Oh please! Quality control for a mosaic image? Good enough free tools for stitching mosaics will do just as good a job as any 'scientific' application for the same purpose.
Unfortunately I read many articles on this subject in magazines and journals printed on paper. I don't know where to find free evidence to support my case.
I'm not singling out DNA evidence in particular - in fact I don't think I even mentioned it. There was a great/. story a few months ago about someone who was fired because they failed a random drug test at work. But then the lawyer hired a mathematician who basically showed that even though the test was fairly reliable there was still a low probability that the person was a drug user. Even though the drug test said yes there was a low a priori probability that any individual person at this workplace was a user and by time you add up the relevant probabilities the test was basically worthless. If the lawyer hadn't had the foresight to get a mathematician involved the guy would be out of work. The default assumption in court is that forensic tests work and are reliable indicators unless you get your own expert to prove otherwise. Limited use of DNA evidence can be OK. But if you use DNA testing by trawling databases for 'cold hits' you get the same issue as with the drug test.
with torturing people would confess things they had not even done
(1) Pick up a history of medieval law or some such book. Torture was more effective than that. While it wasn't 100% reliable, or even 90% reliable, it probably wasn't a completely crap tool for determining the truth. It wasn't just used indicriminately - most countries in Europe had laws governing its use. For example, after a confession, the victim often was allowed an opportunity to retract their confession and further torture wasn't allowed until new evidence was found. This gave time for the innocent to deny their confession but the truly guilty often had the fear of God put into them and wouldn't retract at this stage.
(2) The forensic evidence bluff is probably better than torture. As long as people believe forensic science works the threat of it gets positive results.
(3) See the references I listed in one of my other posts. I'm not talking TV forensics here.
This is closer to the truth. The fact is, people get caught for crimes, not because of smart investigation, but the perpetrators are idiots. They show off about their achievements, or spend the money too quickly, or fail to realize that because they were married to the victim they're the obvious suspect - and so on.
There have been a few investigations of forensic science over the years. They don't get much publicity. It'd take me a while to track them down. Here are some typical examples from one source: here, here, here.
The point is that very little consistency checking goes on. For example forensic evidence is used to convict someone. Then the fact that they were convicted is used as evidence to support the accuracy of the forensic evidence without external validation. This is a very common theme. And it's interesting to listen to forensic experts speaking. Some are incredibly sure of themselves and seem to be unaware of the existence of statistical variation. That maggot story is a prime example. Maggots aren't an entirely stupid way to date the death of a corpse. But it's all about chance - like how many flies just happen to be in a particular area at any one time. But courts sem to be particularly poor at dealing with probabilistic information. (And there are numerous publications on that subject.)
The primary purpose of forensic science is to pretend to be a real science so as to fool people into thinking that there are reliable ways of tracking down who you are if you commit a crime. It's the myth of forensic science that keeps the crime rate down, not actual forensic science successes. Ask N forensic scientists when someone died, how long it was since their last meal and so on and you'll get N differet answers. But programs like CSI help to mislead people into thinking this stuff actually works.
In the days of yore the torture was used much leass than people imagine. Just the threat of torture was enough to make people confess. The same goes with forensic science. A cop says: "we have your DNA and we know it's you for sure" and that's enough to make someone confess. And as long as programs like CSI keep airing people will continue to fall for it.
In fact, the fact that forensic science is 90% bull is probably one of the best kept secrets left in the Western world.
If you want to say 'Huygens' and have the person you're speaking to understand that you'rr speaking about a space probe, and not clearing your throat, then ignore what you hear on that web site. It only applies to Dutch people.
I bought my wife a PC a year ago. I gave her some simple guidelines: don't view attachments unless you know what they are, don't give out your email address to unknown companies and so on. I downloaded the MS spyware beta the other day and ran it. It found zero spyware.
I have a friend who lives a few doors away. He's been having a lot of trouble with spyware. Porn ads popping up all day long. It was making his PC a pain to use even after a neighbor spent a few hours trying to clean it out. This same friend also likes to tell me about how much audio and video he's downloaded he's downloaded using the web and p2p.
It's pretty obvious what's going on. Your machine gets cluttered with spyware if you spend time on sketchy web sites downloading all and sundry and doesn't otherwise.
My solution is pretty simple - if I'm going to download porn off the web I use a Mac.
