They aren't hunting, certainly, they're target shooting. That doesn't mean they aren't hunters, though. They could be practicing for real bird hunts; most people do that by shooting clay pigeons, but real birds provide more realistic practice. It's not ideal because the pigeons don't behave the same as the real game bird, but better than clays. Pigeons are more erratic but slower than pheasants, and are not only slower but don't have the deceptively low flight of chukars. Actually, pigeons probably aren't bad stand-ins for doves.
If released in groups the flocking patterns of pigeons would probably provide very useful training for hunting other flocking birds -- a common mistake for novice hunters is to "flock shoot", just pointing the gun in the direction of the flock and pulling the trigger, relying on the spread of the shot pattern to find a bird. That doesn't work, so hunters need to train themselves to pick a bird and aim for it as though it were flying solo if they want to get a hit.
(Disclaimer: I've never shot live pigeons. I'm just extrapolating based on my observations of their flight patterns, as compared to the those of birds I have hunted. I have wished I could shoot the annoying things, though.)
A rifle would be stupid... hard to hit a flying target and do less damage.
More importantly, using a rifle would be dangerous.
The bullet must come down somewhere and it could injure or kill someone. This is part of the reason that rifles are not used for hunting birds. Shot also must come down somewhere but small round shot loses its velocity and kinetic energy pretty quickly. After 200 yards or so it's basically harmless. Rifle rounds, on the other hand, are dangerous for miles. Even.22LR can still be dangerous in excess of one mile away. Something like the.17HMR is probably not too dangerous beyond a few hundred yards, since it begins tumbling in as little as 100 yards, which ruins its ballistic coefficient and causes it to quickly lose energy, but I wouldn't recommend shooting even that at flying objects.
at 13 points, even if there's only ~100 options per point
Are there that many options per point?
A brief search didn't turn up any estimates of theoretical false positive rate for CODIS, but Wikipedia did have this to say about SGM+, the system used in the UK:
Since 1998, the DNA profiling system supported by The National DNA Database in the UK is the SGM+ DNA profiling system which includes 10 STR regions and a sex indicating test. STRs do not suffer from such subjectivity and provide similar power of discrimination (1 in 10^13 for unrelated individuals if using a full SGM+ profile). It should be noted that figures of this magnitude are not considered to be statistically supportable by scientists in the UK, for unrelated individuals with full matching DNA profiles a match probability of 1 in a billion is considered statistically supportable.
So even the theoretical maximum resolution for SGM+ isn't 100^13, but only 10^13. which means you only need about 3.7M people to have a 50% probability of a false positive, and the statistically-supportable resolution is four orders of magnitude lower than that -- meaning you only need 3700 people for even odds of a false positive.
CODIS may be a little better than SGM+ (or it may be worse), but I strongly doubt that it's 13 orders of magnitude better. I doubt it's even two orders of magnitude better.
If someone is disrupting your ability to legally use your private property but they are not committing a crime in the process, we have a remedy:
A civil suit seeking an injunction against future similar behavior and damages for past losses.
But why? Shooting the drones down is undoubtedly more fun, and it's not like the activists have any legal recourse, since their drone is trespassing.
What might be worth some thought is attempting to find a way to capture the drone with minimal or no damage. Then offer to sell the drone back to the activists for, say, half its value. Repeat that a few times and the hunting club is well-funded for the year.
since it's not clear when to apply it, vs. "if at first you don't succeed, try, try again."
I don't see any contradiction between the two. If at first you don't succeed, try again, but don't continue trying to do exactly the same thing. There has to be some variable involved that gives you reason to think your next attempt may be different than what has come before. If the variable is your skill or ability, then repeated attempts may ultimately lead to success, so try again. Even if there's just an element of randomness which assures different outcomes, and the degree of possible variation is sufficient that some trial may have success, then persistence makes sense. But if it's clear that there are no variables capable of significantly changing the outcome then it's absolutely true that expecting a different result is a useful definition of "insanity", in the sense of a disconnection from reality.
Umm, they're already out there "wasting" ammo and time shooting pigeons. The drone is just a free (and probably quite satisfying) target. It's probably much easier to hit than the pigeons, BUT you get to watch the bird's-eye video on the Internet later.
You should find it scary. The odds of a false positive are probably much higher.
The DNA database in the US has found matches between a black man and a white man if you point it at itself.
Matching a database against itself is a great way to calibrate the accuracy of a test, but not a good way to understand the probability of false positives when comparing a specific sample (from a crime scene, for example) against a population. The mathematical relationship between these two match probabilities is straightforward but not very intuitive.
The most well-known example of this counter-intuitiveness is what's called the "Birthday Paradox", which is that while most people would think that in a room of, say, 25 people, the odds would be fairly low that two of them have the same birthday (month and day), in fact the odds are quite high, nearly 60%. With 30 people it's 70%. With 35 people over 80%. By the time you get 60 people, it's a near-certainty.
Even if your DNA test has only a one-in-a-billion probability of a false positive when matching two specific samples (and I don't think any DNA test commonly used is that precise), if you point it at any database with more than 37,000 people you have a 50% chance of finding a false match between some pair of them. If you have a database of 100,000 people you have a 99% chance of at least one false positive. Expand it to a million entries and you'd need a DNA test with an astronomically low false positive rate in order not to have a false positive.
