Actually the statistics imply that odds are you are alive *now*, not that you'll live forever (I know, brutal abuse). But actually thats a pretty interesting statistic to think about since it gives you a developmental perspective on the human species.
Actually, sunspots are cold! That is why they appear dark on a photograph. Solar flares on the other hand are *very* hot.
Anyways the sunspots are caused by a local irregularity in the magnetic field, and those irregularities become very small during a solar minimum, which is why there are no sunspots right now. The solar minimum is what makes the sun cold, not the lack of sunspots. During a solar maximum the magnetic field gets very "twisty" and there are lots of sunspots in addition to flares, mass ejections and so on.
Re:Read an article to this effect....
on
Cracking Go
·
· Score: 1
Statistical methods are no less "theoretically sound", and in addition can handle a lot of situations where branch analysis is impossible (e.g. continuous domains). For example, simulated annealing is guaranteed to find the global optimum given infinite running time, and many of the latest algorithms in statistical learning theory have quantifiable bounds on the error.
There are lots of things that humans do better than computers, some of which we regard as fairly simple every-day activities -- is the human ability to play Go really *unusual*?
But if they are inexperienced(usually young managers) they unwisely say in a recordable way that they are firing the person because they are "sick, old, black, female, etc." Even in an "at will" state, this will get you sued.
Yeah, the *truth* is a weasely little thing... it can be quite hard to contain!
You have to stay abrest of trends to be relevant, that is true for any subfield of engineering, and for any creative field also. But the good news is that this task is actually easier now than it ever has been -- e.g., Wikipedia will give you an un-hyped description of basically any buzzword. You can stay on top of the alphabet soup with just a few minutes of reading.
It works if you really want to learn the subject, and it helps if the professor has a good lecture style. Also, the webcast course materials are almost always entry-level material so you could take a few survey courses in different areas but don't expect to get anywhere near the depth that an actual degree would entail. In other words, if you don't understand something in the material its probably just because you were not paying attention--its just not that complicated.
Oh and UCB has had public webcast courses for years, just not through YouTube.
America has done a pretty decent job at encourage people to hate us, so to some extent the crazy security is actually justified. And attacks on airplanes and other transit systems are one of the most effective ways to cause infrastructure damage and inspire fear. The fact is, our foreign policy makes air travel less safe. I have no love for the current political trends, but I actually don't find airport security to be overbearing, but maybe that is just because I'm white... nonetheless, some of my friends who are just as white love to complain about it. I just don't see what the big deal is... its not like having to take my shoes off is some huge infraction of my rights.
By the way I have recently taken experimental/prototype electronic devices in carry-on and its not a problem as long as it is well packaged, does not look like a bomb on the x-ray and you are willing to answer a few questions. Refusing to answer a question is just about the worst thing you can do. The exception is US customs, they are extremely paranoid so expect some delays if you are connecting from an international flight to a domestic one.
This counter argument rests on the assumption that solving a crime is simply a matter of having more police resources available. However we know that isn't the only factor, because if you have crimes where you can't ID the perp, no amount of extra police resources will help, short of the gestapo.
Anyways, for moderate changes in the total amount of crime this effect is insignificant, which is why I didn't mention it.
To be fair, it is possible that the cops simply are overworked, and that is why the video cameras don't help -- maybe they just don't have time to review the tapes. If that is the case then its going to be a pretty big embarrassment for the people who spent 200M on cameras when they should have spent the money hiring more officers. That is why the results of this study merit follow-up research into *why* the cameras don't help, because we don't know why and we shouldn't jump to conclusions without looking at the data first.
Personally I don't think people really consider what motivates criminals when considering prevention strategies. There are lots of types of crimes, but street crime, the type that the camera system targets, its the hardest type to enforce due to the free movement of the criminal. Basically all of street crime is motivated by two things: poverty and drug addiction, and its a shame that politicians are willing to just throw money at the problem rather than actually try to find ways to help people.
