Questioning scientific results is part of the Scientific Method. Unfortunately, the current administration doesn't use science to criticize science: it uses the politics of wishmaking.
It's not just SciAm that has observed this creeping Lysenkoism either -- see also the International Herald Tribune, and that bastion of left-leaning reporting, the Washington Post (with the sub-head, "Changes Renew Criticism That the President Puts Politics Ahead of Science").
And by the way, do you consider any and all criticism of the President in an election year invalid by virtue of perceived politicking? Sometimes things are just wrong at any time.
Does it [fair use] also not cover reverse engineering on items you own, eg. examining something to see how it's made?
The doctrine of fair use relates only to copyright infringement, in other words, to the act of copying a protected work.
Examining something to see how it's made doesn't usually entail copying anything (assuming you already have a legal copy), and so fair use simply doesn't apply here. Other legal niceties (e.g. DMCA) may well apply though.
Put another way, choice is an abdication of responsibility on the behalf of the programmer.
When an interface gives you dozens or hundreds of different choices, it's because the programmer (or designer) was lazy. Instead of trying to figure out -- in advance, or by context -- what options would be best for the user, the programmer throws his hands up in the air and says: "YOU figure it out, loser!".
There are SO MANY instances where programs insist on making you make irrelevant, useless choices. Just look at the typical installation program, for example. Like 95% of users, I don't CARE where the program is installed, what the application is named, or what skin I'd like the interface to use. I just want the damn thing installed -- and stop bothering me, dammit!
An interface with fewer options is easier to use, no doubt about it. An interface with fewer, well-selected options also makes a BETTER program.
Read overlawyered.com for a great discussion of this and other cases. (I don't have anything to do with this site, but it's a great read for anyone with an interest in how the legal profession is taking this country down the toilet.)
Here are their conclusions about the case:
To summarize, the problems with the Liebeck case are as follows:
1) a product that, through open and obvious consequences, injures one in 24 million people is not "unreasonably dangerous";
2) the fact that billions of cups of McDonald's coffee are sold should be per se proof that it was serving its coffee at a temperature that consumers desired, rather than "too hot." No one was forced to buy the hotter McDonald's coffee instead of the lukewarm coffee supposedly served elsewhere; if McDonald's coffee was really undesireably "too hot," it would be punished in the marketplace for this flaw. Instead, there is public policy by jury, and the millions of customers who, for whatever reason, prefer McDonald's coffee, are out in the cold;
3) a defendant who is not the proximate cause of an injury should not be held liable for that injury;
4) there is no principled construction of tort law that holds McDonald's liable for failing to prevent injury in the case of a foreseeable coffee spill, but not a clothing manufacturer for failing to prevent injury in the case of a foreseeable coffee spill, and one can agree that the latter scenario is an absurd proposition for liability;
5) a defendant should not be subject to punitive damages because the jury did not understand that "statistically insignificant" is a technical statistical term, and not an insult; and
6) punitive damages were assessed against McDonald's based on their coffee sales, which is a punishment for selling a lot of coffee, rather than because of their behavior.
I can entirely believe the cited case is true, despite your claim that:
More to the point in this case, though, this doc got horrible representation if he or she lost this case.
The problem is, in today's legal practice, negligence or otherwise is not what determines the jury's decision. It is only degree of harm. Juries, especially in certain districts where lawyers choose to have cases heard, are wont to find for the plaintiff even when they know it's not the doctor's fault. Read this article for background on this problem. Check out this example:
The same year, a jury in Sharkey County, where I lived and practiced for eight years, awarded $10 million to the family of a man who had electrocuted himself by touching a pipe to a power line. As the treating physician in that case, as well as a resident of the county, I was interested in knowing what culpability the jury felt the defendant electric company had in the electrocution. One of the jurors told me, "Oh, we didn't think the electrical company did anything wrong, but this way the children will be taken care of."
Poor representation or not, sometimes the jury just can't be persuaded to do the right thing.
Re:Remember Slashdot's History
on
iPod Mini Sells Out
·
· Score: 5, Interesting
Now that's funny. Thanks for posting it!
