At this time WHEN CONDITIONS GET BAD WE DON'T LET THE AI DRIVE.
This is meaningless without some definition of what "bad" means. There are times when conditions get so bad that I don't let myself drive but that is pretty rare so unless we know that "bad" for the AI means a bit or rain or the sun going behind a cloud there is no way to draw any conclusion from that statement.
However, the comment you replied to also lacks sufficient information to be able to judge its validity. An AI may be better than someone texting on their phone or drunk-driving but, given that both acts are illegal in most places, how many people do this? Being better than the worst 5% of drivers would still make driverless cars far more dangerous than human drivers on average. I suspect that the data to evaluate this exists by now though. What we need is the average distance driven per fatality which should be well established for human drivers and will give us an idea whether driverless cars are ahead or behind this.
In this era of computers and CPU's and constant distraction, he wouldn't have managed to get to even first realization.
....and yet there are thousands of papers published by theorists each year which suggest that people still manage to come up with abstract new ideas in fundamental physics for us to test in our experiments. While it is true that none of these have been as significant as Einstein's papers that's not surprising: if papers this significant came up on a regular basis it would mean that we were doing a really bad job figuring out how the universe works. There were 200 years between Newton and Einstein and another hundred years later we are still only just seeing some of Einstein's predictions for the first time with gravitational waves being the latest discovery.
You make an unfounded assumption, namely that "each country's individual accuracy rate (as a fraction) is constant" - that's not supported by the article.
It is for the US, China and India all of which at various points the article states have constant retraction or investigation rates. Admittedly the statements are made without any data to support them but they are made. The data they do give suggests that the rate may actually show a small overall rise as well if you judge the statistical uncertainty by eye from the plot - providing error bars would have been better.
This article appears to contradict itself. It claims that the rate of inaccurate papers as a fraction of the total is not increasing. However, it also notes that countries with weaker misconduct policies, like China and India, have far higher rates of problem papers. It also notes that while the fraction of papers in these countries with issues is approximately constant the overall share of papers coming from these countries is increasing. Hence, the overall fraction of papers with problems must be increasing too because more and more papers are coming from countries with higher rates of inaccuracies while each country's individual accuracy rate (as a fraction) is constant.
It also seems very narrowly focussed on deliberate attempts to mislead since it concentrates on discipline procedures and investigations. However, the reproducibility problem is generally acknowledged to be mainly due to poor scientific practice, e.g. claiming that correlation implies causation or not understanding statistics, and not due to deliberate malfeasance.
The data also show that there does appear to be a slight increase in the number of corrections per journal - although this is only small and the plot fails to provide error bars so it is impossible to know whether or not this is statistically meaningful. It also cryptically mentions that this is for journals which issue corrections suggesting that there are journals which do not issue them.
The number of invitations I get to be an editor on new journals by predatory publishers has markedly increased over the past few years so, at least based on my experience, that there appear to be many more predatory journals than there used to be and I would be amazed if any cared enough to publish errata given that there is no money in it for them so, if the fraction of junk publications has increased this might entirely hide a large source of irreproducible papers from this study.
I don't know why this is particularly worrying. There are lots of children's stories that are more horrific and less believable: Hansel and Gretel, Little Red Riding Hood, Jack and the Beanstalk etc. That's providing you look at the originals and not the heavily sanitized Disney versions.
Most kids are smart enough to realize that the world is not run by reptilian aliens in the same way that they don't worry about running into an evil witch who eats children and lives in a gingerbread house every time they go into a forest. As for the few that are not perhaps it is worth exposing them to this sort of thing while there is an adult around to point out that it is just a fairy story so they can learn that not everything they hear is true.
So, just because current science prefer marbles and mattresses, it doesn't make it particularly funny if someone else uses turtles and oceans.
Science prefers "marbles and mattresses" because if you put a marble on a mattress it deforms the surface so that a smaller marble nearby will fall towards it. If you put a turtle in an ocean it will not suddenly cause all the smaller turtles nearby to be pulled towards it. So the reason one analogy is preferred over the other is that marbles and mastresses work and the turtles do not and if you change the story to have the person reply "It's swimming in an ocean of milk" they don't come across as any less foolish!
'Oh the land is going down a millimetre a year,' and that can be kind of unimpressive," says William Hammond, a researcher at the University of Nevada Reno who studies subsidence (but was not involved in this particular project). "But we know as scientists that these motions, especially if they come from plate tectonics, that they are relentless and they will never stop, at least as long as we're alive on this planet."
