Did they also forbid to change anything when better ideas come up, because all I hear is how people who have died 200 years ago had a good idea. I hear it in such a way that it sounds as if since then no smart people where born.
To me it looks as if they tried, but failed. Otherwise the US would not be in that situation. Change e.g. the "Winner takes all/first past the post". There are better ways to do it.
It might mean that you need to change a LOT. But if that is what it takes, why not? Those smart people 200 years ago did it.
There is a proper mechanism for this, which is called a 'Constitutional amendment' and has a well-defined process.
Have you considered that we've had a lot of rather motivated smart people who are not necessarily acting in the interests of the people? Many of whom have devoted that intelligence towards goals that are, ultimately, not desirable, such as circumventing or just plain breaking what had been intended as built-in safety measures? It is very, very rare that it is in any sense desirable to have somebody doing that, and it's prudent to be careful and attentive in those rare cases when it is desirable.
Nearly all governments have realized that this is a certainty, and have instituted various measures--which in some cases could be summarized as "Hey, as long as we're the ones in power, who gives a fuck? Yay corruption (as long as it benefits us)!" This has had some nasty, nasty results, such as the ruling class scrambling to find scapegoats when the majority is showing signs of being less than happy with them--"No, no, it's all the fault of the [slur]s!" is a classic. The important part to them is, as always, to make sure that the rioting peasants aren't killing them.
It's not just that, you also see these protests happening without anybody actually checking to make sure what the actual religious beliefs of the locals say about sacred ground--the automatic assumption is pretty much "Build nothing on it anywhere, ever" which is, when you think about it, weird in a racist way given that very few cultures actually have that as their approach to sacred ground...and most of the ones I can think of with that approach are ones which didn't have the ability to do even a temporary devotional structure. (Pooling with a fellow anthro geek gets it to "Some Native American tribes?" but we cannot even be sure there because it may be more that the knowledge of the correct thing to build has been lost...)
It gets really ironic here--Poli'ahu might be actually perfectly fine with a telescope on her mountain, but there ought to be no question if the proposal had been a meteorological, geological, or ecological research station.
But, really, the question of the truth of a religion should be completely irrelevant to the whole issue of being polite about things. The history of science is littered with dead theories that linger on well past when they ought to have been abandoned, even to this day...while many primitive peoples' observations are quite accurate even if couched in their religious terminology. Taboos against building things on a site can easily be because they know it's a bad site--but because of how they understood the universe, they attributed something like 'likely to result in mudslide down the side of the mountain' to 'wrath of deity' instead of 'the soil here just is gonna do that if you build here.' Even with modern tech, some of this is things we only can know in hindsight.
So, does it include the necessary infrastructure commitments to go with banning the internal combustion engine? Stuff like ensuring there's enough EV charging stations? What about sufficient power plants as to keep up with the demand and keep it affordable? Or, hey, how about efforts to ensure access to the necessary infrastructure for having EVs will not be limited to, say, rich majority-white neighborhoods?
Even if it is limited to new-new cars, it's going to fuck the used car market after it's been around for a bit--if you can't do an EV, you're going to have to be able to afford to get something brought in from outside of California or find something older and typically less safe. The commitment to infrastructure improvements is necessary to prevent the proposal from being yet another classist and racist measure with a green coat of paint.
Are you kidding me? On-screen keyboards flip out if you're a fast typist. I can have my physical keyboard sounding like it's a musical instrument with minimal trouble, but I try typing those kinds of speeds on an on-screen keyboard and suddenly letters drop out and autocowreckt joins in by guessing (very wrongly) what letters I 'obviously meant' to have hit. It gets worse if any amount of technical terms turn up.
As far as I can tell, Facebook may be currently convinced I'm a hummingbird. Employers where I live have a tendency towards discriminating against people without Facebook accounts, so...
There's a place in the area which sells angel statues that are utterly perfect if you ever want to stick an angel statue outside a friend's window at night. I'm thinking of getting a picture of one and replacing my profile pic with that, possibly talking some friends into taking some pictures so we can tag 'em as me before we give the statue...
About the only jobs I see this actually displacing is that poor night shift cashier's--and that actually circulates among the staff, at many locations. I'm inclined to expect to see these turn up in the minimart, and possibly even just so you don't lose sales merely because somebody's gotten rid of your cashier...so the whole store's still gotta be open even at 3AM, the cashier just isn't needed for many of the transactions.
That's not an assumption in science--the rule is you write the paper so an undergrad in the field should be able to understand it with some work, and you, the reader are expected to educate your own sorry ass. Expecting readers to be capable of obtaining the background information for themselves is not and should not be a 'horrible assumption,' especially now when the internet makes it eminently easy for somebody to learn a good amount of the background they might need on their own if they care to do so.
Marketing and political activists will believe whatever they want, regardless of the research, so there's really no point in trying to do anything about this beyond encouraging people to be properly skeptical of the claims those two groups make. As for reporters, well, you're not going to improve the quality of science journalism until you start insisting that science journalists know the field they're reporting on sufficiently well to do good work.
The first step in that? Do not click the clickbait.
Actually, the question that seems to matter here is if all the ills of a sedentary lifestyle can be attributed to obesity, which actually is important--and, well, Coca-cola et all will cheerfully sell you low- and no-calorie versions of their products. They're not going to be harmed by people reducing calories. The place where you should be concerned about their involvement in a study would be things like studies on artificial sweeteners or high-fructose corn syrup are actually metabolized, places where their sales are very likely to be harmed if the results aren't positive.
I take your point, but if your recipe called for 1/3 of a tablespoon of cinnamon, I'd call that a dodgy recipe.
I've met a few modern recipes like that--they're ones meant for large-batch cooking. If I'm having to multiply one of those by 3.5? I don't want to have to be thinking about that particular math, I've got other problems that are requiring my attention, which hopefully includes "How to use this borrowed industrial kitchen?" and not "How do I pull this off in a home kitchen?"
