I also pointed out that spending has no correlation on education quality.
Since your spending whine is entirely based on the false argument that more money would mean better education... it really doesn't matter.
Again, teachers in private schools are often paid WORSE than public teachers. And yet the educational attainment is higher.
You often have more spending in urban areas than in rural areas and yet rural students on average preform much better despite spending less money.
Often various regions will have uniform per student spending and yet performance will vary widely from school to school.
All of this makes clear that spending is not the controlling factor on literacy, graduation rates, test scores, or any other metric you might pretend to care about.
Your indifference to the lack of correlation calls into question your integrity in the discussion given your continuing fixation on who is spending more or less.
Not only is your spending argument wrong but again... EVEN IF IT WERE RIGHT you'd still have no point because there is no correlation between spending and education quality.
And keep in mind... I'm saying correlation. Even if you had correlation... which you don't... that wouldn't necessarily grant causation.
But you clearly don't care. You've got a programmed narrative you're going with and like some deranged chat bot you're going to keep repeating it no matter how irrelevant the it is.
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 only obtain a racial discrepancy by ignoring ability.
You get this number by taking the total population of race X and race Y and then looking at the hiring distribution at the company.
That is how you get your base line.
The counter position is that if you exclude everyone in the population that doesn't have the skills... such as a medical degree to become a doctor... then the comparison between the remaining population of X and Y and the hiring pattern of the company makes perfect sense.
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.
Your entire position would be laughable if there weren't enough fools in society that would attempt to virtue signal the society into destruction.
... you're just not looking at anything are you...
"This example is also important because of its real world consequences. After the article was published, a team of statisticians at Stanford decided to study the cost of fairness. It is possible to take the COMPAS algorithm and manipulate it to be fair. But in the process of doing this, accuracy is reduced. The Stanford team shows that if this were done, the (mostly black) high risk convicts that the manipulated algorithm would release would then commit 9 percent more violent crimes. Furthermore, 17 percent of the people in jail would be (mostly white) individuals at a very low risk of re-offending."
I linked to the actual study itself. The PDF of the study. And it basically completely contradicted your position.
You keep linking to click bait articles.
Link to your study. And I'll then quote the other study that debunked it if that is even required.
You have no case. Arguing with people like you is like arguing with flatearthers. Evidence is irrelevant. You just go to "i'm right dot com" and cite your priest as if that means anything.
I cited the ACTUAL STUDY. I provided a direct link to the PDF. I then quoted its conclusion. It concluded there was no racial bias.
Black defendants who were predicted to recidivate (i.e., given a âoeNot Lowâ score) actually did recidivate at a higher rate (63%) than the white defendants (59%). This finding provides evidence of predictive parity for the GRRS for blacks and whites in the target population.
2. PP Conclusion: âoe
White defendants who were predicted not to recidivate (i.e., given a âoeLowâ score) actually did not recidivate at a higher rate (71%) than the
We use the Positive and Negative Predictive Values for stating the conclusions instead of their complements to follow standard practice in the field. black defendants (65%). This finding provides evidence of predictive parity for the GRRS for blacks and whites in the target population.
3. PP Conclusion: âoe
The black defendants were not assigned inappropriately high risk scores. PPâ(TM)s conclusion is based on their results from their misspecified reverse logistic regression model, which is contradicted by their results from their correctly specified Cox survival model. PPâ(TM)s own Cox survival analysis showed that a variable coding White (0) versus Black (1) had a positive effect for predicting recidivism over and beyond the COMPAS Low, Medium, and High levels. There was an interaction showing that this effect was smaller for the Medium and High levels, but for all three levels, the predictions from the model were higher for black defendants. The PP authors are essentially arguing that the COMPAS risk levels overpredict for blacks. If that were the case, then the predictions from a regression model that included the risk levels and race (black vs. white) designed to predict recidivism would need to be adjusted down for blacks vs whites. However, when Black is added to the Cox survival model, the predictions from the model for all three risk levels are adjusted in the opposite direction. Thus, the risk levels donâ(TM)t overpredict for blacks. A more detailed technical discussion is provided in section 3.4.
4. PP Conclusion: âoe
Black defendants who were predicted to recidivate for violent recidivism (i.e., given a âoeNot Lowâ score) actually did recidivate at a marginally higher rate (21%) than the white defendants (17%). And white defen- dants who were predicted not to recidivate (i.e., given a âoeLowâ score) actually did not recidivate at a marginally higher rate (93%) than the black defendants (91%). These findings provide evidence of predictive parity for the VRRS for blacks and whites in the target population.
