With consistency a great soul has simply nothing to do. He may as well concern himself with his shadow on the wall. Out upon your guarded lips! Sew them up with packthread, do. Else if you would be a man speak what you think to-day in words as hard as cannon balls, and to-morrow speak what to-morrow thinks in hard words again, though it contradict every thing you said to-day. Ah, then, exclaim the aged ladies, you shall be sure to be misunderstood! Misunderstood! It is a right fool's word. Is it so bad then to be misunderstood? Pythagoras was misunderstood, and Socrates, and Jesus, and Luther, and Copernicus, and Galileo, and Newton, and every pure and wise spirit that ever took flesh. To be great is to be misunderstood.
No - one of the points is to make yourself feel better.
If I can demonstrate how completely clueless and beneath me the object of my affection is, that fulfils the purpose. Why just call someone a moron when you can demonstrate it as well? This works even better with an audience. If your inamorata has to stop and think to work out whether you were insulting them or not, even better - once again, they look like morons.
The cursing has an effect - it makes me feel better - why should I care what the clueless moron I'm insulting feels?
That's much of the point of British cursing. It sounds nicer, but really it's worse. I'll give you a hint, buggering is generally considered worse than fucking. It is pretty funny to insult someone and have them smile in ignorance or, better yet, laugh - the joke is usually on them.
We have to change our passwords every month and can't recycle the past 17.
Naturally, everyone uses [simple password][number] and just increments the number on each iteration. Now those are some secure passwords.
I got so annoyed with it recently that I just keep the same password all the time and once every month I spend some time rotating through 16 junk passwords until I get back to my password.
Forcing users to change passwords frequently inherently forces insecure passwords on people. People can't remember that many essentially random passwords so people don't try - they subvert the system. I wonder how many IT managers who set these policies actually realise what they are doing - you'd have to think not many. Is it that they don't follow their own password policy?
If you are using Excel for any kind of statistics or curve fitting you are not doing serious statistics.
You may be a hardcore spreadsheet user but in the world of statistics Excel is about as racy as the Christian Women's Guild June pinup of Mary-Lou Bonita in a provocative dress that exposes both her shins and her forearms. We're not even close to talking Pirelli calendars here let alone 'hardcore'.
At any rate, I agree with you - definitiely not a killer app.
The main point of the paper is, essentially, that there are lies, damn lies and statistics. Statistics give a false sense of precision.
If researchers only ran one regression and reported the results of that regression, then standard tests of significance would be fairly accurate.
Instead, researchers run lots of regressions, tweaking variables and data to refine their model and get significant and meaningful results. But this then means that the tests of significance are now bogus. As the summary states, if I run 20 regressions I'll probably get one significant one by chance.
A consequence is that a strong test of a model is whether it can survive the addition of genuinely new data. Importantly, this is not the same as the common test of withholding some of the data set that is available to the researcher and then validating the model by showing that it still works with the new data. The problem is that the researcher knows about this data and will obviously only report a model that works on the full data set. The write up will give the impression that the model was robust to the new data - but the whole process is such that it is practically the same as just finding a model that works on the full data set. The 'validation' is just smoke and mirrors.
A consequence of this is that a lot of papers will be falsified by people getting more data that was unavailable to the original researchers. In essence, despite the appearance of objectivity, the data analysis process and interpretation of statistical significance tests are corrupted by actual practice in paper writing.
Survival of the fittest is also non-falsifiable in the same way that (I think) much of ID is. It is, indeed, tautological. By definition, those that survive are called the fittest. (I say I think because I haven't really looked at ID.)
Notwithstanding this, the theory of evolution is mostly about the mechanism and these theories are falsifiable (and indeed have been falsified as you note about the original Darwinian or Lamarckian conception).
Just because the information can be discovered by anyone skilled and motivated enough, doesn't make it alright to collate and publicly broadcast it. How would you feel if someone did the same to you?
This is precisely what Google does to everyone, everyday. The article demonstrated this in the most apropos way possible by showing what Google had already done with respect to their own CEOs information. CNet isn't the one that collated this information - Google did!
I see your point but my understanding of the gist is that Mann et al have been evasive in the past in providing the data necessary to duplicate the results. Barton asks for evidence of how they have responded to such requests. I think that is relevant - if the only way to get the information out of the scientists is by a senator asking for it, that is telling. If, however, they have supplied full information to all who requested it (which I believe is part of the social contract that scientists publishing their results enter into) then that is also important.