A quick search says that, no, minors in the US are legally capable of consenting to a contract: citation, citation, citation, and if you need more I suggest trying Google. The bottom line is that "I am a minor" cannot allow you to void certain types of contracts, it's necessary for the minor to act before they're of age to have the contract nullified, and courts tend to not let you use the "I'm a minor!" card when the contract in question is with a bank.
The problem actually is more one of finding somebody willing to take the risks involved with forming a contract with a minor--as a general rule of thumb, you shouldn't do so unless the law in your state explicitly says that these contracts will be binding anyway, and some laws exist pretty much entirely to keep from fucking minors over by having them unable to form necessary contracts. (For example, car insurance is a contract & some states refuse to issue a driver's license without proof of insurance.) That said, the details will vary from state to state right down to the list of what types of contracts are going to be explicitly non-voidable and details thereof, consult a lawyer experienced with contract law in the jurisdiction(s) in question or avoid the problem entirely.
Most credit card companies seem to go with Option #2.
Also, the whole 'debt ends up on the credit of cosigner' issue is something that happens normally, and most minors have the basic problem of 'no income.' A 16-year-old who is already running their own successful business isn't going to need a cosigner for their credit card unless local law requires it.
It took wandering all over Tokyo for a couple months to find one (possible) alcohol vending machine--I was hiking across Shinjuku and didn't have the time to go check it out, and didn't have a chance to go back later. A tobacco vending machine was located close to where I lived, but I never saw somebody using it and have a sneaking suspicion that nobody actually kept the thing filled anymore & it just hadn't been deemed worth removing from the sidewalk in which it was embedded.
I did find several ice cream vending machines, but I don't think I actually ever managed to find one that did chips or any of the other salty snacks that are rather ubiquitous among US vending machine fare. I certainly had been looking rather hard at the time. (The lemon ice cream was nice, though.)
Oh, and there are some things that can be easily obtained in the US from a vending machine which you cannot get at all from such in Japan.
new terms pretty much are going to have to be found simply to keep from adding to the problem
Until then, doctors who care about things like sickle-cell disease, will have to pick a word and go with it.
No, if they care about things they'll be willing to test everybody and anybody if there's reason to suspect it--sickle-cell is in fact one of the specific reasons why we're having to push doctors to stop being using bad, lazy approximations. About the only thing your obvious ancestry does is modify the odds of you having the gene, though it will remain a non-zero number.
training it to identify somebody's ethnicity and origins would also be distinctly more useful--especially since that's actually something you can do using headshots, as bone structure is more reliable than skin tone and most of the variation can be found in the face.
I don't know about 'distinctly more useful', but sure, that would be a valid research project.
It'd also have much fewer ethical issues and be using concepts with good validity--and, well, humans have an error rate there because they do tend to be lazy assholes and using bad shortcuts. All you'd really need to do is remember that Asians and Native Americans exist, which is not so much of a problem
The minimum set needed to actually make a good operational definition
Who cares?
They weren't setting out to precisely and exactly categorise each person's sexuality, they were setting out to make a comparatively simple, but still meaningful, binary classifier.
Again, all you're really saying is that you can come up with a more precise taxonomy than the gay-or-straight binary model. Well of course you can, but this has no bearing on the basic idea of the research.
If I come up with a classifier that can tell cats from dogs, it would be absurd to try to dismiss the project on the grounds that it can't tell a Labrador from a spaniel.
Okay, I've been trying to be nice and assume you're not a bigot, but I give up here. You're a bigot.
You keep insisting that you can totally get away with saying that you can fucking make a good study by treating a system that includes a set that contains both options and the null set work. No. This is not good. This is a really fucking bad thing to reduce to a binary, and I really don't know what the fucking hell you're doing here on/.
if you cannot grasp sufficiently well what the damned problem is from simply a mathematical perspective.
I couldn't give two fucks what the categorization they used were, aside from the ethical issues other people pointed out. The problem I'm trying to point out where is that how you deem the accuracy of the AI's guess on cases where it's not 0 or 1 but {0,1} is completely arbitrary when you don't allow {0,1} as a response, and that kind of thing is considered abuse when done to a human--and I generally want to avoid doing shit that would be considered abusive if done to a human to an AI.
No, we probably haven't managed to get any of them to the point where it'll actually cause a problem, I just somehow doubt we're going to get that much warning. The time to start good practices is now.
You're only going to have problems when you try lumping the last two into either, unless you deliberately make sure your sample is made up entirely of the first two groups--and even then, you really should note that up front when defining your sample.
Sure, clarity and completeness in a research paper is always desirable. (In true Slashdot fashion, I've not read it.)
I generally make a point of skipping reading research papers where the abstract leaves me with questions about the journal's peer review process. I've limited time, about the only thing of note here is that machine learning research may need to be required to start going through ethics review boards--it usually takes something that just makes everybody rather offended.
I studied precisely the part of science that means I'd know more in detail about those medical contexts than you do, and I know this because I know we're shifting to DNA testing because it's a godsfuckingdamned social construct
Sure, that's a good point: 'race' can mean whatever politically-charged categorisation people want it to mean, and it's important to be clear what we mean by terms like 'white'. The actual biology of the matter (I think we're meant to call it 'ethnicity') is somewhat separate from the identity politics, but it does exist.
(The oft-touted idea that biologically, there's no such thing as race is utter nonsense. Dawkins, among others, has railed against this silliness.)
The concept most people talk about when they say race does not exist biologically--if you want biological race, you need to talk to physical anthropologists. Ethnicity is a more usable concept.
