As far as I know, HFS+ is neither patented nor proprietary. It's documented both as part of the open-source kernel and in TN1150. (Caveat: some newer feature are only documented in the former. A few very new features are not documented at all.)
because it is relatively easy right now to craft collisions with MD5.
Only if you control the original file as well as the new file. (Or, more simply, it's easy to craft two files with the same MD5 hash provided you're able to modify both files.)
Crafting a new file that has a chosen MD5 hash (or, equivalently, has the same hash as a preexisting file you cannot modify) is still hard.
Since we're talking about sociology here, the "rich" should be referring to income and wealth of individuals (the objects being studied). So, something like GDP or GNI per capita, probably adjusted for purchasing-power parity.
You're talking about government debt, which probably doesn't influence sociology much, unless you ask how people feel about the government's debt. But regardless, the US national debt is about 100% of GDP, depending on how you count it, which is a bit above Canada and the EU nations with good finances and a bit below some EU nations with bad finances. It's about half of Japan's debt (per GDP). (But, to be fair, it's 200 times North Korea's debt. It must be nice to be that rich.)
It's not "barely effective". The stated 9% is only for one strain in one population group that historically doesn't see as much benefit from the flu vaccine. The effectiveness in the general population this year -- which is a bit worse than average -- is over 50%.
It's 9% effectiveness against one of the strains in one age group. It's 56% effective in general. This year's effectiveness is particularly low relative to other years, so isn't representative of "value for dollar", which should use effectiveness averaged across multiple years.
Very rough back-of-the-envelope calculations suggest that the cost of saving lives through flu vaccines is on the order of a few tens of thousands of dollars per life. By insurance calculation standards, that makes it worthwhile.
Or should we just stop now and deal with our little flu bouts like grown-ups?
Considering it kills 24,000 people a year in the US, I'm not sure I'd characterize it as a "little bout" that you can just "deal with like a grown-up".
You're switching between topics randomly because, it seems, you understand nothing about statistics and are imagining scenarios from your gut.
My example earlier was only to illustrate that total population size doesn't matter.
What you're talking about now is sample bias. In your case, the result from the 10,000 tweets from the same person are highly correlated, giving you strong sample bias and weakening your sample. Whereas tweets from different people are less-correlated and have less bias.
In neither case did the number of people in the area matter.
No. You seem to be applying some wildly incorrect, reductionist logic that if the total population size doesn't matter, then magically the sample size doesn't matter. The sample size is important. What's not important is how large the population that sample is drawn from. What dictates the sample size is how high you want your confidence and how large the effect size is.
Let's say you're measuring the fraction of tweets containing at least one happy word for a geographical area. If, in reality, 50% of tweets from that area contain happy words, you don't need a very large sample size to accurately measure that. If, on the other hand, 0.1% of tweets contain happy words, you need a much larger sample size.
Note that this is only really true for large populations -- which is most surveyed populations. For ludicrously small sample sizes, all statistics are bad, but then, statistics tells you that tiny sample sizes are very bad. So stop using that as an example.
You don't even get to the point of computing error since it's larger than the sample.
It's a population sample, so the error on a single flip is undefined.
So, statistics correctly says that flipping a coin once is not valid. Population size doesn't matter, it's already true that a single sample doesn't do you any good, statistically. Incidentally, the population size for coin flips is infinite. Yet, you can get an excellent estimate for whether a coin is biased by testing it. Your sensitivity to how much bias you can measure is entirely determined by how many flips you perform. In practice, doing statistics on random samples models the population as infinite. Another way you can tell the population size doesn't matter -- only the sample size and the effect size.
They looked at 10 million whole tweets (out of a population of 300million)
Population size doesn't actually matter for statistics, even though people often think it does. Systematic bias in selecting your random sample matters (as does sample size and size of effect being measured), but not total population size.
Actually, better SSD controllers sense that a page has reached its rewrite limit. The end effect of this is that the size of the overprovisioned space gets reduced by one page. (The controller stops ever writing to the used-up page.) The write performance of the SSD degrades until it goes below a certain amount of overprovisioned space, at which point it refuses to write any more. The disk is still entirely readable, so it's a binary failure mechanism, but a pretty safe one.
Gradual failure over time means either you have a crap controller or that your electronics are failing in ways other than running out of write cycles.
Minting your own currency also is legal in the US. I assume their are restrictions (and participation is of course voluntary), but there are a large number of active local currencies in the US.
... and that's why those of us in civilised countries...
I'm going to go out on a limb and guess that the actual facts of the matter are that most of you in civilized countries don't appreciate how your own legal system works.
The average high school graduate in the US can only read on a 6th grade reading level.
Do you have some source for this? The literacy statistics aren't good (about 2/3 of high school students read below grade level), but I haven't seen anything that bad.
