Sorry if you were offended by something in my post (I reckon part of the first sentence of it, should have been written in another way).
You said I am grasping straws. How is falling back to mathematical formalism "grasping straws"? From my point of view, you are the one "waving hands".
The reason the test is normally applied to a selected group of people was nowhere addressed in my post. In fact you actually concede my point in your email.
My point was:
In this second sample the frequency of people that have disease X, should be quite different than on the first sample taken from "anyone in USA".
It is wrong to take statistical frequencies measured against a sample taken from distribution Dn, and then proceed to make assertions against a sample taken from another distribution Ds (where Ds != Dn).
Perhaps you never studied statistics. Please get a decent statistics book and try to study it, instead of relying on high schoolers writing on Wikipedia.
What is wrong with this analysis is the following:
You measure frequency F (freq of disease X), and a given test (T) effectiveness on sample A taken from a statistical distribution Dn (where Dn ~ "samples randomly anyone: 1. in the USA"). The test effectiveness being T_ef(A) where A was sampled from Dn.
If you apply the test against sample A then you would have that retarded frequency relations you felt necessary to copy from wikipedia. Are you surprised? You just measured that sample A was large enough to reasonably mimic the events frequency to be expected from the distribution the sample was taken from.
My point being, that illustration is deceiving because in practice that rare disease tests are not routinely applied against the general population (aka sample A). The tests are normally applied to a sample of people that do not get sampled from Dn. The test subjects get sampled from distribution Ds where
Ds ~ samples randomly anyone that
is in the USA
presenting medical symptoms to the point they searched for medical assistance
medical symptoms which are specifically like those of disease X
that have already been tested (negatively) for other more frequent diseases with the same symptoms
to the point that a trained MD after analyzing the whole scenario decided to give a go on tests of disease X.
In this second sample the frequency of people that have disease X, should be quite different than on the first sample taken from "anyone in USA".
It is wrong to take statistical frequencies measured against a sample taken from distribution Dn, and then proceed to make assertions against a sample taken from another distribution Ds (where Ds != Dn).
That metaphor sort of holds for the terrorist case because the "is a terrorist?" test is applied to samples from the distribution "people that travel on airplanes", and the frequency of terrorists is probably just as low in this distribution as it in the distribution "people in the USA".
Paradox of the false positive
Statisticians speak of something called the Paradox of the False Positive. Here's how that works: imagine that you've got a disease that strikes one in a million people, and a test for the disease that's 99% accurate. You administer the test to a million people, and it will be positive for around 10,000 of them - because for every hundred people, it will be wrong once (that's what 99% accurate means). Yet, statistically, we know that there's only one infected person in the entire sample. That means that your "99% accurate" test is wrong 9,999 times out of 10,000!
Honestly I find this a rather retarded metaphor. (Specially under the light of the IDEAL metaphor for rare events statistics: terrorist screening at US airports).
The metaphor is retarded because it actually leads to idiots saying things like: so when a patient gets a positive for this test, odds are he/she doesn't have it. (once saw a professor reaching this conclusion).
No "regular citizen going through its daily life is subjected to rare disease tests". People
showing medical symptoms for the disease,
after being tested for more frequent ailments,
after background check made by a MD
Get tested for a rare disease. When you find yourself in that position, the odds that you actually have it are much higher.
Oh and, Mozilla, if you read this: you DO know how stupid it is that FF allows only one firefox process at the time, right? I would really like to use FF on BOTH screens.
You can open another window. Or start Firefox with a different profile.
Yes, but one could say these are all infrastructure software, as opposed to Firefox being a "end user software"; and that users of infrastructure packages will normally know what they are doing to a MUCH higher extent than end users.
How many people (not in IT) know what Firefox stands for? How many people (who are not into IT) know what Xen, MySql and Zope are?
>Mozilla is quite understandably protective of its Firefox trademark,
I think you mean "insanely overprotective of it's Firefox trademark". Mozilla has restrictions that no other FOSS project I know of has, all to "defend their trademark". But Linux, Apache, Gnome and KDE, to name a few, are all trademarked and they don't have those restrictions.
AFAIK None of the other projects you quoted are making a a hell lot of money from their trademarks. The Linux Foundation used to license the "Linux" name in a particular way, but the money they got from it has always been peanuts.