That said, Dawkins is not god, and not really the person I'd want to go with here, anyway. I've been mostly in the 'munch popcorn' section of the argument, since the issue of the biology resembles a primate house which has had a diarrhea outbreak fail to curb their enthusiasm for flinging. The politically-charged and historically bad work on race--check into 'scientific racism' for some of the history here--pretty much mean that if there's something going on biologically, new terms pretty much are going to have to be found simply to keep from adding to the problem.
The question is, is this one of them?
Yes, it is. It's a first-effort research paper, and they're claiming to have achieved very good success with their classifier (ignoring the aforementioned issues with the paper). A 10% nitpick makes for a valid footnote, but doesn't undermine the basic ideas.
Actually, that'd also mean that training it to identify somebody's ethnicity and origins would also be distinctly more useful--especially since that's actually something you can do using headshots, as bone structure is more reliable than skin tone and most of the variation can be found in the face. The gold standard is genetic markers, but these two things are part of how we know the definitions have changed, sometimes in ways to us moderns would be absurd.
(Guess what kinds of things get noticed when you've got graveyards needing digging up when the records tell you who is buried where? There's still archeological work done--you can learn a lot about history, some of which we can't really do otherwise because there's a reason science is careful to be clear about names of things now.)
they needed to choose something wherein a binary classifier would be sufficient--and definitely not one where either option of your binary is equally right (or wrong, depending on how you're scoring it) as an answer.
What are you talking about, 'equally wrong'? The hard numbers of the classifier's effectiveness, are right there in the summary.
The fact that you can come up with a more detailed taxonomy of human sexuality doesn't invalidate the effectiveness of their coarse-grain, first-effort classifier. In the real world, almost nothing falls into neat categories, and no number of highly detailed categories can ever capture all detail of the individual.
Categories are all about making useful generalisations. That's why we use them.
Where does that kind of thinking end, anyway? The classifier fails to distinguish bisexuals from pansexuals, therefore the whole project is incoherent?
The minimum set needed to actually make a good operational definition for 'human sexuality' is to have a four-option system of 'same sex,' 'opposite sex,' 'both sexes,' and 'no.' You can generally get away with lumping everybody who is pretty meh on their partner's sex into
Which doesn't escape from the fact that those orientations are, well, social constructs
You say that as if it invalidates the concept. It doesn't. Nationality, authority, prestige, liberty, justice, progress, and money, are all social constructs, but they certainly matter. Same goes for sexuality.
You can say that race too is 'just' a social construct, but it still matters in certain medical contexts.
More generally: the reason we use imperfect labels (with 'rounding errors' as it were) is that they're useful in practice.
Fun fact: I studied precisely the part of science that means I'd know more in detail about those medical contexts than you do, and I know this because I know we're shifting to DNA testing because it's a godsfuckingdamned social construct--it's not a good shorthand for ancestry, it's got a very high error rate because the definitions change and in fact there's a lot of people here who are old enough that the definitions changed in their lifetimes. In fact, in the past century, there's been at least two major changes in how the definitions were, each of which would change how somebody got counted--and some people lied and forgot to let their kids in on things, and some people will find themselves being counted differently depending on if they're in England, the US, or Australia...
But there is a lag in getting doctors to actually switch off of this bad lazy approximation, which is...pretty much normal.
how you define the categories will throw things off even more than only using Caucasian faces
This sounds a lot like a continuum fallacy. That the classifications are imperfect doesn't mean the whole project is completely invalidated. Subjects' self-reports are easily good enough to serve as a starting-point.
(I can see major issues with the study though, like that they used photos from dating sites, which as others have said likely aren't going to be representative of most photos.)
10% of people who check the 'White' tickbox when asked what their race is aren't actually White, according to current common US definitions of the term. (Yes, all of those specifics are needed.) I suppose a 90% accuracy rate is acceptable for some purposes. The question is, is this one of them?
The odds are pretty good that they did engage in erasure of bisexuals, since that's normal for such research, and I'd be amazed if they even considered the existence of asexuals for this
That wouldn't make these researchers bigots, as you seem to be implying, it would just make them pragmatists. Fewer categories simplifies the classification problem.
There is nothing to apologise for when machine-learning researchers decide to make a binary classifier rather than a multiclass classifier.
The implication exists entirely in your head, and it doesn't actually make them pragmatists. If they were pragmatists wanting to use a binary classifier, they needed to choose something wherein a binary classifier would be sufficient--and definitely not one where either option of your binary is equally right (or wrong, depending on how you're scoring it) as an answer.
I'm not calling them bigots, I'm calling them bad scientists, using bad conceptualizations in their machine-learning research.
A quick check of recent research done using the Kinsey scale supports the ~5% rate--not all bisexuals get counted as straight when people are trying to ignore that bisexuality exists.
I can dig out the citation if you really want, but it's relatively easily found and some of its flaws would require an extensive discussion of...well...all the various problems of using self-report. (Things like how you ask the question can have an impact on the numbers--and yes, sometimes this is set up deliberately because what's being studied are things like how many people will only answer yes when you ask indirectly.) The thing that concerns me about this is that it really sounds like the AI is being trained with the assumption that bisexuality doesn't exist--which means that regardless of which state you consider to be a false positive, it'll be wrong with every single bisexual regardless of if it deems them homosexual or heterosexual.
What is the need to distinguish gay from straight?
It's important if you are looking for a mate.
Which doesn't escape from the fact that those orientations are, well, social constructs--and I'm curious but unwilling to give the article a clickthrough to find out if those were the only two orientations they used, because how you define the categories will throw things off even more than only using Caucasian faces. The odds are pretty good that they did engage in erasure of bisexuals, since that's normal for such research, and I'd be amazed if they even considered the existence of asexuals for this--never mind that asexual=/=aromantic so you definitely should be finding some on dating sites.
I never claimed authority over him, just the right of armchair quarterbacking.
