"My god, given a person's name, I can find her address! And given her address, I can get a satellite image of her neighborhood! That must imply that that person is an undercover CIA agent!"
It's been a while since I learned about this stuff, but hasn't it been mathematically proven that it's futile to try to write a program that, presented with an arbitrary (expressable) problem, will write a program to solve that problem?
Yes, but the problem is that the portion of the universe that is in our airspace is constantly changing. And at some certain radius up, you would have to go faster than light to remain within our airspace.
But I go through this argument with my colleagues, who say that using short, descriptive variable names 'should' be enough as long as the code is well-organized.
And chances are, the average American would consider me wacky far liberal. But this is just ridiculous.
He's not prohibited to travel. He's prohibited to travel via the services of a private company without accepting their policies for use of their services.
He could drive. He could buy a plane. He could walk. He could hitchhike, for goodness' sake. He could even start a private airline company that doesn't require ID.
Probably neither. Are you familiar with the word "approximation"?
And even in the extremely unlikely case that exactly one terabit exactly fits in exactly one square inch, the answer to your question is contained in the sentence you quoted anyway: "one trillion bits of data".
(1) For Microsoft to change the WinAPI in such a way that Firefox will not work under Longhorn while ensuring that the vast majority of other programs continue to work under Longhorn would require a phenomenal amount of time and effort, and therefore money.
(2) It would be overcome in the next version of Firefox, which would come out within a week or so of Longhorn.
Dislike Microsoft all you want, but they're not idiots.
The relative dearth of drivers has been an issue ever since Linux existed. It hasn't killed Linux yet, why would it kill Linux now? Especially considering that it's becoming less of an issue as time passes - Linux is catching up on the driver front (relative to five years ago or whatever).
Of course, this doesn't mean that it's not an important issue, and it doesn't even mean that it won't cause some cap on Linux' marketshare at some point in the future. But kill Linux? Uh, no.
"As of the 1st of March 2005, Australian ISPs and web hosts will face fines of up to $55,000 if they can be used to access child pornography and do not refer the information to the police. Yikes. How on earth are the ISPs (and web hosts -- like my own very small-time and humble company) supposed to enforce this?"
Dispite my last post being rather inncohrant(4 white russians will do that to you), I made a perfectly resonable argument throughout the thread. An argument which you would not address.
This will be my last post attempting to explain this to you.
Namley that theese people have A 8% SUCCSESS RATE!
That is your "perfectly reasonable argument"? That they have an eight percent success rate? And, I assume, therefore they're not seeing something real?
Well, that is, quite simply, wrong. Explanation below.
I don't care that it should of been 5% or what ever number their magical algorythm says it should be.
First of all, their "algorithm" is not "magical". They're simply counting the number of zeroes generated versus the number of ones generated. For each possible such result, there is a percent chance that it will happen in any given try. This chance is calculable - easily - by elementary probability theory. There is some level of ones generated, versus zeroes generated, that will be met or exceeded, according to probability, in five percent of the cases, on average, in the long run. This is straightforward and undeniable probability theory. That's the number that they're using. This is not "magic", it is simple, elementary mathematics that no one with any mathematical clue whatsoever will disagree with.
A 8% success rate is not a vaild scientific result.
That's an incredible overgeneralization.
When the predicted success rate is anything - anything - besides eight percent, then an eight percent success rate is a valid scientific result, given enough attempts. You are, plain and simply, overgeneralizing, and drawing a conclusion based upon your overgeneralization.
Consider this: Assume that you have a pack of cards. You pick a card from it, at random, and don't show it to me. I successfully predict what card it is about eight percent of the time, in twelve attempts.
Math argues that, on average, in the long run, I should be able to predict it correctly only about two percent of the time. But this was not "the long run", this was twelve attempts. It's easily believable that that eight percent was a fluke.
But what if I do it another twelve? And another twelve? And many, many, many more twelves of times? And we're up to some incredibly huge number of attempts, and my success rate is still eight percent, when math says it should, on average, in the long run, be less than that?
At no point, do you begin suspecting that something is going on? I'm cheating, I'm psychic, the deck is rigged, you're drunk, whatever?
At no point, that eight percent becomes "scientifically valid"?
If you still believe that, all I'm going to say is, well, sorry, but you don't have an understanding of what "scientifically valid" means.
