funny but unfortunately you were modded insightful so I feel compelled to respond if not to you then to the moderators who thought your comment was insightful.
I've recently been re-reading E. T. Jaynes' wonderful book, Probability Theory : The Logic of Science which gives a mathematically rigorous treatment of plausible reasoning using, among other things, Bayes Theorem.
One of the things he makes perfectly clear is that new relevant evidence will always affect the decisions of a rational/perfect reasoner unless that evidence is totally redundant with respect to evidence that was already known.
The book was published posthumously in tree form but there are still .pdf and .ps available on the web. I think the world would be a much better place if everyone were to read this book. Unfortunately it has a lot of math in it that makes it un-readable for people without a technical background. But certainly anyone who uses probability theory or statistics really owes it to themselves to read this book.
http://www.epinions.com/hmgdToolsCordlessDrills AndScrewdrivers9BlackBlackDecker72vVers apakCordlessDrillKitWKeylessChuck
The email linked to and the quote from Linus are significant since they show that Linus was advocating that someone should make a BK --> CVS gateway which is just what Tridge was working on.
This whole bruhaha hit Slashdot when Larry McVoy withdrew all the free Bitkeeper licenses to all employees of the OSDL where both Linus and Tridge are working.
If Linus kept using Bitkeeper then he would have to pay for it and anyone else at the OSDL who wanted to work on the kernel would also have to pay to use Bitkeeper. These seem like pretty compelling reasons for Linus to drop Bitkeeper.
Perhaps Linus is not being a dunce in this situation. Perhaps he sees that his friend Larry is enraged and irrational. Perhaps Linus is willing to put the welfare of the kernel above his own public image.
If all of the above is true (and that is a big if), then it could be that Linus is blaming Tridge and praising Larry in order to ensure that there is a smooth transition from Bitkeeper. If Linus came out rooting for Tridge isn't it likely that Larry would yank Bitkeeper immediately and not allow a smooth transition to some other solution?
Adobe Reader 7 JavaScript Bug.
You can disable Javascript support in Acroread but it then nags you. The above patch turns off the nag dialog box.
Second, for all those posting that a 10 second gamma ray burst won't be lethal to all of us:
They don't RTFA, and they don't read all the other posts saying the same stupid thing. What do they think this is? Slashdot?
A couple of things. The reason that energy is not released when a virtual pair recombine is because that energy was already "borrowed" from the universe in order to create the pair. This is why they are called "virtual" -- they don't exist in a form that allows us to extract energy from them. The length of time a virtual pair can exist is controlled by the uncertainly principle and is thus inversely proportional to the energy of the pair:
/(2 pi)
o le-evaporation.html
Energy x Time = h
Theoretically, if you can break the laws of physics as your device does then an infinite amount of energy would become available. The limit should be set by some physical limitation of your device. If you want to know more about background fluctuations google around for "casimir effect".
The idea of your device only allowing one sort of matter to be created might be very unappealing to physicists because you are breaking all sorts of conservation laws that are dear to their hearts. A more appealing device might be some sort of a trap for anti-matter. Remember that current atomic bombs don't destroy atoms, they just convert neutrons and protons from one grouping to another. The energy released is the difference in binding energy of the groupings. You would get much more bang for you buck if you were able to achieve total annihilation using anti-matter. But alas, this idea seems worn and hackneyed. Although there might be some interesting ideas to explore in the trap itself that holds the anti-matter.
If you really want to harness the vacuum fluctuations then I suggest using some sort of sub-sub-atomic mirrors that harness the casimir effect. The mirrors exist for a VERY short period of time, but they are so flat and so perfectly reflecting that they slam together at high speed due to the casimir effect. The mirrors should probably be made up out of strings in some configuration that is not found in nature. If fact, you would probably want to use 'branes instead of strings. This idea is probably just as ridiculous as the first but the details can be swept under a larger and perhaps more appealing rug.
The rules for creating black holes is most simply expressed in terms of escape velocity. A black hole is achieved when the escape velocity is greater than the speed of light.
1/2 mv^2 = GMm/r
v^2 = 2GM/r
On earth the escape velocity is about 11 km/s. The speed of light is roughly 300,000 km/s. So something with the mass of the earth would need to be roughly (300,000/11)^2 times smaller than the earth to form a black hole, roughly 1 cm across.
For evaporation, the follow page contains the simple formulas it sounds like you are looking for:
http://www.alcyone.com/max/writing/essays/black-h
Will the laptop make a screaming sound when it becomes weightless after slipping out of your hands?