If you're responsible for making a business work and you walk away while it is floundering, that's not really fulfilling your responsibilities. Whether that's fine with the stakeholders or not for some reason doesn't make it optimal.
That's not class envy.
Not just that, but it's something worth knowing about the person--both if you're in a position to give them the job of C*O or in a position where you might be working under them, because a bad chief whatever officer can do a number on a company. The sooner you know that you need to start sending out the resumes,
the better your chances of escape. So, basically, only people who don't have to worry about money don't need to care about these things.
That said, doing something like haring off for Burning Man is perfectly fine if your company is stable--or if you're the main reason it's floundering and you're wanting to make sure everybody knows it.
cayenne8 stated that President Obama did not have the authority under the US Constitution to enact DACA unilaterally. He says that in order to enact DACA, an act of Congress is required.
And, frankly, if it's really got the support people are acting like it does, then there should be no particular issue with getting that act of Congress--especially since that would also provide distinctly more security for those who might use the program.
This is a lot of whining, really, over being asked "Hey, can we do this properly?" It makes as much sense as complaining about somebody saying you should replace the brittle crock in your code with...something that at least can pass as proper code & be maintained.
Apparently the plantiffs were not sent a recall notice and Amazon had insisted that anybody who didn't get one had an authentic pair. Your Point A doesn't apply, and odds are that the plantiff's lawyers already know if the pair of glasses were counterfeit, defective, or misused. Failure to check those things before filing a suit can cost a lawyer their license.
Also, you're completely ignoring my point that the court may decide that Amazon fucked up beyond what a recall can cover: Amazon was known at least a month before it issued the recall as having many many counterfeits to the point that it wasn't worth trying to get a pair of good eclipse glasses from them. If you know, or should have known if you've two brain cells in communication, that you've got defective products going out the warehouse--the degree of protection you can expect to get from a recall is proportional to the speed with which you issue it once you ought to have been aware of the problem, assuming you made an actual effort to detect/prevent the problem quickly & the recall is sufficiently broad. (This second one is why recalls tend to be pretty broad--and yes, it's a bad sign for the company responsible for the recall when it has to be broadened anyway.)
Pretty much nobody in industries where this is a significant liability risk have gone with the Amazon's combination of "LOL what's quality control???" and laziness in issuing a recall once the problem became public knowledge.
None of this liability risk would exist if Amazon had asked for the paperwork from vendors that it did when the recall finally happened back before they would let them sell eclipse glasses, or made a point of not co-mingling products, or recalled all of the styles that they knew had counterfeits mixed in regardless of the vendor precisely because they could not be bothered to track the actual source for any given specific pair. Odds are that their legal department is trying its best to get this fact through the pointy-haired skulls of upper management. (If they can track it back to a specific vendor, they can throw that person under the bus, especially if they can show that they made an effort, no matter how lazy, to prevent counterfeits.)
The problem is more that the standards for getting those degrees have dropped--there was a movement to make them easier to get people people with degrees earned more, without really grasping that this was because the degree represented proof that the person had a general skillset.
I have nothing against working to prevent financial barriers to getting diplomas--but the educational standards should never have been dropped just to improve graduation rates. When you do that, you're not much different from a diploma mill.
It also can very much depend on the person. Anybody whose body is screwing up blood glucose in the opposite direction is going to benefit from fruit juice--in fact, it's pretty much what the treatment of preference is for any form of hypoglycemia that hasn't sent you to the hospital. (Hospitals tend to take the more direct approach to raising blood glucose.)
Much to my annoyance, the usual juice recommended for this purpose is orange juice...and I absolutely loathe it.
I'd agree with you, if Amazon had done a better job here--because they aren't careful to prevent comingling of vendors' stock from happening and didn't go with a strict "Must prove authenticity/quality/suitability to even be listed" policy which would have mitigated problems the problems that would be caused by comingling? Yeah, this isn't okay, and this is a place where they need to have been tighter and will have to be tighter if they're wanting to sell industrial and scientific equipment. It also raises some unfortunate questions on just how well they really can verify that somebody is shipping more than 1% of defective products--that's the sort of thing that requires you have your inventory set up to keep track of who sent what, and if you can keep track of that, then you should be 100% capable of making sure that if I order something from a specific vendor I get that thing from that vendor.
Actually, recalls don't provide automatic or even complete protection--it's in the public interest to ensure that efforts are made to ensure that defective products don't hit the market at all & recalls are done in a timely manner. If you issue the recall after the problem's become known, and especially if you knew when you shipped out the product that it was dangerous? The recall is roughly a tiny bandaid on a sucking chest wound--at best.
With Amazon's known issues in sending you the products provided their warehouses by a specific vendor, the by-vendor recall done well after it became known that Amazon had counterfeit eclipse glasses might not really be enough to protect them from liability--especially since it may well be determined in court that the problem only existed because Amazon failed at due diligence.
Amazon may not be able to shift blame, especially if this is 100% the result of their habit of commingling stock--if the plantiffs bought from a vendor who had been selling authentic eclipse glasses and Amazon shipped them fakes, it's Amazon's fuck up. The recall itself is probably not going to protect them here, precisely because Amazon doesn't practice what everybody else would consider good basic practices on inventory: You don't mix things unless you're absolutely certain they're the exact same thing. With something like eclipse glasses? Having vendors have to jump through hoops to even get listed at all would have been a Good Idea, with extra care given to being able to track back past you any malfunctioning or dysfunctional items back to whomever put them on your shelves.
Basically? Amazon fucked up due diligence, and has ensured that they cannot reliably shift liability back to the supplier.
Some of it, at least, is the result of places which are too small to sell to the usual overstock buyers--those tend to want truckloads and you may only ever have a box or two at a time--who just want to recover a decent % of the cost while having it stop haunting their shelves.