But none of that has much to do with the chance that your DNA matches that of a specific sample from a crime scene, other than the fact that searching for false positives in pairwise database matches is a good way to compute the average false positive rate.
Given that DNA has a whole lot more than 366 possibilties, how many people do you need to have two that were born on the same day, for a properly randomized group of people?
Any biometric can be roughly characterized with a "resolution" number, which is basically how many separate "bins" it will divide people into, given a specific set of testing parameters. Birthday has a resolution of 366. So, a particular DNA test which analyzes a certain number of base pairs and has a particular degree of fidelity might, for example, categorize everyone into one of a million categories. One of those categories is already occupied by the killer, so if you test two million people, on average (and assuming the test distributes people relatively uniformly, rather than clustering larger numbers into some of the buckets) you'll expect to find two matches.
Applying the Birthday Paradox, then, if n is the resolution, you need approximately 1.177 * sqrt(n) people in your sample to get a 50% chance that at least two of them are considered the same. Assuming n = 1,000,000, this means you need 1.177*1000 = 1,177 people, roughly.
Note, though, that the "resolution" view of a biometric measurement is a pretty rough approximation because the "edges" of the bins are fuzzy. Any measurement has some number of "false positives" (individuals whose measurements are sufficiently close that for a given match threshold they're considered the same) and some number of "false negatives" (multiple readings from the same individual whose variance is sufficient to fall outside the match threshold). There are various parameters in most systems that can adjust the false positive and false negative rates, but decreasing one increases the other. The "equal error rate" (ERR -- yeah, I know the letters don't match the words) is found by adjusting the parameters until false positive and false negative rates are the same, and this rate provides another measure of system accuracy, but it's not common that ERR is used in real systems.
I think DNA testing is different from most other biometrics in that it has a zero false negative rate -- that is, two non-degraded and uncontaminated samples from the same individual, subjected to the same testing process, will always indicate a match, but I'm not 100% certain about that. And degradation and contamination could still produce false positives. But "resolution" is mostly a measure of false positive rate, so it's still a useful way to talk about DNA testing.
Oh, it's also worth pointing out that the math behind the Birthday Paradox is not in any way specific to measurements of people. I often use the same calculations when deciding how many bits of a hash I need in order to be certain that collisions are sufficiently unlikely.
Even if they are aware of what GIFs are how many of them are using the word on a daily basis let alone using GIF as a verb. I'd imagine JPEG and PNG are going to be used far more.
Nope. My kids know what a GIF is, it's an animated picture, and they talk about GIFs as such, and about "giffing" a clip of a YouTube video. JPEG they've heard of, but aren't really sure about. Non-animated images are just called "pictures" or "images". They have no idea what a PNG is.
random teens on Facebook would still have no idea what a GIF is.
They think they do. Random teens on Facebook know that a GIF is a short animated video clip that is easier to paste into a post than a YouTube video. My teenagers explained this to me since they were sure their dumb old dad couldn't possibly have any understanding of this cool new video technology.
Funny, I interviewed with Google and when I had the first engineer phone interview he sounded burnt-out and disinterested. All he wanted me to do was solve a code problem. No interview questions beyond that.
I'm sorry you thought he sounded burnt-out -- maybe he was having a bad day for some reason -- but yes, all you'll do in a phone screen is a code problem. The purpose of the phone screen is to determine if the candidate is likely enough to succeed to be brought into a face to face interview, and the quickest way to determine that is with a code problem. I should point out that most of the face-to-face interview questions will also be code problems, or at least problems that involve some amount of coding.
I did a lot of interviewing during the 15 years I worked for IBM, too, and some before that, and while I never found an approach which is as effective as what Google does, I was gravitating towards focusing primarily on coding in my interviews. It really is the most effective way to separate those who can talk from those who can do -- and it's astonishing how many candidates can't do something trivial like inserting a node into a sorted linked list.
The interviewers also had no idea what position they wanted to fill, it seemed more like they were just fishing.
Correct. As I mentioned in my post Google doesn't generally hire for specific positions. The interview and hiring process just tries to find good, capable people, and placement is a separate process after that.
This does create something of a problem in that candidates have to be sold on the company rather than on the specific position or the team. In fact, I think most hires don't even know what they're going to be working on when they accept the job offer (I didn't). Countering that is the fact that mobility is high within Google and it's expected that most employees won't stay in their original position very long anyway. People can gravitate to what they enjoy the most, and there is more than enough cool stuff going on in Google that anyone can find an awesome job for themselves.
The one caveat is location. If the team that does what you really want to do is located in another city, changing jobs means relocating. Personally, out of all the stuff Google does I'd most prefer to work on Android security, and Android security team has offered me a job... but I don't want to live in California. So instead I do something which isn't quite as fun but lets me live in Colorado.
I've done more than 100 interviews and know that this approach will lose too many good candidates. If you can't engage with them, you'll never find out whether they fit the position or not.
As I said, there is no "fit with the position" to be determined. That's for later. I realize this isn't what people are accustomed to. Losing good candidates isn't generally a big problem for Google, however, because the company (rightfully) has a reputation as a fantastic place to work. It does happen, and that's unfortunate, but it's rare enough that there are bigger problems that need to be addressed first -- like figuring out how to stop outright rejecting so many good candidates, without letting poor candidates through.