Here's a great quote from a comment in that article:
Agree with the article poster - Lame. Not only is this a lackluster MP3 unit (which by virtue of being firewire will be limited to Apple Mac owners), but it has virtually no UI wizardry that might define it as an Apple product.
This about a product which has garnered more admiration from its interface than just about anything else.
It is legally acceptable to own a trademark on a color.
I believe Paul Klee trademarked the particular shade of blue used in many of his paintings, calling it Klee Blue(TM). I think I learned this at a gallery sometime, but I can't find any indication of this fact on Google. Maybe the fact itself is copyrighted.:)
I'm no math geek, but the way I read it is that over 10,000 tosses, 51% (5100) would be heads and 49% (4900) tails.
That's only true on average (assuming the article's claim of 51% bias is correct). If you actually sat down and made 10,000 tosses yourself, you might find 5087 or 5122 or some other random number of heads.
But in actual fact, seeing 5100 heads or more out of 10,000 tosses has only about 2.3% chance of occurring when the toss is fair, so that's pretty good evidence in itself.
Looked at another way, just 5082 heads or more from 10,000 is significant evidence (at the 5% level), and there's a 63.6% chance of seeing that happen if the coin tossed is really biased as the article claims. So by that criterion, you're more likely than not to detect bias in 10,000 tosses.
...how on earth did they quantify that "10,000 tosses" number? Methinks they might have pulled that number out of thin air.
Not quite. The key phrase is: "would become aware of such a small bias".
How do you become aware of a bias in a sequence of coin tosses? (Let's assume the coin starts at heads when tossed.) 6 heads in 10 tosses would not be cause for suspicion, but more than 8 heads in 10 would be: you'd have only a 1.07% chance of that happening if the toss were fair, so you'd be more likely to conclude that the toss is biased.
Let's say, arbitrarily, that you'd decide a coin toss was biased if you flipped it 10,000 times and saw a proportion of heads that was really unlikely -- say, less than 1% chance of seeing that many tails or fewer if the toss was fair. Using the example from the article, if you make 10,000 fair tosses and saw less than 4884 tails, that's sufficient evidence for bias by that criterion.
Making a 10,000 biased tosses (with 51% probability of heads) has a about a 37.8% chance of deailing 4884 tails of fewer. So that's what the article means: if you made 10,000 of these biased tosses, you'd have about a 1 in 3 chance of observing a result that was pretty suspicions (by the criterion of seeing something that had only 1 1% chance of occurring if the toss were fair).
Er, you take my comments out of context then proceed to go on a rant about something that I didn't say
Fair comment. I'm a bit sensitive on the issue, where it seems with most of these cases, there is an automatic presumption of liability, even where no negligence is apparent or even proven.
Most of these cases get settled out of court because the company knows that even if they aren't negligent, if someone has been injured, a jury is likely to award compensation. In that case, it's only the lawyers, and perhaps a few "lucky" individuals that win. The rest of us pay, and pay dearly.
If people can't trust that professionals in a field will do the right thing...
But what is right thing? And more importantly, how does one determine this without the benefit of hindsight?
Our health, and the health of our loved ones, is an intensely emotional issue because it's so very personal. If you're treated for an illness by a medical professional, and you don't get better, or even get worse, you're obviously going to be upset. The human reaction is to look for someone to blame for your woes. Many people, with the help of their lawyers, look to the doctor.
It's easy in hindsight to say that if only the doctor had checked for this, or prescribed that, or done this other thing then I'd be fine today. But that's the 20/20 vision of hindsight. In retrospect, the doctor's actions might have been a mistake. But the doctor weighted the evidence she had at the time and chose the action she felt was the most appropriate. A mistake is not negligence, and mistakes, unlike negligence, do not need punishment, whatever the emotion involved.
There's a choice quote from the article I linked to:
In order to have a claim, a defendant must have been negligent, and that negligence must have caused injury to a plaintiff... [but] only the degree of injury, not negligence, predicts how a jury will decide a malpractice case.
It's the emotion of the jury that leads to the degree of injury determining whether a suit on a doctor succeeds or not. Whether there is negligence or not is irrelevant in the practice of law (if not in the letters). Who would want to be a doctor in that kind of environment? And what does that mean for the rest of us seeking medical care?
