I have tried several solutions and find org-mode to be the most useful one. I have it set up in essentially a GTD structure, with TODO items in one big list, a separate list of active projects, and a third list of potential future projects.
org-mode is extremely configurable, which is a definite plus for software you intend to organize your life. I recommend the following add-ons as well:
org-caldav to synchronize appointments and scheduled TODO tasks with Google calendar. (two-way sync that is!)
MobileOrg for Android, which will let your Nexus 7 work with your org-files.
Now you just take text notes in emacs and org-mode does the rest. I happen to write LaTeX and some code, and org-mode also supports literal programming. This is an extremely useful feature since I can write rich outlines of all my projects in org-mode and have the TODO items be placed naturally where they would have gone in my hand-written notes.
For a simple example of how this might work, suppose I'm drafting a research paper and don't remember some detail I don't care about currently. I might write, "Smith first published this theory in the 80s * TODO Look up the year for Smith". org-mode will find these todo items and tell me about them when I look at my todo list. Then when I find the appropriate date and want to insert it, it will take me directly to where I put the TODO item so I can insert it into the draft with minimal effort. In this way, outlines naturally progress toward finished projects, just as GTD wants them to.
To prove that all they should have to do is put twenty programmers on the stand and if ten of them can come up with that solution.
Why is that the standard of obviousness? When it comes to algorithm design, algorithms could be manifestly obvious to mathematicians and algorithm specialists but not to ordinary programmers. The obvious criterion should take into account the corporate structure, or else it will be profitable to keep a stable of uninformed programmers in the dark while they implement algorithms designed by specialists with a higher level of technical knowledge. Then any algorithm that is obvious to someone with technical knowledge will still be patentable because it's implemented by less informed programmers.
Because smart people, who think stuff up, ought to be able to get paid for their ideas. Right, and they're not because somebody filed an overly broad patent that covered the obvious portions of their clever ideas. So instead of being paid for their ideas and work, they are paying extortionary licensing fees for work they never used when coming up with their ideas.
Patents don't work for massively distributed fields like software development. The solutions are often obvious next steps that are waiting only for hardware advances or other external events. Solutions and algorithms are often simultaneously discovered by multiple parties independently. Patents are not needed to spur or reward innovation in software and processes, but they are useful as ways of forcing competitors to refrain from implementing obvious improvements.
That is why the rules for patents used by the Supreme Court disallow software patents, even though the lower courts have allowed them in certain circumstances thereby overriding the rules used by the SC (until the SC decides to hear a case that challenges the lower court rulings).
I haven't looked carefully at "abstract idea" and how that applies to patents (or doesn't).
Hi, thanks for the response.
I'm a mathematician and I honestly find a lot of the "all algorithms are math" argument to be rather weak for reasons I'd be happy to go into. I think "abstract idea" is the real issue. The best analysis of the situation that I've seen so far is Ben Klemens' "The Rise of the Information Processing Patent", the pdf of which can be found here http://www.bu.edu/law/central/jd/organizations/journals/scitech/volume141/documents/Klemens.pdf.
That's right. Your gene sequencing computer program should not be patentable because it's an abstract idea. That's independent of whether you implemented that abstract idea on a computer.
Anyone who is actually voting for wikileaks will likely be well informed and voting below the line anyways.
For those not familiar with australian voting, we have preferential instant runoff first past the pole voting.
You can either vote "above the line," where you select ONE party, and that party decides how your preferences fall if they don't win a seat, or you can vote "below the line," where you number individual candidates "1, 2, 3.....".
So, effectively this ensures that anybody voting for Wikileaks is voting their preferences. Since the above the line preferences are unpopular, voters can't just vote on the party line. Or else they'll possibly get white supremacists in office. Instead they have to rank the candidates *as they'd actually like to see them governing* instead of a more accurate approximation than otherwise. Even though they want votes, they can't rationally want votes from people who just follow a party line, since that creates an environment where party politics thrive and transparency is reduced.
That's just an academic point though, I have no idea whether it is relevant to their thought process.
I think you shouldn't be immune to the observation that in calling the parent's argument "based on emotion" and "deliberate propaganda" you occuse your opponent of "promoting... dictatorship and slavery".