I'll say it again -- the probability of catching a criminal is independent of the total amount of crime. This a result of Bayes theorem and involves dividing out the total amount to create a marginal probability. As an aside you are correct that the type (or severity) of crime has to be considered also, which is basically the one extra variable that makes actually doing this study rather complicated. However for the moment lets ignore the problem of type and look at the two basic questions.
Consider this: The first question is the marginal probability, independent of amount, and the second concerns directly the amount. Here are the possible outcomes:
1) Cameras do not improve the probability of catching criminals, and crime rates after installing cameras went up or stayed the same. Likely interpretation: the cameras are a waste of money.
2) Cameras do not improve the probability of catching criminals, but crime rates went down. Likely interpretation: arguably inconclusive, maybe the criminals just moved on to the next town for a reason unrelated to the cameras. This outcome would merit further study into correlation of crime rates with other variables, like crime rates in the next town over, economic data, etc. However, its not likely that this outcome would support installing *more* cameras as there isn't any direct evidence that they help.
3) Cameras DO improve the probability of catching criminals, but crime rates went up or stayed the same. Likely interpretation: The cameras help catch criminals! Great! Lets install *more* so we can fight this crime wave!
4) Cameras DO improve the probability of catching criminals, AND crime rates went down! Likely interpretation: Whomever bought into this system is assured for re-election, its the greatest thing on earth.
Now if you go over those four cases you notice that outcomes 3 and 4 are both good press for the camera supporters, 1 is a disaster, and 2 is basically inconclusive.
Typically, "misuse" of statistics exploits the correlation = causation fallacy, but the misuse is usually more on the side of the media, i.e. the reporting of statistics, not the statistics themselves -- this could have been done, for example, by reporting a drop in total crime rates without really looking into what caused it. Oh and it *is* still possible that the cameras are not effective for some other, unforseen reason, like maybe the cops are too busy porking out on donuts to bother to review the tapes, or maybe the system is just cumbersome to use because it has a crappy UI or the image is too grainy to ID the perps.. Maybe the criminals throw towels over the cameras, or they are pointed in the wrong direction -- there are a lot of possibilities and it certainly merits some investigation. However, my intuition is that video evidence just isn't that useful because its too hard to get a positive ID much less find the perp in any moderate sized city of some half million people or more -- its chance at best. Its hardly surprising, really -- the cops have a hard enough time finding stolen cars even when they are in plain view, and those have a giant unique identifier stuck on both ends...
I personally don't believe that statistics are inherently subject to manipulation just because agendas are at stake. Statisticians are not stupid, they know how to do a proper analysis. However, I tend to be more skeptical of correlations that look at very large scale effects as opposed to narrowly defined questions. Looking directly at efficacy in terms of probability of catching a criminal is, in my estimation, is a good analysis -- it looks directly at the core function that the camera is supposed to perform and doesn't depend on socio-economic factors, ebb and flow of overall crime rates, etc.
Its a marginal probability so the percentages *are* directly comparable. In other words, according to this study fake cameras would be just as effective as real ones for the purpose of solving crimes.
Whether or not cameras -- real OR fake vs none at all has any effect on crime rate is a separate and independent analysis, but I'd say its highly unlikely, since the basic function of a deterrent is to increase risk, which clearly isn't the case here. Presumably that question has also been addressed, since it is an obvious one to raise.
Incidentally, one of the criticisms of the death penalty is that it isn't an effective deterrent either, but that analysis is based entirely on large scale trends, which is a much harder point to argue because the correlation isn't as clear.
I hope that we are a bit better off than that, these days. Modern mechanical engineering is based around finite element simulation and computational fluid dynamics that allows the designer to predict the stress on any component, even for very complex systems. For example the use of these methods is directly responsible for the huge increase in reliability of cars that we have seen over the last 25 years or so.
Of course people are constantly finding new and original ways to crash things...
I personally like best the extremes -- at low levels you have extremely precise control and a small domain of responsibility, and at the very highest level of programing (e.g. mathematical modeling) you have only incidental concern for machine optimization, unlimited structural possibilities and again a small domain of responsibility.