There's no doubt some negligence exists in the medical industry, and wanton negligence should be punished. But the US legal system has turned medicine into a jackpot lottery enabled by sleazy lawyers. The result? Worsening medical care for us all. Read the essay in that link, and tell me you're not scared.
It seems harsh... that they were denied some compensation
Whatever makes you think that everyone has a right to compensation when "bad things" happen? For there to be compensation, there must be fault. Unfortunately, the American legal system is littered with examples of juries finding blame where none exists, because they think that someone "ought" to be compensated. And we as a society pay, in increased costs, restricted services, and restricted freedom.
Among the many highlights in which the plaintiffs tried to prove their case: lawyers showed the jury a 1976 medical form in which Moore complained to IBM of "allergies, dizziness and blurred vision", and blamed this on "systemic chemical poisoning"; the defense showed the jury the full form, which revealed that this was a reaction to a 1955 tetanus shot and seasonal hay fever.
An IBM witness testified that the trichloroethylene (TCE) that plaintiffs blamed for "system chemical poisoning" was frequently used as an anesthetic for surgery. Plaintiffs tried to turn that around: "Hawes asked Whysner if putting a patient to sleep using TCE would have an effect on the entire system, a systemic effect. `Yes,' Whysner agreed."
Given that this is a discussion of Fahrenheit versus Centigrade your mention of -40 is amusing. I know it doesn't matter, but what type of degrees did you mean?
Actually, in this specific case it doesn't matter. -40 degrees is the same temperature in both the Centigrade and Fahrenheit scales!
You've hit the nail on the head there. Even though the iPod has fewer buttons, it's more usable than other similar devices that have more. Why? Even though some lesser-used functions aren't super-intuitive, the interface does encourage experimentation, and more importantly, doesn't penalize the user for it. You can try out most buttons and the song will keep on playing. Soon enough, you'll discover some of the handier features for yourself. Your anecdote is great vindication of that design.
Other devices solve the problem by adding more buttons, but that just makes experimenting harder. I like the iPod's approach better: use it intuitively when you first pick it up, learn how to use it better with time.
Re:Put more information on your website!
on
KISS
·
· Score: 2, Insightful
How was this modded insightful?
The point of the article is that most gadgets these days are unusable, despite the documentation. And you think this problem can be fixed with more documentation? Sheesh.
The problem is lack of design, not lack of documentation.
Umm, what's quality control got to do with this article? All the testing in the world isn't going to make a poorly designed, feature-packed and misoriented product any easier to use.
I'll give you one word: design.
The quote at the end of the article gets it right: "The simpler it looks," Nielsen said, "the harder it is to build." Great design exudes simplicity, but it's surprising hard to get right. The iPod did a good job, by focusing on making music, and music alone, available through a simple interface. (I despaired to find you could maintain a calendar and play games on an iPod, but who does that? Fortunately these unnecessary features didn't interfere with the design too much.) My DirecTV DVR gets it mostly right too -- I shudder to think of all the things they could have added (partial show recording? a trashcan? games?) and I'm glad they didn't.
On the other hand (and as the article points out) every cellphone I've seen in the last two years has been a failure. The failure is not in QA, and it's not in documentation. It's certainly not in the user. The failure is in design.
... the bottom of the soda can is concave, rather than convex...
Hmm, I always thought the base wasn't convex because if it was rounded out on the bottom it would fall over whenever you attempted to stand it on a flat surface. Not so useful for a container holding liquid really...
Yep. Even if those these studies showed Windows was "better" were valid, how do we know Microsoft didn't commission nine other studies that showed Linux was better for each Windows-positive one?
This problem doesn't just apply to commissioned studies, either. Even in the peer-reviewed academic literature, "publication bias" can be a serious problem. Journals will only publish significant results (or, equivalently, researchers will only submit positive results). For every article which shows drug X is beneficial there could be 10 studies which found there was no effect but were never published, this inflating the apparent benefit of X.
This wouldn't be possible. The glyph is generated in real-time, when you (say) sign up for the email account. You can't even display the correct glyph without beginning the email signup transaction first. So there's always danger the glyph will expire.