Most times I see "community" these days, especially in a "Web 2.0" context, I think it's a euphemism for gathering personal data for marketing purposes.
Outside the "Web 2.0" context, I associate the use of the word "community" to a marketing strategy meant to make people associate their own personal identity with brand names.
My own problem is that I rather like to use applications from KDE and Gnome together. Some from KDE are better than Gnome's, some from Gnome are better than KDE's.
The problem is this huge NIH attitude that pervades both projects.
Why can't these people unify a minimal amount of API so that at least the programs are able to colaborate instead of sabotage each other? I want to use programs, not desktops environments.
That and the fact that KDE people are still so lost as to think that adding a retarded "K" before each program name is a good idea (KAlarm, KNetworkManager, KSysGuard, Kmix, KArm, KXML Editor etc)
I do use KDE these days FWIW but it is most due to Katapult, Konsole and Yakuake than anything else.
DISCLAIMER: I have a science background but not in Medical science.
More often than not, up to date university staff will be able to give students much more insightful tips on break through / newsworthy articles than, say, slashdot. Surprised? I am not.
The fact that slashdot doesn't get that many *up-to-date* good science stories is, most probably, due to the lack of more up-to-date stories submissions. It certainly isn't due to lack of slow news days anyway...
In any case, I often like slashdot better than the average popular science magazine. Once you scroll the kids making silly jokes out, you often do get a couple of PhDs (in areas I know nothing of)
making well informed comments,
explaining things,
giving books and articles suggestions.
This interaction with people in other areas, actually discussing a subject and giving literature suggestions is (for me) the best thing in the Science section.
I reckon articles in areas you know a lot of, sound boring and uninformative. But otherwise, 3 or 4 year old articles are still interesting for those of us who like science, but are not from the specific field.
Either way, I'm inclined to say, "Two wrongs don't make a right." However bad the US is, that shouldn't be an excuse for other countries to act recklessly.
Two wrongs don't make a right, but it is brutally hypocritical to just point to the issue in China, when the US has been, and still is, the leading polluting nation on Earth.
I find disgusting when people say that poorer nations should learn from our mistakes, when nobody in developed nations talks seriously about lowering their own standards of living.
You people should learn from us and keep being poor, while we, well, will just keep being rich...
Dude, I can see why some people want some symbolic ritual to take place, but perhaps you could encrypt the data with multiple keys, and then just preserve the data as usual backup gets preserved.
With multiple keys people would be prevented from looking at the data before hand, and a pgp signature would make it difficult for it to be altered (tough 25 years is a long time...)
And, no matter what Fred Brook's sacred book says, there really is a magic bullet for software development.
Did you actually read that book?
Because he specifically says that he was talking about a single magic bullet. One and only one magical bullet that would cause a ten-fold increase in productivity.
Yet you say he is wrong, and the proof we are offered for it is the use of positive and totally vague adjectives with software development factors: "sound", "proper", "quality". In other words: "hand waving".
Then you start complaining that the field is
obscured by vapour, hype and gas...
When everything thing you did was to play (your own) buzzword bingo
But in typical slashdot fashion, you whine a lot, misrepresents a classical proposition, talks with confidence, yet without any content, and all the clueless mods jump in.
What makes you think the Chinese Government actually cares what the rest of the World thinks? They have well over a trillion dollars in cash, they are getting more and more business from the West regardless of their human rights record, they know we won't do anything about it
I couldn't agree more. Change in China will have to come from inside.
For as long as anything I find on sale in Western Europe is made in China, the Chinese won't give a shit when politicians pretend to make some protest in order to make their own voters feel better.
How many people in the US knows that Clinton bombed a very large pharmaceutical plant in Africa leading to deaths in the hundreds of thousands? Saying "ops, my bad" afterwards, and blocking UN investigations in the actual number of dead people (due to lack of medicine).
How many people in the US knows that the US supported Saddam gassing his own people in the 80s?
Compare that to China where people who are Falun Gong (a religious movement) get to suffer round-the-clock torture, including mutilation and permanent debilitation, simply to force them to renounce their religious beliefs. I invite you to read up on it, and then tell us how that compares to fake menstrual blood, loud music, and all the other horrible things that go on at Gitmo.
