By 'irreducibly complex' I take it you mean (by analogy) 'given the completed building, it is hard
or impossible to deduce the scaffolding configurations and prototypes that existed during the design and building stages'
Given the fuzzy-ness of 'X is true' and 'Y is true' in practice, combined with the fact that in many cases, one does not have control of X,
and the fact that there needn't be an unchanging relationship between X and Y,
is classical predicate logic the best model for reasoning? (See modern research on uncertain reasoning to see how things could be done...)
For example, suppose that in 200 trials, you get that X holds in 120 of them. When X holds (in 120 trials), Y holds in 110 of them, and when X fails to hold (80 trials), Y holds in 20 of them. What would you deduce then? In the classical logic sense, (where you see such things as contrapositive and modus ponens), there is NO VALID IMPLICATION between X and Y, yet there is a clear correlation.
Lets change the figures so that we have 2000000 trials, X holding in 1200000 of them. In the 1200000 times X holds, Y holds 1199999 times and when X fails, Y holds once. There is still no implication in the strict sense, yet if you discard a certain two of the trials, all of a sudden the strict implication holds in your data.
In short, you should be looking at the likelihood of your theory being accurate, rather than the 'truth' or 'falsity' of it.
p.s. You cannot prove 'for sure' that a theory (say theory X) is false --- only show that it's unlikely to be true some/most/all of the time.
Face it --- uncertainty is a fact of life in science. Philosophers may wish to tie themselves in knots arguing with the question of truth, most people just want to get on with the fun part of finding things out.
I guess 'disproving things' would have been better stated as 'showing the contrary to seem unlikely'.
I think the previous poster was probably a little confused. Most probably there was some great scientist in history whos ideas took 200-odd years to get accepted. For a similar example, entirely contained in the 20th century, consider Subramanyan Chandrasekhar's (possibly mis-spelt) theories regarding black holes. He was given the nobel prize for physics a few decades after he published the research. But the fact is, no theory is 'too beautiful to be false' and any new theory in a certain area should be able to predict, and to a point explain, the experimental success (and failure) of previous 'accepted' theories.
I think it proves very well the point John Stuart Mill made in On Liberty: any idea should be debated. If it's not true, it will be exposed; if it is, it will be strengthened.
Except for susceptibility to a 'DoS' style attack.
If I put out a simple 'plausible-sounding and easily-believable' claim, for which there is no simple yet correct refutation, any proper argument will be forgotten by many of those witnessing the debate, yet the wrong argument can quite possibly be seen to be true, and remembered, carried away and repeated by those who can't see that its wrong. Whilst the principle holds true between scientists with a healthy level of scepticism, those who haven't learned to pick out the information from the noise will tend to like the good-sounding quick-and-simple arguments. Compare this with how political campaigns go --- it's usually easier to bring your opponent down with scaremonger than to lift yourself up with honesty, hence the campaigning style.
The only real defense against terrorism, is to reduce the motive for terrorism.
So if they want you to die out, and your nation be destroyed, and your existence is their motive (which for some of the very extreme, it probably is), then you should reduce your nation and its people? Alternatively, lets look at 11/9 another way: the people were motivated by Islam. Why not proscribe Islam in its entirety? The problem with your statement is that it can be interpreted in many ways, depending on what you see as 'their' motive for terrorism. Any 'absolute' morals or rights that you may perceive needn't be shared by 'them' (whomever 'them' happens to refer to...)
Consistently abusive foreign policy actions, breed millions of enemies.
Just being richer (as a nation) than the rest of the world will do that. The US' foreign policy only helps to rub salt into the wound. Removing the foreign policy 'problems' wouldn't solve things (what difference not having them in the first place would have made is good debating matter). Many of the terrorists will bear grudges for generations. Just worked up stories of past atrocities are enough to get a small bunch of terrorists (and unfortunately that is enough). Once you have a terrorist problem, counter-terrorism and intelligence is a necessity for whatever solution you want (even appeasement, since whatever you grant them, enough will probably want more so as to continue the threat).