I agree with the point of your examples. Barton's questions are rather aggressive so I would argue against the execution of the requests, but I think the motivation is valid and relevant.
Request 5 has to do with disclosure of the methods used to obtain the results. There has been an ongoing dispute about access to the data; it has been claimed that the publicly archived data is insufficient to allow replication of their results. For example, their initial description of the data series in Nature was inaccurate and they had to issue a corrigendum to correct the problem. This was ultimately prompted by great evasiveness when other researchers who were outside the 'club' attempted to obtain the underlying data to perform their own replication. When the outsiders (McIntyre and McKitrick) eventually obtained sufficient information to get close to replicating the results it became clear that the description of the series used in Nature was wrong. The outsiders then had to issue a materials complaint to Nature in order to get the correct information provided.
So 5 seems to have to do with an impression of evasiveness and incompleteness with regards to providing sufficient information to allow replication of the results.
Speaking personally, I have published in a different field and am generally happy when people contact me for information on the underlying data or methods - it means they are reading the article! I usually have a package of files prepared when the paper is finished. Responding to such requests is then a simple matter of sending the package off to the person who requests it. I think that scientists should spend as much time as necessary responding to such requests because that is one of the foundations of science - it ensures that others can build on the results you have achieved which is fundamentally the point of publication in the first place.
I would advise you to read the actual letters. They don't request personal financial information. They request information on the funding sources for his research and information on disclosure obligations that result from those funding sources.
Simply stated, the newspaper article and the Slashdot summary are wrong. But since when has this been a surprise to anyone?
You simply can not believe everything you read in a paper. The article summary is simply wrong. No personal financial information was requested. You can verify this for yourself if you go and read the actual letters at this link.
You will see that what was requested was:
2. List all financial support you have received related to your research, including, but not limited to, all private, state, and federal assistance, grants, contracts (including subgrants or subcontracts), or other financial awards or honoraria.
3. Regarding all such work involving federal grants or funding support under which you were a recipient of funding or principal investigator, provide all agreements relating to those underlying grants or funding, including, but not limited to, any provisions, adjustments, or exceptions made in the agreements relating to the dissemination and sharing of research results.
That is not personal financial information - that is information that bears directly on his disclosure responsibilities. NSF grants require disclosure of the resultant products (data and algorithms). Asking about funding serves to establish what disclosure obligations result.
the #1 rule with GUI design is to minimize mouse travel
You don't know much about GUI design do you?
I suppose people who believe that decided to put the quit box right next to the maximise box on all windows and default the mouse to be hyperactive so that they could minimise mouse travel. Problem is - move the mouse by a pixel or two and you do the exact opposite of what you wanted to do (maximise instead of quit or quit instead of maximise).
I have also struggled with programs that are so poorly designed that the only way you can do something is by right clicking to bring up a contextual menu in some specified place. There are absolutely no clues given to the user that this is the way to do whatever it is you want to do. They don't put the option in the obvious menu (or even inobvious sub-menu) and they don't put a button on a toolbar to tell you this is an option available to you. No, they hide it in a contextual right-click menu that only appears when you happen to have your mouse over some random icon. That is just bad GUI design and it is caused by the designers going hog wild on your #1 principle of GUI design. I hope I never have to use a program that you design.
You might want to clarify that statement - because as it stands it's kind of meaningless. It's almost akin to saying I can do nothing at the speed of light. In fact, I did nothing 50,000 times between that last sentence and this sentence, that is blazingly fast.
At a slightly less ridiculous level, whose fault is it that just installing (but not using?!) software makes the OS performance molasses like? Not being a Windows guy I don't get it. Does this affect OS X as well? I load and open all sorts of software, have multiple users logged in and, unless the process is actually doing something, I don't notice it in OS X - am I missing something?
I feel like exploring the nuances so I'll make some response.
Central Limit Theorem's mean that it doesn't really matter what the underlying distribution is, the mean will asymptotically have a normal distribution. (Couldn't be bothered worrying about t-distributions.)
As for the statements, I think it becomes one more of semantics and reducing ambiguity than anything else. The true value is not a random variable, but one could interpret a statment such as "there is a 95% chance that the true value lies within the interval" as implying that is was one. A more correct statement is that "there is a 95% chance the random interval includes the true value". Mathematically they are equivalent (becaue mathematics does not allow confusion about what is a parameter and what is a random variable) and, provided the context is clear, very few people would interpret the parameter as a random variable.