Slashdot Mirror
Did they also forbid to change anything when better ideas come up, because all I hear is how people who have died 200 years ago had a good idea. I hear it in such a way that it sounds as if since then no smart people where born.
To me it looks as if they tried, but failed. Otherwise the US would not be in that situation. Change e.g. the "Winner takes all/first past the post". There are better ways to do it.
It might mean that you need to change a LOT. But if that is what it takes, why not? Those smart people 200 years ago did it.
There is a proper mechanism for this, which is called a 'Constitutional amendment' and has a well-defined process.
Have you considered that we've had a lot of rather motivated smart people who are not necessarily acting in the interests of the people? Many of whom have devoted that intelligence towards goals that are, ultimately, not desirable, such as circumventing or just plain breaking what had been intended as built-in safety measures? It is very, very rare that it is in any sense desirable to have somebody doing that, and it's prudent to be careful and attentive in those rare cases when it is desirable.
Nearly all governments have realized that this is a certainty, and have instituted various measures--which in some cases could be summarized as "Hey, as long as we're the ones in power, who gives a fuck? Yay corruption (as long as it benefits us)!" This has had some nasty, nasty results, such as the ruling class scrambling to find scapegoats when the majority is showing signs of being less than happy with them--"No, no, it's all the fault of the [slur]s!" is a classic. The important part to them is, as always, to make sure that the rioting peasants aren't killing them.
It's not just that, you also see these protests happening without anybody actually checking to make sure what the actual religious beliefs of the locals say about sacred ground--the automatic assumption is pretty much "Build nothing on it anywhere, ever" which is, when you think about it, weird in a racist way given that very few cultures actually have that as their approach to sacred ground...and most of the ones I can think of with that approach are ones which didn't have the ability to do even a temporary devotional structure. (Pooling with a fellow anthro geek gets it to "Some Native American tribes?" but we cannot even be sure there because it may be more that the knowledge of the correct thing to build has been lost...)
It gets really ironic here--Poli'ahu might be actually perfectly fine with a telescope on her mountain, but there ought to be no question if the proposal had been a meteorological, geological, or ecological research station.
But, really, the question of the truth of a religion should be completely irrelevant to the whole issue of being polite about things. The history of science is littered with dead theories that linger on well past when they ought to have been abandoned, even to this day...while many primitive peoples' observations are quite accurate even if couched in their religious terminology. Taboos against building things on a site can easily be because they know it's a bad site--but because of how they understood the universe, they attributed something like 'likely to result in mudslide down the side of the mountain' to 'wrath of deity' instead of 'the soil here just is gonna do that if you build here.' Even with modern tech, some of this is things we only can know in hindsight.
So, does it include the necessary infrastructure commitments to go with banning the internal combustion engine? Stuff like ensuring there's enough EV charging stations? What about sufficient power plants as to keep up with the demand and keep it affordable? Or, hey, how about efforts to ensure access to the necessary infrastructure for having EVs will not be limited to, say, rich majority-white neighborhoods?
Even if it is limited to new-new cars, it's going to fuck the used car market after it's been around for a bit--if you can't do an EV, you're going to have to be able to afford to get something brought in from outside of California or find something older and typically less safe. The commitment to infrastructure improvements is necessary to prevent the proposal from being yet another classist and racist measure with a green coat of paint.
Are you kidding me? On-screen keyboards flip out if you're a fast typist. I can have my physical keyboard sounding like it's a musical instrument with minimal trouble, but I try typing those kinds of speeds on an on-screen keyboard and suddenly letters drop out and autocowreckt joins in by guessing (very wrongly) what letters I 'obviously meant' to have hit. It gets worse if any amount of technical terms turn up.
As far as I can tell, Facebook may be currently convinced I'm a hummingbird. Employers where I live have a tendency towards discriminating against people without Facebook accounts, so...
There's a place in the area which sells angel statues that are utterly perfect if you ever want to stick an angel statue outside a friend's window at night. I'm thinking of getting a picture of one and replacing my profile pic with that, possibly talking some friends into taking some pictures so we can tag 'em as me before we give the statue...
About the only jobs I see this actually displacing is that poor night shift cashier's--and that actually circulates among the staff, at many locations. I'm inclined to expect to see these turn up in the minimart, and possibly even just so you don't lose sales merely because somebody's gotten rid of your cashier...so the whole store's still gotta be open even at 3AM, the cashier just isn't needed for many of the transactions.
That's not an assumption in science--the rule is you write the paper so an undergrad in the field should be able to understand it with some work, and you, the reader are expected to educate your own sorry ass. Expecting readers to be capable of obtaining the background information for themselves is not and should not be a 'horrible assumption,' especially now when the internet makes it eminently easy for somebody to learn a good amount of the background they might need on their own if they care to do so.
Marketing and political activists will believe whatever they want, regardless of the research, so there's really no point in trying to do anything about this beyond encouraging people to be properly skeptical of the claims those two groups make. As for reporters, well, you're not going to improve the quality of science journalism until you start insisting that science journalists know the field they're reporting on sufficiently well to do good work.
The first step in that? Do not click the clickbait.
Actually, the question that seems to matter here is if all the ills of a sedentary lifestyle can be attributed to obesity, which actually is important--and, well, Coca-cola et all will cheerfully sell you low- and no-calorie versions of their products. They're not going to be harmed by people reducing calories. The place where you should be concerned about their involvement in a study would be things like studies on artificial sweeteners or high-fructose corn syrup are actually metabolized, places where their sales are very likely to be harmed if the results aren't positive.
I take your point, but if your recipe called for 1/3 of a tablespoon of cinnamon, I'd call that a dodgy recipe.