I do admit that some people just struggle with the on-the-spot nature, and might be able to devise great solutions given time to go off and think, but such people wouldn't do well in Google's fast-paced technical culture anyway.
I had a good laugh when I read this.:)
Of course, I've never been to Google so probably shouldn't laugh. But it's funny anyway.:)
I'm not sure what aspect of it you're laughing at... but it's nothing less than the truth. Technical discussions at Google, whether water-cooler discussions, or design review meetings, or whatever else tend to be very fast-paced, with lots of ideas being raised and discarded in rapid succession. This is pretty typical of engineering discussions wherever I've worked in my ~25-year career, but the culture at Google and the nature of people Google hires (those who do well in rapid-fire, on-the-spot technical discussions) turn it up. It's a lot of fun, IMO, but not for everyone.
I'd say the real reason Google asks the sort of questions it does is because it's the only way the company has found to get a handle on what it really wants to find out about candidates: Problem-solving ability.
There are two problems with this. The first is that even Google admits that it doesn't work. The most successful people inside Google are the ones that had one or more negative reports during the interview process: exactly the ones that would be rejected in the normal process.
Cite? I've seen the internal stats on correlation between interview ratings and subsequent performance and while the correlation is imperfect, I've never seen anything that would indicate what you're saying.
Second, Google isn't short of people who can solve problems, and being able to solve problems isn't even an especially rare skill. Google is short of people who can identify the problems that are worth solving, and the interview process does nothing to address that.
Hmm. This is a more interesting assertion.
First, I'm not sure I agree that identifying the problems that are worth solving is either particularly hard or particularly rare. However, if we adjust your statement slightly to "people who can identify the problems that are most worth solving and can achieve the most value for the least effort", (note that "value" needn't be -- and at Google mostly isn't -- financial) then I'd agree that such ability is extremely valuable and very hard to find. I think Google probably has more of it than most companies, but more would always be better.
Second, if good problem solvers are hard to identify (and they are... else Google's process wouldn't reject so many of them), good problem choosers are almost impossible to identify.
Almost everything I had to do in the interviews involved stuff you're supposed to learn when studying Computer Science at a university that deserves its name, and I think that's a very good and reasonable thing. I've always been a fan of the "concepts, not implementations/products"-kind-of-education. I think that's especially important at Google - their infrastructure is so vast and powerful and unlike any other in the industry that the overwhelming majority of people who take a position there won't have seen anything even remotely like it in terms of scale, and they will probably find very little there that's overly "familiar" to them
This is very true. Pretty much everything in Google's tech stack is homegrown. Most of it because there is (or was) nothing out there that was capable of doing the job. Some of it because it doesn't even occur to Google engineers to look. Google doesn't sneer at technologies not invented at Google... Google doesn't even notice them.:-)
That's only part of the reason for the CS-heavy interview approach, though. I'm an interviewer at Google (though a relatively inexperienced one, at least in interviewing the Google way), and I'd say the real reason Google asks the sort of questions it does is because it's the only way the company has found to get a handle on what it really wants to find out about candidates: Problem-solving ability. Technical jobs at Google all require people who can think on their feet, who can quickly absorb the salient points of a problem, rapidly identify areas that need to be defined, then define, implement and analyze a solution. That ability could perhaps be tested with other sorts of problems, but CS provides a wealth of potential problems for discussion along with a well-defined common set of concepts and language which both interviewer and candidate are (or should be) intimately familiar with.
Some experienced candidates (like loufoque, apparently), find it insulting to be asked questions a kid straight out of school should be able to answer. They want the interviewer to give due deference to the value of their experience. The problem with that is that experience can be fudged, and it is simply not true that you can judge a candidate's real experience by asking about their previous work. I've met many who can talk the talk with the best of them, but when you start asking them to solve problems on the spot their weaknesses start to become very apparent. I do admit that some people just struggle with the on-the-spot nature, and might be able to devise great solutions given time to go off and think, but such people wouldn't do well in Google's fast-paced technical culture anyway.
But don't think this means Google doesn't value experience. It does, a lot, because of the judgment that comes with experience. But experience can easily be judged by reading the candidate's resume, so there's really no value in spending time in the interview trying to evaluate it.
So, the interview focuses on evaluating ability and cultural fit. CS theory is a useful tool for evaluating the former, and it's not unrelated to the latter. Assuming the candidate does well in the interview, experience becomes relevant later in determining compensation and placement (Google doesn't generally hire for specific positions; Google hires good people, then figures out where to put them).
One final caveat about Google's interviewing approach: It rejects a lot of good people, and everyone at Google knows it. It's broadly accepted among engineers at Google that virtually any one of us could be interviewed again and have maybe a 30% chance of being rejected. Maybe 50%.
This is decidedly sub-optimal.
The problem is that no one knows how to identify top talent accurately other than by hiring them and putting them to work for a few months. Doing exactly that is a big focus of Google's internship program -- it's one of the very best sources of good permanent hires -- but trial peri
Because of all the downtime on Google docs, the communication with the C&C server is intermittent and therefore difficult to pinpoint by law enforcement. Security by instability.