Reducing the expiration time of the glyphs would help alleviate the problem, at least. Unless the porn site has massive, constant traffic.
And if you're having a bad day, you can entertain yourself by throwing the little buggers into fire, drowning them, getting them eaten or squished... the possibilities are endless. After a hard day at work it's very cathartic.
Has anyone ever made a list of all the different ways a Pikmin can die? I want to make sure I haven't missed one:)
Well, you were partly right (and I was partly wrong);) -- I didn't look closely enough to realize you were referring not to a T test for a Normal repsonse but to an approximate Z test for independent Bernoulli trials. And 30 subjects is indeed enough for the Z approximation to hold.
But even the test you referred to has its own assumptions, particularly that each subject represents an independent replicate. These two conditions -- independence and replication -- hardly hold in practice. Independence is pretty obvious, but the replicate issue is usually ignored. In your example, it's pretty unlikely that each person in the first group has a probablity of positive response of 0.6 as the model requires. For each person, it's most likely a different value, dependent on unmeasured (latent) factors. So again, the assumptions are not upheld and the P-value is therefore suspect.
I stand by my conclusion that the P-values reported in most small-sample studies in the psychological and social sciences literature are generally too small.
Z statistic = (p1-p2) / (sigma * sqrt(1/n1+1/n2)) =2.99
p-value = 0.0014.
However, that P-value relies on some pretty hefty assumptions (most importantly, Normality), which aren't often upheld outside of physical scientific experiments. For experiments on people you cen be pretty sure they're not. When you get larger numbers in your sample the Central Limit Theorem starts to kick in and you can stop worrying so much about the assumptions, but 30 per sample isn't even in the ballpark.
For small samples like these you'd prefer nonparametric tests. These don't have the underlying assumptions of Normality, but do require a bigger difference between the groups before you can declare significance (the tests are of a much lower power).
You see a lot of bogus conclusions in the psychological and social sciences literature where they've run T-tests or similar parametric tests like this on small samples where the distributional assumptions are unjustified.
Have you ever tried to design a user interface? If so, would you ever have claimed to designed a good one? Truly good user interfaces are few and far between. See Alan Cooper's book for plenty of counterexamples. In particular, see Chapter 9 "Designing for Pleasure" for another good use of a wheel in interface design. (Use Amazon's "Search Inside the Book" feature around page 138.)
But isn't there prior art? The iPod's wheel is for navigating music, unlike a VCR jog wheel or the cellphone examples others have given, and that makes it different for a start. The way the scrolling accelerates to navigate a long list, as another poster pointed out, is also unique. IANAL, but I think there are enough unique features here to warrant a patent.
But the final verdict, for me, comes from this: plenty of other manufacturers have attempted to produce an iPod killer, but by universal consent, none are as easy to use. I think that's enough to prove that the iPod's design is worth of recognition.
Once again, I remind you: all good design is obvious, but only in retrospect.
Slashdot Mirror
It's not just SciAm that has observed this creeping Lysenkoism either -- see also the International Herald Tribune, and that bastion of left-leaning reporting, the Washington Post (with the sub-head, "Changes Renew Criticism That the President Puts Politics Ahead of Science").
And by the way, do you consider any and all criticism of the President in an election year invalid by virtue of perceived politicking? Sometimes things are just wrong at any time.
The doctrine of fair use relates only to copyright infringement, in other words, to the act of copying a protected work.
Examining something to see how it's made doesn't usually entail copying anything (assuming you already have a legal copy), and so fair use simply doesn't apply here. Other legal niceties (e.g. DMCA) may well apply though.
Put another way, choice is an abdication of responsibility on the behalf of the programmer.
When an interface gives you dozens or hundreds of different choices, it's because the programmer (or designer) was lazy. Instead of trying to figure out -- in advance, or by context -- what options would be best for the user, the programmer throws his hands up in the air and says: "YOU figure it out, loser!".
There are SO MANY instances where programs insist on making you make irrelevant, useless choices. Just look at the typical installation program, for example. Like 95% of users, I don't CARE where the program is installed, what the application is named, or what skin I'd like the interface to use. I just want the damn thing installed -- and stop bothering me, dammit!