Its the mid-level and "glue" coding that sucks the most; that is where your manager wants the application to include everything and the kitchen sink, which in my experience leads to burn out.
Good vision is quite helpful for looking at very small things; for example soldering SMT components without the aid of a magnifying glass, reading fine print text and so on. Since the astronaut's tasks include running experiments they might need to be able to work with miniaturized components, read text on interfaces, graduations and other markings, etc. And a corrective lens can't restore both near and far sighted vision at the same time, and since it isn't very easy to "change glasses" in the middle of a spacewalk... I suppose they probably also require normal color vision, which excludes about 6% of the male population (and currently is not correctable).
Well if you want your family to be well off you could just buy life insurance and then kill yourself. If you want to be famous in addition, you could do the same but go on a killing spree and then kill yourself. So yeah, there are people who are willing to do that, but I wouldn't say that any of them are of sufficient mental fitness to serve as an astronaut for NASA.
Furthermore the question is basically moot because it is quite simple to design a manned mission to Mars that includes a return vehicle with fairly low risk (really, it is).
USB isn't a decentralized network like ethernet. There are no collisions because the host controller manages every transaction. This feature, by the way, makes possible a number of interesting things such as isochronous endpoints that have dedicated bandwidth and bounded latency.
In order to achieve that low overhead firewire controllers require a significant amount of dedicated silicon which translates to a higher manufacturing cost. However with the proliferation of additional general-purpose and specialized cores on the modern CPU I expect the overhead problem will eventually be moot.
Slashdot Mirror
Actually the statistics imply that odds are you are alive *now*, not that you'll live forever (I know, brutal abuse). But actually thats a pretty interesting statistic to think about since it gives you a developmental perspective on the human species.
Actually, sunspots are cold! That is why they appear dark on a photograph. Solar flares on the other hand are *very* hot.
Anyways the sunspots are caused by a local irregularity in the magnetic field, and those irregularities become very small during a solar minimum, which is why there are no sunspots right now. The solar minimum is what makes the sun cold, not the lack of sunspots. During a solar maximum the magnetic field gets very "twisty" and there are lots of sunspots in addition to flares, mass ejections and so on.
Statistical methods are no less "theoretically sound", and in addition can handle a lot of situations where branch analysis is impossible (e.g. continuous domains). For example, simulated annealing is guaranteed to find the global optimum given infinite running time, and many of the latest algorithms in statistical learning theory have quantifiable bounds on the error.
There are lots of things that humans do better than computers, some of which we regard as fairly simple every-day activities -- is the human ability to play Go really *unusual*?
I for one would much prefer to see money diverted from bomb-making and gratuitous eavesdropping to nerds who like to think about number theory.
But if they are inexperienced(usually young managers) they unwisely say in a recordable way that they are firing the person because they are "sick, old, black, female, etc." Even in an "at will" state, this will get you sued.
Yeah, the *truth* is a weasely little thing... it can be quite hard to contain!
You have to stay abrest of trends to be relevant, that is true for any subfield of engineering, and for any creative field also. But the good news is that this task is actually easier now than it ever has been -- e.g., Wikipedia will give you an un-hyped description of basically any buzzword. You can stay on top of the alphabet soup with just a few minutes of reading.
"In the broadest sense, a fraud is a deception made for personal gain." (wikipedia). "Fraud" not *strictly* a legal term.
It works if you really want to learn the subject, and it helps if the professor has a good lecture style. Also, the webcast course materials are almost always entry-level material so you could take a few survey courses in different areas but don't expect to get anywhere near the depth that an actual degree would entail. In other words, if you don't understand something in the material its probably just because you were not paying attention--its just not that complicated.
Oh and UCB has had public webcast courses for years, just not through YouTube.
Why not just encrypt the entire link?