Thus, I would argue that the statements are equivalent provided one does not interpret the true value as a random variable. The first statement permits an ambiguous interpretation that is not equivalent, but otherwise they are equivalent.
It's a whole lot messier than that. There is not enough information to make any meaningful conclusions like that.
Notwithstanding that, and as you note, in a perfect world 1 in 20 studies using normal statistical standards would be 'wrong'. However, if the confidence interval is, say, 5-6 and all you care about is that the thing is positive it may not make much difference that it is 'wrong'.
No - the standard way to report statistics is with a 95% confidence interval (i.e. there is a 95% chance that the true value lies within the interval quoted). It's just a norm that has nothing to do with overblown results (at least not directly) and nothing to do with the study.
It's marketing - no more, no less. By creating anticipation and supply restrictions they are seeking to drive more sales and make more money. They probably get evening news coverage of the midnight launch and more people will buy the book to see what all the fuss is about.
Ahh! The 'music industry' is a nebulous beast. (Just ask Apple Corps and Apple Computer about whether they are in the music industry and let the fireworks begin.)
I could give them to my girlfriend, but she works for the music industry and gets her music for free
No wonder the music industry has no clue about why on-line music is so popular and so many people look for cheaper alternatives than buying a $30 CD (Australia).
Being a Peeping Tom is also illegal in a large number of places for, I would have thought, obvious reasons. Given that this entire discussion is about bad analogies, why would you argue this is different than a Peeping Tom looking at your TV (or worse) from the laneway behind your house?
Given all the bad analogies floating around on this thread (the Microsoft one in the article has to take the cake though) I thought I'd add my own.
What this guy did was like walking into your house when you leave the door unlocked and drinking the milk in your fridge. That is both illegal and creepy.
We do not yet live in a society where failure to lock your doors is the same as an invitation for everyone to come in and do whatever they will. The attitude that just because someone doesn't secure their Wi-Fi network it is an open invitation to wardrivers to use your network is fundamentally mixed up.
Slashdot Mirror
With consistency a great soul has simply nothing to do. He may as well concern himself with his shadow on the wall. Out upon your guarded lips! Sew them up with packthread, do. Else if you would be a man speak what you think to-day in words as hard as cannon balls, and to-morrow speak what to-morrow thinks in hard words again, though it contradict every thing you said to-day. Ah, then, exclaim the aged ladies, you shall be sure to be misunderstood! Misunderstood! It is a right fool's word. Is it so bad then to be misunderstood? Pythagoras was misunderstood, and Socrates, and Jesus, and Luther, and Copernicus, and Galileo, and Newton, and every pure and wise spirit that ever took flesh. To be great is to be misunderstood.
No - one of the points is to make yourself feel better.
If I can demonstrate how completely clueless and beneath me the object of my affection is, that fulfils the purpose. Why just call someone a moron when you can demonstrate it as well? This works even better with an audience. If your inamorata has to stop and think to work out whether you were insulting them or not, even better - once again, they look like morons.
The cursing has an effect - it makes me feel better - why should I care what the clueless moron I'm insulting feels?
bugger off sounds a whole lot nicer than fuck off
That's much of the point of British cursing. It sounds nicer, but really it's worse. I'll give you a hint, buggering is generally considered worse than fucking. It is pretty funny to insult someone and have them smile in ignorance or, better yet, laugh - the joke is usually on them.
We have to change our passwords every month and can't recycle the past 17.
Naturally, everyone uses [simple password][number] and just increments the number on each iteration. Now those are some secure passwords.
I got so annoyed with it recently that I just keep the same password all the time and once every month I spend some time rotating through 16 junk passwords until I get back to my password.
Forcing users to change passwords frequently inherently forces insecure passwords on people. People can't remember that many essentially random passwords so people don't try - they subvert the system. I wonder how many IT managers who set these policies actually realise what they are doing - you'd have to think not many. Is it that they don't follow their own password policy?
My music collection is about 210GB - but I keep it on silvered plastic discs with lots of pretty printing on one side.
I've also got a low quality, highly compressed copy of it sitting on my hard-drive somewhere...
If you are using Excel for any kind of statistics or curve fitting you are not doing serious statistics.
