That's an unusual amount of space to have in the radiator support of a fully assembled car, but that's irrelevant because the ECU is usually mounted right behind the firewall, a decently thick steel plate with just a few small openings into the cabin for wires, hoses and control linkages. It's the panel that makes up the back of the engine bay that extends from the base of the windshield to the cabin floor.
So it seems that this microwave gun is the electromagnetic equivalent of spraying bullets fast and thick enough to kill someone hiding behind an impenetrable wall with ricochet.
There are other sensors in the engine bay of an EFI'd car that could be targeted to cause an engine malfunction though.
You're defining racism as hiring more of race X than race Y even though such hiring is required if qualified people are required.
You've already gone wrong. I'm defining it as using race (or proxy factors for race in a way that has an effect similar to using race) as a factor in the hiring process.
For the hiring pattern to be racist, the hiring requirements would have to be racist.
Given that the hiring requirements are work experience and education, that is not a credible argument on your part.
And if you continue to question the point, I'll point out that there are a lot of people of any race that don't have medical degrees. Do you want just such a person to be your doctor?
Say education and job experience is racist again, and you deserve to have a person picked at random and appointed to be your doctor... especially when you're very sick. Possibly when you need surgery.
You seem to be denying the use of those proxy factors, or again arguing that the use of proxy factors cannot amount to racism. Are you?
An excellent article. However I think that we actually should "force algorithms to reflect wishful thinking" as the author puts it, because the alternative is to use algorithms to justify enacting racism for utilitarian ends such as maximizing profitability or safety. For example, if black Americans are statistically more likely to default on a loan, that doesn't mean it's fine to charge someone higher interest rates in the US because they're black, even if a program makes the association only through proxy factors. Everyone should be given a fair chance regardless of their immutable traits even if it isn't the most efficient solution.
Yes I think we should sacrifice accuracy for fairness, to avoid using a person's race (or other immutable tratis) as a factor. Would you like to argue against this?
Again, the ProPublica analysis and the vendor's study basically agree on the numbers but define "bias" in different ways. The vendor's PDF says there was no racial bias for a statistician's definition of "bias" which in practical terms, is extremely utilitarian. Again, it basically says that it's fine to discriminate against people based on race as long as the discrimination results in outcomes consistent with the overall numbers for that race. It says that if black people are statistically more likely to reoffend, it's fine to apply higher risk numbers to any and all black people in response to this to produce a statistically unbiased outcome. This is unfair to any individual, who will suffer or benefit from the averages for their race. Propublica calls that biased, rather than using the statistician's definition.
I would ask you to read the article linked in this comment and my response to it to improve your understanding on the difference, and why I disagree with using the statistician's definition to argue that racist algorithms are "unbiased" and therefore acceptable:
An excellent article. However I think that we actually should "force algorithms to reflect wishful thinking" as the author puts it, because the alternative is to use algorithms to justify enacting racism for utilitarian ends such as maximizing profitability or safety. For example, if black Americans are statistically more likely to default on a loan, that doesn't mean it's fine to charge someone higher interest rates in the US because they're black, even if a program makes the association only through proxy factors. Everyone should be given a fair chance regardless of their immutable traits even if it isn't the most efficient solution.
We shouldn't let algorithms be conduits and laundries for utilitarian racism. We wouldn't allow such discrimination if done directly by observation, and we shouldn't allow it if done with complex algorithms that greatly obscure the relationship between race and biased outputs but effectively act in the same way.
I mean, it should be obvious that something is wrong when the counterpoint basically boils down to "we're discriminating against people in a way that turns out to be statistically correct overall, therefore it's not wrong." So much for treating people like individuals!
How can you even aspire to achieve colorblindness (or as I like to call it, "babby's first attempt at not being a racist asshat through pretending that different ethnicities and cultures with histories aren't a thing") if you're willing to justify discrimination by painting whole ethnic groups with one broad brush? In your previous post you casually brushed away the legacy of centuries of slavery and segregation, can't you see that this thinking is sending you down a bad path?
Was thinking the same thing. I am sure there is no satellite (other than perhaps a modern amateur microsat) running anything bearing any resemblance to a desktop operating system. The control system may be running Windows 95, but that's a different problem.
First link is denialist bunk. Not surprised, I mark Slashdot users as foes as a way to keep track of the denialists. Accumulated cyclone energy hasn't increased but it sure as hell hasn't decreased either:
It's not about ability, it's about race. I think you know you're full of crap because you don't seem to have read the ProPublica article about the criminal sentencing program which also suffers from racial biases. It has nothing to do with what skills happen to be commonly found in what ethnic groups, it has to do with programs making wildly different decisions between people who have very similar relevant data when only race proxy information is significantly different. People who are arguing in good faith don't ignore information and harp on about strawman points like that.