America has done a pretty decent job at encourage people to hate us, so to some extent the crazy security is actually justified. And attacks on airplanes and other transit systems are one of the most effective ways to cause infrastructure damage and inspire fear. The fact is, our foreign policy makes air travel less safe. I have no love for the current political trends, but I actually don't find airport security to be overbearing, but maybe that is just because I'm white... nonetheless, some of my friends who are just as white love to complain about it. I just don't see what the big deal is... its not like having to take my shoes off is some huge infraction of my rights.
By the way I have recently taken experimental/prototype electronic devices in carry-on and its not a problem as long as it is well packaged, does not look like a bomb on the x-ray and you are willing to answer a few questions. Refusing to answer a question is just about the worst thing you can do. The exception is US customs, they are extremely paranoid so expect some delays if you are connecting from an international flight to a domestic one.
This counter argument rests on the assumption that solving a crime is simply a matter of having more police resources available. However we know that isn't the only factor, because if you have crimes where you can't ID the perp, no amount of extra police resources will help, short of the gestapo.
Anyways, for moderate changes in the total amount of crime this effect is insignificant, which is why I didn't mention it.
To be fair, it is possible that the cops simply are overworked, and that is why the video cameras don't help -- maybe they just don't have time to review the tapes. If that is the case then its going to be a pretty big embarrassment for the people who spent 200M on cameras when they should have spent the money hiring more officers. That is why the results of this study merit follow-up research into *why* the cameras don't help, because we don't know why and we shouldn't jump to conclusions without looking at the data first.
Personally I don't think people really consider what motivates criminals when considering prevention strategies. There are lots of types of crimes, but street crime, the type that the camera system targets, its the hardest type to enforce due to the free movement of the criminal. Basically all of street crime is motivated by two things: poverty and drug addiction, and its a shame that politicians are willing to just throw money at the problem rather than actually try to find ways to help people.
I'll say it again -- the probability of catching a criminal is independent of the total amount of crime. This a result of Bayes theorem and involves dividing out the total amount to create a marginal probability. As an aside you are correct that the type (or severity) of crime has to be considered also, which is basically the one extra variable that makes actually doing this study rather complicated. However for the moment lets ignore the problem of type and look at the two basic questions.
Consider this: The first question is the marginal probability, independent of amount, and the second concerns directly the amount. Here are the possible outcomes:
1) Cameras do not improve the probability of catching criminals, and crime rates after installing cameras went up or stayed the same. Likely interpretation: the cameras are a waste of money.
2) Cameras do not improve the probability of catching criminals, but crime rates went down. Likely interpretation: arguably inconclusive, maybe the criminals just moved on to the next town for a reason unrelated to the cameras. This outcome would merit further study into correlation of crime rates with other variables, like crime rates in the next town over, economic data, etc. However, its not likely that this outcome would support installing *more* cameras as there isn't any direct evidence that they help.
3) Cameras DO improve the probability of catching criminals, but crime rates went up or stayed the same. Likely interpretation: The cameras help catch criminals! Great! Lets install *more* so we can fight this crime wave!
4) Cameras DO improve the probability of catching criminals, AND crime rates went down! Likely interpretation: Whomever bought into this system is assured for re-election, its the greatest thing on earth.
Now if you go over those four cases you notice that outcomes 3 and 4 are both good press for the camera supporters, 1 is a disaster, and 2 is basically inconclusive.
Typically, "misuse" of statistics exploits the correlation = causation fallacy, but the misuse is usually more on the side of the media, i.e. the reporting of statistics, not the statistics themselves -- this could have been done, for example, by reporting a drop in total crime rates without really looking into what caused it. Oh and it *is* still possible that the cameras are not effective for some other, unforseen reason, like maybe the cops are too busy porking out on donuts to bother to review the tapes, or maybe the system is just cumbersome to use because it has a crappy UI or the image is too grainy to ID the perps.. Maybe the criminals throw towels over the cameras, or they are pointed in the wrong direction -- there are a lot of possibilities and it certainly merits some investigation. However, my intuition is that video evidence just isn't that useful because its too hard to get a positive ID much less find the perp in any moderate sized city of some half million people or more -- its chance at best. Its hardly surprising, really -- the cops have a hard enough time finding stolen cars even when they are in plain view, and those have a giant unique identifier stuck on both ends...