Furthermore, the methology used is absolute insanity. A wiccan spell on bin laden? Nothing more important happened on that day? How far do they look back? A minute? an hour? A week? Until they see the magical 5% threshold? How do they define the timeframe under which the event is taking place?
These questions were answered while you were drunk. Several times. Go back and read.
It is plain old psedo-scienece, and I find it hard to beleive that anyone would give it credence.
Look. On a gut level, I have a very hard time believing these people. In fact, I'm very inclined to believe that they're wrong.
However, that doesn't mean they're wrong. And neither does just screaming "EIGHT PERCENT IS NOT SCIENTIFICALLY VALID!!!" over and over again, especially without an understanding of what it means to be scientifically valid.
What will show that they're wrong is one of these things:
(1) Proof that they rigged their data.
(2) A demonstrable mistake in their mathematics.
(3) Many more trials which show things coming back into line with mathematical expectations.
Do not agree with you. I have not seen any evidence to support what you claim. Having read the article and having known about this even before the/. posting. There are possibly plenty of times a major deviation happens and there are no major events and there are also times events happen with no major deviation.
Of course there are plenty of times a major event deviation happens and there are no blah blah blah. That's what the article is saying, and that's what my post - the one you're directly responding to - is saying that nobody denies.
But it has nothing to do with anything. They're not claiming "major deviation == major event". They're claiming "major event implies a greater than expected chance of major deviation". And again, the "debunking" article doesn't address that claim.
Until I see all the data noting any kind of event that could be seen as "major" it's appears to be a case of going back into a huge amount of data and trying to connect the dots by highlighting what you want to find and ignoring that which you don't. Every time there's a deviation they go and actively look to match it to something. Not very scientific to me.
Again, like many people here, you're assuming, and in fact you're assuming incorrectly.
They do not look for deviations and then go look for something to match it to. For exactly the reason you (and others) describe. Their website is explicit about this.
Rather, what they claim to do is:
(1) In general, not look at the data.
(2) Decide that they arbitrarily consider an event sufficiently "major".
(3) Decide upon a timeframe around that event which they arbitrarily consider to be sufficiently linked to the event.
(4) Upon such a decision, then, and only then, go back and look at the data in that timeframe, and count the number of zeroes and the number of ones generated during that timeframe.
And their claim is, that so far, they've found major deviations using this method significantly more often than is predicted by assuming that steps (2) and (3) are meaningless.
I think that this article debunks it pretty well.
Again, the article debunks the claim that you have assumed that they're making. Which is not the claim that they're making.
But when you only partially show your hand, giving people what you want to show them, it becomes sensationalistic.
In what way have they only "partially" showed their hand? Exactly what way? Before you answer, please be advised that their website is chock full of information.
The article points out that the selection of events as "major" seems to be done after looking at the RNG output ("You are selecting your data.") rather than blind.
Well, then, the article is claiming things contrary to what the GCP itself is claiming.
The Global Consciousness Project's web site does not emphasize what you say it claims. It emphasizes the data points that support the affirmative hypothesis.
Perhaps it "emphasizes" what you say. But its methodology is what I described, and not debunked by the article.
Those (in the first link I gave you) are the timeframes that they decided to look at, after the event happened, and before looking at the data.
In any of those particular time frames, a certain number of zeroes were generated, and a certain number of ones were generated.
The particular number of zeroes generated versus ones generated has a certain chance of happening (assuming all is truly random). That chance is described in the second link that I gave you, for each event in question.
Every one of those events have those numbers listed. Not all of those events are "spikes" in the sense that we have been using - i.e. a five percent or less chance of happening. The ones listed in bright red are those. And those are the "successes" that we have been discussing.
You(and they) stll have not awnsred where the cut-off line is for a spike being part of a SWE.
You're misunderstanding (again). They're not saying "a spike of such-and-such a size indicates a significant world event".
But ignoring that, and assuming that you instead stated that I have not answered what size spike is considered a success:
Yes, I have. Multiple times. Five percent. One in twenty.
Thank you. It's to bad it took 3 hours and god knows how many comments, but, finally, you did it.
I'm sorry I didn't respond to that particular request of yours, multiple times. I was busy responding to other requests of yours. Okay? And would it have been so difficult to just look it up yourself in the first place? Did you need me to do it?