Although your shameless use of textbook"model minority" arguments and colorblindness suggests that you could indeed be utterly clueless, mindlessly supporting scientific racism through sheer ignorance.
But then you also suggest that race "isn't why people are poor or rich" right in the face of a history of slavery and systematic oppression of black people primarily for the benefit of white people. It's hard to be that wrong by accident, suggesting darker intent.
Horseshit. If a program makes decisions that are indistinguishable from racist decisions, it's racist. You could even build a program or neural network for the explicit (but only in external documentation) purpose of discriminating against certain races without that program having anything about the concept of race in its code. That wouldn't excuse its racist results.
You're less than half right. There will be nightmare scenarios within a year, AND the social media megacorps will have a tighter grip on online discourse than ever before, because they'll be the only companies wealthy enough to get into zero-rating deals with all the big ISPs. Smaller sites including deplorable cesspools like Gab and Mind will have their traffic count towards your data cap or theoretically could be blocked altogether.
To weigh them in the manner you suggest they would have to have built in associations between places and these racial groups.
Not in the program, no.
Cite proof. The source code can be audited. Surely court cases could demand it. Where is the evidence?
Like I said, it's not in the program. The program doesn't know that race is a thing, it's not programmed explicitly to be racist, it's just learned to make decisions similar to the ones it's been trained on, which have a racial bias that is only reflected in proxy factors. The program only "knows" that certain values for these proxy factors are bad, especially in combination. But it's not so easy to audit the "source code" of neural networks anyway.
That means the correlating element is not race... its ability.
This would only be true if the training data was completely free of racial bias. If you want to argue that, we'll have to agree to disagree, which I'd be glad to do considering that you're ignoring a well-known problem with machine learning algorithms trained on biased human decisions which I can link to again. But you won't read it, you'll stick your head in the sand and call it a liberal witch hunt. Why, I wonder.
If you'd read my link you might've been tipped off that we're not talking about a small script disqualifying people solely on address or employment history, but a complex program weighting different factors and using them together. For example it might've been trained on data that shows people from a certain geographic area being less likely to be hired (because they're black) so when the program sees that a person is from that area, it decreases that person's desirability score. It sees that certain names are less likely to result in a hire (because they're more common among the black population), and again it decides that person is less desirable. Again with work experience, it might find that working in a warehouse as a first job is less desirable than working at a Starbucks as a first job, although both are equally irrelevant to a programming job, due to the racial proxy factor and being trained on data created with human-generated biases. And of course multiple factors together would have an even stronger effect due to a closer match to what it was trained on.
If you understand how it unintentionally uses this proxy data to replicate the racism it was trained on, it's obvious that the real problem is in the training data and the only way to avoid it in machine learning is to carefully sanitize the training data, a mammoth task that could make the technology impractical. Racism in, racism out.
Yes the program can make racist decisions without even being aware of the concept of race, because it was trained on racist decisions which it blindly emulates. Educate yourself: https://www.propublica.org/art...
To turn a profit, by definition you must take more than you produce. So either Openbook will be a premium social networking site for the privacy-aware that takes your money instead of your privacy (cryptocurrency is just a hyper-inefficient and criminal-friendly form of money) effectively making them a charity-supported operation, or the best they could hope for is something approaching "neutral."
When I first heard that Trump would be choosing someone to head NASA, my first thought was "Well he'll definitely be a climate conspiracy theorist, I wonder if he'll be a flat earther?"
I didn't expect him to be a homophobe though. Hopefully we won't make first contact with the Moclans while this guy is in charge...
Slashdot Mirror
That's an unusual amount of space to have in the radiator support of a fully assembled car, but that's irrelevant because the ECU is usually mounted right behind the firewall, a decently thick steel plate with just a few small openings into the cabin for wires, hoses and control linkages. It's the panel that makes up the back of the engine bay that extends from the base of the windshield to the cabin floor.
So it seems that this microwave gun is the electromagnetic equivalent of spraying bullets fast and thick enough to kill someone hiding behind an impenetrable wall with ricochet.
There are other sensors in the engine bay of an EFI'd car that could be targeted to cause an engine malfunction though.
You're defining racism as hiring more of race X than race Y even though such hiring is required if qualified people are required.
You've already gone wrong. I'm defining it as using race (or proxy factors for race in a way that has an effect similar to using race) as a factor in the hiring process.
For the hiring pattern to be racist, the hiring requirements would have to be racist.
Given that the hiring requirements are work experience and education, that is not a credible argument on your part.
