"Aren't those rednecks funny with their redneck culture, they're not like real people" "Aren't those gays funny with their gay culture, they're not like real people" "Aren't those Jews funny with their Jewish culture, they're not like real people" "Aren't those Blacks funny with their Black culture, they're not like real people"
None of these statements is OK. None of those jokes are funny. It is never OK to "unpeople" someone. It's not a fair tool in a political argument.
Anyone who reads history has seen what lies at the end of that path, and it's not a destination we want to revisit.
PETA is on board the global warming hoax, apparently.
It is an amazing coincidence how the sentence that begins "Because of global warming we must..." always seems to end with furthering the leftwing agenda. "Ha! We can finally make you stop driving that SUV!" "Ha! No more meat!" . I was trying to think of one with plastic straws, but that one's too stupid to even make a coherent joke about.
Wall's only a few billion, trivial compared to the economic effect of 30 million people (whether you think that's positive or negative). When you're determined to oppose something, you can always come up with a reason, but the wall is worth it just for the symbolism.
You are asserting that women and non-whites have their "bell curve" such that they're not as good as white men.
Ah, I see: you're replying to the wrong thread. You're looking for the post by strawman. Were you replying to my thread, you'd have noticed I claimed the opposite.
Sure. Feed it nothing but red cars, and it will never pick a blue car. Because you limited the input. Really easy to prove in court too
I see you've never worked with "big data". You can't make such firm claims about what's not in a very large data set - pretty much like most attempts to prove a negative.
But that has nothing to do with how machine leaning works: except in trivially simple cases, it's very difficult, often impossible, to prove what its criteria are, as they're entirely arbitrary. You can remove all direct indication of gender from the training data, and it could still be using statistical distribution of punctuation or something equally bizarre, which turns out to be an 80% proxy for gender.
"We don't want false information out there." - he had a mouse in his pocket?
More like, he's in the NSA's pocket. He meant "the NSA is trying to subvert these chips to spy on all Americans. If they fail, we'll be sure to notify everyone affected so they can replace these servers. In the meantime, forget we ever mentioned it."
Yes, you have to establish that such bell curves differ based on gender and not-whiteness. And again, if you're just looking at outcome as your proof, you're falling for the same incomplete data as the AI in TFS.
The point is, it doesn't matter how you establish it. All the details you go into are irrelevant. The people you interview can't be less representative than the candidate pool of protected classes, regardless of why - even if you had ironclad proof, doesn't matter.
It's the nature of machine learning systems that you can't really prove what criteria they've "learned" from your training data
Sure you can. The machine learning system will produce a result that closely matches the training data. When your training data is the result of bias, the machine learning system will strive to continue that bias.
Man, you really love to say "bias". But you've missed the point here: * You cannot, from the output of a machine learning system, prove that it isn't using X as criteria in screening candidates. * You might need to be able to prove that legally. * Therefore, you can't use machine learning to screen candidates.
You need to distinguish between "deciding who to interview" and "deciding who to hire". There's no real defense on the former: you had better be interviewing at least as many people in each protected class, proportionally, as you have in your candidate pool. You can defend yourself on the latter, but you'd better be prepared to show that each and every interview was decided on objective criteria that don't include protected class. But you have a much bigger risk of a lawsuit than an EEOC fine on the latter, these days, and that's what companies tend to worry about.
That interpretation is very silly. They're trying to figure out "as a percentage of candidates with resumes like X that we interview, how many do well at the job", and make all such statistic correlations that they can.
It's not about "they have fewer women on staff", obviously. It's about "how do female candidates fare". Amazon, like all of the Big 5, is trying to address gender balance, and so will interview women with less chance of getting hired than their male candidates. They want more women hired, so they take a bigger risk, an accept the elevated cost in interviewer time. That's great, as it lets you hire more women without lowering the bar for them (accusations about Google's process aside).
But of course the side effect of that is that fewer women who get interviewed get hired. The AI of course found that correlation, and probably dozens of other correlations associated with protected classes, and used it. Pretty obvious in hindsight, but then lots of things are obvious in hindsight.
Until you can demonstrate that women or non-whites are not physically capable of being "the best for the job", then this reasoning is flawed.
And if your "proof" is the lack of not-white-men in these positions, you're falling for the same incomplete data as the AI in TFS.