Let's take a step back here. We are all geeks and we can approach governance from an engineering point of view. If I read you correctly, you're advocating a somewhat extreme social contract viewpoint in which the only legitimate function of government is to take us out of the state of nature by granting the government as minimal a monopoly on violence and power as possible. This view isn't unreasonable on its face, but there is more to it.
In any society, including humans, there is a distribution of power and resources that comes from natural survival of the fittest. The basic idea of government is to voluntarily abdicate much of this power to a central authority (where central can mean tribal, state, federal, world, etc). This creates the natural problem of a single point of attack: any group that can infiltrate the government will be able to use that centralized power for personal gain. This creates the need to impose laws on things like (1) who can serve in government (probably not felons, e.g.), (2) what limits members of government can have in their personal influence (e.g. can anybody unilaterally declare war?), (3) limitations on the influence other powerful social members have on the government (e.g. campaign funds, revolving doors) and many others. The point here is that it's a question of engineering. How do we organize the central authority so that the people have maximal freedom to do anything they want and they don't have to fear being pushed around by anyone.
Where you and your opponent disagree most is in determing the strongest threat to government infiltration. The social contract tradition, of which you seem to subscribe, was concerned primarily about landed aristocracies using the power of government to do anything they like. This was historically opposed by a class of business owners and farmers who favored a representative democracy. But in the course of history many things have changed. America has no landed aristocrats, so we no longer worry about that. Instead, it has a class of people who believe that selfishness is a positive virtue that is much more important than telling the truth or helping others. This belief flourishes in America's political structure because it is easy for such people to lie in order to gain votes but then to ignore the voters once in office. The parent believes that these people are using government to their advantage to the detriment of the vast majority of the public.
From this point of view, the "free market" rhetoric is a device to convince the population that there is a moral imperative to turn every part of society into a money-making enterprise. Naturally, if you do this you end up with something very much like what you see in, say, Boardwalk Empire or any other popular depiction of organized crime. In fact, it should be rather obvious that organized crime is only crime because there are laws that prevent business men from engaging in certain profitable businesses, such as human trafficking, drug exportation, war profiteering, and so forth. It raises the question of why any of these things is illegal in the first place.
And, quite obviously, the reason they are illegal is that in a truly free market, labor has the right to organize and the poor have the right to protest and vote. So in a free market, a government will voluntarily implement policies that limit the ways that businesses can exploit others for profit. The only way to get an "ideal" free market of the sort Milton Friedman advocates is to sneak pro-corporate laws onto the books or into the courts since naturally those laws will be opposed by anybody not directly benefited by them.
The mistake you make is that you don't see a free market as a process, you see it as a set of laws. But historically you never get that set of law
Hi, I posted the question. Maybe it's worth pointing out that I've used Linux and open source exclusively for well over a decade and have no interest in smearing anybody. I'm positive they have backdoors to closed source programs, because it has been leaked that they have access to MS exploits before they're fixed, and one of the Snowden slides implied the UK has the ability to break BlackBerry encryption from devices owned by heads of state and diplomats. I assume open source is the *safer* option, but I want to know how safe.
That said, the link you posted to *confirms* that US intelligence has tried to put back doors in encryption libraries! That's really all the information we need. My understanding is that the NSA is far more advanced in cryptography than the FBI. It seems almost negligence for the head of the NSA not to attempt to put back doors in open encryption libraries. Plus they've had 13 years since the FBI attempt to learn from their mistakes. So if we haven't heard of the NSA doing it, it's reasonable to wonder if that's because they're doing it extremely well.
As stated in the submission, although NSA hardened the algorithm to most attacks, they lobbied to reduce the key length. Specifically they wanted 48 bit keys instead of 64 bit. Perhaps there is a good reason for this, but on the face of it, weaker keys would seem to weaken the algorithm to brute force attacks. It may have just been that at the time computing power was the best advantage NSA had.
Hi, I wrote the submission. To fearmonger is to exagerrate some threat to use fear in order to promote some specific ends. This question is me asking to what extent caution should be justified so that I as a user can know what to do. I'm sure you can see how those things are extremely different and in fact the opposite. One is an attempt to drive action with fear you know is unjustified, the other is an attempt to systematically determine the appropriate amount of caution.
"This is fearmongering" seems inappropriate as a response to a submission that contains only links to Wikipedia documenting known facts and that even goes so far as to call some proponents of this theory paranoid.