Compare that Falun Gong to destroying an entire country on an illegal invasion for no actual reason.
I'm serious. You talk about mutilation, and only mentions Gitmo. Amazing.
May I remind you of the war Bush started? I wonder what kind of mental gymnastics one goes through to forget all the people who got bombed to death (or mutilation) in Iraq.
And that is after forgetting all the people who had died already due to that embargo...
What lack of readability? All my Perl stuff is very readable, and it's easy to follow what's going on.
Myself I almost never have a problem reading my own code. It is generally other people's code that are hard to read. You will only have an honest idea of how readable your code is when somebody else has to start maintaining it.
I think a robust python codebase requires strong committment to quality coding standards, and an automatic checker to enforce at least some of them. I'm not sure if pychecker cuts it. Otherwise it's just too easy to do potentially evil things (like adding members to a class from outside the class, which is convenient but very confusing too, or getting into the whole import mess with too many from import *). Oh and unit tests, unit tests...
Your assertion that Java 'dominates the web' is laughable, as applets have been a near-total failure, replaced wholesale by Flash- and AJAX-based systems, while on the backend PHP, Perl and more recently Python and Ruby have eaten Java's lunch.
There are servlets, and there is also stuff like GWT abstracting javascript into java. Can you get anything like GWT using Python or Ruby? (it's a serious question, if there is stuff like that I would like to look at it)
Re:Sometimes the correct answer is the simplest
on
Why Corporates Hate Perl
·
· Score: 3, Insightful
"You have a problem, and you discover you can solve it by using regular expressions. Well, now you have another problem"
I read this in the opening of a regular expression chapter of a Python book
by Tim Peters (though the quote is certainly not exact); so I am unsure whether he was quoting somebody, or not. Anyway, the point stands.
Look I actually agree with you, but the point was that the GP was making was that if we had a landmark case in an area where people can't be so easily taken by emotions, it would be easier to have a rational discussion about it, and establish a sensible precedent on crypto keys.
As Prof.Lessig learned when he went to SCOTUS to try to shorten the copyright expiration, in real life, being able to present compelling arguments is a big piece of the process. (just google for Lessig's comments on the case, it is too late over here).
Having a perfect candidate for the "think about the children" argument is/not/ good, if anything because we all know that so many people buy it.
The problem with this case is that it is the textbook example of the think about the children argument to bypass regular civil rights. As such it could just as well end up being used to throw (more) smoke at the public understanding of the merits of being entitled to privacy.
In any case, it is good to see judges in the US (or anywhere else) making into the news for taking the right stand regarding governmental search limits.
Slashdot Mirror
You said I am grasping straws. How is falling back to mathematical formalism "grasping straws"? From my point of view, you are the one "waving hands".
The reason the test is normally applied to a selected group of people was nowhere addressed in my post. In fact you actually concede my point in your email.
My point was:
In this second sample the frequency of people that have disease X, should be quite different than on the first sample taken from "anyone in USA".
It is wrong to take statistical frequencies measured against a sample taken from distribution Dn, and then proceed to make assertions against a sample taken from another distribution Ds (where Ds != Dn).
Care to point what is wrong with that?
Following from your post on http://slashdot.org/comments.pl?sid=973141&cid=25129053
What is wrong with this analysis is the following:
You measure frequency F (freq of disease X), and a given test (T) effectiveness on sample A taken from a statistical distribution Dn (where Dn ~ "samples randomly anyone: 1. in the USA"). The test effectiveness being T_ef(A) where A was sampled from Dn.
If you apply the test against sample A then you would have that retarded frequency relations you felt necessary to copy from wikipedia. Are you surprised? You just measured that sample A was large enough to reasonably mimic the events frequency to be expected from the distribution the sample was taken from.
My point being, that illustration is deceiving because in practice that rare disease tests are not routinely applied against the general population (aka sample A). The tests are normally applied to a sample of people that do not get sampled from Dn. The test subjects get sampled from distribution Ds where
Ds ~ samples randomly anyone that
In this second sample the frequency of people that have disease X, should be quite different than on the first sample taken from "anyone in USA".
It is wrong to take statistical frequencies measured against a sample taken from distribution Dn, and then proceed to make assertions against a sample taken from another distribution Ds (where Ds != Dn).