It's not all-or-nothing, it's bell curves. If you allow race as a consideration for e.g. software development, you'll get a statistical model that tells you to mostly interview Asians. That's an accurate predictive factor (one of many) of how well a candidate is likely to succeed at the job. It's also illegal (well, not the discrimination against whites of course, but against all the non-pariah races).
It's not about "incomplete data as the AI" unless the coders were complete idiots. Amazon has a huge population of developers. It's about the fact that protected classes cannot legally be used to screen interview candidates regardless of why, and you have to prove you're not doing so. It's the nature of machine learning systems that you can't really prove what criteria they've "learned" from your training data, which makes them very legally dangerous.
You can't use any sort of "proof" of someone (because of a protected class) being not physically capable of doing the job, let alone slightly less likely to succeed, as a reason not to interview them. IMO that's not unreasonable: if someone for whatever reason actually can't do the job, surely that comes out in the interview itself.
In other words, if your only criteria is hiring whomever best for the job, you will likely be operating illegally and subject to fines and lawsuits. This is the product of laws that are designed to create social engineering based restrictions based on someones religious idea that any measurable discrepancy in minority placement must be corrected.
That's not quite true. The rules for who you interview are different from the rules on who you hire.
What you say is true for the former: if the people you interview don't at least match the candidate pool for any protected class, you're boned regardless of why. It's a bit different for hiring, where you can defend yourself by showing that your process is objective.
In practice, the government tend to focus a lot on who gets interviewed, as that's strictly numbers and easy to audit and enforce. The fear for biased hiring decisions is mostly about lawsuits (ditto promotion decisions and termination decisions). If you're interviewing lots of women, but not hiring a proportional amount, you're asking for lawsuits, even class actions, and had better be ready to objectively justify every single "no hire" decision, perhaps years after the fact. That leads to lots of paperwork around the interview process at companies that have been burned by this. But if your hiring decisions are objective, that is a defense (similarly promotion and firing, but that's usually lots of paperwork anyhow).
As hard as you want to say, sometimes you still need an actual person doing the job. That person will be biased in some way other another too, so I guess it's not a perfect system any way you look at it.
Amazon goes to extremes though for the hiring process for any professional job: the interview loop must include someone not associated with the hiring team, and that person runs the interview process (and gets lots of extra training and auditing). They're going to great lengths to avoid "we're desperate to hire anyone, so we'll take someone almost good enough", but the side effect is strict objectivity.
But the law is pretty clear that you can't use any protected category in your resume sorting even if it's statistically predictive, which pretty much screws any machine-learning approach: by their nature, you can't really prove how they work, and they'll use anything statistically predictive.
Humans can work differently: overfill the input of the candidate pipeline with women, for example. That's normal for the left coast software companies. E.g., for college recruiting you'll only see men with a very narrow profile get interviewed (specific schools, degrees, school performance, etc), but you'll see women with a wide range of backgrounds (adjacent degrees, non-traditional background, really anything that shows promise). That lets them get closer to the hiring output they want without having a lower bar during the interview itself.
Well, as soon as you say "RGB values", you've lost a wide color gamut. Your monitor looking off without the right color profile is just a part of the picture, so to speak. The "red" and "blue" in the normal RGB encoding aren't at the ends of the spectrum, as that diagram of standards shows. If any part of the pipe is lossy, obviously the result is lossy. The initial recording, the image format, the rendering, the transmission format, and the monitor must all be wide color gamut, or there's no win.
And don't get me started on how "pixels" are done wrong everywhere.
Absolutely. But why would any sane person want that when shopping? It's appalling.
And the hilarious thing for cars is that everyone gets taken, yet walks away thinking they got the best deal ever, and totally outsmarted that car salesman.
we're not building a wall to keep 45,000 Mexicans from marching into Texas with guns and tanks and brigadier generals.
There are still 45,000 coming through though, probably each month. How is that not an invasion again? Tanks and generals are not needed now, as 20-30 million people have just walked in without any resistance. We've already taken "open borders" to a historically unprecedented extreme, and I'm for any attempt to control the border even a symbolic one.
Why did China build the Great Wall? It didn't 100% keep the Mongols out - the just climbed the wall. But their horses didn't, and it limited the amount of loot that could carry back with them after a raid.
Physical security, like digital security, isn't about "all or nothing". Making it harder makes it harder. It's harder to walk across a desert than to drive.
Meh, it's mostly symbolic anyway, and what people are actually arguing about is whether they like the symbolism. Why not just say "globalism is good; no borders" instead of pretending your objection is to the effectiveness of the wall?