Because many email providers -- such as gmail, hotmail, yahoo, etc -- want to read your email to serve you ads. Encryption runs counter to the profit motive.
I'm just a little confused about your wording. I'm not sure what it means for an average to be better or for betterness to apply. I can see at least two reasons to want to know about standard deviation. One is that you want to know about how ISP speeds differ according to other variables, like neighborhood. Another is to see whether the difference in mean ISP speeds is "significant."
If you're a frequentist, then the probability that ISP A and ISP B have different speeds (so that one is larger than the other) is 1. So sigma here is 0. P(A = B) = P(A- B = 0) corresponds to doing an integral along the rectangle [0,0], which must be 0. (More generally, it's a measure 0 set). But people report sample standard deviation estimates (which are not sigma, but estimates of sigma) to show how precise the measurement is. In this case, since there are so many data points, I would imagine that the sample standard deviations aren't that relevant.
If you're more curious about what the average individual might see in terms of speed, then you probably want more info about speed broken up by, say, location. And that is more about giving conditional distributions than about standard deviation.
The content of what you're saying is clear, but it's less clear how sigma will tell you about neighborhood-by-neighborhood variation. What they're reporting is a sample statistic, and sample variance estimates approach zero with large sample sizes. So their observed standard deviation is probably extremely tiny. If you subscribe to the view that there is a true mean out there to be measured, then sigma -- the true standard deviation -- is literally zero.
Data about variation would be useful, but that would mean publishing more than just standard deviation data. It basically means publishing the whole data set, or computing means conditioned on neigborhood.
Your points are all valid, but the problem isn't as bad as it seems. If the noise in the errors are uncorrelated (e.g. all cars throw off readings, but in a random way), then the noise disappears in the average (by the Mean Value Theorem). Otherwise, you just model the correlation and account for it (e.g. with Bayesian methods). More generally, there are numerous algorithms specifically designed to extract useful aggregate data from a large number of networked noisy/low quality sensors.
Statistical significance is a binary phenomenon, and there is no such thing as "high statistical significance."
That's like claiming that a result is "extremely true" because the contradiction you get by assuming it is "absolutely crazy".
Significance is not and never has been an in indicator of the probability of a theory, It's only an indication that you've passed the an extremely low threshold to make a claim (i.e. your claim is not plima facie absurd).
there are other human being starving and dying and suffering
Yeah, like the many people dead or wounded due to gang violence fueled by the street drug trade, or the many people addicted to drugs who can't get medical or treatment help because they will get arrested or simply ignored, the people dying in Afghanistan and Iraq due to terrorist groups funded largely by the heroin trade.
I could go on, but you're an idiot if you think the current US policy toward narcotics doesn't cause starving, dying and suffering.
People who think caring about drug policy is for bong-toting fratboys sicken me.
I notice in myself that music can distract or enhance concentration depending on music style
False. The empirical studies indicate that many people believe something along these lines (or more generally, "I work better when I'm multitasking"), but it's just not true. Although it *is* true that many people are addicted to being interrupted all the time and crave it in the form of multitasking. It takes time to get used to concentrating on a single task, but once you do, you're noticeably more efficient.
If the submitter is using music to isolate him/herself from the talking, they find a better means. In my experience neither ear muffs nor earplugs alone are good enough to drown out talking, but both in combination work. Noise-cancelling headphones will not work
I wasn't attacking the fMRI community, I was just replying to the portion of your original post that I quoted.
I don't know what standard practice in fMRI research is as far as bonferonni corrections so I can't comment on that. I wasn't defending the claim that such voxels don't exist.
Technically, all comparisons are significantly different from 0, so the probability that voxels passing any arbitrary criterion is 100% given sufficiently precise measuring devices.
If you really do know of a flaw in statistical practice that affects many thousands of studies and many
millions (probably billions) of dollars of grant money, but that has escaped the notice of everyone in the
field, and you haven't taken the time to submit your insight to a decent journal, then you are the worst
kind of bad scientist.
