I think a good analogy would be a post office making all its PO boxes open when you knock on them. He opened his box and noticed that they were horribly designed, so then he knocked on all of them and took picture of the contents, which he sent to a local journalist as proof of the poor design that he had discovered.
Sure, what he did was overboard. But having such a poor security mechanism on their mail boxes is most certainly the fault of the post office. He should be blamed for the publicising (unless it can be shown that he first went to the post office and gave them reasonable warning), and the post office blamed for the poor design of the mail boxes.
Truth is, law enforcement isn't going to take you seriously if you say you have tracked down your laptop. Rather just get it insured and have your files backed up somewhere else.
It's all about whether the sampling is *actually* random, and when you have a very diverse population, increasing the sample size to cover a broader range of situations is a good way of making the samples more random.
You can argue stats all day, but unless the sampling isn't biased (which it is, because we live in a real world and we don't control the population) they will never be as good as the theory says they are.
Let's take an extreme example. Let's say that we have a sample size of 1. Now, I think you'll agree, even if we have a perfectly random selection process, our conclusions that we draw won't be particularly accurate. Unless, of course, our population is also 1.
On the other hand, let's look at a population of Infinity. The trouble here is that with our finite sampling process, because we live in the real world we have no way of knowing if our sample is actually random, because there are infinite location we could choose from. We would need a sampling process capable of dealing with an infinite population, which we don't have.
In the middle, let's look at a population of eg. the USA, 300,000,000. Now the problem becomes, "Is our sampling process representative of the actual population?" Here is the difficult part, because when you actually try to collect the data, you find that people don't always answer polls, so immediately your sampling is biased towards people who *do* answer polls. Then say they are collected by telephone. This limits the poll to people who a) have phones b) YouGov has their number c) are near their phone and not eg. at work, looking after kids, driving etc when YouGov calls d) answer their phone to unknown numbers e) answer polls. I think you'll agree that this is no longer an accurate representation of the population, because it biases against people who don't have phones, who are busy, who avoid telemarketers and who are willing to give their opinion on a touchy subject to a stranger.
One way you can reduce the amount of bias is by using a larger number of methods to sample opinions. Eg internet, street interviews, TV call-in hotlines. Of course, all of these have their own sampling bias just like the phone one. So, to reduce the amount of bias the number of methods should be very large. Let's say we have 50 different ways of getting peoples' opinions, and we weight them all equally. Suddenly, this means just 20 phone calls, 20 street interviews, 20 TV hotline callins etc. which doesn't seem nearly as robust now, does it, when you're trying to accurately portray the opinion of the entire population of the USA. Which is why people are saying that it is *only* 1000 samples, and even then there will still be residual bias towards people who don't want to answer polls.
The modern Tea Party movement actually started fairly similarly to OWS, but was derailed and turned into a "hur hur evolution sucks and global warming is false" spectacle.
OWS wasn't even about political control over corporations, they just wanted the bankers who had caused and benefited from the stock market collapse to be brought to justice.
Because then he would have been a "coward" for being anonymous, and probably "not standing up for what he believes in". He did what he had to to get his info taken as seriously as possible. He essentially sacrificed himself for a higher cause. The thing is, he can still do more 'good' by remaining free, so there is no problem with him doing his best to avoid US law enforcement.
These calculations are assuming a unimodal distribution, which isn't necessarily true (in fact, almost certainly false) when it comes to something like political affiliation or opinion, which is multimodal. People aren't going to fit a nice bell curve for something like "on a scale of 0 to 10, with 0 being ultra-left and 10 being ultra-right, where do your political affiliations lie?", there are going to be distict bumps on each side.
I think the real question is how did they make their sampling 'random'. It is extremely easy to (accidentally or on purpose) introduce bias based on where and when the samples were taken. This is why using a larger sample helps, because it means that you can choose from a broader spectrum of locations.
Are you forgetting the Occupy Wall Street and the (original) Tea Party movements? The government is just much better at derailing protest movements these days than they were back in Nixon's time.
I think six, with two control systems each running a set of 3 in 60-degree-offset triangles. That way even a control system failure would still result in a recoverable (hopefully) situation.
Oh really, and what would you have done if you were the military and the current president was trying to turn the country into a theocracy, replacing police, judges and school principles with his cronies?
Also, apparently the officials being arrested and the TV stations being closed are because they were encouraging their supporters to be violent.
True, but it does complicate things. Seriously, if you're going to provide a tool with incompatible functionality, at least name it something different. At the minimum append a version number.
Slashdot Mirror
I think the real issue drawing the credibility into account is the timing of this news story...
Deflate has a maximum theoretical compression ratio of 1032:1, so I don't think a few people zipping things would be enough.
And possibly close their US embassy.
I think a good analogy would be a post office making all its PO boxes open when you knock on them. He opened his box and noticed that they were horribly designed, so then he knocked on all of them and took picture of the contents, which he sent to a local journalist as proof of the poor design that he had discovered.