Saturn used to do the "the price is the price" thing, too. Seems like American car buyers like to haggle, despite getting taken every time. Millennials may be different though - maybe it's an idea whose time has come.
As a car buyer, Tesla is dead to me without a volume knob. It's far from alone in that doghouse, of course. My current car has enough tactile cues on the center console to do the basic stuff without looking: volume, audio source, radio station, temp control. I do wish there was a knob for temp control, but you can't have everything.
Tesla isn't competing with Tata, you know? The market for a car in the Model 3's price range is certainly big enough to sustain current sales levels. Whether the car can really compete with similarly priced cars in the long run is an open question, but that's a large enough market for once car model to do OK.
IMO, Tesla's long term future in the US depends on coming up with a good pickup truck. There's a huge untapped market there, but Tesla will need to enter it carefully, with the semi selling at least a few units, to use it for marketing, and a $40k model.
Or would you like to classify what most non-US governments would call hate speech, or outright abuse from someone on the right as "conservative stuff?"
Color gamut is limited by the highest and lowest peak frequency your monitor emits. The farther apart the red and blue, the more real colors can look. However, it doesn't do much good if your display is wider than your signal. High-end TVs fake it, by making assumptions about how red a "red" pixel in the signal really is, with various results. There is a wide color gamut protocol standards but I don't know whether e.g. a consumer video card and HDMI can make any use of them.
Yeah, conservative stuff is banned constantly, but the one time something progressive accidentally gets caught up in the conservative-banning-machine, it's a big story. That link reads remarkably like this one. "Unfortunately, our automation accidentally banned some fake news that was progressive, merely because the headline was a blatant lie. Any new system will have the occasional bug, and we humbly apologize for this one. We're taking steps to better train our algorithms."
Slashdot Mirror
"Aren't those rednecks funny with their redneck culture, they're not like real people"
"Aren't those gays funny with their gay culture, they're not like real people"
"Aren't those Jews funny with their Jewish culture, they're not like real people"
"Aren't those Blacks funny with their Black culture, they're not like real people"
None of these statements is OK. None of those jokes are funny. It is never OK to "unpeople" someone. It's not a fair tool in a political argument.
Anyone who reads history has seen what lies at the end of that path, and it's not a destination we want to revisit.
PETA is on board the global warming hoax, apparently.
It is an amazing coincidence how the sentence that begins "Because of global warming we must ..." always seems to end with furthering the leftwing agenda. "Ha! We can finally make you stop driving that SUV!" "Ha! No more meat!" . I was trying to think of one with plastic straws, but that one's too stupid to even make a coherent joke about.
Wall's only a few billion, trivial compared to the economic effect of 30 million people (whether you think that's positive or negative). When you're determined to oppose something, you can always come up with a reason, but the wall is worth it just for the symbolism.
Embrace the healing power of "and". We build the wall and we do the other things to close the border and remove 30 million illegal aliens from the US.
You are asserting that women and non-whites have their "bell curve" such that they're not as good as white men.
Ah, I see: you're replying to the wrong thread. You're looking for the post by strawman. Were you replying to my thread, you'd have noticed I claimed the opposite.
Sure. Feed it nothing but red cars, and it will never pick a blue car. Because you limited the input. Really easy to prove in court too
I see you've never worked with "big data". You can't make such firm claims about what's not in a very large data set - pretty much like most attempts to prove a negative.
But that has nothing to do with how machine leaning works: except in trivially simple cases, it's very difficult, often impossible, to prove what its criteria are, as they're entirely arbitrary. You can remove all direct indication of gender from the training data, and it could still be using statistical distribution of punctuation or something equally bizarre, which turns out to be an 80% proxy for gender.
"We don't want false information out there." - he had a mouse in his pocket?
More like, he's in the NSA's pocket. He meant "the NSA is trying to subvert these chips to spy on all Americans. If they fail, we'll be sure to notify everyone affected so they can replace these servers. In the meantime, forget we ever mentioned it."
What bullshit. Fewer people will come across a wall. It won't stop everyone, but it will stop some. "Some" is a good start.
But you don't care anyhow, you object to the concept of border enforcement, right?
Yes, you have to establish that such bell curves differ based on gender and not-whiteness. And again, if you're just looking at outcome as your proof, you're falling for the same incomplete data as the AI in TFS.