Umm are you not familiar with the hundreds upon hundreds of articles dating from the 1950s to today
explaining why statistical testing is unscientific and harmful to psychology? Here's a list of 402 of them
fMRI research is often
(although clearly not always) conducted by people using the same statistical techniques. Sure the actual
generation of images involves a lot of complicated math, but the analysis does not. In many cases, they simply perform a significance test by voxel to determine whether two images are "different". Here's a PDF of a paper that appeared in Science where they do this: http://www.wjh.harvard.edu/~jgreene/GreeneWJH/Greene-et-al-Science-9-01.pdf
What is the response to the criticisms? These points are ignored. The Association for Psychological Science (APS), one
of the two main psychology associations recently defined a replication as anything that obtains a non-zero
effect in the same direction. Why? Because it allows psychologists to make a simple transformation of the
p-value without actually changing standard operating practice.
You greatly overestimate the extent to which psychologists care how much they are wasting grant money
conducting bad research.
One of the authors of the cognitive dissonance paper has told me more than once that any study that requires math more complicated than a t-test is too complicated.
I've also been told by several graduate students at Yale that their research doesn't require statistics. I wouldn't say that all psychologists don't know math, but I will say that Yale's psychology department is pretty anti-math.
It's not much older than our planet, which is only about 6000 years old. Also, Jesus created the dinosaurs to test our faith. He also red-shifted a bunch of light to test our faith. Praise be.
I'm not going to say that Internets will replace academic publishing houses, but they are in somewhat of a precarious position. I imagine they are facing pressure to be open more than would publishers in other fields, since openness can rightly be claimed as essential to the scientific process. And the Internets were at least partially created to publish scientific research in a useful and organized way, and people seem to be using it for that (albeit by posting pdfs rather than writing articles in HTML). So it's not clear to me what these publishing houses do (other of course than conferring prestige on authors) that is essential to the scientific process these days.
I imagine the publishing houses make a lot of their money from licensing access to their online databases of articles, which tend to have terrible user interfaces and be disconnected from each other (e.g. each database has a fixed number of journals that it serves). But Google scholar is much more pleasant to use than these databases so many scholars may not even be aware of which database Google draws the articles from. If enough people routinely post PDFs of their articles online, Google may preferentially link to these (since you don't have to have a subscription to read those versions) which might further dig into profits for the publishers.
Slashdot Mirror
I have tried several solutions and find org-mode to be the most useful one. I have it set up in essentially a GTD structure, with TODO items in one big list, a separate list of active projects, and a third list of potential future projects.
org-mode is extremely configurable, which is a definite plus for software you intend to organize your life. I recommend the following add-ons as well:
Now you just take text notes in emacs and org-mode does the rest. I happen to write LaTeX and some code, and org-mode also supports literal programming. This is an extremely useful feature since I can write rich outlines of all my projects in org-mode and have the TODO items be placed naturally where they would have gone in my hand-written notes.
For a simple example of how this might work, suppose I'm drafting a research paper and don't remember some detail I don't care about currently. I might write, "Smith first published this theory in the 80s * TODO Look up the year for Smith". org-mode will find these todo items and tell me about them when I look at my todo list. Then when I find the appropriate date and want to insert it, it will take me directly to where I put the TODO item so I can insert it into the draft with minimal effort. In this way, outlines naturally progress toward finished projects, just as GTD wants them to.
To prove that all they should have to do is put twenty programmers on the stand and if ten of them can come up with that solution.
Why is that the standard of obviousness? When it comes to algorithm design, algorithms could be manifestly obvious to mathematicians and algorithm specialists but not to ordinary programmers. The obvious criterion should take into account the corporate structure, or else it will be profitable to keep a stable of uninformed programmers in the dark while they implement algorithms designed by specialists with a higher level of technical knowledge. Then any algorithm that is obvious to someone with technical knowledge will still be patentable because it's implemented by less informed programmers.
Because smart people, who think stuff up, ought to be able to get paid for their ideas.
Right, and they're not because somebody filed an overly broad patent that covered the obvious portions of their clever ideas. So instead of being paid for their ideas and work, they are paying extortionary licensing fees for work they never used when coming up with their ideas.
Patents don't work for massively distributed fields like software development. The solutions are often obvious next steps that are waiting only for hardware advances or other external events. Solutions and algorithms are often simultaneously discovered by multiple parties independently. Patents are not needed to spur or reward innovation in software and processes, but they are useful as ways of forcing competitors to refrain from implementing obvious improvements.
