"The problem is that the cost is paid by the First World, especially the US, and the citizens of those nations."
Um, no, that's the point. Think of it as redistribution of wealth on a global scale. Yes, it's "socialism" which Americans seem to have some kind of superstitious aversion to, despite engaging in such socialist activities as having police forces, firefighters and public highways.
Bringing the third world up to speed is good for us all, and MOST of us first worlders think it's reasonable to kick in a bit of our wealth to help those less fortunate. Particularly when the cost is jobs most of us don't want to do, and we get cheap stuff in exchange.
"The scandal provoked discussion in the scientific community about the degree of responsibility of coauthors and reviewers of scientific papers. The debate centered on whether peer review, traditionally designed to find errors and determine relevance and originality of papers, should also be required to detect deliberate fraud."
Nothing to do with publish or perish. The guy wasn't even faking things because he was up for review and didn't have any papers - he was faking things to get famous.
If that's the best example you can come up with....
Almost all data is ugly, capricious and vile. Occasionally, with lots of work, you can dress it up to be moderately attractive. If you want to publish all data then start a Journal of Inconclusive Results and Lazy Statistics. The mainstream journals have too much trouble publishing what they get now.
"'Inconclusive' is subjective. It depends on the prior probability you ascribe to the hypothesis."
It is not. Your very next sentence suggests how it can be objectively assessed. If you're too lazy to calculate confidence intervals on your non-significant result and make an argument about why the maximum likely difference is too small to care about then you're definitely too lazy to properly calculate (and justify) prior probabilities.
In light of your post, let me revise the statement of my thesis slightly: "publication bias" as generally described is a bias against publishing papers with inadequate data and/or inadequate statistics.
Yes. Is it right? I don't know. Is it state-sanctioned/conducted industrial espionage? Absolutely.
"I should also note that in the US, it is illegal to bribe foreign governments."
That doesn't have anything at all to do with it. US laws don't apply except in the US. If American companies don't like those laws, or can't compete under them, they can try to have them changed.
I'm familiar with the situation. Most granting agencies, and most universities, at least the ones I interact with, have reasonable limits for minimum productivity. If you're going multiple years without producing any publications you're not contributing to the scientific community and need to reexamine the way you do science. On the flip side, if you're publishing a hundred papers a year, the university needs to take a look at exactly how you're doing it.
Debate in the literature is being killed by people (silly reviewers included) who think that everything should be perfect before it's published. Someone in the Slashdot story about cancer cures today posted that scientists shouldn't publish animal research because the results might not translate to humans. I had a reviewer on my last paper actually say "method should be perfect before it is published" because we mentioned some potential improvements we planned to look at as future work.
I have yet to see evidence that there is any publication bias, at least of the kind that most people talk about. There is an ANALYSIS bias - everybody (thinks) they know how to analyze for positive results, but few few researchers have any clue at all how to actually analyze negative results. When you hear the vast majority of (non-particle physics) researchers talk about "negative results" they're actually talking about inconclusive results - p-values that are not significant, with no discussion of beta, confidence intervals, or minimum significant effects. Inconclusive results shouldn't be published, unless it's to provide required sample size estimates for future studies.
Most researchers' poor stats skills are indeed a problem, but not a scientific method one. Errors due to poor stats will be discovered, in time, by the scientific method, just like actual fraud.
"science shouldn't allow it, but as we know, theory and practice rarely align in practice"
It looks like science isn't allowing it, which isn't really surprising. The fakers are caught out eventually, whether it's being explicitly identified and their papers retracted, or their results disproved.
The problem is non-scientific - we'd like the system to work more efficiently by discouraging the fakery and other dirty tricks in the first place, using means unrelated to the scientific method.
Publish or perish is good. As a scientist you MUST communicate your ideas or you're a failure. What's wrong is the use of simple metrics like paper count or journal "quality." As usual, if you want to properly evaluate someone's worth you need to use your brain, not your calculator.
What ecological damage? You mean the "damage" of there being a lake there? Maybe we should drain all the natural lakes because they're all causing "damage?"
Publishing in scientific journals is part of the scientific process. Communication and sharing your results is critical to science, particularly when the next steps (human trials) will involve way more resources than your little wet lab probably has. You also want everyone you can get examining your work before you go giving experimental drugs to people.