Actually as scientists we know that's not true provided we hang around for long enough (several hundred million years) as some other species have managed to. Plate tectonics can also uplift land and is responsible for mountain building. As new plates form and others merge the effects in one location can change....but you will need to hold your breath, literally, for a few million years or more (or, on that timescale, evolve gills) so it's not particularly helpful!
MS is not saying that handing over the data would be illegal for them to do on the EU side
Hence, as stated, European law says that MS cannot hand the data over to US authorities in this situation. That does not rule out that there are other, legal ways to get the data only that in this situation it would be illegal for them to do so.
It is utterly unlawful to create a situation where it is no longer possible for somebody to comply with all the laws that are over them.
No, it's stupid, shortsighted and undermines everyone's confidence in the law but unless there is a law which says that this is unlawful it is perfectly legal for this situation to arise. If it happens within the laws of a single country then a judge would be able to resolve it be ruling what the correct course of action is. However, when the laws of two independent countries collide there is no single authority which can resolve the contradiction. The usual way this is avoided is by not extending your laws beyond your national boundaries but since the US seems to be completely ignoring this constraint now this type of problem is likely to keep cropping up for US companies and citizens.
Linux is way too fucking big and popular to be squashed.
Nothing is too big or popular to be relatively quickly replaced by a superior successor. However, to do this you need something new which out performs the software it replaces. Just look at how quickly Blackberry crashed and burned under the onslaught from iPhone and Andriod.
This is why Linux has nothing to fear from Microsoft. MS have tried and failed to get Windows into the mobile device market because the OS is too big, clunky and resource intensive to run well on a small device. Similarly, their inroads on the server market have similarly had very limited success for the same reasons: Windows is a desktop OS and does not cope well with other uses.
On the flip side Linux has repeatedly tried to get in on the desktop with a similar lack of success (unless 2018 really is the year of Linux on the desktop;-). Given that each OS clearly outperforms the other in very different areas it makes a huge amount of sense for MS to embrace Linux. All fighting Linux has done is drive users to MacOS which is unix-based and works well with Linux. Embracing Linux is the best way to get developers, scientists etc to move to Windows by providing them the same desktop+unix environment MacOS provides. The recent string of minor and dud innovations from Apple plus their high and rising prices for no-longer cutting edge hardware also helps but that's not something MS can control!
So I don't see this as "embrace, extend and extinguish" because that will simply not work: they already fought Linux in servers and mobile and lost. This is more playing to their strengths on the desktop: it's "embrace, encourage and expand".
Ohhhhh, patently offensive material! That could be like bleeding heart politicians and bullshit bills.
Oh, I don't know I think it might be quite amusing at the moment given that your current president is someone who a good fraction of your country finds patently offensive and attempting to ban every mention of him on the internet will be a fun exercise to watch from a safe distance, especially when he finds out what they are trying to do. With a bit of luck it may distract him from his usual business of stirring up a nuclear/trade/cold/... (depending on the flavour of the week) war.
Microsoft owns that data, thus it is Microsoft property, and since Microsoft is an american citizen, it must adhere to American laws.
Yes, but any American citizen in Europe must obey European laws. If you happen to be a holocaust denier and get arrested in Germany explaining about the US constitution's protections on free speech will get you nowhere.US companies have the choice not to go to Europe but, if they do, they must follow the law there. European law says that Microsoft cannot hand the data over to foreign (US) authorities. There is no law that the US Congress can pass that can relieve them of this responsibility and, if the US forces MS to hand over the data, it will make it close to impossible for US companies to do business in Europe since they will be unable to follow the law and so be liable for financial and possibly criminal penalties.
No, I know no magic technologies are required but the so-called scientific paper specifically mentions that they may threaten to "make the sun go supernova" and that would require "magic" technologies and new physics. This was my point: it is bad enough that someone has written a supposedly scientific paper about hypothetical alien threats but, not content with that, they go out of their way to include non-scientific threats when there is no need to do so.
In a science fiction novel, it is fine to assume that aliens have magic technology which allows them to do things which we cannot do. However, when writing what purports to be a science paper invoking aliens with magic technology tends to be somewhat frowned on. Even then, though, this premise is flawed because 2010 needed "magic" alien technology to warm a planet for new life to evolve and flourish while current technology is sufficient to provide existential threats to humanity e.g. genetically engineered plague, nuclear bombardment etc. Inventing new, "magic" technology when none is needed is bad science fiction. In something claiming to be a scientific paper, it is appalling!
Even worse they get the science wrong. Stars the mass of the sun do not go supernova because they lack sufficient mass. Overcoming this would be a monumental task: you have to exhaust 5 billion years worth of hydrogen and then somehow hold the star together while it fuses all the way up to iron.