Unforutnatly, Let's take a look at how they define a signficant world event shall we?
That is simply not relevant.
What is relevant is that out of the times that they chose to check whether there was a statistically significant spike or not, there was a statistically significant spike a larger percentage of the time than would be predicted as average by the laws of probability. And the degree to which that larger percentage was larger is claimed to be surprising.
Why they chose to check at the times that they chose to check is not relevant to the fact that they successfully predicted at a greater than expected rate when they did choose to predict.
Admittedly, their hypotheses is about this "major event" stuff, but that is not strictly what they are measuring. They are measuring their ability to predict when statistically significant spikes occur. And, so far, they have done so at a higher rate than mathematically normal.
All that and they can only manage a 8% success rate?
I have no idea what this is supposed to imply.
That "eight percent success rate" is expected, mathematically, to be five percent. Are you saying that with the more events they consider to be "significant", the higher you would expect their actual success rate to diverge from the mathematically expected success rate of five percent?
If so, well, mathematics disagrees with you. If not, what do you mean?
Slashdot Mirror
How is this at all relevant?
"My god, given a person's name, I can find her address! And given her address, I can get a satellite image of her neighborhood! That must imply that that person is an undercover CIA agent!"
Morons.
Kludge.
It's been a while since I learned about this stuff, but hasn't it been mathematically proven that it's futile to try to write a program that, presented with an arbitrary (expressable) problem, will write a program to solve that problem?
(1) It would have to be Episode Zero.
(2) The scrolling intro always gives the episode number in Roman numerals.
(3) The Romans had no numeral for zero.
Yes, but the problem is that the portion of the universe that is in our airspace is constantly changing. And at some certain radius up, you would have to go faster than light to remain within our airspace.
But I go through this argument with my colleagues, who say that using short, descriptive variable names 'should' be enough as long as the code is well-organized.
Your colleagues are idiots.
Give me the right to vote, and I'll pay your damn taxes. Till then, up yours. I've got tea, you've got a harbor.
People seem to be interpreting this as "Sun, which made UNIX, which is X, and which is Y".
However, it can also be parsed as "Sun, which made UNIX (which is X) into Y".
The latter, while highly debatable to say the least, is at least opinion rather than misstatement of fact.
And chances are, the average American would consider me wacky far liberal. But this is just ridiculous.
He's not prohibited to travel. He's prohibited to travel via the services of a private company without accepting their policies for use of their services.
He could drive. He could buy a plane. He could walk. He could hitchhike, for goodness' sake. He could even start a private airline company that doesn't require ID.
Probably neither. Are you familiar with the word "approximation"?
And even in the extremely unlikely case that exactly one terabit exactly fits in exactly one square inch, the answer to your question is contained in the sentence you quoted anyway: "one trillion bits of data".
(1) For Microsoft to change the WinAPI in such a way that Firefox will not work under Longhorn while ensuring that the vast majority of other programs continue to work under Longhorn would require a phenomenal amount of time and effort, and therefore money.
(2) It would be overcome in the next version of Firefox, which would come out within a week or so of Longhorn.
Dislike Microsoft all you want, but they're not idiots.
The relative dearth of drivers has been an issue ever since Linux existed. It hasn't killed Linux yet, why would it kill Linux now? Especially considering that it's becoming less of an issue as time passes - Linux is catching up on the driver front (relative to five years ago or whatever).
Of course, this doesn't mean that it's not an important issue, and it doesn't even mean that it won't cause some cap on Linux' marketshare at some point in the future. But kill Linux? Uh, no.
... I am unsure, from the trailer, whether it's overly self-serious or whether it's faux-overly-self-serious humor.
Could someone please enlighten me?
"As of the 1st of March 2005, Australian ISPs and web hosts will face fines of up to $55,000 if they can be used to access child pornography and do not refer the information to the police. Yikes. How on earth are the ISPs (and web hosts -- like my own very small-time and humble company) supposed to enforce this?"
Easy:
Dear Police,
My ISP can be used to access child pornography.
Thanks,
Every ISP on Earth
Base score: 1.
First mod: Overrated. Total: 0.
Next mod: Overrated. Total: -1.
"Overrated" down to GNAA levels, without ever having been rated up in the first place? What is this, pre-emptive overrating?