That is why the rules for patents used by the Supreme Court disallow software patents, even though the lower courts have allowed them in certain circumstances thereby overriding the rules used by the SC (until the SC decides to hear a case that challenges the lower court rulings).
I haven't looked carefully at "abstract idea" and how that applies to patents (or doesn't).
Hi, thanks for the response.
I'm a mathematician and I honestly find a lot of the "all algorithms are math" argument to be rather weak for reasons I'd be happy to go into. I think "abstract idea" is the real issue. The best analysis of the situation that I've seen so far is Ben Klemens' "The Rise of the Information Processing Patent", the pdf of which can be found here http://www.bu.edu/law/central/jd/organizations/journals/scitech/volume141/documents/Klemens.pdf.
That's right. Your gene sequencing computer program should not be patentable because it's an abstract idea. That's independent of whether you implemented that abstract idea on a computer.
"instead of a more accurate" should be "or at least a more accurate."
Anyone who is actually voting for wikileaks will likely be well informed and voting below the line anyways.
For those not familiar with australian voting, we have preferential instant runoff first past the pole voting.
You can either vote "above the line," where you select ONE party, and that party decides how your preferences fall if they don't win a seat, or you can vote "below the line," where you number individual candidates "1, 2, 3.....".
So, effectively this ensures that anybody voting for Wikileaks is voting their preferences. Since the above the line preferences are unpopular, voters can't just vote on the party line. Or else they'll possibly get white supremacists in office. Instead they have to rank the candidates *as they'd actually like to see them governing* instead of a more accurate approximation than otherwise. Even though they want votes, they can't rationally want votes from people who just follow a party line, since that creates an environment where party politics thrive and transparency is reduced.
That's just an academic point though, I have no idea whether it is relevant to their thought process.
I think you shouldn't be immune to the observation that in calling the parent's argument "based on emotion" and "deliberate propaganda" you occuse your opponent of "promoting ... dictatorship and slavery".
Let's take a step back here. We are all geeks and we can approach governance from an engineering point of view. If I read you correctly, you're advocating a somewhat extreme social contract viewpoint in which the only legitimate function of government is to take us out of the state of nature by granting the government as minimal a monopoly on violence and power as possible. This view isn't unreasonable on its face, but there is more to it.
In any society, including humans, there is a distribution of power and resources that comes from natural survival of the fittest. The basic idea of government is to voluntarily abdicate much of this power to a central authority (where central can mean tribal, state, federal, world, etc). This creates the natural problem of a single point of attack: any group that can infiltrate the government will be able to use that centralized power for personal gain. This creates the need to impose laws on things like (1) who can serve in government (probably not felons, e.g.), (2) what limits members of government can have in their personal influence (e.g. can anybody unilaterally declare war?), (3) limitations on the influence other powerful social members have on the government (e.g. campaign funds, revolving doors) and many others. The point here is that it's a question of engineering. How do we organize the central authority so that the people have maximal freedom to do anything they want and they don't have to fear being pushed around by anyone.
Where you and your opponent disagree most is in determing the strongest threat to government infiltration. The social contract tradition, of which you seem to subscribe, was concerned primarily about landed aristocracies using the power of government to do anything they like. This was historically opposed by a class of business owners and farmers who favored a representative democracy. But in the course of history many things have changed. America has no landed aristocrats, so we no longer worry about that. Instead, it has a class of people who believe that selfishness is a positive virtue that is much more important than telling the truth or helping others. This belief flourishes in America's political structure because it is easy for such people to lie in order to gain votes but then to ignore the voters once in office. The parent believes that these people are using government to their advantage to the detriment of the vast majority of the public.
From this point of view, the "free market" rhetoric is a device to convince the population that there is a moral imperative to turn every part of society into a money-making enterprise. Naturally, if you do this you end up with something very much like what you see in, say, Boardwalk Empire or any other popular depiction of organized crime. In fact, it should be rather obvious that organized crime is only crime because there are laws that prevent business men from engaging in certain profitable businesses, such as human trafficking, drug exportation, war profiteering, and so forth. It raises the question of why any of these things is illegal in the first place.
And, quite obviously, the reason they are illegal is that in a truly free market, labor has the right to organize and the poor have the right to protest and vote. So in a free market, a government will voluntarily implement policies that limit the ways that businesses can exploit others for profit. The only way to get an "ideal" free market of the sort Milton Friedman advocates is to sneak pro-corporate laws onto the books or into the courts since naturally those laws will be opposed by anybody not directly benefited by them.