If you have a Bachelor's degree, that basically brings you up to par with the high school graduates in other countries.
This is a gross generalization that in many respects is simply not true. "Up to par" in what sense? Reading ability, maybe I believe. Are you seriously claiming that the average graduate with a science or engineering degree has no more knowledge, skill, or experience than a high school graduate from Germany, Japan, or Russia?
That also brings you up to the level your grandparents in the US had when they finished High School.
That's simply completely false, at least if you restrict your collection of high school students to those who did or would have gone to college. The availability and popularity of college-level instruction in high school on subjects like mathematics (to say nothing of engineering and science) is far and away higher than in even my parents' generation. Then, it was uncommon to learn any calculus before college; now, it's uncommon to see science and engineering students start college without it. In my grandparents' generation, such instruction was unavailable.
Damn, I came here to say this. There is a direct real-world analog: Google is operating the store and selling products on behalf of suppliers. They're the only ones that have direct customers.
Are you kidding? Lots of processes still use that damn indefinite progress bar. I think a recent version of IE might, even. As time progresses, even if no real progress is made, it fills up some small fraction of the remaining space. The next block of no-progress "time" uses up the same small fraction of the *now-remaining* space, so the progress bar creeps forever toward the end, even though no progress is made.
It puzzles me when I see that people work really hard to come up with difficult passwords for their bank accounts, but not for their personal accounts on their own computers. They really need to think about what value those passwords have to other people - in particular what could someone else do with those passwords if they had them?
The attack scenarios are very different. Most home PCs do not have any remote access service enabled. So your password is safeguarding your computer from someone who is physically present but has neither the time nor the technical skill to bypass it. That's often, but not always, a very rare occurrence. (If it's a laptop and you were prudent enough to use full-disk encryption, that password is actually being useful.) Your banking password is protecting you from anyone who is (a) on the Internet and (b) discovers that your (bank, username) pair is valid.
Slashdot Mirror
As far as I know, HFS+ is neither patented nor proprietary. It's documented both as part of the open-source kernel and in TN1150. (Caveat: some newer feature are only documented in the former. A few very new features are not documented at all.)
because it is relatively easy right now to craft collisions with MD5.
Only if you control the original file as well as the new file. (Or, more simply, it's easy to craft two files with the same MD5 hash provided you're able to modify both files.)
Crafting a new file that has a chosen MD5 hash (or, equivalently, has the same hash as a preexisting file you cannot modify) is still hard.
Since we're talking about sociology here, the "rich" should be referring to income and wealth of individuals (the objects being studied). So, something like GDP or GNI per capita, probably adjusted for purchasing-power parity.
You're talking about government debt, which probably doesn't influence sociology much, unless you ask how people feel about the government's debt. But regardless, the US national debt is about 100% of GDP, depending on how you count it, which is a bit above Canada and the EU nations with good finances and a bit below some EU nations with bad finances. It's about half of Japan's debt (per GDP). (But, to be fair, it's 200 times North Korea's debt. It must be nice to be that rich.)
The car was on loan for testing and a reviewer should not assume they have privacy rights...
Particularly since apparently they're informed of this in advance.
It's not "barely effective". The stated 9% is only for one strain in one population group that historically doesn't see as much benefit from the flu vaccine. The effectiveness in the general population this year -- which is a bit worse than average -- is over 50%.
I agree; automobile accidents are one place where relatively cheap changes could save a lot of lives. It's not an either-or situation, though.
It's 9% effectiveness against one of the strains in one age group. It's 56% effective in general. This year's effectiveness is particularly low relative to other years, so isn't representative of "value for dollar", which should use effectiveness averaged across multiple years.
Very rough back-of-the-envelope calculations suggest that the cost of saving lives through flu vaccines is on the order of a few tens of thousands of dollars per life. By insurance calculation standards, that makes it worthwhile.
Or should we just stop now and deal with our little flu bouts like grown-ups?
Considering it kills 24,000 people a year in the US, I'm not sure I'd characterize it as a "little bout" that you can just "deal with like a grown-up".
You're switching between topics randomly because, it seems, you understand nothing about statistics and are imagining scenarios from your gut.
My example earlier was only to illustrate that total population size doesn't matter.
What you're talking about now is sample bias. In your case, the result from the 10,000 tweets from the same person are highly correlated, giving you strong sample bias and weakening your sample. Whereas tweets from different people are less-correlated and have less bias.
In neither case did the number of people in the area matter.
No. You seem to be applying some wildly incorrect, reductionist logic that if the total population size doesn't matter, then magically the sample size doesn't matter. The sample size is important. What's not important is how large the population that sample is drawn from. What dictates the sample size is how high you want your confidence and how large the effect size is.