Two separate moderators were worried that some future mod might potentially mod it as "funny", or something?
Bleh.
"IsNot" IsNot "Patentable"
Perhaps if he can solve a few NP-Complete problem
If you solve one NP-Complete problem, you have solved all NP-Complete problems. That's what the "Complete" part of "NP-Complete" means.
Dispite my last post being rather inncohrant(4 white russians will do that to you), I made a perfectly resonable argument throughout the thread. An argument which you would not address.
This will be my last post attempting to explain this to you.
Namley that theese people have A 8% SUCCSESS RATE!
That is your "perfectly reasonable argument"? That they have an eight percent success rate? And, I assume, therefore they're not seeing something real?
Well, that is, quite simply, wrong. Explanation below.
I don't care that it should of been 5% or what ever number their magical algorythm says it should be.
First of all, their "algorithm" is not "magical". They're simply counting the number of zeroes generated versus the number of ones generated. For each possible such result, there is a percent chance that it will happen in any given try. This chance is calculable - easily - by elementary probability theory. There is some level of ones generated, versus zeroes generated, that will be met or exceeded, according to probability, in five percent of the cases, on average, in the long run. This is straightforward and undeniable probability theory. That's the number that they're using. This is not "magic", it is simple, elementary mathematics that no one with any mathematical clue whatsoever will disagree with.
A 8% success rate is not a vaild scientific result.
That's an incredible overgeneralization.
When the predicted success rate is anything - anything - besides eight percent, then an eight percent success rate is a valid scientific result, given enough attempts. You are, plain and simply, overgeneralizing, and drawing a conclusion based upon your overgeneralization.
Consider this: Assume that you have a pack of cards. You pick a card from it, at random, and don't show it to me. I successfully predict what card it is about eight percent of the time, in twelve attempts.
Math argues that, on average, in the long run, I should be able to predict it correctly only about two percent of the time. But this was not "the long run", this was twelve attempts. It's easily believable that that eight percent was a fluke.
But what if I do it another twelve? And another twelve? And many, many, many more twelves of times? And we're up to some incredibly huge number of attempts, and my success rate is still eight percent, when math says it should, on average, in the long run, be less than that?
At no point, do you begin suspecting that something is going on? I'm cheating, I'm psychic, the deck is rigged, you're drunk, whatever?
At no point, that eight percent becomes "scientifically valid"?
If you still believe that, all I'm going to say is, well, sorry, but you don't have an understanding of what "scientifically valid" means.
Furthermore, the methology used is absolute insanity. A wiccan spell on bin laden? Nothing more important happened on that day? How far do they look back? A minute? an hour? A week? Until they see the magical 5% threshold? How do they define the timeframe under which the event is taking place?
These questions were answered while you were drunk. Several times. Go back and read.
It is plain old psedo-scienece, and I find it hard to beleive that anyone would give it credence.
Look. On a gut level, I have a very hard time believing these people. In fact, I'm very inclined to believe that they're wrong.
However, that doesn't mean they're wrong. And neither does just screaming "EIGHT PERCENT IS NOT SCIENTIFICALLY VALID!!!" over and over again, especially without an understanding of what it means to be scientifically valid.
What will show that they're wrong is one of these things:
(1) Proof that they rigged their data.
(2) A demonstrable mistake in their mathematics.
(3) Many more trials which show things coming back into line with mathematical expectations.
Do not agree with you. I have not seen any evidence to support what you claim. Having read the article and having known about this even before the /. posting. There are possibly plenty of times a major deviation happens and there are no major events and there are also times events happen with no major deviation.
Of course there are plenty of times a major event deviation happens and there are no blah blah blah. That's what the article is saying, and that's what my post - the one you're directly responding to - is saying that nobody denies.
But it has nothing to do with anything. They're not claiming "major deviation == major event". They're claiming "major event implies a greater than expected chance of major deviation". And again, the "debunking" article doesn't address that claim.
Until I see all the data noting any kind of event that could be seen as "major" it's appears to be a case of going back into a huge amount of data and trying to connect the dots by highlighting what you want to find and ignoring that which you don't. Every time there's a deviation they go and actively look to match it to something. Not very scientific to me.
Again, like many people here, you're assuming, and in fact you're assuming incorrectly.