That metaphor sort of holds for the terrorist case because the "is a terrorist?" test is applied to samples from the distribution "people that travel on airplanes", and the frequency of terrorists is probably just as low in this distribution as it in the distribution "people in the USA".
Got it?
Paradox of the false positive Statisticians speak of something called the Paradox of the False Positive. Here's how that works: imagine that you've got a disease that strikes one in a million people, and a test for the disease that's 99% accurate. You administer the test to a million people, and it will be positive for around 10,000 of them - because for every hundred people, it will be wrong once (that's what 99% accurate means). Yet, statistically, we know that there's only one infected person in the entire sample. That means that your "99% accurate" test is wrong 9,999 times out of 10,000!
Honestly I find this a rather retarded metaphor. (Specially under the light of the IDEAL metaphor for rare events statistics: terrorist screening at US airports).
The metaphor is retarded because it actually leads to idiots saying things like: so when a patient gets a positive for this test, odds are he/she doesn't have it. (once saw a professor reaching this conclusion).
No "regular citizen going through its daily life is subjected to rare disease tests". People
Get tested for a rare disease. When you find yourself in that position, the odds that you actually have it are much higher.
Oh and, Mozilla, if you read this: you DO know how stupid it is that FF allows only one firefox process at the time, right? I would really like to use FF on BOTH screens.
You can open another window. Or start Firefox with a different profile.
How many people (not in IT) know what Firefox stands for? How many people (who are not into IT) know what Xen, MySql and Zope are?
>Mozilla is quite understandably protective of its Firefox trademark,
I think you mean "insanely overprotective of it's Firefox trademark". Mozilla has restrictions that no other FOSS project I know of has, all to "defend their trademark". But Linux, Apache, Gnome and KDE, to name a few, are all trademarked and they don't have those restrictions.
AFAIK None of the other projects you quoted are making a a hell lot of money from their trademarks. The Linux Foundation used to license the "Linux" name in a particular way, but the money they got from it has always been peanuts.
>help build community
Most times I see "community" these days, especially in a "Web 2.0" context, I think it's a euphemism for gathering personal data for marketing purposes.
Outside the "Web 2.0" context, I associate the use of the word "community" to a marketing strategy meant to make people associate their own personal identity with brand names.
The problem is this huge NIH attitude that pervades both projects.
Why can't these people unify a minimal amount of API so that at least the programs are able to colaborate instead of sabotage each other? I want to use programs, not desktops environments.
That and the fact that KDE people are still so lost as to think that adding a retarded "K" before each program name is a good idea (KAlarm, KNetworkManager, KSysGuard, Kmix, KArm, KXML Editor etc)
I do use KDE these days FWIW but it is most due to Katapult, Konsole and Yakuake than anything else.
More often than not, up to date university staff will be able to give students much more insightful tips on break through / newsworthy articles than, say, slashdot. Surprised? I am not.
The fact that slashdot doesn't get that many *up-to-date* good science stories is, most probably, due to the lack of more up-to-date stories submissions. It certainly isn't due to lack of slow news days anyway...
In any case, I often like slashdot better than the average popular science magazine. Once you scroll the kids making silly jokes out, you often do get a couple of PhDs (in areas I know nothing of)
This interaction with people in other areas, actually discussing a subject and giving literature suggestions is (for me) the best thing in the Science section.
I reckon articles in areas you know a lot of, sound boring and uninformative. But otherwise, 3 or 4 year old articles are still interesting for those of us who like science, but are not from the specific field.
There was also something about ReiserFS being able to avoid writing on the disk very often, which made it more suitable for laptops.
Either way, I'm inclined to say, "Two wrongs don't make a right." However bad the US is, that shouldn't be an excuse for other countries to act recklessly.
Two wrongs don't make a right, but it is brutally hypocritical to just point to the issue in China, when the US has been, and still is, the leading polluting nation on Earth.
I find disgusting when people say that poorer nations should learn from our mistakes, when nobody in developed nations talks seriously about lowering their own standards of living.
You people should learn from us and keep being poor, while we, well, will just keep being rich...
They create stories, "spin" them I believe is the term they use, all the time and we love 'em for it.
Speak for yourself, dude.