Sure, what he did was overboard. But having such a poor security mechanism on their mail boxes is most certainly the fault of the post office. He should be blamed for the publicising (unless it can be shown that he first went to the post office and gave them reasonable warning), and the post office blamed for the poor design of the mail boxes.
Standard english grammar has 1.1 bits of information per character (at least in larger text bodies).
Aaannnd it's Slashdotted.
Truth is, law enforcement isn't going to take you seriously if you say you have tracked down your laptop. Rather just get it insured and have your files backed up somewhere else.
It's all about whether the sampling is *actually* random, and when you have a very diverse population, increasing the sample size to cover a broader range of situations is a good way of making the samples more random.
You can argue stats all day, but unless the sampling isn't biased (which it is, because we live in a real world and we don't control the population) they will never be as good as the theory says they are.
The new Sony Xperia Z is waterproof to 1 meter for 30 minutes. Try again.
To be honest, what I'd really like is anti-shatter.
https://yourlogicalfallacyis.com/appeal-to-authority
Let's take an extreme example. Let's say that we have a sample size of 1. Now, I think you'll agree, even if we have a perfectly random selection process, our conclusions that we draw won't be particularly accurate. Unless, of course, our population is also 1.
On the other hand, let's look at a population of Infinity. The trouble here is that with our finite sampling process, because we live in the real world we have no way of knowing if our sample is actually random, because there are infinite location we could choose from. We would need a sampling process capable of dealing with an infinite population, which we don't have.
In the middle, let's look at a population of eg. the USA, 300,000,000. Now the problem becomes, "Is our sampling process representative of the actual population?" Here is the difficult part, because when you actually try to collect the data, you find that people don't always answer polls, so immediately your sampling is biased towards people who *do* answer polls. Then say they are collected by telephone. This limits the poll to people who a) have phones b) YouGov has their number c) are near their phone and not eg. at work, looking after kids, driving etc when YouGov calls d) answer their phone to unknown numbers e) answer polls. I think you'll agree that this is no longer an accurate representation of the population, because it biases against people who don't have phones, who are busy, who avoid telemarketers and who are willing to give their opinion on a touchy subject to a stranger.
One way you can reduce the amount of bias is by using a larger number of methods to sample opinions. Eg internet, street interviews, TV call-in hotlines. Of course, all of these have their own sampling bias just like the phone one. So, to reduce the amount of bias the number of methods should be very large. Let's say we have 50 different ways of getting peoples' opinions, and we weight them all equally. Suddenly, this means just 20 phone calls, 20 street interviews, 20 TV hotline callins etc. which doesn't seem nearly as robust now, does it, when you're trying to accurately portray the opinion of the entire population of the USA. Which is why people are saying that it is *only* 1000 samples, and even then there will still be residual bias towards people who don't want to answer polls.
The modern Tea Party movement actually started fairly similarly to OWS, but was derailed and turned into a "hur hur evolution sucks and global warming is false" spectacle.
OWS wasn't even about political control over corporations, they just wanted the bankers who had caused and benefited from the stock market collapse to be brought to justice.
In theory, yes. Unfortunately, we live in the real world where trying to get a truly random sample of an uncontrolled population is impossible.
Because then he would have been a "coward" for being anonymous, and probably "not standing up for what he believes in". He did what he had to to get his info taken as seriously as possible. He essentially sacrificed himself for a higher cause. The thing is, he can still do more 'good' by remaining free, so there is no problem with him doing his best to avoid US law enforcement.
These calculations are assuming a unimodal distribution, which isn't necessarily true (in fact, almost certainly false) when it comes to something like political affiliation or opinion, which is multimodal. People aren't going to fit a nice bell curve for something like "on a scale of 0 to 10, with 0 being ultra-left and 10 being ultra-right, where do your political affiliations lie?", there are going to be distict bumps on each side.
I think the real question is how did they make their sampling 'random'. It is extremely easy to (accidentally or on purpose) introduce bias based on where and when the samples were taken. This is why using a larger sample helps, because it means that you can choose from a broader spectrum of locations.
I think the reason he went to HK then RU is because they are the only countries with the balls to stand up to a US extradition request.
The smaller the proportion of population sampled the larger the likelihood that the sampling wasn't *truly* random, or representative.
Are you forgetting the Occupy Wall Street and the (original) Tea Party movements? The government is just much better at derailing protest movements these days than they were back in Nixon's time.
Swallows.
I think six, with two control systems each running a set of 3 in 60-degree-offset triangles. That way even a control system failure would still result in a recoverable (hopefully) situation.
There are the 4 little stability-only rockets out on the legs.
Even in a hex losing 2 adjacent struts would cause failure, because the rockets only fire downwards.
Oh really, and what would you have done if you were the military and the current president was trying to turn the country into a theocracy, replacing police, judges and school principles with his cronies?
Also, apparently the officials being arrested and the TV stations being closed are because they were encouraging their supporters to be violent.
Fine, in Fascist America, Bing searches you!
FTFY
True, but it does complicate things. Seriously, if you're going to provide a tool with incompatible functionality, at least name it something different. At the minimum append a version number.