The mistake you make is that you don't see a free market as a process, you see it as a set of laws. But historically you never get that set of law
Thanks. I'm the one who posted the original question, and your answer does the best job of the ones I've read so far in allaying my concerns.
Hi, I posted the question. Maybe it's worth pointing out that I've used Linux and open source exclusively for well over a decade and have no interest in smearing anybody. I'm positive they have backdoors to closed source programs, because it has been leaked that they have access to MS exploits before they're fixed, and one of the Snowden slides implied the UK has the ability to break BlackBerry encryption from devices owned by heads of state and diplomats. I assume open source is the *safer* option, but I want to know how safe.
That said, the link you posted to *confirms* that US intelligence has tried to put back doors in encryption libraries! That's really all the information we need. My understanding is that the NSA is far more advanced in cryptography than the FBI. It seems almost negligence for the head of the NSA not to attempt to put back doors in open encryption libraries. Plus they've had 13 years since the FBI attempt to learn from their mistakes. So if we haven't heard of the NSA doing it, it's reasonable to wonder if that's because they're doing it extremely well.
As stated in the submission, although NSA hardened the algorithm to most attacks, they lobbied to reduce the key length. Specifically they wanted 48 bit keys instead of 64 bit. Perhaps there is a good reason for this, but on the face of it, weaker keys would seem to weaken the algorithm to brute force attacks. It may have just been that at the time computing power was the best advantage NSA had.
Hi, I wrote the submission. To fearmonger is to exagerrate some threat to use fear in order to promote some specific ends. This question is me asking to what extent caution should be justified so that I as a user can know what to do. I'm sure you can see how those things are extremely different and in fact the opposite. One is an attempt to drive action with fear you know is unjustified, the other is an attempt to systematically determine the appropriate amount of caution.
"This is fearmongering" seems inappropriate as a response to a submission that contains only links to Wikipedia documenting known facts and that even goes so far as to call some proponents of this theory paranoid.
That said, thanks for the link.
Because many email providers -- such as gmail, hotmail, yahoo, etc -- want to read your email to serve you ads. Encryption runs counter to the profit motive.
I'm just a little confused about your wording. I'm not sure what it means for an average to be better or for betterness to apply. I can see at least two reasons to want to know about standard deviation. One is that you want to know about how ISP speeds differ according to other variables, like neighborhood. Another is to see whether the difference in mean ISP speeds is "significant."
If you're a frequentist, then the probability that ISP A and ISP B have different speeds (so that one is larger than the other) is 1. So sigma here is 0. P(A = B) = P(A- B = 0) corresponds to doing an integral along the rectangle [0,0], which must be 0. (More generally, it's a measure 0 set). But people report sample standard deviation estimates (which are not sigma, but estimates of sigma) to show how precise the measurement is. In this case, since there are so many data points, I would imagine that the sample standard deviations aren't that relevant.
If you're more curious about what the average individual might see in terms of speed, then you probably want more info about speed broken up by, say, location. And that is more about giving conditional distributions than about standard deviation.
The content of what you're saying is clear, but it's less clear how sigma will tell you about neighborhood-by-neighborhood variation. What they're reporting is a sample statistic, and sample variance estimates approach zero with large sample sizes. So their observed standard deviation is probably extremely tiny. If you subscribe to the view that there is a true mean out there to be measured, then sigma -- the true standard deviation -- is literally zero.
Data about variation would be useful, but that would mean publishing more than just standard deviation data. It basically means publishing the whole data set, or computing means conditioned on neigborhood.
Your points are all valid, but the problem isn't as bad as it seems. If the noise in the errors are uncorrelated (e.g. all cars throw off readings, but in a random way), then the noise disappears in the average (by the Mean Value Theorem). Otherwise, you just model the correlation and account for it (e.g. with Bayesian methods). More generally, there are numerous algorithms specifically designed to extract useful aggregate data from a large number of networked noisy/low quality sensors.
Statistical significance is a binary phenomenon, and there is no such thing as "high statistical significance."
That's like claiming that a result is "extremely true" because the contradiction you get by assuming it is "absolutely crazy".