They do not look for deviations and then go look for something to match it to. For exactly the reason you (and others) describe. Their website is explicit about this.
Rather, what they claim to do is:
(1) In general, not look at the data.
(2) Decide that they arbitrarily consider an event sufficiently "major".
(3) Decide upon a timeframe around that event which they arbitrarily consider to be sufficiently linked to the event.
(4) Upon such a decision, then, and only then, go back and look at the data in that timeframe, and count the number of zeroes and the number of ones generated during that timeframe.
And their claim is, that so far, they've found major deviations using this method significantly more often than is predicted by assuming that steps (2) and (3) are meaningless.
I think that this article debunks it pretty well.
Again, the article debunks the claim that you have assumed that they're making. Which is not the claim that they're making.
But when you only partially show your hand, giving people what you want to show them, it becomes sensationalistic.
In what way have they only "partially" showed their hand? Exactly what way? Before you answer, please be advised that their website is chock full of information.
The article points out that the selection of events as "major" seems to be done after looking at the RNG output ("You are selecting your data.") rather than blind.
Well, then, the article is claiming things contrary to what the GCP itself is claiming.
The Global Consciousness Project's web site does not emphasize what you say it claims. It emphasizes the data points that support the affirmative hypothesis.
Perhaps it "emphasizes" what you say. But its methodology is what I described, and not debunked by the article.
Wait....I thought we argeed that they where corasponding current events to past spikes way back 10 or so comments ago. Bah.
In a sense, but not in the sense that you're harping on.
I am too drunk to continue.
That explains a lot.
No, that's simply not the case.
Those (in the first link I gave you) are the timeframes that they decided to look at, after the event happened, and before looking at the data.
In any of those particular time frames, a certain number of zeroes were generated, and a certain number of ones were generated.
The particular number of zeroes generated versus ones generated has a certain chance of happening (assuming all is truly random). That chance is described in the second link that I gave you, for each event in question.
Every one of those events have those numbers listed. Not all of those events are "spikes" in the sense that we have been using - i.e. a five percent or less chance of happening. The ones listed in bright red are those. And those are the "successes" that we have been discussing.
You(and they) stll have not awnsred where the cut-off line is for a spike being part of a SWE.
You're misunderstanding (again). They're not saying "a spike of such-and-such a size indicates a significant world event".
But ignoring that, and assuming that you instead stated that I have not answered what size spike is considered a success:
Yes, I have. Multiple times. Five percent. One in twenty.
Oh good god.
Thank you. It's to bad it took 3 hours and god knows how many comments, but, finally, you did it.
I'm sorry I didn't respond to that particular request of yours, multiple times. I was busy responding to other requests of yours. Okay? And would it have been so difficult to just look it up yourself in the first place? Did you need me to do it?
Unforutnatly, Let's take a look at how they define a signficant world event shall we?
That is simply not relevant.
What is relevant is that out of the times that they chose to check whether there was a statistically significant spike or not, there was a statistically significant spike a larger percentage of the time than would be predicted as average by the laws of probability. And the degree to which that larger percentage was larger is claimed to be surprising.
Why they chose to check at the times that they chose to check is not relevant to the fact that they successfully predicted at a greater than expected rate when they did choose to predict.
Admittedly, their hypotheses is about this "major event" stuff, but that is not strictly what they are measuring. They are measuring their ability to predict when statistically significant spikes occur. And, so far, they have done so at a higher rate than mathematically normal.
All that and they can only manage a 8% success rate?
I have no idea what this is supposed to imply.
That "eight percent success rate" is expected, mathematically, to be five percent. Are you saying that with the more events they consider to be "significant", the higher you would expect their actual success rate to diverge from the mathematically expected success rate of five percent?
If so, well, mathematics disagrees with you. If not, what do you mean?
You: Regardless, it's detail that appears to be missing from their description of their experimental method,
Me: Does it? Have you actually read their material, or just the brief summary linked to by Slashdot?
You: Yes, I've read as much as I could bear of their material.
Well, you couldn't bear very much, then. Because that detail that you claim "appears to be missing" actually appears quite early in their material.
Via what Time frame? I have asked you this three times. I have recived no response.
Jesus Christ. Look on their friggin' website.
Here is each event, and the time frames used for each.
Here is a listing of the observed spike's probability for each event.