And if you continue to question the point, I'll point out that there are a lot of people of any race that don't have medical degrees. Do you want just such a person to be your doctor?
Say education and job experience is racist again, and you deserve to have a person picked at random and appointed to be your doctor... especially when you're very sick. Possibly when you need surgery.
You seem to be denying the use of those proxy factors, or again arguing that the use of proxy factors cannot amount to racism. Are you?
I saw that. Did you see this, from my response?
An excellent article. However I think that we actually should "force algorithms to reflect wishful thinking" as the author puts it, because the alternative is to use algorithms to justify enacting racism for utilitarian ends such as maximizing profitability or safety. For example, if black Americans are statistically more likely to default on a loan, that doesn't mean it's fine to charge someone higher interest rates in the US because they're black, even if a program makes the association only through proxy factors. Everyone should be given a fair chance regardless of their immutable traits even if it isn't the most efficient solution.
Yes I think we should sacrifice accuracy for fairness, to avoid using a person's race (or other immutable tratis) as a factor. Would you like to argue against this?
So the actual scientists at skepticalscience are cranks, and some dude on a climate conspiracy blog is trustworthy? Off to a great start.
Which graph shows a slight negative slope? Certainly not the '70-'17 global ACE graph.
Beware of "skeptical" slope interpretation, and good luck with that Santa Ana house.
You seem to be willing to ignore not only my sources but also my arguments. But I'll try linking more sources anyway.
https://www.propublica.org/art...
https://github.com/propublica/...
Again, the ProPublica analysis and the vendor's study basically agree on the numbers but define "bias" in different ways. The vendor's PDF says there was no racial bias for a statistician's definition of "bias" which in practical terms, is extremely utilitarian. Again, it basically says that it's fine to discriminate against people based on race as long as the discrimination results in outcomes consistent with the overall numbers for that race. It says that if black people are statistically more likely to reoffend, it's fine to apply higher risk numbers to any and all black people in response to this to produce a statistically unbiased outcome. This is unfair to any individual, who will suffer or benefit from the averages for their race. Propublica calls that biased, rather than using the statistician's definition.
I would ask you to read the article linked in this comment and my response to it to improve your understanding on the difference, and why I disagree with using the statistician's definition to argue that racist algorithms are "unbiased" and therefore acceptable:
https://slashdot.org/comments....
An excellent article. However I think that we actually should "force algorithms to reflect wishful thinking" as the author puts it, because the alternative is to use algorithms to justify enacting racism for utilitarian ends such as maximizing profitability or safety. For example, if black Americans are statistically more likely to default on a loan, that doesn't mean it's fine to charge someone higher interest rates in the US because they're black, even if a program makes the association only through proxy factors. Everyone should be given a fair chance regardless of their immutable traits even if it isn't the most efficient solution.
We shouldn't let algorithms be conduits and laundries for utilitarian racism. We wouldn't allow such discrimination if done directly by observation, and we shouldn't allow it if done with complex algorithms that greatly obscure the relationship between race and biased outputs but effectively act in the same way.
Not the messenger, the interpretation. Your crank sees a decline where there is none.
I'm really bad at this? You linked to a deflection from the author and vendor of the system. Here's why it actually doesn't disprove the problems pointed out in the propublica article, it merely tries to reframe them in a way that looks less damning.
I mean, it should be obvious that something is wrong when the counterpoint basically boils down to "we're discriminating against people in a way that turns out to be statistically correct overall, therefore it's not wrong." So much for treating people like individuals!
How can you even aspire to achieve colorblindness (or as I like to call it, "babby's first attempt at not being a racist asshat through pretending that different ethnicities and cultures with histories aren't a thing") if you're willing to justify discrimination by painting whole ethnic groups with one broad brush? In your previous post you casually brushed away the legacy of centuries of slavery and segregation, can't you see that this thinking is sending you down a bad path?
Was thinking the same thing. I am sure there is no satellite (other than perhaps a modern amateur microsat) running anything bearing any resemblance to a desktop operating system. The control system may be running Windows 95, but that's a different problem.
First link is denialist bunk. Not surprised, I mark Slashdot users as foes as a way to keep track of the denialists. Accumulated cyclone energy hasn't increased but it sure as hell hasn't decreased either:
https://www.skepticalscience.c...
Of course you had to cherry pick for your second argument. "Last year's Santa Ana fires in particular weren't made worse by global warming! Everything's fine, nothing to see here, pay no attention to all the other wildfires linked in the same article!"
It's not about ability, it's about race. I think you know you're full of crap because you don't seem to have read the ProPublica article about the criminal sentencing program which also suffers from racial biases. It has nothing to do with what skills happen to be commonly found in what ethnic groups, it has to do with programs making wildly different decisions between people who have very similar relevant data when only race proxy information is significantly different. People who are arguing in good faith don't ignore information and harp on about strawman points like that.