Significance is not and never has been an in indicator of the probability of a theory, It's only an indication that you've passed the an extremely low threshold to make a claim (i.e. your claim is not plima facie absurd).
Yeah, like the many people dead or wounded due to gang violence fueled by the street drug trade, or the many people addicted to drugs who can't get medical or treatment help because they will get arrested or simply ignored, the people dying in Afghanistan and Iraq due to terrorist groups funded largely by the heroin trade.
I could go on, but you're an idiot if you think the current US policy toward narcotics doesn't cause starving, dying and suffering.
People who think caring about drug policy is for bong-toting fratboys sicken me.
I notice in myself that music can distract or enhance concentration depending on music style
False. The empirical studies indicate that many people believe something along these lines (or more generally, "I work better when I'm multitasking"), but it's just not true. Although it *is* true that many people are addicted to being interrupted all the time and crave it in the form of multitasking. It takes time to get used to concentrating on a single task, but once you do, you're noticeably more efficient.
If the submitter is using music to isolate him/herself from the talking, they find a better means. In my experience neither ear muffs nor earplugs alone are good enough to drown out talking, but both in combination work. Noise-cancelling headphones will not work
I wasn't attacking the fMRI community, I was just replying to the portion of your original post that I quoted.
I don't know what standard practice in fMRI research is as far as bonferonni corrections so I can't comment on that. I wasn't defending the claim that such voxels don't exist. Technically, all comparisons are significantly different from 0, so the probability that voxels passing any arbitrary criterion is 100% given sufficiently precise measuring devices.
Umm are you not familiar with the hundreds upon hundreds of articles dating from the 1950s to today explaining why statistical testing is unscientific and harmful to psychology? Here's a list of 402 of them fMRI research is often (although clearly not always) conducted by people using the same statistical techniques. Sure the actual generation of images involves a lot of complicated math, but the analysis does not. In many cases, they simply perform a significance test by voxel to determine whether two images are "different". Here's a PDF of a paper that appeared in Science where they do this: http://www.wjh.harvard.edu/~jgreene/GreeneWJH/Greene-et-al-Science-9-01.pdf
What is the response to the criticisms? These points are ignored. The Association for Psychological Science (APS), one of the two main psychology associations recently defined a replication as anything that obtains a non-zero effect in the same direction. Why? Because it allows psychologists to make a simple transformation of the p-value without actually changing standard operating practice.
You greatly overestimate the extent to which psychologists care how much they are wasting grant money conducting bad research.
One of the authors of the cognitive dissonance paper has told me more than once that any study that requires math more complicated than a t-test is too complicated.
I've also been told by several graduate students at Yale that their research doesn't require statistics. I wouldn't say that all psychologists don't know math, but I will say that Yale's psychology department is pretty anti-math.
Errr, I meant younger. Jesus also occasionally switches the meanings of words to their opposites.
It's not much older than our planet, which is only about 6000 years old. Also, Jesus created the dinosaurs to test our faith. He also red-shifted a bunch of light to test our faith. Praise be.
I'm not going to say that Internets will replace academic publishing houses, but they are in somewhat of a precarious position. I imagine they are facing pressure to be open more than would publishers in other fields, since openness can rightly be claimed as essential to the scientific process. And the Internets were at least partially created to publish scientific research in a useful and organized way, and people seem to be using it for that (albeit by posting pdfs rather than writing articles in HTML). So it's not clear to me what these publishing houses do (other of course than conferring prestige on authors) that is essential to the scientific process these days.
I imagine the publishing houses make a lot of their money from licensing access to their online databases of articles, which tend to have terrible user interfaces and be disconnected from each other (e.g. each database has a fixed number of journals that it serves). But Google scholar is much more pleasant to use than these databases so many scholars may not even be aware of which database Google draws the articles from. If enough people routinely post PDFs of their articles online, Google may preferentially link to these (since you don't have to have a subscription to read those versions) which might further dig into profits for the publishers.
Also free pre-print servers like arxiv.org seem to doing just fine in terms of fueling scientific progress. In some cases, posting to arxiv.org is done instead of publishing in a paper journal. There are even some questions about how important peer-review as practiced by journals really is. E.g.: http://medicine.plosjournals.org/perlserv/?request=get-document&doi=10.1371%2Fjournal.pmed.0040058&ct=1