...come with a feature that detects skin tones and applies a blur in those areas to enhance the look of skin. The camera I was demoed could be set to recognize up to 3 different colors as skin tones. Anyway, as the US has a fairly consistent policy of waging war on people who have dark skins maybe they could use this technology in a 'smart' gun to avoid friendly fire. Hey, they could even give it to cops.
Note: In no way do I advocate racism. I'm just suggesting this technology for the use of those people who do.
I don't get it - did they all get steamrollered or something? And how come they always face precisely towards me whatever direction I move in? It's a weird universe this game is set in.
What does 'should' mean? There are groups of people: workers, company owners, geeks, consumers, not necessarily mutually exclusive, all of whom have their own different interests. You can never answer the question 'should' without knowing whose interests you are talking about. For a manual worker robots shouldn't take their jobs. For a company owner maybe they should. If he isn't prepared to even touch on this issue how can this guy think his article about what robots should and shouldn't do have any value?
I have a 3.2GHz machine under my desk and even that was because Santa was grumpy when he delivered my stuff. You probably just forgot to leave Santa a bottle of whisky by the chimney. I'm not sure what the fastest PC around is, say 4GHz.
Remember, growth of computing power is exponential and when comparing trends it often makes sense to switch to a logarithmic scale. log(10GHz/4GHz) isn't too much of a deviation from the trend.
That's like saying there's nothing wrong with murdering people, it's just really hard to pull off without killing them. If A implies B, and B is bad, then A is probably bad too, regardless of the existence of fantasies in which A is true and B is false.
I counter your Occam with a Voltaire: "A witty saying proves nothing". It matters nothing what Occam said unless you have a reason for thinking Occam's Razor is good. Why is it good?
It's as sensible as measuring distance travelled/max acceleration of a car. There simply isn't a linear relationship between these things and so dividing one by the other doesn't give you anything interesting. If we start dividing random variables by each other and reporting the result on/. we'd never get to read any interesting news.
...made by the usual bag companies and the computer manufacturer logos on them. If these were popular enough then I could safely carry round my laptop in a laptop bag because thieves would just think "nah! that's just one of those fancy high end pizza boxes".
...in around 1977 at around 10 or 11 yrs old programming mainframes in Fortran with punched cards. We had a computer club after school and each evening the teacher who ran it would take our cards down to Imperial College (London) to run our programs.
My very first program was to compute the relativistic time dilation you get when moving at a walking pace. Imagine my disappointment when the result was 1.00000000. I thought that these big machines that could compute thousands (if not millions) of operations per second would at least give me enough digits to see a slight deviation from 1. The answer was no better than what I could get from my Casio calculator.
By the way, I used to punch those cards with a little pin on the end of a dowel. It was practically medieval!
A year or two later I joined the mainstream: programming ZX80s and BBC Micros in BASIC.
Slashdot Mirror
Oh please! Quality control for a mosaic image? Good enough free tools for stitching mosaics will do just as good a job as any 'scientific' application for the same purpose.
I'm not singling out DNA evidence in particular - in fact I don't think I even mentioned it. There was a great /. story a few months ago about someone who was fired because they failed a random drug test at work. But then the lawyer hired a mathematician who basically showed that even though the test was fairly reliable there was still a low probability that the person was a drug user. Even though the drug test said yes there was a low a priori probability that any individual person at this workplace was a user and by time you add up the relevant probabilities the test was basically worthless. If the lawyer hadn't had the foresight to get a mathematician involved the guy would be out of work. The default assumption in court is that forensic tests work and are reliable indicators unless you get your own expert to prove otherwise. Limited use of DNA evidence can be OK. But if you use DNA testing by trawling databases for 'cold hits' you get the same issue as with the drug test.
(2) The forensic evidence bluff is probably better than torture. As long as people believe forensic science works the threat of it gets positive results.
(3) See the references I listed in one of my other posts. I'm not talking TV forensics here.
This is closer to the truth. The fact is, people get caught for crimes, not because of smart investigation, but the perpetrators are idiots. They show off about their achievements, or spend the money too quickly, or fail to realize that because they were married to the victim they're the obvious suspect - and so on.