Will it then make the obligatory loud car crash sounds when the weightlessness ends right before it makes loud laptop crashing sounds?
No, I didn't get the wrong datatype. The datatype I tried to use was called "long" (as you so kindly point out). It was their 32 bit integer which is twice (or double) the size of the standard 16 bit integer. They (Microsoft) implemented their "long" 32 bit integer datatype by converting to and from double precision floats for every operation.
I realize this sounds totally insane. That is why I posted the original anecdote. **sigh**
That makes everything 16 bit integers by default. Not all real world problems can be solved with 16 bit arithmetic. I needed a 32 bit integer. They had a datatype that was officially a 32 bit integer but when I tried to use it, it turned out that the 32 bit integers were converted to and from double precision floats for every arithmetic operation. The answers it gave were correct but about 1,000 times too slow.
/.
Next time save all of us some time by actually reading a post before replying to it. Oops, for a moment there I forgot that this is
Dead? Oh my God! I thought she was English.
Many moons ago, I got called in to put out fires in a project that had a DOS based computer in part of the feedback loop of an industrial cutting machine. If the feedback wasn't fast enough the loop would open and things would break.
The software was written in QBASIC, which had just recently come out. I needed double precision (32 bit) integers for the control loop. QBASIC had this type built-in. Problem was that when I switched to 32 bit integers the program ran about 1,000 times slower and things in the real world got broken.
I couldn't figure it out. After carefully checking and re-checking my code, I did an assembly level debug. Turns out the brainiac billionaires at Microsoft had decided to "save" about 10 minutes of programming time by using floating point double precision for all their 32 bit calculations, even though 32 bit add and subtract were either already part of the machine language instruction set or took just two or three instructions at worst case. Instead, for every math operation the 32 bit values were converted to double precision floats, the calculation was done in floating point and then the answer was converted back to 32 bits. To make matters worse, the hardware didn't have a floating point co-proccessor (because the designer knew that no floating point calculations were needed) so all the floating point stuff was done in software emulation. Of course, there wasn't a word or a warning about this in any of the manuals.
Once I figured out the problem (morons had written the 32 bit integer support) I was able to write my own 32 bit routines in QBASIC that were 100's of times faster than Microsoft's built in routines, even without dipping down into assembly and taking advantage of the carry flag.
Quick Basic indeed! If it were any quicker it would be running backwards.
When I was a young kid, very many years ago, I was introduced to computers and programming in a summer school for bright students.
We were each given (or made, I don't remember) a long strip of paper about an inch or two wide, with lines making it into a long row of squares, like a single row from a page of graph paper. Each square was like a memory location. We were also given a paper clip that acted like the instruction pointer.
There was a simple instruction set, I think they associated numbers with instructions, like move forward or backward so many squares and simple arithmetic. The whole class together worked through a simple program.
There were some downsides to this. Some of the kids were totally lost on what we were trying to do. So we had to go real slow which then made it very boring for me. But the upside was that when it was over, I understood how computers worked and that simple model has helped my throughout my life as a programmer.
The key is to be really prepared and work it all out ahead of time. Make sure you have the instruction set and the simple programs all worked out ahead of time. You might want to practice it a bit. Keep it simple. Treat it like a game with rules. Forget about being true to Turing, instead focus on giving the kids something they can do and understand.
Not all of the kids will "get it". But those that do will really understand what programming is all about. Especially when you combine the simple Turing machine with gee-whiz-bang stuff on your laptop.
Somebody hired by HBS screws up and makes information that should have been kept private accessible on a public web server.
Instead of firing the people who made the boo-boo, the powers that be at HBS decide to punish anyone they can find who looked at their own admission letter.
First of all, it is not at all clear to me that it is ethically wrong to look at your own admission letter when it is posted on a public web site where *many* other people can already see it. For example, if I had heard about something like this I would probably try it just to see if it was really true. I would trust that HBS was not so bone-headed as to allow such a thing to happen.
Second, even if it were established that it was ethically wrong or questionable to peek, that is one heck of a temptation to put in front of someone since so much of their future plans depend upon what is in that letter.
Finally, I don't see that any harm is done by someone just peeking at the letter. If they act upon that information then that is another matter, for example by starting apartment hunting a month early. But just looking doesn't hurt anyone. According to my own ethics, if I am not hurting someone then I am not doing something bad.
I hope some of those people who got rejected band together and sue the pants off of HBS.
Apple et. al. can just move to Eur ... Opps. Never mind.
A lot of speculation and conviction has been thrown around in this thread. The parent links to some very interesting facts.
Of course not! We are all training to become corporate media journalists.