FYI, if you'd like to know how often Google docs (or any other Google Apps service) is unavailable, Google provides an on-line status dashboard with both current and historical information going back two months.
Googling for overall uptime stats shows that in 2010, Apps achieved 99.984% uptime and in 2011 99.9949% uptime, even after changing the methodology to count all downtimes, not just those lasting more than 10 minutes.
in almost no states is it legal to travel around with it in a useful condition, i.e. fully loaded and close at hand.
According to the map compiled by opencarry.org, ~35 states allow unlicensed vehicle carry, though some have some restrictions (must be visible, must be concealed, must be in glove box, may be restricted by local law, etc.).
That's an interesting solution. A Leaf's battery (24kWh at 400V) will actually power a whole house for a couple of days, but that would require getting direct access to the main battery output to sustain whole-house amperages. What he did was to connect his fridge to a small inverter, which he connected in turn to the Leaf's 12V battery. The 12V battery in a Leaf is pretty much an ordinary car battery (a little smaller than most), whose normal use is to power the interior amenities of the car. Like any other 12V car battery, it would be depleted before too long... but the Leaf automatically recharges the 12V battery from the main battery as needed.
You're technically right, but wrong in a broader and more important sense.
The excessive scope and duration of copyright, along with the theory of copyright that is pushed by the content industry, give individuals very little moral reason not to infringe.
As the brief points out, the true purpose of copyright is to advance the progress of science and the useful arts, by providing some temporary incentives. In other words, it's a dynamic balance: society voluntarily agrees to restrict its own freedom with respect to expressive materials for a period of time in order to encourage the publication of material and its flow into the public domain. Under that theory, people have a moral incentive to hold up their side of the bargain.
But that's not the theory implied by the content industry's preferred view of copyright. They push the notion that creators have some natural right to control their creations, even after they've chosen to publish them to the world, and regardless of the fact that 99% of "their" creations were given to them by society. This notion is perfectly consistent with long and broad copyright, in fact it's easy to argue that if it's a right of the creator, and a heritable right, then it really should be perpetual. The view may not be completely consistent with the idea that the right can be sold, or that works made for hire belong to the employer, etc., but they're happy to ignore that point.
Under that theory, why should individuals not infringe? Obviously there are the potential legal penalties, but from a moral perspective, the calculation becomes one of percentages. How much harm does it really do if I pirate a copy of a song written and performed by a multi-millionaire musician? A dollar or two... a sum so small that the musician won't even notice. Even more, what if the choice is not between paying or not paying, but between listening without paying or not listening -- if in either case the musician doesn't get paid, what's the harm in making a copy?
And in fact the content industry has been enormously successful at pushing its view. If you ask the average man on the street who owns the copyright on Shakespeare's plays, he will see it as an obscure, trivial question, and one whose answer may be unknowable, but he's unlikely to realize that it's a nonsensical question, that not only are the plays long past any expiration, but that they should be in the public domain, that they're an important part of our shared culture and belong to us all.
Big media counts that as a success, but in fact it's a big part of what causes that same man on the street to feel that copyright infringement really isn't wrong in any deep sense. Curtailing the excessive scope and duration of copyright would enable people to understand why it is (or would be). Even better, decreasing copyright terms to the point where the average person sees a real choice between infringing now or waiting a few years for the work to enter the public domain would make the point even more sharply.
Ideally, we should set copyright duration to a point where a large majority of profitable works return a substantial portion of their lifetime expected profitability, and no longer. For films, for example, we could really set the term as short as two years, maybe less. It's a rare film that becomes profitable but hasn't done it within that time frame. Books probably need somewhere between 5 and 10 years, music is similar.
Shorter terms would not only help people understand why they shouldn't infringe, they would also serve the true purpose of copyright better, by making it less likely that successful artists, authors, etc. will produce one or two works and then give up creating, living on the the royalty stream.
Sorry, I got hung up on your "cannot be viewed" bit and didn't read the rest carefully enough. There's no need to encrypt it to make it non-modifiable and non-erasable, though. The easiest way to achieve "non-erasable" is probably just to require that storage be done by a third party who contracts to ensure that the data is never erased. Tamper resistance (or at least tamper evidence) is easily achieved with chained hashes. There may be some value in encryption for privacy, but that's a separate issue.
Asking after isn't a very good way to go. A lot of hormones are released that tend to make women forget how miserable they were. One of my friends wanted to have 5 or 6 kids until he had to put up with his wife while she was pregnant so it might not just be a boon for women.
She always said the same thing during, as well. Except for the last three weeks or so.
If they can view it, they can choose to 'lose/destroy' the physical video if it does not help their side of the story. I'd rather both sides of a trial get the video sight-unseen without knowing if it'll help them or not.
Not necessarily. Depends on how the video is stored. I can think of many ways to make it accessible/viewable but make it impossible for them to modify or erase it.
Slashdot Mirror
I'm sorry, but these arn't hunters.
They aren't hunting, certainly, they're target shooting. That doesn't mean they aren't hunters, though. They could be practicing for real bird hunts; most people do that by shooting clay pigeons, but real birds provide more realistic practice. It's not ideal because the pigeons don't behave the same as the real game bird, but better than clays. Pigeons are more erratic but slower than pheasants, and are not only slower but don't have the deceptively low flight of chukars. Actually, pigeons probably aren't bad stand-ins for doves.