You may be a hardcore spreadsheet user but in the world of statistics Excel is about as racy as the Christian Women's Guild June pinup of Mary-Lou Bonita in a provocative dress that exposes both her shins and her forearms. We're not even close to talking Pirelli calendars here let alone 'hardcore'.
At any rate, I agree with you - definitiely not a killer app.
The main point of the paper is, essentially, that there are lies, damn lies and statistics. Statistics give a false sense of precision.
If researchers only ran one regression and reported the results of that regression, then standard tests of significance would be fairly accurate.
Instead, researchers run lots of regressions, tweaking variables and data to refine their model and get significant and meaningful results. But this then means that the tests of significance are now bogus. As the summary states, if I run 20 regressions I'll probably get one significant one by chance.
A consequence is that a strong test of a model is whether it can survive the addition of genuinely new data. Importantly, this is not the same as the common test of withholding some of the data set that is available to the researcher and then validating the model by showing that it still works with the new data. The problem is that the researcher knows about this data and will obviously only report a model that works on the full data set. The write up will give the impression that the model was robust to the new data - but the whole process is such that it is practically the same as just finding a model that works on the full data set. The 'validation' is just smoke and mirrors.
A consequence of this is that a lot of papers will be falsified by people getting more data that was unavailable to the original researchers. In essence, despite the appearance of objectivity, the data analysis process and interpretation of statistical significance tests are corrupted by actual practice in paper writing.
Survival of the fittest is correct
Survival of the fittest is also non-falsifiable in the same way that (I think) much of ID is. It is, indeed, tautological. By definition, those that survive are called the fittest. (I say I think because I haven't really looked at ID.)
Notwithstanding this, the theory of evolution is mostly about the mechanism and these theories are falsifiable (and indeed have been falsified as you note about the original Darwinian or Lamarckian conception).
Just because the information can be discovered by anyone skilled and motivated enough, doesn't make it alright to collate and publicly broadcast it. How would you feel if someone did the same to you?
This is precisely what Google does to everyone, everyday. The article demonstrated this in the most apropos way possible by showing what Google had already done with respect to their own CEOs information. CNet isn't the one that collated this information - Google did!
I see your point but my understanding of the gist is that Mann et al have been evasive in the past in providing the data necessary to duplicate the results. Barton asks for evidence of how they have responded to such requests. I think that is relevant - if the only way to get the information out of the scientists is by a senator asking for it, that is telling. If, however, they have supplied full information to all who requested it (which I believe is part of the social contract that scientists publishing their results enter into) then that is also important.
I agree with the point of your examples. Barton's questions are rather aggressive so I would argue against the execution of the requests, but I think the motivation is valid and relevant.
Request 5 has to do with disclosure of the methods used to obtain the results. There has been an ongoing dispute about access to the data; it has been claimed that the publicly archived data is insufficient to allow replication of their results. For example, their initial description of the data series in Nature was inaccurate and they had to issue a corrigendum to correct the problem. This was ultimately prompted by great evasiveness when other researchers who were outside the 'club' attempted to obtain the underlying data to perform their own replication. When the outsiders (McIntyre and McKitrick) eventually obtained sufficient information to get close to replicating the results it became clear that the description of the series used in Nature was wrong. The outsiders then had to issue a materials complaint to Nature in order to get the correct information provided.
So 5 seems to have to do with an impression of evasiveness and incompleteness with regards to providing sufficient information to allow replication of the results.
Speaking personally, I have published in a different field and am generally happy when people contact me for information on the underlying data or methods - it means they are reading the article! I usually have a package of files prepared when the paper is finished. Responding to such requests is then a simple matter of sending the package off to the person who requests it. I think that scientists should spend as much time as necessary responding to such requests because that is one of the foundations of science - it ensures that others can build on the results you have achieved which is fundamentally the point of publication in the first place.
If they asked for personal financial information then you might have a point. But they didn't. They asked for information on funding for the research.
Don't believe me? Prefer to believe what you read in the newspaper?
Go and read the letters for yourself.
I would advise you to read the actual letters. They don't request personal financial information. They request information on the funding sources for his research and information on disclosure obligations that result from those funding sources.
Simply stated, the newspaper article and the Slashdot summary are wrong. But since when has this been a surprise to anyone?