The problem seems to be overeager laymen. I guess we could close all the scientific journals to non-scientists and only announce final, ready to market developments. Personally I prefer the open approach and educating the public, but if you don't, please feel free not to read any scientific publications and avoid any news articles about them.
Very rarely does anyone hold 11x14s in their lap. Those are framed, on the wall, and most people don't get up to the six to twelve inches you might look at a smaller print from. 4x6 - 8x12, maybe in the same ballpark.
If that were the case they'd insist that the best print is the one from the highest resolution source possible and the best TV is the one with the highest resolution possible.
These people realize that after features are smaller than your eye can resolve there's no point in making the resolution any higher, but they completely neglect viewing distance. Do they all image we're looking at things from an inch away no matter how big they are? Or do they just not understand the difference between angular resolution and dots per inch?
Here's the key (that everyone who talks about it seems to forget to mention): interference is a Fourier transform.
Interferometry is nothing more than letting two (or more) sources interfere with each other, measuring the interference pattern (i.e. the Fourier spectrum), inverse Fourier transforming, and there's your image.
The tricky bits come when you do things like synthetic aperture radar (a moving receiver instead of two receivers), or VLBI (the interference is virtual, inside a computer).
Now, if you want your mind blown, what is the interference pattern in a double slit experiment?
Go the other way. Lower the pressure. Everyone passes out and you wake up at your destination. Would be nice if they could figure out how to prevent the headache though.
Only because you don't understand Shannon's law. It applies to a channel. Shannon has nothing to say about how many channels you can pack into a given bit of spectrum.
Slashdot Mirror
"Not entirely true. Here's but one web page describing laws that restrict individual and corporate action outside the US:
http://www.bu.edu/globalprograms/global-toolkit/getting-established/us-laws-abroad/ [bu.edu]"
Only for US citizens or others with assets in the US. This doesn't apply to Airbus, the Saudi airline or the Saudi government.
"Also, certain parts of the IRS code apply to US citizens with foreign income, even if they are no longer US residents."
Ditto.
"And, various laws regarding sex with underage minors, even when legal in the foreign country, still apply to US citizens when abroad."
Ditto.
"Not surprisingly, children born to US citizens while abroad are eligible for US citizenship, by US law."
Ditto.
And ditto for the rest of your examples.
If motivation isn't the issue, then what link are you claiming between publish or perish and Schon?
"The problem is that the cost is paid by the First World, especially the US, and the citizens of those nations."
Um, no, that's the point. Think of it as redistribution of wealth on a global scale. Yes, it's "socialism" which Americans seem to have some kind of superstitious aversion to, despite engaging in such socialist activities as having police forces, firefighters and public highways.
Bringing the third world up to speed is good for us all, and MOST of us first worlders think it's reasonable to kick in a bit of our wealth to help those less fortunate. Particularly when the cost is jobs most of us don't want to do, and we get cheap stuff in exchange.
"The scandal provoked discussion in the scientific community about the degree of responsibility of coauthors and reviewers of scientific papers. The debate centered on whether peer review, traditionally designed to find errors and determine relevance and originality of papers, should also be required to detect deliberate fraud."
Nothing to do with publish or perish. The guy wasn't even faking things because he was up for review and didn't have any papers - he was faking things to get famous.
If that's the best example you can come up with....
"All data is beautiful and should be published."
Almost all data is ugly, capricious and vile. Occasionally, with lots of work, you can dress it up to be moderately attractive. If you want to publish all data then start a Journal of Inconclusive Results and Lazy Statistics. The mainstream journals have too much trouble publishing what they get now.
"'Inconclusive' is subjective. It depends on the prior probability you ascribe to the hypothesis."
It is not. Your very next sentence suggests how it can be objectively assessed. If you're too lazy to calculate confidence intervals on your non-significant result and make an argument about why the maximum likely difference is too small to care about then you're definitely too lazy to properly calculate (and justify) prior probabilities.
In light of your post, let me revise the statement of my thesis slightly: "publication bias" as generally described is a bias against publishing papers with inadequate data and/or inadequate statistics.
"Do you call that industrial espionage?"
Yes. Is it right? I don't know. Is it state-sanctioned/conducted industrial espionage? Absolutely.
"I should also note that in the US, it is illegal to bribe foreign governments."
That doesn't have anything at all to do with it. US laws don't apply except in the US. If American companies don't like those laws, or can't compete under them, they can try to have them changed.
