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.
Right, but that was my whole point - it does correlate with ethnicity, which is enough to show it's not just something we made up, as some people like to try to argue.
All you'd really need to do is remember that Asians and Native Americans exist, which is not so much of a problem
No, you don't need to 'remember' that any particular group exists.
Whether we're talking gender, race, sexuality, or anything else, it doesn't matter whether the classifier's taxonomy aligns perfectly with the one used by today's liberals. Unless the researchers are lying about what they built, there is no ethical issue here.
You could make a classifier that distinguishes people of Han Chinese ethnicity from ethnically Nigerian people. That wouldn't make it racist against any other ethnicities, it would just make it a more limited classifier.
Okay, I've been trying to be nice and assume you're not a bigot, but I give up here. You're a bigot.
Ah, modern liberalism: where anyone who doesn't completely agree with you is to be denounced as a bigot.
This is a really fucking bad thing to reduce to a binary
It's a machine-learning classifier with limitations, it's not a model for a new world order. Are you unable to tell the difference?
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.
There isn't any problem here from a mathematical perspective. Again, it's a rather imperfect machine-learning classifier, it's not a template for a new, reductionist, society.
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
No, it's not 'completely arbitrary', it's an imperfect but not unreasonable approximation. It's essentially a clamping function, no?
They could make a follow-up paper where they add bisexuals to the mix. And another where they add asexuals. And another, and another. At which point would you stop your accusations?
that kind of thing is considered abuse when done to a human
We're discussing a machine-learning paper, for heaven's sake. No harm is being done to anyone. I don't care what sort of rounding errors it suffers from.
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.
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.
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.
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 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 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.
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?
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.
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.)
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.
never mind that asexual=/=aromantic so you definitely should be finding some on dating sites.
Oh come on. You realise this kind of minutia-obsession alienates ordinary people, right? The numbers there are so tiny that it just isn't a problem worth worrying about, either for the researchers or for users of dating sites.
You're right that's meant to be part of the point of Java as a development platform, but there are also other reasons to use Java.
It scales well, it does a reasonable job helping out on correctness and security, there's no shortage of Java developers, there's no real chance the platform will die overnight, and there's the No-one ever got fired for choosing Java effect.
People are not supposed to be happy, at least not for very long. If people are happy they will not change anything to risk it going away.
That's pretty braindead, even by the standards of conspiracy thinking.
In the same paragraph, you say that happy people are less likely to want a revolution, but that the ruling elite don't want the masses to be happy. Am I missing something?
Not to mention that the topic of discussion is clinical depression, not happiness.
It's not so bad. It's a proper sporting event, even if it's a weird novelty match. Yes, it's a violent contact sport, but I have no issue with that.
I'm not disheartened to see it receive so much attention. It's a good deal better than the vacuous garbage they normally print in the tabloids, or show on crapholes like msn.com.
Slashdot Mirror
Legally it's not.
Here in the UK, taxi driving and private-hire driving both require a special driving licence, with a background-check. Uber drivers don't have these.
Combined with severe and possibly violent, example-setting punishments
Yes yes, this will be the time that barbaric punishments will finally work as effective deterrents, right?
You ACs and your poorly-thought-out ideas.
Security can 'fall off the radar' for your average idiot CEO - they're sure security incidents only happen to other people.
I think you have to have an independent team of security specialists perform a serious security audit at least every 6 months would go a long way.
Perhaps even require the tiger team to report directly to shareholders. That would force the company to own the specifics of their security concerns.
Because as far as I can tell it's a Chrome extension, but for some reason neither the summary nor the linked articles bother to make this clear.
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.
Right, but that was my whole point - it does correlate with ethnicity, which is enough to show it's not just something we made up, as some people like to try to argue.
All you'd really need to do is remember that Asians and Native Americans exist, which is not so much of a problem
No, you don't need to 'remember' that any particular group exists.
Whether we're talking gender, race, sexuality, or anything else, it doesn't matter whether the classifier's taxonomy aligns perfectly with the one used by today's liberals. Unless the researchers are lying about what they built, there is no ethical issue here.
You could make a classifier that distinguishes people of Han Chinese ethnicity from ethnically Nigerian people. That wouldn't make it racist against any other ethnicities, it would just make it a more limited classifier.
Okay, I've been trying to be nice and assume you're not a bigot, but I give up here. You're a bigot.
Ah, modern liberalism: where anyone who doesn't completely agree with you is to be denounced as a bigot.
This is a really fucking bad thing to reduce to a binary
It's a machine-learning classifier with limitations, it's not a model for a new world order. Are you unable to tell the difference?
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.
There isn't any problem here from a mathematical perspective. Again, it's a rather imperfect machine-learning classifier, it's not a template for a new, reductionist, society.
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
No, it's not 'completely arbitrary', it's an imperfect but not unreasonable approximation. It's essentially a clamping function, no?
They could make a follow-up paper where they add bisexuals to the mix. And another where they add asexuals. And another, and another. At which point would you stop your accusations?
that kind of thing is considered abuse when done to a human
We're discussing a machine-learning paper, for heaven's sake. No harm is being done to anyone. I don't care what sort of rounding errors it suffers from.
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.
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.
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.
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 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 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.
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?
[Citation needed]
As far as I know, Kaspersky doesn't have a reputation for being hard to remove. Or is this 'just in case' thinking?
Also YouTube.
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.
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.)
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.
never mind that asexual=/=aromantic so you definitely should be finding some on dating sites.
Oh come on. You realise this kind of minutia-obsession alienates ordinary people, right? The numbers there are so tiny that it just isn't a problem worth worrying about, either for the researchers or for users of dating sites.
I disagree.
With what? That privacy is important, or that Google have inflexible terms? Be specific.
The fact that some Fortune 500 company is trusting Google with sensitive data, doesn't mean they should be doing so.
It won't have this affect on algorithms trained with this inherent bias in them.
What?
You're right that's meant to be part of the point of Java as a development platform, but there are also other reasons to use Java.
It scales well, it does a reasonable job helping out on correctness and security, there's no shortage of Java developers, there's no real chance the platform will die overnight, and there's the No-one ever got fired for choosing Java effect.
So it has the same origins as HotSpot? Curious.
It's called antitrust law, genius.
Nothing is original!
Not quite fair. V8 is original.
it is hard for us to be strategic in our planning when everyone can see what code we are committing
What?
Also, since when did reddit do any meaningful work on their code? Have they added a feature at some point in the last few years? Did I miss something?
B-. Good bluster, but not enough substance.
Looks a lot like a very awkward attempt at redirection.
You were talking about the will of the wealthy elites, remember? 'Permanent revolution' has nothing to do with it.
Quite right, I must've latched on too literally to Opportunist's post.
People are not supposed to be happy, at least not for very long. If people are happy they will not change anything to risk it going away.
That's pretty braindead, even by the standards of conspiracy thinking.
In the same paragraph, you say that happy people are less likely to want a revolution, but that the ruling elite don't want the masses to be happy. Am I missing something?
Not to mention that the topic of discussion is clinical depression, not happiness.
ACs never fail to disappoint.
It's not so bad. It's a proper sporting event, even if it's a weird novelty match. Yes, it's a violent contact sport, but I have no issue with that.
I'm not disheartened to see it receive so much attention. It's a good deal better than the vacuous garbage they normally print in the tabloids, or show on crapholes like msn.com.
eight times
So the blockchain still slows linearly with time?
The title stands to lose the first word. Clickbait-style title writing.
Would you like me to clap like a seal?