I've met a few modern recipes like that--they're ones meant for large-batch cooking. If I'm having to multiply one of those by 3.5? I don't want to have to be thinking about that particular math, I've got other problems that are requiring my attention, which hopefully includes "How to use this borrowed industrial kitchen?" and not "How do I pull this off in a home kitchen?"
A quick search says that, no, minors in the US are legally capable of consenting to a contract: citation, citation, citation, and if you need more I suggest trying Google. The bottom line is that "I am a minor" cannot allow you to void certain types of contracts, it's necessary for the minor to act before they're of age to have the contract nullified, and courts tend to not let you use the "I'm a minor!" card when the contract in question is with a bank.
The problem actually is more one of finding somebody willing to take the risks involved with forming a contract with a minor--as a general rule of thumb, you shouldn't do so unless the law in your state explicitly says that these contracts will be binding anyway, and some laws exist pretty much entirely to keep from fucking minors over by having them unable to form necessary contracts. (For example, car insurance is a contract & some states refuse to issue a driver's license without proof of insurance.) That said, the details will vary from state to state right down to the list of what types of contracts are going to be explicitly non-voidable and details thereof, consult a lawyer experienced with contract law in the jurisdiction(s) in question or avoid the problem entirely.
Most credit card companies seem to go with Option #2.
Also, the whole 'debt ends up on the credit of cosigner' issue is something that happens normally, and most minors have the basic problem of 'no income.' A 16-year-old who is already running their own successful business isn't going to need a cosigner for their credit card unless local law requires it.
It took wandering all over Tokyo for a couple months to find one (possible) alcohol vending machine--I was hiking across Shinjuku and didn't have the time to go check it out, and didn't have a chance to go back later. A tobacco vending machine was located close to where I lived, but I never saw somebody using it and have a sneaking suspicion that nobody actually kept the thing filled anymore & it just hadn't been deemed worth removing from the sidewalk in which it was embedded.
I did find several ice cream vending machines, but I don't think I actually ever managed to find one that did chips or any of the other salty snacks that are rather ubiquitous among US vending machine fare. I certainly had been looking rather hard at the time. (The lemon ice cream was nice, though.)
Oh, and there are some things that can be easily obtained in the US from a vending machine which you cannot get at all from such in Japan.
new terms pretty much are going to have to be found simply to keep from adding to the problem
Until then, doctors who care about things like sickle-cell disease, will have to pick a word and go with it.
No, if they care about things they'll be willing to test everybody and anybody if there's reason to suspect it--sickle-cell is in fact one of the specific reasons why we're having to push doctors to stop being using bad, lazy approximations. About the only thing your obvious ancestry does is modify the odds of you having the gene, though it will remain a non-zero number.
training it to identify somebody's ethnicity and origins would also be distinctly more useful--especially since that's actually something you can do using headshots, as bone structure is more reliable than skin tone and most of the variation can be found in the face.
I don't know about 'distinctly more useful', but sure, that would be a valid research project.
It'd also have much fewer ethical issues and be using concepts with good validity--and, well, humans have an error rate there because they do tend to be lazy assholes and using bad shortcuts. All you'd really need to do is remember that Asians and Native Americans exist, which is not so much of a problem
The minimum set needed to actually make a good operational definition
Who cares?
They weren't setting out to precisely and exactly categorise each person's sexuality, they were setting out to make a comparatively simple, but still meaningful, binary classifier.
Again, all you're really saying is that you can come up with a more precise taxonomy than the gay-or-straight binary model. Well of course you can, but this has no bearing on the basic idea of the research.
If I come up with a classifier that can tell cats from dogs, it would be absurd to try to dismiss the project on the grounds that it can't tell a Labrador from a spaniel.
Okay, I've been trying to be nice and assume you're not a bigot, but I give up here. You're a bigot.
You keep insisting that you can totally get away with saying that you can fucking make a good study by treating a system that includes a set that contains both options and the null set work. No. This is not good. This is a really fucking bad thing to reduce to a binary, and I really don't know what the fucking hell you're doing here on /.
if you cannot grasp sufficiently well what the damned problem is from simply a mathematical perspective.
I couldn't give two fucks what the categorization they used were, aside from the ethical issues other people pointed out. The problem I'm trying to point out where is that how you deem the accuracy of the AI's guess on cases where it's not 0 or 1 but {0,1} is completely arbitrary when you don't allow {0,1} as a response, and that kind of thing is considered abuse when done to a human--and I generally want to avoid doing shit that would be considered abusive if done to a human to an AI.
No, we probably haven't managed to get any of them to the point where it'll actually cause a problem, I just somehow doubt we're going to get that much warning. The time to start good practices is now.
You're only going to have problems when you try lumping the last two into either, unless you deliberately make sure your sample is made up entirely of the first two groups--and even then, you really should note that up front when defining your sample.
Sure, clarity and completeness in a research paper is always desirable. (In true Slashdot fashion, I've not read it.)
I generally make a point of skipping reading research papers where the abstract leaves me with questions about the journal's peer review process. I've limited time, about the only thing of note here is that machine learning research may need to be required to start going through ethics review boards--it usually takes something that just makes everybody rather offended.
I studied precisely the part of science that means I'd know more in detail about those medical contexts than you do, and I know this because I know we're shifting to DNA testing because it's a godsfuckingdamned social construct
Sure, that's a good point: 'race' can mean whatever politically-charged categorisation people want it to mean, and it's important to be clear what we mean by terms like 'white'. The actual biology of the matter (I think we're meant to call it 'ethnicity') is somewhat separate from the identity politics, but it does exist.
(The oft-touted idea that biologically, there's no such thing as race is utter nonsense. Dawkins, among others, has railed against this silliness.)
The concept most people talk about when they say race does not exist biologically--if you want biological race, you need to talk to physical anthropologists. Ethnicity is a more usable concept.