If released in groups the flocking patterns of pigeons would probably provide very useful training for hunting other flocking birds -- a common mistake for novice hunters is to "flock shoot", just pointing the gun in the direction of the flock and pulling the trigger, relying on the spread of the shot pattern to find a bird. That doesn't work, so hunters need to train themselves to pick a bird and aim for it as though it were flying solo if they want to get a hit.
(Disclaimer: I've never shot live pigeons. I'm just extrapolating based on my observations of their flight patterns, as compared to the those of birds I have hunted. I have wished I could shoot the annoying things, though.)
A rifle would be stupid... hard to hit a flying target and do less damage.
More importantly, using a rifle would be dangerous.
The bullet must come down somewhere and it could injure or kill someone. This is part of the reason that rifles are not used for hunting birds. Shot also must come down somewhere but small round shot loses its velocity and kinetic energy pretty quickly. After 200 yards or so it's basically harmless. Rifle rounds, on the other hand, are dangerous for miles. Even .22LR can still be dangerous in excess of one mile away. Something like the .17HMR is probably not too dangerous beyond a few hundred yards, since it begins tumbling in as little as 100 yards, which ruins its ballistic coefficient and causes it to quickly lose energy, but I wouldn't recommend shooting even that at flying objects.
at 13 points, even if there's only ~100 options per point
Are there that many options per point?
A brief search didn't turn up any estimates of theoretical false positive rate for CODIS, but Wikipedia did have this to say about SGM+, the system used in the UK:
Since 1998, the DNA profiling system supported by The National DNA Database in the UK is the SGM+ DNA profiling system which includes 10 STR regions and a sex indicating test. STRs do not suffer from such subjectivity and provide similar power of discrimination (1 in 10^13 for unrelated individuals if using a full SGM+ profile). It should be noted that figures of this magnitude are not considered to be statistically supportable by scientists in the UK, for unrelated individuals with full matching DNA profiles a match probability of 1 in a billion is considered statistically supportable.
So even the theoretical maximum resolution for SGM+ isn't 100^13, but only 10^13. which means you only need about 3.7M people to have a 50% probability of a false positive, and the statistically-supportable resolution is four orders of magnitude lower than that -- meaning you only need 3700 people for even odds of a false positive.
CODIS may be a little better than SGM+ (or it may be worse), but I strongly doubt that it's 13 orders of magnitude better. I doubt it's even two orders of magnitude better.
If someone is disrupting your ability to legally use your private property but they are not committing a crime in the process, we have a remedy:
A civil suit seeking an injunction against future similar behavior and damages for past losses.
But why? Shooting the drones down is undoubtedly more fun, and it's not like the activists have any legal recourse, since their drone is trespassing.
What might be worth some thought is attempting to find a way to capture the drone with minimal or no damage. Then offer to sell the drone back to the activists for, say, half its value. Repeat that a few times and the hunting club is well-funded for the year.
since it's not clear when to apply it, vs. "if at first you don't succeed, try, try again."
I don't see any contradiction between the two. If at first you don't succeed, try again, but don't continue trying to do exactly the same thing. There has to be some variable involved that gives you reason to think your next attempt may be different than what has come before. If the variable is your skill or ability, then repeated attempts may ultimately lead to success, so try again. Even if there's just an element of randomness which assures different outcomes, and the degree of possible variation is sufficient that some trial may have success, then persistence makes sense. But if it's clear that there are no variables capable of significantly changing the outcome then it's absolutely true that expecting a different result is a useful definition of "insanity", in the sense of a disconnection from reality.
Why waste ammo/time shooting the drone.
Umm, they're already out there "wasting" ammo and time shooting pigeons. The drone is just a free (and probably quite satisfying) target. It's probably much easier to hit than the pigeons, BUT you get to watch the bird's-eye video on the Internet later.
Someone else will just fly another one.
Another free target!
Your estimate of the probability assumes that all results are correct, and none have been fudged to match a pre-determined answer.
Yes, I thought that went without saying. Fraud and human error introduce additional uncertainty.
P.S. There is no way using animated gifs for videos is good for the internet. ;)
I agree!
You should find it scary. The odds of a false positive are probably much higher.
The DNA database in the US has found matches between a black man and a white man if you point it at itself.
Matching a database against itself is a great way to calibrate the accuracy of a test, but not a good way to understand the probability of false positives when comparing a specific sample (from a crime scene, for example) against a population. The mathematical relationship between these two match probabilities is straightforward but not very intuitive.
The most well-known example of this counter-intuitiveness is what's called the "Birthday Paradox", which is that while most people would think that in a room of, say, 25 people, the odds would be fairly low that two of them have the same birthday (month and day), in fact the odds are quite high, nearly 60%. With 30 people it's 70%. With 35 people over 80%. By the time you get 60 people, it's a near-certainty.