With multiple keys people would be prevented from looking at the data before hand, and a pgp signature would make it difficult for it to be altered (tough 25 years is a long time...)
GNU Hurd was announced years before Linux, and look how far that project got.
In computing what counts is shipping / release date.
And, no matter what Fred Brook's sacred book says, there really is a magic bullet for software development.
Did you actually read that book?
Because he specifically says that he was talking about a single magic bullet. One and only one magical bullet that would cause a ten-fold increase in productivity.
Yet you say he is wrong, and the proof we are offered for it is the use of positive and totally vague adjectives with software development factors: "sound", "proper", "quality". In other words: "hand waving".
Then you start complaining that the field is
obscured by vapour, hype and gas...
When everything thing you did was to play (your own) buzzword bingo
But in typical slashdot fashion, you whine a lot, misrepresents a classical proposition, talks with confidence, yet without any content, and all the clueless mods jump in.
What makes you think the Chinese Government actually cares what the rest of the World thinks? They have well over a trillion dollars in cash, they are getting more and more business from the West regardless of their human rights record, they know we won't do anything about it
I couldn't agree more. Change in China will have to come from inside.
For as long as anything I find on sale in Western Europe is made in China, the Chinese won't give a shit when politicians pretend to make some protest in order to make their own voters feel better.
How many people in the US knows that the US supported Saddam gassing his own people in the 80s?
Compare that to China where people who are Falun Gong (a religious movement) get to suffer round-the-clock torture, including mutilation and permanent debilitation, simply to force them to renounce their religious beliefs. I invite you to read up on it, and then tell us how that compares to fake menstrual blood, loud music, and all the other horrible things that go on at Gitmo.
Compare that Falun Gong to destroying an entire country on an illegal invasion for no actual reason.
I'm serious. You talk about mutilation, and only mentions Gitmo. Amazing.
May I remind you of the war Bush started? I wonder what kind of mental gymnastics one goes through to forget all the people who got bombed to death (or mutilation) in Iraq.
And that is after forgetting all the people who had died already due to that embargo...
1. Any game requires rules, even if you don't like them. Breaking the rules, even if it's a stupid rule, still means you broke the rules.
Like the danish sailing due? Who switched boats because theirs was broken, and got the golden medal even though the rules forbid repairing any boat.
Last I checked the protest regarding that one rule ended into nothing... I mean, its not like this is a fairly straight rule violation, right?
What lack of readability? All my Perl stuff is very readable, and it's easy to follow what's going on.
Myself I almost never have a problem reading my own code. It is generally other people's code that are hard to read. You will only have an honest idea of how readable your code is when somebody else has to start maintaining it.
I think a robust python codebase requires strong committment to quality coding standards, and an automatic checker to enforce at least some of them. I'm not sure if pychecker cuts it. Otherwise it's just too easy to do potentially evil things (like adding members to a class from outside the class, which is convenient but very confusing too, or getting into the whole import mess with too many from import *). Oh and unit tests, unit tests...
When I was coding python, I used to love pylint.
Your assertion that Java 'dominates the web' is laughable, as applets have been a near-total failure, replaced wholesale by Flash- and AJAX-based systems, while on the backend PHP, Perl and more recently Python and Ruby have eaten Java's lunch.
There are servlets, and there is also stuff like GWT abstracting javascript into java. Can you get anything like GWT using Python or Ruby? (it's a serious question, if there is stuff like that I would like to look at it)
s!(?:^|\w*/|\.\./|\./)[^\s,@;:]*(?<=/)([^\s,@;:]+?)(?=[\s,@;:]|$)!$1!g;
"You have a problem, and you discover you can solve it by using regular expressions.
Well, now you have another problem"
I read this in the opening of a regular expression chapter of a Python book by Tim Peters (though the quote is certainly not exact); so I am unsure whether he was quoting somebody, or not. Anyway, the point stands.
As Prof.Lessig learned when he went to SCOTUS to try to shorten the copyright expiration, in real life, being able to present compelling arguments is a big piece of the process. (just google for Lessig's comments on the case, it is too late over here).
Having a perfect candidate for the "think about the children" argument is /not/ good, if anything because we all know that so many people buy it.
In any case, it is good to see judges in the US (or anywhere else) making into the news for taking the right stand regarding governmental search limits.