Yes. The 'should be' bit is the important bit.
The problem is that, while TiVo can remove any information about who watches what before reporting, the national security guys can come in and get data earlier on. i.e. if they see people watching a news channel that isn't popular with the government (read: Al-Jazera or however its spelt) they can quietly monitor WHO is making up those statistics, and the public will be none the wiser.
The point of the pro-GPL people from the outset is that people should not have the right to put the kind of restrictions on software that they do. They see no problem with the restrictions of the GPL because they obstruct an 'author's control' that they believe shouldn't exist in the first place. Unfortunately the whole GPL/LGPL/MPL/whateverPL/kitchensinkPL/etc. mess is the best attempt to date.
Public domain would be nice if the IP laws were significantly weaker (e.g. no barring from reverse engineering and disseminating what's found, no barring from modifying your copy, nor from explaining to someone else how to modify theirs), unfortunately this isn't the case, and so it isn't the case that software released as public domain will stay that way (if company A forks the code making theirs proprietary, the main distribution of the PD code vanishes, and those that still have copies can't stand up to the onslaught of lawyers launched by company A, then the PD version will vanish). In any case, if it is rapidly obsoleted, (which is what proprietary forks do---far more so than necessary) then it will fade from view.
As things stand, if all software today was public domain, but I could see a big market for 'my little modificaton', I could add some extra 'inoperability' features and use my killer feature to bring over customers. Everyone else is then playing catch-up. If a large number of people do this, then properietary software will overtake public domain software --in the short term-- (note that long-term consequences have no bearing whatsoever). In a short space of time, the needs of shareholders take precedence over all else.
In particular, the right to enforce licenses in the way that they do (i.e. running around with detector vans, suspecting EVERY household without a license of possibly watching TV without one, and requiring the address of every purchaser of every TV capable device being sent to TV licensing within 1 month of the sale...) is granted by the state (read: act of parliament).
Luckily, it's just that, and would require an act of parliament at the very least to allow any other public broadcast media to be licensed in a similar way. (Allowing anybody to use such methods to enforce e.g. copyright would be a disaster, but TV watching seems to work OK with the current system)
Though bear in mind that for a Chinese person to travel to and from Hong Kong they require permission -- it isn't as simple as driving from, say, Somerset to Norfolk (UK) or Texas to Tenessee (US). China and Hong Kong, though now owned as a whole by China, do not act like a single country. If and when that changes, only time will tell.
Factoring primes is DEFINATELY in NP (since
you can multiply and compare in polynomial time).
An NP-complete problem that can be done in
polynomial time means that ALL NP problems
(including factoring) can be done in P time
(not just all NP-complete problems).
See Cook's theorem.
In a similar fashion, in Dune 2000, you sneek an
engineer into the enemy base, nick one of the buildings, then put a concrete slab next to it.
The you do the old build-barracks, capture expensive building, sell barracks and building and build a few more concrete blocks... until the enemy has no infrastructure, then you trundle in your task force and finish him off.
Re:Tiny operating systems
on
Tiny Apps
·
· Score: 1
In fact, this scenario gives a nice example when actual physical space is plentiful (i.e. the 600-ish Mb on the CDR), yet small utilities are important, since we want to cache a large number of programs and bits in a small amount of RAM, so as to avoid going to CD that often (since CD's aren't exactly in the same speed category as a HD)
But network transparency at the window/pixel level isn't the only way to do it. Given directFB to handle actually putting things on screen, you could experiment with other ways of achieving network transparent windowing.
(cf. what the Berlin lot are doing).
Essentially, when there is a quick way to do something involving a GUI, but it doesn't fit with the way the X protocol works, X gets in your way. That's the problem.
I've read things about bandwidth shaping, but this is not whats required here. What would be an idea for someone to write (if no-one has already done so) is a Linux kernel module that adds an aliased ip address with a configurable delay (i.e. take the ip-alias module and make a few modifications).