Even if your DNA test has only a one-in-a-billion probability of a false positive when matching two specific samples (and I don't think any DNA test commonly used is that precise), if you point it at any database with more than 37,000 people you have a 50% chance of finding a false match between some pair of them. If you have a database of 100,000 people you have a 99% chance of at least one false positive. Expand it to a million entries and you'd need a DNA test with an astronomically low false positive rate in order not to have a false positive.
But none of that has much to do with the chance that your DNA matches that of a specific sample from a crime scene, other than the fact that searching for false positives in pairwise database matches is a good way to compute the average false positive rate.
Given that DNA has a whole lot more than 366 possibilties, how many people do you need to have two that were born on the same day, for a properly randomized group of people?
Any biometric can be roughly characterized with a "resolution" number, which is basically how many separate "bins" it will divide people into, given a specific set of testing parameters. Birthday has a resolution of 366. So, a particular DNA test which analyzes a certain number of base pairs and has a particular degree of fidelity might, for example, categorize everyone into one of a million categories. One of those categories is already occupied by the killer, so if you test two million people, on average (and assuming the test distributes people relatively uniformly, rather than clustering larger numbers into some of the buckets) you'll expect to find two matches.
Applying the Birthday Paradox, then, if n is the resolution, you need approximately 1.177 * sqrt(n) people in your sample to get a 50% chance that at least two of them are considered the same. Assuming n = 1,000,000, this means you need 1.177*1000 = 1,177 people, roughly.
Note, though, that the "resolution" view of a biometric measurement is a pretty rough approximation because the "edges" of the bins are fuzzy. Any measurement has some number of "false positives" (individuals whose measurements are sufficiently close that for a given match threshold they're considered the same) and some number of "false negatives" (multiple readings from the same individual whose variance is sufficient to fall outside the match threshold). There are various parameters in most systems that can adjust the false positive and false negative rates, but decreasing one increases the other. The "equal error rate" (ERR -- yeah, I know the letters don't match the words) is found by adjusting the parameters until false positive and false negative rates are the same, and this rate provides another measure of system accuracy, but it's not common that ERR is used in real systems.
I think DNA testing is different from most other biometrics in that it has a zero false negative rate -- that is, two non-degraded and uncontaminated samples from the same individual, subjected to the same testing process, will always indicate a match, but I'm not 100% certain about that. And degradation and contamination could still produce false positives. But "resolution" is mostly a measure of false positive rate, so it's still a useful way to talk about DNA testing.
Oh, it's also worth pointing out that the math behind the Birthday Paradox is not in any way specific to measurements of people. I often use the same calculations when deciding how many bits of a hash I need in order to be certain that collisions are sufficiently unlikely.
Even if they are aware of what GIFs are how many of them are using the word on a daily basis let alone using GIF as a verb. I'd imagine JPEG and PNG are going to be used far more.
Nope. My kids know what a GIF is, it's an animated picture, and they talk about GIFs as such, and about "giffing" a clip of a YouTube video. JPEG they've heard of, but aren't really sure about. Non-animated images are just called "pictures" or "images". They have no idea what a PNG is.
random teens on Facebook would still have no idea what a GIF is.
They think they do. Random teens on Facebook know that a GIF is a short animated video clip that is easier to paste into a post than a YouTube video. My teenagers explained this to me since they were sure their dumb old dad couldn't possibly have any understanding of this cool new video technology.
Funny, I interviewed with Google and when I had the first engineer phone interview he sounded burnt-out and disinterested. All he wanted me to do was solve a code problem. No interview questions beyond that.
I'm sorry you thought he sounded burnt-out -- maybe he was having a bad day for some reason -- but yes, all you'll do in a phone screen is a code problem. The purpose of the phone screen is to determine if the candidate is likely enough to succeed to be brought into a face to face interview, and the quickest way to determine that is with a code problem. I should point out that most of the face-to-face interview questions will also be code problems, or at least problems that involve some amount of coding.
I did a lot of interviewing during the 15 years I worked for IBM, too, and some before that, and while I never found an approach which is as effective as what Google does, I was gravitating towards focusing primarily on coding in my interviews. It really is the most effective way to separate those who can talk from those who can do -- and it's astonishing how many candidates can't do something trivial like inserting a node into a sorted linked list.
The interviewers also had no idea what position they wanted to fill, it seemed more like they were just fishing.
Correct. As I mentioned in my post Google doesn't generally hire for specific positions. The interview and hiring process just tries to find good, capable people, and placement is a separate process after that. This does create something of a problem in that candidates have to be sold on the company rather than on the specific position or the team. In fact, I think most hires don't even know what they're going to be working on when they accept the job offer (I didn't). Countering that is the fact that mobility is high within Google and it's expected that most employees won't stay in their original position very long anyway. People can gravitate to what they enjoy the most, and there is more than enough cool stuff going on in Google that anyone can find an awesome job for themselves.
The one caveat is location. If the team that does what you really want to do is located in another city, changing jobs means relocating. Personally, out of all the stuff Google does I'd most prefer to work on Android security, and Android security team has offered me a job... but I don't want to live in California. So instead I do something which isn't quite as fun but lets me live in Colorado.
I've done more than 100 interviews and know that this approach will lose too many good candidates. If you can't engage with them, you'll never find out whether they fit the position or not.