The point is, it doesn't matter how you establish it. All the details you go into are irrelevant. The people you interview can't be less representative than the candidate pool of protected classes, regardless of why - even if you had ironclad proof, doesn't matter.
It's the nature of machine learning systems that you can't really prove what criteria they've "learned" from your training data
Sure you can. The machine learning system will produce a result that closely matches the training data. When your training data is the result of bias, the machine learning system will strive to continue that bias.
Man, you really love to say "bias". But you've missed the point here:
* You cannot, from the output of a machine learning system, prove that it isn't using X as criteria in screening candidates.
* You might need to be able to prove that legally.
* Therefore, you can't use machine learning to screen candidates.
QED.
You need to distinguish between "deciding who to interview" and "deciding who to hire". There's no real defense on the former: you had better be interviewing at least as many people in each protected class, proportionally, as you have in your candidate pool. You can defend yourself on the latter, but you'd better be prepared to show that each and every interview was decided on objective criteria that don't include protected class. But you have a much bigger risk of a lawsuit than an EEOC fine on the latter, these days, and that's what companies tend to worry about.
That interpretation is very silly. They're trying to figure out "as a percentage of candidates with resumes like X that we interview, how many do well at the job", and make all such statistic correlations that they can.
It's not about "they have fewer women on staff", obviously. It's about "how do female candidates fare". Amazon, like all of the Big 5, is trying to address gender balance, and so will interview women with less chance of getting hired than their male candidates. They want more women hired, so they take a bigger risk, an accept the elevated cost in interviewer time. That's great, as it lets you hire more women without lowering the bar for them (accusations about Google's process aside).
But of course the side effect of that is that fewer women who get interviewed get hired. The AI of course found that correlation, and probably dozens of other correlations associated with protected classes, and used it. Pretty obvious in hindsight, but then lots of things are obvious in hindsight.
Until you can demonstrate that women or non-whites are not physically capable of being "the best for the job", then this reasoning is flawed.
And if your "proof" is the lack of not-white-men in these positions, you're falling for the same incomplete data as the AI in TFS.
It's not all-or-nothing, it's bell curves. If you allow race as a consideration for e.g. software development, you'll get a statistical model that tells you to mostly interview Asians. That's an accurate predictive factor (one of many) of how well a candidate is likely to succeed at the job. It's also illegal (well, not the discrimination against whites of course, but against all the non-pariah races).
It's not about "incomplete data as the AI" unless the coders were complete idiots. Amazon has a huge population of developers. It's about the fact that protected classes cannot legally be used to screen interview candidates regardless of why, and you have to prove you're not doing so. It's the nature of machine learning systems that you can't really prove what criteria they've "learned" from your training data, which makes them very legally dangerous.
You can't use any sort of "proof" of someone (because of a protected class) being not physically capable of doing the job, let alone slightly less likely to succeed, as a reason not to interview them. IMO that's not unreasonable: if someone for whatever reason actually can't do the job, surely that comes out in the interview itself.
In other words, if your only criteria is hiring whomever best for the job, you will likely be operating illegally and subject to fines and lawsuits. This is the product of laws that are designed to create social engineering based restrictions based on someones religious idea that any measurable discrepancy in minority placement must be corrected.
That's not quite true. The rules for who you interview are different from the rules on who you hire.
What you say is true for the former: if the people you interview don't at least match the candidate pool for any protected class, you're boned regardless of why. It's a bit different for hiring, where you can defend yourself by showing that your process is objective.
In practice, the government tend to focus a lot on who gets interviewed, as that's strictly numbers and easy to audit and enforce. The fear for biased hiring decisions is mostly about lawsuits (ditto promotion decisions and termination decisions). If you're interviewing lots of women, but not hiring a proportional amount, you're asking for lawsuits, even class actions, and had better be ready to objectively justify every single "no hire" decision, perhaps years after the fact. That leads to lots of paperwork around the interview process at companies that have been burned by this. But if your hiring decisions are objective, that is a defense (similarly promotion and firing, but that's usually lots of paperwork anyhow).
As hard as you want to say, sometimes you still need an actual person doing the job. That person will be biased in some way other another too, so I guess it's not a perfect system any way you look at it.
Amazon goes to extremes though for the hiring process for any professional job: the interview loop must include someone not associated with the hiring team, and that person runs the interview process (and gets lots of extra training and auditing). They're going to great lengths to avoid "we're desperate to hire anyone, so we'll take someone almost good enough", but the side effect is strict objectivity.