The result is that this paper reads more like the plot of a second-rate Hollywood science-fiction movie where they get the science horribly wrong.
We are talking about a _wooden_ highrise here. If you have a _wooden_ frame, not a steel one, then it certainly can burn and collapse. I used 9/11 as an example of what happens when the stairs fail (in this case due to a plane hitting the building). People were not trapped in the stairwell they were trapped by a lack of a stairwell which prevented them from getting out of the building. Those below could, and did, leave. However, I suspect this reply is a waste of time because if you don't believe that anyone died in 9/11 then you are delusional.
So it's quite possible for a concrete building to become a fire trap;
True, but how many more would have died if instead of just the cladding the main structural support material was inflammable? Most highrise buildings have staircases which are made of solid concrete to provide a safe, non-flammable escape route from most fires. When that fails, for example in the 911 attacks, the death toll can be one or more orders of magnitude larger because there is no safe escape route and the building will eventually collapse killing everyone who is trapped.
Once again I've had big updates like that on my machines. At home I run Windows Pro and at work Windows Education and neither have ever done this. The annoyance for me is if I have left something running in the Windows Linux subsystem the job gets killed if the update reboots the machine. However, it has never installed any new apps or uninstalled old ones. Indeed in my experience the update procedure for Win 10 has been a lot less hassle than it was for my previous mac which would fail the update a good fraction of the time when told to do it at night.
I agree that auto installing or uninstalling apps is completely unacceptable. However, I've been using Windows 10 for about a year now and this has never happened to me once. Is there some hidden "stop being an arsehole" option which needs to be set?
Improvements to improve capacity also increase their revenue since there are per-passenger and per-plane usage fees that they already charge. Adding a separate "improvement fee" so they can increase capacity and make more money is what is disingenuous.
You already pay most of these fees except for the facetious ones at the end plus others that you have not included like "airport improvement" despite the fact that airports almost never seem to improve no matter how much you pay.
Most of these fees are hidden by being included in your ticket price either directly as a fee you pay or as fees which the airline pays and so passes on indirectly (like landing fees) to you. Have a look at the list of fees on your next air ticket. Typical charges on mine are things like: Air Travellers Security Charge; Passenger Service Charge; Air Passenger Duty; Airport Improvement Fee etc. The exact names vary based on the countries you are flying to/from but they are always there and are becoming a considerable fraction of the cost of flying.
Slashdot Mirror
At this time WHEN CONDITIONS GET BAD WE DON'T LET THE AI DRIVE.
This is meaningless without some definition of what "bad" means. There are times when conditions get so bad that I don't let myself drive but that is pretty rare so unless we know that "bad" for the AI means a bit or rain or the sun going behind a cloud there is no way to draw any conclusion from that statement.
However, the comment you replied to also lacks sufficient information to be able to judge its validity. An AI may be better than someone texting on their phone or drunk-driving but, given that both acts are illegal in most places, how many people do this? Being better than the worst 5% of drivers would still make driverless cars far more dangerous than human drivers on average. I suspect that the data to evaluate this exists by now though. What we need is the average distance driven per fatality which should be well established for human drivers and will give us an idea whether driverless cars are ahead or behind this.
In this era of computers and CPU's and constant distraction, he wouldn't have managed to get to even first realization.
You make an unfounded assumption, namely that "each country's individual accuracy rate (as a fraction) is constant" - that's not supported by the article.
It is for the US, China and India all of which at various points the article states have constant retraction or investigation rates. Admittedly the statements are made without any data to support them but they are made. The data they do give suggests that the rate may actually show a small overall rise as well if you judge the statistical uncertainty by eye from the plot - providing error bars would have been better.
This article appears to contradict itself. It claims that the rate of inaccurate papers as a fraction of the total is not increasing. However, it also notes that countries with weaker misconduct policies, like China and India, have far higher rates of problem papers. It also notes that while the fraction of papers in these countries with issues is approximately constant the overall share of papers coming from these countries is increasing. Hence, the overall fraction of papers with problems must be increasing too because more and more papers are coming from countries with higher rates of inaccuracies while each country's individual accuracy rate (as a fraction) is constant.
It also seems very narrowly focussed on deliberate attempts to mislead since it concentrates on discipline procedures and investigations. However, the reproducibility problem is generally acknowledged to be mainly due to poor scientific practice, e.g. claiming that correlation implies causation or not understanding statistics, and not due to deliberate malfeasance.
The data also show that there does appear to be a slight increase in the number of corrections per journal - although this is only small and the plot fails to provide error bars so it is impossible to know whether or not this is statistically meaningful. It also cryptically mentions that this is for journals which issue corrections suggesting that there are journals which do not issue them.