As I said, there is no "fit with the position" to be determined. That's for later. I realize this isn't what people are accustomed to. Losing good candidates isn't generally a big problem for Google, however, because the company (rightfully) has a reputation as a fantastic place to work. It does happen, and that's unfortunate, but it's rare enough that there are bigger problems that need to be addressed first -- like figuring out how to stop outright rejecting so many good candidates, without letting poor candidates through.
I do admit that some people just struggle with the on-the-spot nature, and might be able to devise great solutions given time to go off and think, but such people wouldn't do well in Google's fast-paced technical culture anyway.
I had a good laugh when I read this. :)
Of course, I've never been to Google so probably shouldn't laugh. But it's funny anyway. :)
I'm not sure what aspect of it you're laughing at... but it's nothing less than the truth. Technical discussions at Google, whether water-cooler discussions, or design review meetings, or whatever else tend to be very fast-paced, with lots of ideas being raised and discarded in rapid succession. This is pretty typical of engineering discussions wherever I've worked in my ~25-year career, but the culture at Google and the nature of people Google hires (those who do well in rapid-fire, on-the-spot technical discussions) turn it up. It's a lot of fun, IMO, but not for everyone.
I'd say the real reason Google asks the sort of questions it does is because it's the only way the company has found to get a handle on what it really wants to find out about candidates: Problem-solving ability.
There are two problems with this. The first is that even Google admits that it doesn't work. The most successful people inside Google are the ones that had one or more negative reports during the interview process: exactly the ones that would be rejected in the normal process.
Cite? I've seen the internal stats on correlation between interview ratings and subsequent performance and while the correlation is imperfect, I've never seen anything that would indicate what you're saying.
Second, Google isn't short of people who can solve problems, and being able to solve problems isn't even an especially rare skill. Google is short of people who can identify the problems that are worth solving, and the interview process does nothing to address that.
Hmm. This is a more interesting assertion.
First, I'm not sure I agree that identifying the problems that are worth solving is either particularly hard or particularly rare. However, if we adjust your statement slightly to "people who can identify the problems that are most worth solving and can achieve the most value for the least effort", (note that "value" needn't be -- and at Google mostly isn't -- financial) then I'd agree that such ability is extremely valuable and very hard to find. I think Google probably has more of it than most companies, but more would always be better.
Second, if good problem solvers are hard to identify (and they are... else Google's process wouldn't reject so many of them), good problem choosers are almost impossible to identify.
Strangely Google don't understand that not everyone wants to sell their house and move to a stupidly expensive capital city just to work at Google.
Did you ask about other locations? Google has many.
Almost everything I had to do in the interviews involved stuff you're supposed to learn when studying Computer Science at a university that deserves its name, and I think that's a very good and reasonable thing. I've always been a fan of the "concepts, not implementations/products"-kind-of-education. I think that's especially important at Google - their infrastructure is so vast and powerful and unlike any other in the industry that the overwhelming majority of people who take a position there won't have seen anything even remotely like it in terms of scale, and they will probably find very little there that's overly "familiar" to them
This is very true. Pretty much everything in Google's tech stack is homegrown. Most of it because there is (or was) nothing out there that was capable of doing the job. Some of it because it doesn't even occur to Google engineers to look. Google doesn't sneer at technologies not invented at Google... Google doesn't even notice them. :-)
That's only part of the reason for the CS-heavy interview approach, though. I'm an interviewer at Google (though a relatively inexperienced one, at least in interviewing the Google way), and I'd say the real reason Google asks the sort of questions it does is because it's the only way the company has found to get a handle on what it really wants to find out about candidates: Problem-solving ability. Technical jobs at Google all require people who can think on their feet, who can quickly absorb the salient points of a problem, rapidly identify areas that need to be defined, then define, implement and analyze a solution. That ability could perhaps be tested with other sorts of problems, but CS provides a wealth of potential problems for discussion along with a well-defined common set of concepts and language which both interviewer and candidate are (or should be) intimately familiar with.
Some experienced candidates (like loufoque, apparently), find it insulting to be asked questions a kid straight out of school should be able to answer. They want the interviewer to give due deference to the value of their experience. The problem with that is that experience can be fudged, and it is simply not true that you can judge a candidate's real experience by asking about their previous work. I've met many who can talk the talk with the best of them, but when you start asking them to solve problems on the spot their weaknesses start to become very apparent. I do admit that some people just struggle with the on-the-spot nature, and might be able to devise great solutions given time to go off and think, but such people wouldn't do well in Google's fast-paced technical culture anyway.
But don't think this means Google doesn't value experience. It does, a lot, because of the judgment that comes with experience. But experience can easily be judged by reading the candidate's resume, so there's really no value in spending time in the interview trying to evaluate it.
So, the interview focuses on evaluating ability and cultural fit. CS theory is a useful tool for evaluating the former, and it's not unrelated to the latter. Assuming the candidate does well in the interview, experience becomes relevant later in determining compensation and placement (Google doesn't generally hire for specific positions; Google hires good people, then figures out where to put them).
One final caveat about Google's interviewing approach: It rejects a lot of good people, and everyone at Google knows it. It's broadly accepted among engineers at Google that virtually any one of us could be interviewed again and have maybe a 30% chance of being rejected. Maybe 50%.
This is decidedly sub-optimal.