But the law is pretty clear that you can't use any protected category in your resume sorting even if it's statistically predictive, which pretty much screws any machine-learning approach: by their nature, you can't really prove how they work, and they'll use anything statistically predictive.
Humans can work differently: overfill the input of the candidate pipeline with women, for example. That's normal for the left coast software companies. E.g., for college recruiting you'll only see men with a very narrow profile get interviewed (specific schools, degrees, school performance, etc), but you'll see women with a wide range of backgrounds (adjacent degrees, non-traditional background, really anything that shows promise). That lets them get closer to the hiring output they want without having a lower bar during the interview itself.
Well, as soon as you say "RGB values", you've lost a wide color gamut. Your monitor looking off without the right color profile is just a part of the picture, so to speak. The "red" and "blue" in the normal RGB encoding aren't at the ends of the spectrum, as that diagram of standards shows. If any part of the pipe is lossy, obviously the result is lossy. The initial recording, the image format, the rendering, the transmission format, and the monitor must all be wide color gamut, or there's no win.
And don't get me started on how "pixels" are done wrong everywhere.
Absolutely. But why would any sane person want that when shopping? It's appalling.
And the hilarious thing for cars is that everyone gets taken, yet walks away thinking they got the best deal ever, and totally outsmarted that car salesman.
we're not building a wall to keep 45,000 Mexicans from marching into Texas with guns and tanks and brigadier generals.
There are still 45,000 coming through though, probably each month. How is that not an invasion again? Tanks and generals are not needed now, as 20-30 million people have just walked in without any resistance. We've already taken "open borders" to a historically unprecedented extreme, and I'm for any attempt to control the border even a symbolic one.
Yes, but the normal case when shopping is "take it or leave it" pricing. That's a kind of negotiation, to be sure.
Why did China build the Great Wall? It didn't 100% keep the Mongols out - the just climbed the wall. But their horses didn't, and it limited the amount of loot that could carry back with them after a raid.
Physical security, like digital security, isn't about "all or nothing". Making it harder makes it harder. It's harder to walk across a desert than to drive.
Meh, it's mostly symbolic anyway, and what people are actually arguing about is whether they like the symbolism. Why not just say "globalism is good; no borders" instead of pretending your objection is to the effectiveness of the wall?
Saturn used to do the "the price is the price" thing, too. Seems like American car buyers like to haggle, despite getting taken every time. Millennials may be different though - maybe it's an idea whose time has come.
As a car buyer, Tesla is dead to me without a volume knob. It's far from alone in that doghouse, of course. My current car has enough tactile cues on the center console to do the basic stuff without looking: volume, audio source, radio station, temp control. I do wish there was a knob for temp control, but you can't have everything.
Tesla isn't competing with Tata, you know? The market for a car in the Model 3's price range is certainly big enough to sustain current sales levels. Whether the car can really compete with similarly priced cars in the long run is an open question, but that's a large enough market for once car model to do OK.
IMO, Tesla's long term future in the US depends on coming up with a good pickup truck. There's a huge untapped market there, but Tesla will need to enter it carefully, with the semi selling at least a few units, to use it for marketing, and a $40k model.
It's a classic /. troll, one from the early days.
Or would you like to classify what most non-US governments would call hate speech, or outright abuse from someone on the right as "conservative stuff?"
Yes, I object strongly to the ridiculous redefinition by progressives of free speech as "hate speech", as of disagreement as "abuse". Tell me: is this hate speech or abuse? Of course, you can call for the death of all white men without consequence.
Not to mention social media's habit of just outright banning of political speech they disagree with.
Color gamut is limited by the highest and lowest peak frequency your monitor emits. The farther apart the red and blue, the more real colors can look. However, it doesn't do much good if your display is wider than your signal. High-end TVs fake it, by making assumptions about how red a "red" pixel in the signal really is, with various results. There is a wide color gamut protocol standards but I don't know whether e.g. a consumer video card and HDMI can make any use of them.
Yeah, conservative stuff is banned constantly, but the one time something progressive accidentally gets caught up in the conservative-banning-machine, it's a big story. That link reads remarkably like this one. "Unfortunately, our automation accidentally banned some fake news that was progressive, merely because the headline was a blatant lie. Any new system will have the occasional bug, and we humbly apologize for this one. We're taking steps to better train our algorithms."