That said, Dawkins is not god, and not really the person I'd want to go with here, anyway. I've been mostly in the 'munch popcorn' section of the argument, since the issue of the biology resembles a primate house which has had a diarrhea outbreak fail to curb their enthusiasm for flinging. The politically-charged and historically bad work on race--check into 'scientific racism' for some of the history here--pretty much mean that if there's something going on biologically, new terms pretty much are going to have to be found simply to keep from adding to the problem.
The question is, is this one of them?
Yes, it is. It's a first-effort research paper, and they're claiming to have achieved very good success with their classifier (ignoring the aforementioned issues with the paper). A 10% nitpick makes for a valid footnote, but doesn't undermine the basic ideas.
Actually, that'd also mean that training it to identify somebody's ethnicity and origins would also be distinctly more useful--especially since that's actually something you can do using headshots, as bone structure is more reliable than skin tone and most of the variation can be found in the face. The gold standard is genetic markers, but these two things are part of how we know the definitions have changed, sometimes in ways to us moderns would be absurd. (Guess what kinds of things get noticed when you've got graveyards needing digging up when the records tell you who is buried where? There's still archeological work done--you can learn a lot about history, some of which we can't really do otherwise because there's a reason science is careful to be clear about names of things now.)
they needed to choose something wherein a binary classifier would be sufficient--and definitely not one where either option of your binary is equally right (or wrong, depending on how you're scoring it) as an answer.
What are you talking about, 'equally wrong'? The hard numbers of the classifier's effectiveness, are right there in the summary.
The fact that you can come up with a more detailed taxonomy of human sexuality doesn't invalidate the effectiveness of their coarse-grain, first-effort classifier. In the real world, almost nothing falls into neat categories, and no number of highly detailed categories can ever capture all detail of the individual.
Categories are all about making useful generalisations. That's why we use them.
Where does that kind of thinking end, anyway? The classifier fails to distinguish bisexuals from pansexuals, therefore the whole project is incoherent?
The minimum set needed to actually make a good operational definition for 'human sexuality' is to have a four-option system of 'same sex,' 'opposite sex,' 'both sexes,' and 'no.' You can generally get away with lumping everybody who is pretty meh on their partner's sex into
Which doesn't escape from the fact that those orientations are, well, social constructs
You say that as if it invalidates the concept. It doesn't. Nationality, authority, prestige, liberty, justice, progress, and money, are all social constructs, but they certainly matter. Same goes for sexuality.
You can say that race too is 'just' a social construct, but it still matters in certain medical contexts.
More generally: the reason we use imperfect labels (with 'rounding errors' as it were) is that they're useful in practice.
Fun fact: I studied precisely the part of science that means I'd know more in detail about those medical contexts than you do, and I know this because I know we're shifting to DNA testing because it's a godsfuckingdamned social construct--it's not a good shorthand for ancestry, it's got a very high error rate because the definitions change and in fact there's a lot of people here who are old enough that the definitions changed in their lifetimes. In fact, in the past century, there's been at least two major changes in how the definitions were, each of which would change how somebody got counted--and some people lied and forgot to let their kids in on things, and some people will find themselves being counted differently depending on if they're in England, the US, or Australia...
But there is a lag in getting doctors to actually switch off of this bad lazy approximation, which is...pretty much normal.
how you define the categories will throw things off even more than only using Caucasian faces
This sounds a lot like a continuum fallacy. That the classifications are imperfect doesn't mean the whole project is completely invalidated. Subjects' self-reports are easily good enough to serve as a starting-point.
(I can see major issues with the study though, like that they used photos from dating sites, which as others have said likely aren't going to be representative of most photos.)
10% of people who check the 'White' tickbox when asked what their race is aren't actually White, according to current common US definitions of the term. (Yes, all of those specifics are needed.) I suppose a 90% accuracy rate is acceptable for some purposes. The question is, is this one of them?
The odds are pretty good that they did engage in erasure of bisexuals, since that's normal for such research, and I'd be amazed if they even considered the existence of asexuals for this
That wouldn't make these researchers bigots, as you seem to be implying, it would just make them pragmatists. Fewer categories simplifies the classification problem.
There is nothing to apologise for when machine-learning researchers decide to make a binary classifier rather than a multiclass classifier.
The implication exists entirely in your head, and it doesn't actually make them pragmatists. If they were pragmatists wanting to use a binary classifier, they needed to choose something wherein a binary classifier would be sufficient--and definitely not one where either option of your binary is equally right (or wrong, depending on how you're scoring it) as an answer.
I'm not calling them bigots, I'm calling them bad scientists, using bad conceptualizations in their machine-learning research.
A quick check of recent research done using the Kinsey scale supports the ~5% rate--not all bisexuals get counted as straight when people are trying to ignore that bisexuality exists.
I can dig out the citation if you really want, but it's relatively easily found and some of its flaws would require an extensive discussion of...well...all the various problems of using self-report. (Things like how you ask the question can have an impact on the numbers--and yes, sometimes this is set up deliberately because what's being studied are things like how many people will only answer yes when you ask indirectly.) The thing that concerns me about this is that it really sounds like the AI is being trained with the assumption that bisexuality doesn't exist--which means that regardless of which state you consider to be a false positive, it'll be wrong with every single bisexual regardless of if it deems them homosexual or heterosexual.
What is the need to distinguish gay from straight?
It's important if you are looking for a mate.
Which doesn't escape from the fact that those orientations are, well, social constructs--and I'm curious but unwilling to give the article a clickthrough to find out if those were the only two orientations they used, because how you define the categories will throw things off even more than only using Caucasian faces. The odds are pretty good that they did engage in erasure of bisexuals, since that's normal for such research, and I'd be amazed if they even considered the existence of asexuals for this--never mind that asexual=/=aromantic so you definitely should be finding some on dating sites.
I never claimed authority over him, just the right of armchair quarterbacking.