Whilst that would be the problem with a Mozilla/XFS style `take out as little as possible
to make it legal to release as Open Source', what
about the opposite approach: Release what they -are- sure doesn't sit on anyone else's proprietary rights, (possibly) indicating what isn't there and challenging whoever's interested to try and make something out of it.
This is obviously a far cry from the coordinated release-as-Open-Source operations most companies (most-companies-doing-OSS, not most-companies-full-stop) have launched, but would be better than to simply dump BeOS entirely.
As another thought, and IANAL (definately), would it be possible for a bunch of volunteers to
work under NDA, stripping BeOS and clearing IP issues with the relevant IP owners? (obviously erring on the don't release side...)
n.b. I haven't personally given any serious thought to the practicality of these suggestions --- they just popped into my head.
'Release early, release often' can and does produce stable code, as does other approaches.
The problem is the same as with, say, a new Windoze release---the stability, reliability, etc. of a product that has just been released is an unknown quantity. It may have passed tests behind closed doors etc. but there is no substitute for seeing-how-it-works-in-the-field.
Basically the universal rule is that you don't put new software in critical places till its been tried elsewhere.
If you're after stability, then why would you be upgrading(aka going through the download-compile-reboot-rinse-lather-repeat cycle) with every release of a new kernel. Don't go to a new product immediately (the common sense attitude that the industry analysts always adivise for a new Windoze release also applies here), firstly you should see whats changed, and only upgrade once an important change has come about and been run in, secondly you definately should wait for a newer release of the kernel to be run in before downloading and installing it.
someone may like to take this up as a suggestion, but woudln't it be an idea to have a website somewhere comparing newer and older releases of the kernel (i.e. notes about how they work in practice, where someone upgraded from, which version to, and what improvements there were...)
It goes like this (if you ask my old Pentium anyhow...)
We all know that 1.0 = 0.99999... (recurring 9's
). Thus to the precision available in a 64bit double, we have 1.0 = 0.99999999 (some finite number of 9's). So Intel was correct (to double precision) after all!
Whats needed is a better internet infrastructure in the underlying operating system. There is currently nowhere an app can simply say 'share this with these permissions, etc. etc.' and have the requested data immediately available.
Let's face it. At current, we're barely at the point of dynamically configurable webservers. What is needed is a single, uniform way of informing the system of network related requests and letting it carry them out.
to make it available at 'http://MyComputer/hello/world' etc.
Obviously extra stuff needs to be given at runtime, but you get the idea. This is something that needs to be given serious thought. What can be done? (the reasoning behind this should be obvious: you shouldn't have to learn everything over again when moving from, say, apache to Roxen to wuftpd to samba to whatever next...)
The bit of the program that sends the key will have a hardwired public key (specific to you) and only the government (that issued you with the key) has the private key. They could then make it an offense to send an encrypted email that doesn't carry the requisite 'authentication' information that (a) authenticates who sent the email and (b) lets them quickly and efficiently look up the key.
Forcing the use of backdoors to encryption isn't so hard, the problem is eliminating the use of subtle forms of steganography.
Re:What can be done about terrorism?
on
More On Tragedy
·
· Score: 1
The difference between that post and its parent should indicate the multiple different `interpretations' that one can get from the world's religious texts, most probably from any religious text that explains how one should live one's life.
Whilst the first quotes were clearly out of context, that is the sort of thing that can be shoved down the throat of the sort of suggestible people that make suicide bombers---'. You can read it yourself, but you know it to be true anyway, as do all you brothers in arms...'
Eventually one has to oppose that which is against his beliefs, and it is writings about this that are so open to being twisted.
The fact that it has happened with more than one religion (to name two: Islam and Christianity) indicates that it is not a problem confined to one religion.
Computer folks just don't understand 1st April...
on
Parrot: For Real
·
· Score: 1
First we had TCP/IP over carrier pigeon (suggested
in an April fool's RFC). Now this... What next?
In any case, if he wants people to use another name, it had better be two syllables or less.
No one likes the length of GNU/Linux, and would abbreviate it to Linux in any case. (note that GNU would be too ambiguous, since GNU tools are available on more systems).