The point is that very little consistency checking goes on. For example forensic evidence is used to convict someone. Then the fact that they were convicted is used as evidence to support the accuracy of the forensic evidence without external validation. This is a very common theme. And it's interesting to listen to forensic experts speaking. Some are incredibly sure of themselves and seem to be unaware of the existence of statistical variation. That maggot story is a prime example. Maggots aren't an entirely stupid way to date the death of a corpse. But it's all about chance - like how many flies just happen to be in a particular area at any one time. But courts sem to be particularly poor at dealing with probabilistic information. (And there are numerous publications on that subject.)
In the days of yore the torture was used much leass than people imagine. Just the threat of torture was enough to make people confess. The same goes with forensic science. A cop says: "we have your DNA and we know it's you for sure" and that's enough to make someone confess. And as long as programs like CSI keep airing people will continue to fall for it.
In fact, the fact that forensic science is 90% bull is probably one of the best kept secrets left in the Western world.
Now that's an episode I have to see! After the hints dropped in the DS9 tribbles sequel masterpiece I've been wondering about this... :-)
If you want to say 'Huygens' and have the person you're speaking to understand that you'rr speaking about a space probe, and not clearing your throat, then ignore what you hear on that web site. It only applies to Dutch people.
I have a friend who lives a few doors away. He's been having a lot of trouble with spyware. Porn ads popping up all day long. It was making his PC a pain to use even after a neighbor spent a few hours trying to clean it out. This same friend also likes to tell me about how much audio and video he's downloaded he's downloaded using the web and p2p.
It's pretty obvious what's going on. Your machine gets cluttered with spyware if you spend time on sketchy web sites downloading all and sundry and doesn't otherwise.
My solution is pretty simple - if I'm going to download porn off the web I use a Mac.
I'd hate to thing that someone might decide that F=m+a simply because addition is less stressful than multiplication for the average kid's brain.
These are the people who are apparently caught out by being asked 'Are you a terrorist?'. They'd never thing of anything so cunning.
Note: In no way do I advocate racism. I'm just suggesting this technology for the use of those people who do.
I don't get it - did they all get steamrollered or something? And how come they always face precisely towards me whatever direction I move in? It's a weird universe this game is set in.
What does 'should' mean? There are groups of people: workers, company owners, geeks, consumers, not necessarily mutually exclusive, all of whom have their own different interests. You can never answer the question 'should' without knowing whose interests you are talking about. For a manual worker robots shouldn't take their jobs. For a company owner maybe they should. If he isn't prepared to even touch on this issue how can this guy think his article about what robots should and shouldn't do have any value?
Just like the way terrorists respond truthfully when asked "Are you a terrorist?".
Remember, growth of computing power is exponential and when comparing trends it often makes sense to switch to a logarithmic scale. log(10GHz/4GHz) isn't too much of a deviation from the trend.
...pirates the RIAA actually have a chance of catching Grandma.
bash> $PREFERRED_P2P_CLIENT
That's like saying there's nothing wrong with murdering people, it's just really hard to pull off without killing them. If A implies B, and B is bad, then A is probably bad too, regardless of the existence of fantasies in which A is true and B is false.
...had bad working conditions!
I counter your Occam with a Voltaire: "A witty saying proves nothing". It matters nothing what Occam said unless you have a reason for thinking Occam's Razor is good. Why is it good?
It's as sensible as measuring distance travelled/max acceleration of a car. There simply isn't a linear relationship between these things and so dividing one by the other doesn't give you anything interesting. If we start dividing random variables by each other and reporting the result on /. we'd never get to read any interesting news.
(Incidentally, if you're in the SF Bay Area you can play with triangular rollers at the Exploratorium. Amazingly they give a nice smooth ride.)
...made by the usual bag companies and the computer manufacturer logos on them. If these were popular enough then I could safely carry round my laptop in a laptop bag because thieves would just think "nah! that's just one of those fancy high end pizza boxes".
My very first program was to compute the relativistic time dilation you get when moving at a walking pace. Imagine my disappointment when the result was 1.00000000. I thought that these big machines that could compute thousands (if not millions) of operations per second would at least give me enough digits to see a slight deviation from 1. The answer was no better than what I could get from my Casio calculator.
By the way, I used to punch those cards with a little pin on the end of a dowel. It was practically medieval!
A year or two later I joined the mainstream: programming ZX80s and BBC Micros in BASIC.