I personally don't believe that statistics are inherently subject to manipulation just because agendas are at stake. Statisticians are not stupid, they know how to do a proper analysis. However, I tend to be more skeptical of correlations that look at very large scale effects as opposed to narrowly defined questions. Looking directly at efficacy in terms of probability of catching a criminal is, in my estimation, is a good analysis -- it looks directly at the core function that the camera is supposed to perform and doesn't depend on socio-economic factors, ebb and flow of overall crime rates, etc.
Its a marginal probability so the percentages *are* directly comparable. In other words, according to this study fake cameras would be just as effective as real ones for the purpose of solving crimes.
Whether or not cameras -- real OR fake vs none at all has any effect on crime rate is a separate and independent analysis, but I'd say its highly unlikely, since the basic function of a deterrent is to increase risk, which clearly isn't the case here. Presumably that question has also been addressed, since it is an obvious one to raise.
Incidentally, one of the criticisms of the death penalty is that it isn't an effective deterrent either, but that analysis is based entirely on large scale trends, which is a much harder point to argue because the correlation isn't as clear.
Are you trolling or do you really have no female friends?
I hope that we are a bit better off than that, these days. Modern mechanical engineering is based around finite element simulation and computational fluid dynamics that allows the designer to predict the stress on any component, even for very complex systems. For example the use of these methods is directly responsible for the huge increase in reliability of cars that we have seen over the last 25 years or so.
Of course people are constantly finding new and original ways to crash things...
I personally like best the extremes -- at low levels you have extremely precise control and a small domain of responsibility, and at the very highest level of programing (e.g. mathematical modeling) you have only incidental concern for machine optimization, unlimited structural possibilities and again a small domain of responsibility.
Its the mid-level and "glue" coding that sucks the most; that is where your manager wants the application to include everything and the kitchen sink, which in my experience leads to burn out.
You could probably vacuum up the floating blobs quite easily... now that would make for some interesting housework!
Good vision is quite helpful for looking at very small things; for example soldering SMT components without the aid of a magnifying glass, reading fine print text and so on. Since the astronaut's tasks include running experiments they might need to be able to work with miniaturized components, read text on interfaces, graduations and other markings, etc. And a corrective lens can't restore both near and far sighted vision at the same time, and since it isn't very easy to "change glasses" in the middle of a spacewalk... I suppose they probably also require normal color vision, which excludes about 6% of the male population (and currently is not correctable).
Well if you want your family to be well off you could just buy life insurance and then kill yourself. If you want to be famous in addition, you could do the same but go on a killing spree and then kill yourself. So yeah, there are people who are willing to do that, but I wouldn't say that any of them are of sufficient mental fitness to serve as an astronaut for NASA.
Furthermore the question is basically moot because it is quite simple to design a manned mission to Mars that includes a return vehicle with fairly low risk (really, it is).
Actually they *are* attempting to make a gamma ray laser. Its a new area of physics research called "quantum nucleonics".
Oh man... do let me know if you ever find that guy, I'd like to kick him in the stomach after you punch him!
In spite of the standard it seems that the digital camera manufacturers still use their proprietary mini-plugs, which IMHO is rather annoying.
Nooooo!!! Sharks with frickin' USB3 ports on their heads!!
USB isn't a decentralized network like ethernet. There are no collisions because the host controller manages every transaction. This feature, by the way, makes possible a number of interesting things such as isochronous endpoints that have dedicated bandwidth and bounded latency.
In order to achieve that low overhead firewire controllers require a significant amount of dedicated silicon which translates to a higher manufacturing cost. However with the proliferation of additional general-purpose and specialized cores on the modern CPU I expect the overhead problem will eventually be moot.