The number of invitations I get to be an editor on new journals by predatory publishers has markedly increased over the past few years so, at least based on my experience, that there appear to be many more predatory journals than there used to be and I would be amazed if any cared enough to publish errata given that there is no money in it for them so, if the fraction of junk publications has increased this might entirely hide a large source of irreproducible papers from this study.
I don't know why this is particularly worrying. There are lots of children's stories that are more horrific and less believable: Hansel and Gretel, Little Red Riding Hood, Jack and the Beanstalk etc. That's providing you look at the originals and not the heavily sanitized Disney versions.
Most kids are smart enough to realize that the world is not run by reptilian aliens in the same way that they don't worry about running into an evil witch who eats children and lives in a gingerbread house every time they go into a forest. As for the few that are not perhaps it is worth exposing them to this sort of thing while there is an adult around to point out that it is just a fairy story so they can learn that not everything they hear is true.
That right there tells you why linux will never succeed on the desktop.
No, that tells you the sort of people who already use Linux on the desktop.
vim is great because it's on all platforms...
Oh come on. Vim is far more versatile than emacs. Have you ever seen anyone using emacs to clean their toilet?
So, just because current science prefer marbles and mattresses, it doesn't make it particularly funny if someone else uses turtles and oceans.
Science prefers "marbles and mattresses" because if you put a marble on a mattress it deforms the surface so that a smaller marble nearby will fall towards it. If you put a turtle in an ocean it will not suddenly cause all the smaller turtles nearby to be pulled towards it. So the reason one analogy is preferred over the other is that marbles and mastresses work and the turtles do not and if you change the story to have the person reply "It's swimming in an ocean of milk" they don't come across as any less foolish!
'Oh the land is going down a millimetre a year,' and that can be kind of unimpressive," says William Hammond, a researcher at the University of Nevada Reno who studies subsidence (but was not involved in this particular project). "But we know as scientists that these motions, especially if they come from plate tectonics, that they are relentless and they will never stop, at least as long as we're alive on this planet."
Actually as scientists we know that's not true provided we hang around for long enough (several hundred million years) as some other species have managed to. Plate tectonics can also uplift land and is responsible for mountain building. As new plates form and others merge the effects in one location can change....but you will need to hold your breath, literally, for a few million years or more (or, on that timescale, evolve gills) so it's not particularly helpful!
MS is not saying that handing over the data would be illegal for them to do on the EU side
Hence, as stated, European law says that MS cannot hand the data over to US authorities in this situation. That does not rule out that there are other, legal ways to get the data only that in this situation it would be illegal for them to do so.
It is utterly unlawful to create a situation where it is no longer possible for somebody to comply with all the laws that are over them.
No, it's stupid, shortsighted and undermines everyone's confidence in the law but unless there is a law which says that this is unlawful it is perfectly legal for this situation to arise. If it happens within the laws of a single country then a judge would be able to resolve it be ruling what the correct course of action is. However, when the laws of two independent countries collide there is no single authority which can resolve the contradiction. The usual way this is avoided is by not extending your laws beyond your national boundaries but since the US seems to be completely ignoring this constraint now this type of problem is likely to keep cropping up for US companies and citizens.
Linux is way too fucking big and popular to be squashed.
Nothing is too big or popular to be relatively quickly replaced by a superior successor. However, to do this you need something new which out performs the software it replaces. Just look at how quickly Blackberry crashed and burned under the onslaught from iPhone and Andriod.
;-). Given that each OS clearly outperforms the other in very different areas it makes a huge amount of sense for MS to embrace Linux. All fighting Linux has done is drive users to MacOS which is unix-based and works well with Linux. Embracing Linux is the best way to get developers, scientists etc to move to Windows by providing them the same desktop+unix environment MacOS provides. The recent string of minor and dud innovations from Apple plus their high and rising prices for no-longer cutting edge hardware also helps but that's not something MS can control!
This is why Linux has nothing to fear from Microsoft. MS have tried and failed to get Windows into the mobile device market because the OS is too big, clunky and resource intensive to run well on a small device. Similarly, their inroads on the server market have similarly had very limited success for the same reasons: Windows is a desktop OS and does not cope well with other uses.
On the flip side Linux has repeatedly tried to get in on the desktop with a similar lack of success (unless 2018 really is the year of Linux on the desktop
So I don't see this as "embrace, extend and extinguish" because that will simply not work: they already fought Linux in servers and mobile and lost. This is more playing to their strengths on the desktop: it's "embrace, encourage and expand".
Ohhhhh, patently offensive material! That could be like bleeding heart politicians and bullshit bills.