You will see that what was requested was:
That is not personal financial information - that is information that bears directly on his disclosure responsibilities. NSF grants require disclosure of the resultant products (data and algorithms). Asking about funding serves to establish what disclosure obligations result.
the #1 rule with GUI design is to minimize mouse travel
You don't know much about GUI design do you?
I suppose people who believe that decided to put the quit box right next to the maximise box on all windows and default the mouse to be hyperactive so that they could minimise mouse travel. Problem is - move the mouse by a pixel or two and you do the exact opposite of what you wanted to do (maximise instead of quit or quit instead of maximise).
I have also struggled with programs that are so poorly designed that the only way you can do something is by right clicking to bring up a contextual menu in some specified place. There are absolutely no clues given to the user that this is the way to do whatever it is you want to do. They don't put the option in the obvious menu (or even inobvious sub-menu) and they don't put a button on a toolbar to tell you this is an option available to you. No, they hide it in a contextual right-click menu that only appears when you happen to have your mouse over some random icon. That is just bad GUI design and it is caused by the designers going hog wild on your #1 principle of GUI design. I hope I never have to use a program that you design.
You might want to clarify that statement - because as it stands it's kind of meaningless. It's almost akin to saying I can do nothing at the speed of light. In fact, I did nothing 50,000 times between that last sentence and this sentence, that is blazingly fast.
At a slightly less ridiculous level, whose fault is it that just installing (but not using?!) software makes the OS performance molasses like? Not being a Windows guy I don't get it. Does this affect OS X as well? I load and open all sorts of software, have multiple users logged in and, unless the process is actually doing something, I don't notice it in OS X - am I missing something?
I feel like exploring the nuances so I'll make some response.
Central Limit Theorem's mean that it doesn't really matter what the underlying distribution is, the mean will asymptotically have a normal distribution. (Couldn't be bothered worrying about t-distributions.)
As for the statements, I think it becomes one more of semantics and reducing ambiguity than anything else. The true value is not a random variable, but one could interpret a statment such as "there is a 95% chance that the true value lies within the interval" as implying that is was one. A more correct statement is that "there is a 95% chance the random interval includes the true value". Mathematically they are equivalent (becaue mathematics does not allow confusion about what is a parameter and what is a random variable) and, provided the context is clear, very few people would interpret the parameter as a random variable.
Thus, I would argue that the statements are equivalent provided one does not interpret the true value as a random variable. The first statement permits an ambiguous interpretation that is not equivalent, but otherwise they are equivalent.
It's a whole lot messier than that. There is not enough information to make any meaningful conclusions like that.
Notwithstanding that, and as you note, in a perfect world 1 in 20 studies using normal statistical standards would be 'wrong'. However, if the confidence interval is, say, 5-6 and all you care about is that the thing is positive it may not make much difference that it is 'wrong'.
No - the standard way to report statistics is with a 95% confidence interval (i.e. there is a 95% chance that the true value lies within the interval quoted). It's just a norm that has nothing to do with overblown results (at least not directly) and nothing to do with the study.
It's marketing - no more, no less. By creating anticipation and supply restrictions they are seeking to drive more sales and make more money. They probably get evening news coverage of the midnight launch and more people will buy the book to see what all the fuss is about.
We Americans just want pure unadulterated violence, no sex, no drugs!
What about rock 'n roll?
Silly question - ever since Britney (maybe even before the Beatles) it's been no sex, no drugs, no rock 'n roll - but lots of violence...
Ahh! The 'music industry' is a nebulous beast. (Just ask Apple Corps and Apple Computer about whether they are in the music industry and let the fireworks begin.)
I could give them to my girlfriend, but she works for the music industry and gets her music for free
No wonder the music industry has no clue about why on-line music is so popular and so many people look for cheaper alternatives than buying a $30 CD (Australia).
Being a Peeping Tom is also illegal in a large number of places for, I would have thought, obvious reasons. Given that this entire discussion is about bad analogies, why would you argue this is different than a Peeping Tom looking at your TV (or worse) from the laneway behind your house?
Given all the bad analogies floating around on this thread (the Microsoft one in the article has to take the cake though) I thought I'd add my own.
What this guy did was like walking into your house when you leave the door unlocked and drinking the milk in your fridge. That is both illegal and creepy.
We do not yet live in a society where failure to lock your doors is the same as an invitation for everyone to come in and do whatever they will. The attitude that just because someone doesn't secure their Wi-Fi network it is an open invitation to wardrivers to use your network is fundamentally mixed up.