Which ever way you cut it, Linux the name is here to stay. And as noted earlier, the name wasn't Linus' self gratification---it was suggested and adopted by the users very early on.
This appears to be a much earlier program bearing the name Illustrator. I think this was what someone was referring to earlier. It's certainly an example of a product using Illustrator as a generic term in its title (much as KIllustrator does). John
Slashdot Mirror
By 'irreducibly complex' I take it you mean (by analogy) 'given the completed building, it is hard
or impossible to deduce the scaffolding configurations and prototypes that existed during the design and building stages'
Given the fuzzy-ness of 'X is true' and 'Y is true' in practice, combined with the fact that in many cases, one does not have control of X,
and the fact that there needn't be an unchanging relationship between X and Y,
is classical predicate logic the best model for reasoning? (See modern research on uncertain reasoning to see how things could be done...)
For example, suppose that in 200 trials, you get that X holds in 120 of them. When X holds (in 120 trials), Y holds in 110 of them, and when X fails to hold (80 trials), Y holds in 20 of them. What would you deduce then? In the classical logic sense, (where you see such things as contrapositive and modus ponens), there is NO VALID IMPLICATION between X and Y, yet there is a clear correlation.
Lets change the figures so that we have 2000000 trials, X holding in 1200000 of them. In the 1200000 times X holds, Y holds 1199999 times and when X fails, Y holds once. There is still no implication in the strict sense, yet if you discard a certain two of the trials, all of a sudden the strict implication holds in your data.
In short, you should be looking at the likelihood of your theory being accurate, rather than the 'truth' or 'falsity' of it.
p.s. You cannot prove 'for sure' that a theory (say theory X) is false --- only show that it's unlikely to be true some/most/all of the time.
Face it --- uncertainty is a fact of life in science. Philosophers may wish to tie themselves in knots arguing with the question of truth, most people just want to get on with the fun part of finding things out.
I guess 'disproving things' would have been better stated as 'showing the contrary to seem unlikely'.
I think the previous poster was probably a little confused. Most probably there was some great scientist in history whos ideas took 200-odd years to get accepted. For a similar example, entirely contained in the 20th century, consider Subramanyan Chandrasekhar's (possibly mis-spelt) theories regarding black holes. He was given the nobel prize for physics a few decades after he published the research. But the fact is, no theory is 'too beautiful to be false' and any new theory in a certain area should be able to predict, and to a point explain, the experimental success (and failure) of previous 'accepted' theories.
I think it proves very well the point John Stuart Mill made in On Liberty: any idea should be debated. If it's not true, it will be exposed; if it is, it will be strengthened.
Except for susceptibility to a 'DoS' style attack.
If I put out a simple 'plausible-sounding and easily-believable' claim, for which there is no simple yet correct refutation, any proper argument will be forgotten by many of those witnessing the debate, yet the wrong argument can quite possibly be seen to be true, and remembered, carried away and repeated by those who can't see that its wrong. Whilst the principle holds true between scientists with a healthy level of scepticism, those who haven't learned to pick out the information from the noise will tend to like the good-sounding quick-and-simple arguments. Compare this with how political campaigns go --- it's usually easier to bring your opponent down with scaremonger than to lift yourself up with honesty, hence the campaigning style.
The only real defense against terrorism, is to reduce the motive for terrorism.
So if they want you to die out, and your nation be destroyed, and your existence is their motive (which for some of the very extreme, it probably is), then you should reduce your nation and its people? Alternatively, lets look at 11/9 another way: the people were motivated by Islam. Why not proscribe Islam in its entirety? The problem with your statement is that it can be interpreted in many ways, depending on what you see as 'their' motive for terrorism. Any 'absolute' morals or rights that you may perceive needn't be shared by 'them' (whomever 'them' happens to refer to...)
Just being richer (as a nation) than the rest of the world will do that. The US' foreign policy only helps to rub salt into the wound. Removing the foreign policy 'problems' wouldn't solve things (what difference not having them in the first place would have made is good debating matter). Many of the terrorists will bear grudges for generations. Just worked up stories of past atrocities are enough to get a small bunch of terrorists (and unfortunately that is enough). Once you have a terrorist problem, counter-terrorism and intelligence is a necessity for whatever solution you want (even appeasement, since whatever you grant them, enough will probably want more so as to continue the threat).