An interface with fewer options is easier to use, no doubt about it. An interface with fewer, well-selected options also makes a BETTER program.
Here are their conclusions about the case:
I can entirely believe the cited case is true, despite your claim that:
The problem is, in today's legal practice, negligence or otherwise is not what determines the jury's decision. It is only degree of harm. Juries, especially in certain districts where lawyers choose to have cases heard, are wont to find for the plaintiff even when they know it's not the doctor's fault. Read this article for background on this problem. Check out this example: Poor representation or not, sometimes the jury just can't be persuaded to do the right thing.Here's a great quote from a comment in that article:
This about a product which has garnered more admiration from its interface than just about anything else.I believe Paul Klee trademarked the particular shade of blue used in many of his paintings, calling it Klee Blue(TM). I think I learned this at a gallery sometime, but I can't find any indication of this fact on Google. Maybe the fact itself is copyrighted. :)
That's only true on average (assuming the article's claim of 51% bias is correct). If you actually sat down and made 10,000 tosses yourself, you might find 5087 or 5122 or some other random number of heads.
But in actual fact, seeing 5100 heads or more out of 10,000 tosses has only about 2.3% chance of occurring when the toss is fair, so that's pretty good evidence in itself.
Looked at another way, just 5082 heads or more from 10,000 is significant evidence (at the 5% level), and there's a 63.6% chance of seeing that happen if the coin tossed is really biased as the article claims. So by that criterion, you're more likely than not to detect bias in 10,000 tosses.
Not quite. The key phrase is: "would become aware of such a small bias".
How do you become aware of a bias in a sequence of coin tosses? (Let's assume the coin starts at heads when tossed.) 6 heads in 10 tosses would not be cause for suspicion, but more than 8 heads in 10 would be: you'd have only a 1.07% chance of that happening if the toss were fair, so you'd be more likely to conclude that the toss is biased.
Let's say, arbitrarily, that you'd decide a coin toss was biased if you flipped it 10,000 times and saw a proportion of heads that was really unlikely -- say, less than 1% chance of seeing that many tails or fewer if the toss was fair. Using the example from the article, if you make 10,000 fair tosses and saw less than 4884 tails, that's sufficient evidence for bias by that criterion.
Making a 10,000 biased tosses (with 51% probability of heads) has a about a 37.8% chance of deailing 4884 tails of fewer. So that's what the article means: if you made 10,000 of these biased tosses, you'd have about a 1 in 3 chance of observing a result that was pretty suspicions (by the criterion of seeing something that had only 1 1% chance of occurring if the toss were fair).
Remember, kids: math is fun!
Fair comment. I'm a bit sensitive on the issue, where it seems with most of these cases, there is an automatic presumption of liability, even where no negligence is apparent or even proven.
Most of these cases get settled out of court because the company knows that even if they aren't negligent, if someone has been injured, a jury is likely to award compensation. In that case, it's only the lawyers, and perhaps a few "lucky" individuals that win. The rest of us pay, and pay dearly.
But what is right thing? And more importantly, how does one determine this without the benefit of hindsight?
Our health, and the health of our loved ones, is an intensely emotional issue because it's so very personal. If you're treated for an illness by a medical professional, and you don't get better, or even get worse, you're obviously going to be upset. The human reaction is to look for someone to blame for your woes. Many people, with the help of their lawyers, look to the doctor.
It's easy in hindsight to say that if only the doctor had checked for this, or prescribed that, or done this other thing then I'd be fine today. But that's the 20/20 vision of hindsight. In retrospect, the doctor's actions might have been a mistake. But the doctor weighted the evidence she had at the time and chose the action she felt was the most appropriate. A mistake is not negligence, and mistakes, unlike negligence, do not need punishment, whatever the emotion involved.
There's a choice quote from the article I linked to:
It's the emotion of the jury that leads to the degree of injury determining whether a suit on a doctor succeeds or not. Whether there is negligence or not is irrelevant in the practice of law (if not in the letters). Who would want to be a doctor in that kind of environment? And what does that mean for the rest of us seeking medical care?There's no doubt some negligence exists in the medical industry, and wanton negligence should be punished. But the US legal system has turned medicine into a jackpot lottery enabled by sleazy lawyers. The result? Worsening medical care for us all. Read the essay in that link, and tell me you're not scared.