Oh, I don't know I think it might be quite amusing at the moment given that your current president is someone who a good fraction of your country finds patently offensive and attempting to ban every mention of him on the internet will be a fun exercise to watch from a safe distance, especially when he finds out what they are trying to do. With a bit of luck it may distract him from his usual business of stirring up a nuclear/trade/cold/... (depending on the flavour of the week) war.
Microsoft owns that data, thus it is Microsoft property, and since Microsoft is an american citizen, it must adhere to American laws.
Yes, but any American citizen in Europe must obey European laws. If you happen to be a holocaust denier and get arrested in Germany explaining about the US constitution's protections on free speech will get you nowhere.US companies have the choice not to go to Europe but, if they do, they must follow the law there. European law says that Microsoft cannot hand the data over to foreign (US) authorities. There is no law that the US Congress can pass that can relieve them of this responsibility and, if the US forces MS to hand over the data, it will make it close to impossible for US companies to do business in Europe since they will be unable to follow the law and so be liable for financial and possibly criminal penalties.
No, I know no magic technologies are required but the so-called scientific paper specifically mentions that they may threaten to "make the sun go supernova" and that would require "magic" technologies and new physics. This was my point: it is bad enough that someone has written a supposedly scientific paper about hypothetical alien threats but, not content with that, they go out of their way to include non-scientific threats when there is no need to do so.
In a science fiction novel, it is fine to assume that aliens have magic technology which allows them to do things which we cannot do. However, when writing what purports to be a science paper invoking aliens with magic technology tends to be somewhat frowned on. Even then, though, this premise is flawed because 2010 needed "magic" alien technology to warm a planet for new life to evolve and flourish while current technology is sufficient to provide existential threats to humanity e.g. genetically engineered plague, nuclear bombardment etc. Inventing new, "magic" technology when none is needed is bad science fiction. In something claiming to be a scientific paper, it is appalling!
Even worse they get the science wrong. Stars the mass of the sun do not go supernova because they lack sufficient mass. Overcoming this would be a monumental task: you have to exhaust 5 billion years worth of hydrogen and then somehow hold the star together while it fuses all the way up to iron.
The result is that this paper reads more like the plot of a second-rate Hollywood science-fiction movie where they get the science horribly wrong.
You missed the infused humanity!!
No, he was just pointing out that today the bit of humanity that Apple infuses into its product today is greed....and possibly a love of dongles.
We are talking about a _wooden_ highrise here. If you have a _wooden_ frame, not a steel one, then it certainly can burn and collapse. I used 9/11 as an example of what happens when the stairs fail (in this case due to a plane hitting the building). People were not trapped in the stairwell they were trapped by a lack of a stairwell which prevented them from getting out of the building. Those below could, and did, leave. However, I suspect this reply is a waste of time because if you don't believe that anyone died in 9/11 then you are delusional.
So it's quite possible for a concrete building to become a fire trap;
True, but how many more would have died if instead of just the cladding the main structural support material was inflammable? Most highrise buildings have staircases which are made of solid concrete to provide a safe, non-flammable escape route from most fires. When that fails, for example in the 911 attacks, the death toll can be one or more orders of magnitude larger because there is no safe escape route and the building will eventually collapse killing everyone who is trapped.
Once again I've had big updates like that on my machines. At home I run Windows Pro and at work Windows Education and neither have ever done this. The annoyance for me is if I have left something running in the Windows Linux subsystem the job gets killed if the update reboots the machine. However, it has never installed any new apps or uninstalled old ones. Indeed in my experience the update procedure for Win 10 has been a lot less hassle than it was for my previous mac which would fail the update a good fraction of the time when told to do it at night.
I agree that auto installing or uninstalling apps is completely unacceptable. However, I've been using Windows 10 for about a year now and this has never happened to me once. Is there some hidden "stop being an arsehole" option which needs to be set?
I do.
Improvements to improve capacity also increase their revenue since there are per-passenger and per-plane usage fees that they already charge. Adding a separate "improvement fee" so they can increase capacity and make more money is what is disingenuous.
You already pay most of these fees except for the facetious ones at the end plus others that you have not included like "airport improvement" despite the fact that airports almost never seem to improve no matter how much you pay.
Most of these fees are hidden by being included in your ticket price either directly as a fee you pay or as fees which the airline pays and so passes on indirectly (like landing fees) to you. Have a look at the list of fees on your next air ticket. Typical charges on mine are things like: Air Travellers Security Charge; Passenger Service Charge; Air Passenger Duty; Airport Improvement Fee etc. The exact names vary based on the countries you are flying to/from but they are always there and are becoming a considerable fraction of the cost of flying.