Yes. The 'should be' bit is the important bit.
The problem is that, while TiVo can remove any information about who watches what before reporting, the national security guys can come in and get data earlier on. i.e. if they see people watching a news channel that isn't popular with the government (read: Al-Jazera or however its spelt) they can quietly monitor WHO is making up those statistics, and the public will be none the wiser.
The point of the pro-GPL people from the outset is that people should not have the right to put the kind of restrictions on software that they do. They see no problem with the restrictions of the GPL because they obstruct an 'author's control' that they believe shouldn't exist in the first place. Unfortunately the whole GPL/LGPL/MPL/whateverPL/kitchensinkPL/etc. mess is the best attempt to date.
Public domain would be nice if the IP laws were significantly weaker (e.g. no barring from reverse engineering and disseminating what's found, no barring from modifying your copy, nor from explaining to someone else how to modify theirs), unfortunately this isn't the case, and so it isn't the case that software released as public domain will stay that way (if company A forks the code making theirs proprietary, the main distribution of the PD code vanishes, and those that still have copies can't stand up to the onslaught of lawyers launched by company A, then the PD version will vanish). In any case, if it is rapidly obsoleted, (which is what proprietary forks do---far more so than necessary) then it will fade from view.
As things stand, if all software today was public domain, but I could see a big market for 'my little modificaton', I could add some extra 'inoperability' features and use my killer feature to bring over customers. Everyone else is then playing catch-up. If a large number of people do this, then properietary software will overtake public domain software --in the short term-- (note that long-term consequences have no bearing whatsoever). In a short space of time, the needs of shareholders take precedence over all else.
In particular, the right to enforce licenses in the way that they do (i.e. running around with detector vans, suspecting EVERY household without a license of possibly watching TV without one, and requiring the address of every purchaser of every TV capable device being sent to TV licensing within 1 month of the sale...) is granted by the state (read: act of parliament).
Luckily, it's just that, and would require an act of parliament at the very least to allow any other public broadcast media to be licensed in a similar way. (Allowing anybody to use such methods to enforce e.g. copyright would be a disaster, but TV watching seems to work OK with the current system)
Though bear in mind that for a Chinese person to travel to and from Hong Kong they require permission -- it isn't as simple as driving from, say, Somerset to Norfolk (UK) or Texas to Tenessee (US). China and Hong Kong, though now owned as a whole by China, do not act like a single country. If and when that changes, only time will tell.
Factoring primes is DEFINATELY in NP (since
you can multiply and compare in polynomial time).
An NP-complete problem that can be done in
polynomial time means that ALL NP problems
(including factoring) can be done in P time
(not just all NP-complete problems).
See Cook's theorem.
In a similar fashion, in Dune 2000, you sneek an
engineer into the enemy base, nick one of the buildings, then put a concrete slab next to it.
The you do the old build-barracks, capture expensive building, sell barracks and building and build a few more concrete blocks... until the enemy has no infrastructure, then you trundle in your task force and finish him off.
In fact, this scenario gives a nice example when actual physical space is plentiful (i.e. the 600-ish Mb on the CDR), yet small utilities are important, since we want to cache a large number of programs and bits in a small amount of RAM, so as to avoid going to CD that often (since CD's aren't exactly in the same speed category as a HD)
But network transparency at the window/pixel level isn't the only way to do it. Given directFB to handle actually putting things on screen, you could experiment with other ways of achieving network transparent windowing. (cf. what the Berlin lot are doing).
Essentially, when there is a quick way to do something involving a GUI, but it doesn't fit with the way the X protocol works, X gets in your way. That's the problem.
I've read things about bandwidth shaping, but this is not whats required here. What would be an idea for someone to write (if no-one has already done so) is a Linux kernel module that adds an aliased ip address with a configurable delay (i.e. take the ip-alias module and make a few modifications).