Let's say you're measuring the fraction of tweets containing at least one happy word for a geographical area. If, in reality, 50% of tweets from that area contain happy words, you don't need a very large sample size to accurately measure that. If, on the other hand, 0.1% of tweets contain happy words, you need a much larger sample size.
Note that this is only really true for large populations -- which is most surveyed populations. For ludicrously small sample sizes, all statistics are bad, but then, statistics tells you that tiny sample sizes are very bad. So stop using that as an example.
A handful of trivially-found references.
You don't even get to the point of computing error since it's larger than the sample.
It's a population sample, so the error on a single flip is undefined.
So, statistics correctly says that flipping a coin once is not valid. Population size doesn't matter, it's already true that a single sample doesn't do you any good, statistically. Incidentally, the population size for coin flips is infinite. Yet, you can get an excellent estimate for whether a coin is biased by testing it. Your sensitivity to how much bias you can measure is entirely determined by how many flips you perform. In practice, doing statistics on random samples models the population as infinite. Another way you can tell the population size doesn't matter -- only the sample size and the effect size.
You actually pick a good example.
Questions:
What's the population size for "coin flips"?
What's the error on a sample size of one?
It's a completely arbitrary scale, so you have no idea how big a 0.3 difference is.
They looked at 10 million whole tweets (out of a population of 300million)
Population size doesn't actually matter for statistics, even though people often think it does. Systematic bias in selecting your random sample matters (as does sample size and size of effect being measured), but not total population size.
Actually, better SSD controllers sense that a page has reached its rewrite limit. The end effect of this is that the size of the overprovisioned space gets reduced by one page. (The controller stops ever writing to the used-up page.) The write performance of the SSD degrades until it goes below a certain amount of overprovisioned space, at which point it refuses to write any more. The disk is still entirely readable, so it's a binary failure mechanism, but a pretty safe one.
Gradual failure over time means either you have a crap controller or that your electronics are failing in ways other than running out of write cycles.
Minting your own currency also is legal in the US. I assume their are restrictions (and participation is of course voluntary), but there are a large number of active local currencies in the US.
... and that's why those of us in civilised countries...
I'm going to go out on a limb and guess that the actual facts of the matter are that most of you in civilized countries don't appreciate how your own legal system works.
The average high school graduate in the US can only read on a 6th grade reading level.
Do you have some source for this? The literacy statistics aren't good (about 2/3 of high school students read below grade level), but I haven't seen anything that bad.
If you have a Bachelor's degree, that basically brings you up to par with the high school graduates in other countries.
This is a gross generalization that in many respects is simply not true. "Up to par" in what sense? Reading ability, maybe I believe. Are you seriously claiming that the average graduate with a science or engineering degree has no more knowledge, skill, or experience than a high school graduate from Germany, Japan, or Russia?
That also brings you up to the level your grandparents in the US had when they finished High School.
That's simply completely false, at least if you restrict your collection of high school students to those who did or would have gone to college. The availability and popularity of college-level instruction in high school on subjects like mathematics (to say nothing of engineering and science) is far and away higher than in even my parents' generation. Then, it was uncommon to learn any calculus before college; now, it's uncommon to see science and engineering students start college without it. In my grandparents' generation, such instruction was unavailable.
Being unable to brake at 125 mph stands a pretty good chance of destroying more than the transmission.
So, same situation as Rapiscan, then.
Damn, I came here to say this. There is a direct real-world analog: Google is operating the store and selling products on behalf of suppliers. They're the only ones that have direct customers.
How about this for RaPi?
Why would you refer to a product as "rapey"?
Are you kidding? Lots of processes still use that damn indefinite progress bar. I think a recent version of IE might, even. As time progresses, even if no real progress is made, it fills up some small fraction of the remaining space. The next block of no-progress "time" uses up the same small fraction of the *now-remaining* space, so the progress bar creeps forever toward the end, even though no progress is made.
It puzzles me when I see that people work really hard to come up with difficult passwords for their bank accounts, but not for their personal accounts on their own computers. They really need to think about what value those passwords have to other people - in particular what could someone else do with those passwords if they had them?
The attack scenarios are very different. Most home PCs do not have any remote access service enabled. So your password is safeguarding your computer from someone who is physically present but has neither the time nor the technical skill to bypass it. That's often, but not always, a very rare occurrence. (If it's a laptop and you were prudent enough to use full-disk encryption, that password is actually being useful.) Your banking password is protecting you from anyone who is (a) on the Internet and (b) discovers that your (bank, username) pair is valid.
Either your bank sucks or you didn't browbeat them enough. They should reverse the bounced-check charges resulting from the stolen money.
You need to dispute the results of identity theft on your credit rating. If the rating agencies refuse to fix it, you can sue the pants off them.
Of course, this is a lot of trouble and it sucks pretty hard. TFA actually agrees with you on this.