The problem is that no one knows how to identify top talent accurately other than by hiring them and putting them to work for a few months. Doing exactly that is a big focus of Google's internship program -- it's one of the very best sources of good permanent hires -- but trial peri
Because of all the downtime on Google docs, the communication with the C&C server is intermittent and therefore difficult to pinpoint by law enforcement. Security by instability.
FYI, if you'd like to know how often Google docs (or any other Google Apps service) is unavailable, Google provides an on-line status dashboard with both current and historical information going back two months.
Googling for overall uptime stats shows that in 2010, Apps achieved 99.984% uptime and in 2011 99.9949% uptime, even after changing the methodology to count all downtimes, not just those lasting more than 10 minutes.
in almost no states is it legal to travel around with it in a useful condition, i.e. fully loaded and close at hand.
According to the map compiled by opencarry.org, ~35 states allow unlicensed vehicle carry, though some have some restrictions (must be visible, must be concealed, must be in glove box, may be restricted by local law, etc.).
That's an interesting solution. A Leaf's battery (24kWh at 400V) will actually power a whole house for a couple of days, but that would require getting direct access to the main battery output to sustain whole-house amperages. What he did was to connect his fridge to a small inverter, which he connected in turn to the Leaf's 12V battery. The 12V battery in a Leaf is pretty much an ordinary car battery (a little smaller than most), whose normal use is to power the interior amenities of the car. Like any other 12V car battery, it would be depleted before too long... but the Leaf automatically recharges the 12V battery from the main battery as needed.
You're technically right, but wrong in a broader and more important sense.
The excessive scope and duration of copyright, along with the theory of copyright that is pushed by the content industry, give individuals very little moral reason not to infringe.
As the brief points out, the true purpose of copyright is to advance the progress of science and the useful arts, by providing some temporary incentives. In other words, it's a dynamic balance: society voluntarily agrees to restrict its own freedom with respect to expressive materials for a period of time in order to encourage the publication of material and its flow into the public domain. Under that theory, people have a moral incentive to hold up their side of the bargain.
But that's not the theory implied by the content industry's preferred view of copyright. They push the notion that creators have some natural right to control their creations, even after they've chosen to publish them to the world, and regardless of the fact that 99% of "their" creations were given to them by society. This notion is perfectly consistent with long and broad copyright, in fact it's easy to argue that if it's a right of the creator, and a heritable right, then it really should be perpetual. The view may not be completely consistent with the idea that the right can be sold, or that works made for hire belong to the employer, etc., but they're happy to ignore that point.
Under that theory, why should individuals not infringe? Obviously there are the potential legal penalties, but from a moral perspective, the calculation becomes one of percentages. How much harm does it really do if I pirate a copy of a song written and performed by a multi-millionaire musician? A dollar or two... a sum so small that the musician won't even notice. Even more, what if the choice is not between paying or not paying, but between listening without paying or not listening -- if in either case the musician doesn't get paid, what's the harm in making a copy?
And in fact the content industry has been enormously successful at pushing its view. If you ask the average man on the street who owns the copyright on Shakespeare's plays, he will see it as an obscure, trivial question, and one whose answer may be unknowable, but he's unlikely to realize that it's a nonsensical question, that not only are the plays long past any expiration, but that they should be in the public domain, that they're an important part of our shared culture and belong to us all.
Big media counts that as a success, but in fact it's a big part of what causes that same man on the street to feel that copyright infringement really isn't wrong in any deep sense. Curtailing the excessive scope and duration of copyright would enable people to understand why it is (or would be). Even better, decreasing copyright terms to the point where the average person sees a real choice between infringing now or waiting a few years for the work to enter the public domain would make the point even more sharply.
Ideally, we should set copyright duration to a point where a large majority of profitable works return a substantial portion of their lifetime expected profitability, and no longer. For films, for example, we could really set the term as short as two years, maybe less. It's a rare film that becomes profitable but hasn't done it within that time frame. Books probably need somewhere between 5 and 10 years, music is similar.
Shorter terms would not only help people understand why they shouldn't infringe, they would also serve the true purpose of copyright better, by making it less likely that successful artists, authors, etc. will produce one or two works and then give up creating, living on the the royalty stream.
Sorry, I got hung up on your "cannot be viewed" bit and didn't read the rest carefully enough. There's no need to encrypt it to make it non-modifiable and non-erasable, though. The easiest way to achieve "non-erasable" is probably just to require that storage be done by a third party who contracts to ensure that the data is never erased. Tamper resistance (or at least tamper evidence) is easily achieved with chained hashes. There may be some value in encryption for privacy, but that's a separate issue.
Asking after isn't a very good way to go. A lot of hormones are released that tend to make women forget how miserable they were. One of my friends wanted to have 5 or 6 kids until he had to put up with his wife while she was pregnant so it might not just be a boon for women.
She always said the same thing during, as well. Except for the last three weeks or so.
What's the total capacity in bytes?
If they can view it, they can choose to 'lose/destroy' the physical video if it does not help their side of the story. I'd rather both sides of a trial get the video sight-unseen without knowing if it'll help them or not.
Not necessarily. Depends on how the video is stored. I can think of many ways to make it accessible/viewable but make it impossible for them to modify or erase it.