Whilst that would be the problem with a Mozilla/XFS style `take out as little as possible to make it legal to release as Open Source', what about the opposite approach: Release what they -are- sure doesn't sit on anyone else's proprietary rights, (possibly) indicating what isn't there and challenging whoever's interested to try and make something out of it.
This is obviously a far cry from the coordinated release-as-Open-Source operations most companies (most-companies-doing-OSS, not most-companies-full-stop) have launched, but would be better than to simply dump BeOS entirely.
As another thought, and IANAL (definately), would it be possible for a bunch of volunteers to work under NDA, stripping BeOS and clearing IP issues with the relevant IP owners? (obviously erring on the don't release side...)
n.b. I haven't personally given any serious thought to the practicality of these suggestions --- they just popped into my head.
But it helps when the door is fitted with a lock that actually locks into something in the doorframe :-)
'Release early, release often' can and does produce stable code, as does other approaches. The problem is the same as with, say, a new Windoze release---the stability, reliability, etc. of a product that has just been released is an unknown quantity. It may have passed tests behind closed doors etc. but there is no substitute for seeing-how-it-works-in-the-field. Basically the universal rule is that you don't put new software in critical places till its been tried elsewhere.
If you're after stability, then why would you be upgrading(aka going through the download-compile-reboot-rinse-lather-repeat cycle) with every release of a new kernel. Don't go to a new product immediately (the common sense attitude that the industry analysts always adivise for a new Windoze release also applies here), firstly you should see whats changed, and only upgrade once an important change has come about and been run in, secondly you definately should wait for a newer release of the kernel to be run in before downloading and installing it.
someone may like to take this up as a suggestion, but woudln't it be an idea to have a website somewhere comparing newer and older releases of the kernel (i.e. notes about how they work in practice, where someone upgraded from, which version to, and what improvements there were...)
It goes like this (if you ask my old Pentium anyhow...)
Let's face it. At current, we're barely at the point of dynamically configurable webservers. What is needed is a single, uniform way of informing the system of network related requests and letting it carry them out.
In pseudocode, one should be able to say
Obviously extra stuff needs to be given at runtime, but you get the idea. This is something that needs to be given serious thought. What can be done? (the reasoning behind this should be obvious: you shouldn't have to learn everything over again when moving from, say, apache to Roxen to wuftpd to samba to whatever next...)
The bit of the program that sends the key will have a hardwired public key (specific to you) and only the government (that issued you with the key) has the private key. They could then make it an offense to send an encrypted email that doesn't carry the requisite 'authentication' information that (a) authenticates who sent the email and (b) lets them quickly and efficiently look up the key.
Forcing the use of backdoors to encryption isn't so hard, the problem is eliminating the use of subtle forms of steganography.
The difference between that post and its parent should indicate the multiple different `interpretations' that one can get from the world's religious texts, most probably from any religious text that explains how one should live one's life.
Whilst the first quotes were clearly out of context, that is the sort of thing that can be shoved down the throat of the sort of suggestible people that make suicide bombers---'. You can read it yourself, but you know it to be true anyway, as do all you brothers in arms...'
Eventually one has to oppose that which is against his beliefs, and it is writings about this that are so open to being twisted.
The fact that it has happened with more than one religion (to name two: Islam and Christianity) indicates that it is not a problem confined to one religion.
First we had TCP/IP over carrier pigeon (suggested
in an April fool's RFC). Now this... What next?
In any case, if he wants people to use another name, it had better be two syllables or less.
No one likes the length of GNU/Linux, and would abbreviate it to Linux in any case. (note that GNU would be too ambiguous, since GNU tools are available on more systems).
Which ever way you cut it, Linux the name is here to stay. And as noted earlier, the name wasn't Linus' self gratification---it was suggested and adopted by the users very early on.
This appears to be a much earlier program bearing the name Illustrator. I think this was what someone was referring to earlier. It's certainly an example of a product using Illustrator as a generic term in its title (much as KIllustrator does).
John