I'm familiar with the situation. Most granting agencies, and most universities, at least the ones I interact with, have reasonable limits for minimum productivity. If you're going multiple years without producing any publications you're not contributing to the scientific community and need to reexamine the way you do science. On the flip side, if you're publishing a hundred papers a year, the university needs to take a look at exactly how you're doing it.
Debate in the literature is being killed by people (silly reviewers included) who think that everything should be perfect before it's published. Someone in the Slashdot story about cancer cures today posted that scientists shouldn't publish animal research because the results might not translate to humans. I had a reviewer on my last paper actually say "method should be perfect before it is published" because we mentioned some potential improvements we planned to look at as future work.
I have yet to see evidence that there is any publication bias, at least of the kind that most people talk about. There is an ANALYSIS bias - everybody (thinks) they know how to analyze for positive results, but few few researchers have any clue at all how to actually analyze negative results. When you hear the vast majority of (non-particle physics) researchers talk about "negative results" they're actually talking about inconclusive results - p-values that are not significant, with no discussion of beta, confidence intervals, or minimum significant effects. Inconclusive results shouldn't be published, unless it's to provide required sample size estimates for future studies.
Most researchers' poor stats skills are indeed a problem, but not a scientific method one. Errors due to poor stats will be discovered, in time, by the scientific method, just like actual fraud.
So go do it.
"science shouldn't allow it, but as we know, theory and practice rarely align in practice"
It looks like science isn't allowing it, which isn't really surprising. The fakers are caught out eventually, whether it's being explicitly identified and their papers retracted, or their results disproved.
The problem is non-scientific - we'd like the system to work more efficiently by discouraging the fakery and other dirty tricks in the first place, using means unrelated to the scientific method.
Publish or perish is good. As a scientist you MUST communicate your ideas or you're a failure. What's wrong is the use of simple metrics like paper count or journal "quality." As usual, if you want to properly evaluate someone's worth you need to use your brain, not your calculator.
You read the part (in the SUMMARY) about where they were comparing collectors of equivalent area, right?
The peer reviewers, who are professional scientists working in the field, disagree with you.
What ecological damage? You mean the "damage" of there being a lake there? Maybe we should drain all the natural lakes because they're all causing "damage?"
Ah, an emotional appeal. A sure sign that your argument is flawed.
Publishing in scientific journals is part of the scientific process. Communication and sharing your results is critical to science, particularly when the next steps (human trials) will involve way more resources than your little wet lab probably has. You also want everyone you can get examining your work before you go giving experimental drugs to people.
The problem seems to be overeager laymen. I guess we could close all the scientific journals to non-scientists and only announce final, ready to market developments. Personally I prefer the open approach and educating the public, but if you don't, please feel free not to read any scientific publications and avoid any news articles about them.
Flight/space sims take skill. Most MMOs (and their offline breatheren) are like FarmVille with a good story.
I love Diablo, but after one or two play throughs it's as boring as farming.
Very rarely does anyone hold 11x14s in their lap. Those are framed, on the wall, and most people don't get up to the six to twelve inches you might look at a smaller print from. 4x6 - 8x12, maybe in the same ballpark.
If that were the case they'd insist that the best print is the one from the highest resolution source possible and the best TV is the one with the highest resolution possible.
These people realize that after features are smaller than your eye can resolve there's no point in making the resolution any higher, but they completely neglect viewing distance. Do they all image we're looking at things from an inch away no matter how big they are? Or do they just not understand the difference between angular resolution and dots per inch?
Here's the key (that everyone who talks about it seems to forget to mention): interference is a Fourier transform.
Interferometry is nothing more than letting two (or more) sources interfere with each other, measuring the interference pattern (i.e. the Fourier spectrum), inverse Fourier transforming, and there's your image.
The tricky bits come when you do things like synthetic aperture radar (a moving receiver instead of two receivers), or VLBI (the interference is virtual, inside a computer).
Now, if you want your mind blown, what is the interference pattern in a double slit experiment?
And yet you still managed to get the headline wrong. Maybe it's not such a stupid mistake after all.
"I bring my own food on the plane and it tastes just fine"
So the only reasons left are packaging, preprocessing, etc. So in summary, airline food tastes like crap because it is.
Go the other way. Lower the pressure. Everyone passes out and you wake up at your destination. Would be nice if they could figure out how to prevent the headache though.
Only because you don't understand Shannon's law. It applies to a channel. Shannon has nothing to say about how many channels you can pack into a given bit of spectrum.