Although your shameless use of textbook "model minority" arguments and colorblindness suggests that you could indeed be utterly clueless, mindlessly supporting scientific racism through sheer ignorance.
But then you also suggest that race "isn't why people are poor or rich" right in the face of a history of slavery and systematic oppression of black people primarily for the benefit of white people. It's hard to be that wrong by accident, suggesting darker intent.
Horseshit. If a program makes decisions that are indistinguishable from racist decisions, it's racist. You could even build a program or neural network for the explicit (but only in external documentation) purpose of discriminating against certain races without that program having anything about the concept of race in its code. That wouldn't excuse its racist results.
It's screwing us already, the people in Puerto Rico, Barbuda, Dominica, NYC and California wildfire country have already been screwed hard.
You're less than half right. There will be nightmare scenarios within a year, AND the social media megacorps will have a tighter grip on online discourse than ever before, because they'll be the only companies wealthy enough to get into zero-rating deals with all the big ISPs. Smaller sites including deplorable cesspools like Gab and Mind will have their traffic count towards your data cap or theoretically could be blocked altogether.
If you don't like the power of private censorship on the Internet, the NN repeal is your worst fucking nightmare.
Lucky you, here's what you've been missing. You'll soon have another chance to experience them first hand.
To weigh them in the manner you suggest they would have to have built in associations between places and these racial groups.
Not in the program, no.
Cite proof. The source code can be audited. Surely court cases could demand it. Where is the evidence?
Like I said, it's not in the program. The program doesn't know that race is a thing, it's not programmed explicitly to be racist, it's just learned to make decisions similar to the ones it's been trained on, which have a racial bias that is only reflected in proxy factors. The program only "knows" that certain values for these proxy factors are bad, especially in combination. But it's not so easy to audit the "source code" of neural networks anyway.
That means the correlating element is not race... its ability.
This would only be true if the training data was completely free of racial bias. If you want to argue that, we'll have to agree to disagree, which I'd be glad to do considering that you're ignoring a well-known problem with machine learning algorithms trained on biased human decisions which I can link to again. But you won't read it, you'll stick your head in the sand and call it a liberal witch hunt. Why, I wonder.
Excellent point, mod parent up.
If you'd read my link you might've been tipped off that we're not talking about a small script disqualifying people solely on address or employment history, but a complex program weighting different factors and using them together. For example it might've been trained on data that shows people from a certain geographic area being less likely to be hired (because they're black) so when the program sees that a person is from that area, it decreases that person's desirability score. It sees that certain names are less likely to result in a hire (because they're more common among the black population), and again it decides that person is less desirable. Again with work experience, it might find that working in a warehouse as a first job is less desirable than working at a Starbucks as a first job, although both are equally irrelevant to a programming job, due to the racial proxy factor and being trained on data created with human-generated biases. And of course multiple factors together would have an even stronger effect due to a closer match to what it was trained on.
If you understand how it unintentionally uses this proxy data to replicate the racism it was trained on, it's obvious that the real problem is in the training data and the only way to avoid it in machine learning is to carefully sanitize the training data, a mammoth task that could make the technology impractical. Racism in, racism out.
Address, name, employment and education history.
Yes the program can make racist decisions without even being aware of the concept of race, because it was trained on racist decisions which it blindly emulates. Educate yourself: https://www.propublica.org/art...
the robots in many cases aren't even aware of your race or gender so how are they going to select against your race and gender?
They're clearly NOT deciding on that factor as they literally can't because they're literally not given that variable most of the time.
They use proxy factors closely associated with race or gender, care to rework the rest of your post in light of this new information that was in TFS?
To turn a profit, by definition you must take more than you produce. So either Openbook will be a premium social networking site for the privacy-aware that takes your money instead of your privacy (cryptocurrency is just a hyper-inefficient and criminal-friendly form of money) effectively making them a charity-supported operation, or the best they could hope for is something approaching "neutral."
When I first heard that Trump would be choosing someone to head NASA, my first thought was "Well he'll definitely be a climate conspiracy theorist, I wonder if he'll be a flat earther?"
I didn't expect him to be a homophobe though. Hopefully we won't make first contact with the Moclans while this guy is in charge...
I also hear that Soylent gives you awful haunted demon farts, so his wife might just want to eat outside to get some fresh air.
This post brought to you by Post-Truth! Working to put uninformed gut feelings on par with history, science, and math since 2016!
Jesus kind of summed it up:
You will always have the poor...
A self-fulfilling prophecy if there ever was one.