If you're responsible for making a business work and you walk away while it is floundering, that's not really fulfilling your responsibilities. Whether that's fine with the stakeholders or not for some reason doesn't make it optimal.
That's not class envy.
Not just that, but it's something worth knowing about the person--both if you're in a position to give them the job of C*O or in a position where you might be working under them, because a bad chief whatever officer can do a number on a company. The sooner you know that you need to start sending out the resumes, the better your chances of escape. So, basically, only people who don't have to worry about money don't need to care about these things.
That said, doing something like haring off for Burning Man is perfectly fine if your company is stable--or if you're the main reason it's floundering and you're wanting to make sure everybody knows it.
cayenne8 stated that President Obama did not have the authority under the US Constitution to enact DACA unilaterally. He says that in order to enact DACA, an act of Congress is required.
And, frankly, if it's really got the support people are acting like it does, then there should be no particular issue with getting that act of Congress--especially since that would also provide distinctly more security for those who might use the program.
This is a lot of whining, really, over being asked "Hey, can we do this properly?" It makes as much sense as complaining about somebody saying you should replace the brittle crock in your code with...something that at least can pass as proper code & be maintained.
Apparently the plantiffs were not sent a recall notice and Amazon had insisted that anybody who didn't get one had an authentic pair. Your Point A doesn't apply, and odds are that the plantiff's lawyers already know if the pair of glasses were counterfeit, defective, or misused. Failure to check those things before filing a suit can cost a lawyer their license.
Also, you're completely ignoring my point that the court may decide that Amazon fucked up beyond what a recall can cover: Amazon was known at least a month before it issued the recall as having many many counterfeits to the point that it wasn't worth trying to get a pair of good eclipse glasses from them. If you know, or should have known if you've two brain cells in communication, that you've got defective products going out the warehouse--the degree of protection you can expect to get from a recall is proportional to the speed with which you issue it once you ought to have been aware of the problem, assuming you made an actual effort to detect/prevent the problem quickly & the recall is sufficiently broad. (This second one is why recalls tend to be pretty broad--and yes, it's a bad sign for the company responsible for the recall when it has to be broadened anyway.)
Pretty much nobody in industries where this is a significant liability risk have gone with the Amazon's combination of "LOL what's quality control???" and laziness in issuing a recall once the problem became public knowledge.
None of this liability risk would exist if Amazon had asked for the paperwork from vendors that it did when the recall finally happened back before they would let them sell eclipse glasses, or made a point of not co-mingling products, or recalled all of the styles that they knew had counterfeits mixed in regardless of the vendor precisely because they could not be bothered to track the actual source for any given specific pair. Odds are that their legal department is trying its best to get this fact through the pointy-haired skulls of upper management. (If they can track it back to a specific vendor, they can throw that person under the bus, especially if they can show that they made an effort, no matter how lazy, to prevent counterfeits.)
The problem is more that the standards for getting those degrees have dropped--there was a movement to make them easier to get people people with degrees earned more, without really grasping that this was because the degree represented proof that the person had a general skillset.
I have nothing against working to prevent financial barriers to getting diplomas--but the educational standards should never have been dropped just to improve graduation rates. When you do that, you're not much different from a diploma mill.
It also can very much depend on the person. Anybody whose body is screwing up blood glucose in the opposite direction is going to benefit from fruit juice--in fact, it's pretty much what the treatment of preference is for any form of hypoglycemia that hasn't sent you to the hospital. (Hospitals tend to take the more direct approach to raising blood glucose.)
Much to my annoyance, the usual juice recommended for this purpose is orange juice...and I absolutely loathe it.
I'd agree with you, if Amazon had done a better job here--because they aren't careful to prevent comingling of vendors' stock from happening and didn't go with a strict "Must prove authenticity/quality/suitability to even be listed" policy which would have mitigated problems the problems that would be caused by comingling? Yeah, this isn't okay, and this is a place where they need to have been tighter and will have to be tighter if they're wanting to sell industrial and scientific equipment. It also raises some unfortunate questions on just how well they really can verify that somebody is shipping more than 1% of defective products--that's the sort of thing that requires you have your inventory set up to keep track of who sent what, and if you can keep track of that, then you should be 100% capable of making sure that if I order something from a specific vendor I get that thing from that vendor.
Actually, recalls don't provide automatic or even complete protection--it's in the public interest to ensure that efforts are made to ensure that defective products don't hit the market at all & recalls are done in a timely manner. If you issue the recall after the problem's become known, and especially if you knew when you shipped out the product that it was dangerous? The recall is roughly a tiny bandaid on a sucking chest wound--at best.
With Amazon's known issues in sending you the products provided their warehouses by a specific vendor, the by-vendor recall done well after it became known that Amazon had counterfeit eclipse glasses might not really be enough to protect them from liability--especially since it may well be determined in court that the problem only existed because Amazon failed at due diligence.
Amazon may not be able to shift blame, especially if this is 100% the result of their habit of commingling stock--if the plantiffs bought from a vendor who had been selling authentic eclipse glasses and Amazon shipped them fakes, it's Amazon's fuck up. The recall itself is probably not going to protect them here, precisely because Amazon doesn't practice what everybody else would consider good basic practices on inventory: You don't mix things unless you're absolutely certain they're the exact same thing. With something like eclipse glasses? Having vendors have to jump through hoops to even get listed at all would have been a Good Idea, with extra care given to being able to track back past you any malfunctioning or dysfunctional items back to whomever put them on your shelves.
Basically? Amazon fucked up due diligence, and has ensured that they cannot reliably shift liability back to the supplier.
Some of it, at least, is the result of places which are too small to sell to the usual overstock buyers--those tend to want truckloads and you may only ever have a box or two at a time--who just want to recover a decent % of the cost while having it stop haunting their shelves.