Whatever makes you think that everyone has a right to compensation when "bad things" happen? For there to be compensation, there must be fault. Unfortunately, the American legal system is littered with examples of juries finding blame where none exists, because they think that someone "ought" to be compensated. And we as a society pay, in increased costs, restricted services, and restricted freedom.
Lawyers are largely to blame for this problem. There are some classic examples from this very case, as highlighted on overlawyered.com:
Actually, in this specific case it doesn't matter. -40 degrees is the same temperature in both the Centigrade and Fahrenheit scales!
Other devices solve the problem by adding more buttons, but that just makes experimenting harder. I like the iPod's approach better: use it intuitively when you first pick it up, learn how to use it better with time.
The point of the article is that most gadgets these days are unusable, despite the documentation. And you think this problem can be fixed with more documentation? Sheesh.
The problem is lack of design, not lack of documentation.
I'll give you one word: design.
The quote at the end of the article gets it right: "The simpler it looks," Nielsen said, "the harder it is to build." Great design exudes simplicity, but it's surprising hard to get right. The iPod did a good job, by focusing on making music, and music alone, available through a simple interface. (I despaired to find you could maintain a calendar and play games on an iPod, but who does that? Fortunately these unnecessary features didn't interfere with the design too much.) My DirecTV DVR gets it mostly right too -- I shudder to think of all the things they could have added (partial show recording? a trashcan? games?) and I'm glad they didn't.
On the other hand (and as the article points out) every cellphone I've seen in the last two years has been a failure. The failure is not in QA, and it's not in documentation. It's certainly not in the user. The failure is in design.
Hmm, I always thought the base wasn't convex because if it was rounded out on the bottom it would fall over whenever you attempted to stand it on a flat surface. Not so useful for a container holding liquid really ...
This problem doesn't just apply to commissioned studies, either. Even in the peer-reviewed academic literature, "publication bias" can be a serious problem. Journals will only publish significant results (or, equivalently, researchers will only submit positive results). For every article which shows drug X is beneficial there could be 10 studies which found there was no effect but were never published, this inflating the apparent benefit of X.
Reducing the expiration time of the glyphs would help alleviate the problem, at least. Unless the porn site has massive, constant traffic.
Google doesn't have to be boycotting SCO for this to happen. A trivially crafted robots.txt created by SCO could easily be the cause.
Has anyone ever made a list of all the different ways a Pikmin can die? I want to make sure I haven't missed one :)
But even the test you referred to has its own assumptions, particularly that each subject represents an independent replicate. These two conditions -- independence and replication -- hardly hold in practice. Independence is pretty obvious, but the replicate issue is usually ignored. In your example, it's pretty unlikely that each person in the first group has a probablity of positive response of 0.6 as the model requires. For each person, it's most likely a different value, dependent on unmeasured (latent) factors. So again, the assumptions are not upheld and the P-value is therefore suspect.
I stand by my conclusion that the P-values reported in most small-sample studies in the psychological and social sciences literature are generally too small.
For small samples like these you'd prefer nonparametric tests. These don't have the underlying assumptions of Normality, but do require a bigger difference between the groups before you can declare significance (the tests are of a much lower power).
You see a lot of bogus conclusions in the psychological and social sciences literature where they've run T-tests or similar parametric tests like this on small samples where the distributional assumptions are unjustified.
But isn't there prior art? The iPod's wheel is for navigating music, unlike a VCR jog wheel or the cellphone examples others have given, and that makes it different for a start. The way the scrolling accelerates to navigate a long list, as another poster pointed out, is also unique. IANAL, but I think there are enough unique features here to warrant a patent.
But the final verdict, for me, comes from this: plenty of other manufacturers have attempted to produce an iPod killer, but by universal consent, none are as easy to use. I think that's enough to prove that the iPod's design is worth of recognition.
Once again, I remind you: all good design is obvious, but only in retrospect.