5. PP Conclusion: âoe
The black defendants were not assigned inappropriately high risk scores for violent recidivism. In a survival analysis, a variable coding White (0) versus Black (1) had a positive but non-significant main effect for predicting recidivism over and beyond the COMPAS Low, Medium, and High levels. There was no interaction of this variable with levels as reported by PP. This finding indicates that the COMPAS risk levels neither underpredict or overpredict for black defendants. ""
To recap from the conclusion:
1. The "group" predicted to commit further crimes did commit further crimes at a higher rate.
2. The "group" predicted to not commit further crimes did commit fewer crimes.
3. The algorithm did not over predict.
4. Largely a repeat of 1 and 2 with an added cavot that it applies to violent crime as well.
As to neural nets, this is totally irrelevant... why would the neural net be racist if it were simply instructed to hire people that had certain job experience and educational backgrounds?
You're just making things up at this point. Self teaching neural nets? Come on. Who do you think you're fooling with this crap? This is pathetic.
As to your link, the Guardian article wasn't addressing hiring software but rather a language association system. It wasn't actually biasing against anyone. It merely saw certain words used in certain contexts and made associations on that basis. Hiring software doesn't do that.
Strike one.
The New scientist article was talking about stop and frisk in New York as administered by HUMAN BEINGS and not computers. There's no citation in that article of evidence of actual racial bias on the part of a computer in that article. What is more, the methodology of how they determined something was racist by the NYPD was not specified. You clearly didn't read your citations.
Strike two.
The article from the Atlantic is referring to accusations made against a court reporter software... the issue was investigated... https://www.washingtonpost.com...
There are two or three papers I could send you that would explain why the accusation didn't made sense. Fuck it, here they are: https://www.documentcloud.org/...
As to neural networks etc... why would the neural network be consistently biased against race X? I mean, if it is unpredictable then that shouldn't be consistent.
Think about it. You're making an argument that relies on some randomness and chaotic behavior and yet the results are very consistent. That makes your position indefensible. You can't argue that these black boxes programmed by different people, given different data sets, and given different critera are all going to be racist against the same group. That makes no sense. Such systems would be as likely to be biased against different groups or fixate on things entirely unrelated to race. I see no justification for the argument that hiring AIs would be racially obsessed and specifically biased against a given ethnic or racial group absent instruction to do so.
This is a translation of the whole "subconscious racism" argument that was spun up when the civil rights vultures ran out of actual racists to attack. So they had to start attacking non-racists otherwise who would need them? Now that we're moving on to machines you have to claim the robots are racists too.
no correlation between teacher pay and student performance.
Most private school teachers make less money than do public school teachers and yet their students have a much higher academic achievement.
There is no correlation between paying them more and getting better results.
that is not to say that you can pay them zero or that if you offered better compensation you couldn't get better people to teach.
However, that isn't what happens when pay is increased.
When I increase pay at a business the concept would be that I have a hiring shortfall or I want to improve the quality of the type of person I am hiring. that means firing bad teachers and replacing them with teachers you couldn't get before you raised the pay.
What they do instead is keep the existing bad teachers, don't improve standards to be a teacher, and just raise their wages.
if you have a bad employee, paying them more doesn't make them a good employee. If you have an ignorant person, paying them more doesn't cause the to become educated. If you have a lazy person or an immoral person or an unethical person working for you... paying them more doesn't change anything but what they are getting paid.
The concept is very very simple. Consider if you have a hospital with bad doctors. Will paying the doctors more cause them to be better doctors? No.
Offering more pay might help you REPLACE the bad doctors but you have to actually do that.
The teacher's unions that endlessly advocate for more pay never accept mass firings of bad teachers. But if you want to go the "pay them more" route then you MUST fire the bad teachers and replace them with higher quality individuals that could not be attracted to the industry prior to the pay increase. If you keep the existing work force you're just pissing money down the drain.
Our systems were more efficient before the existing government intrusions.
Our healthcare costs are doubling as are our education costs etc as a result of feedback subsidies.
Where in the government offers money to help people buy something, the market responds by raising prices, the government responds by increasing subsidies, and the market responds by increasing prices again... over and over.
It happened with the US housing market, our education, and our healthcare.
large portions of the US population are economically illiterate. This is not unusual, economic illiteracy is actually very common throughout the world and last I checked europeans score worse than Americans on it.
The point is that people don't understand basic economic concepts and so don't understand why given courses of action are obviously stupid.
To prove that it is government mismanagement and incompetence... We have a medical baking soda shortage in the US.
To be very clear here, this is a shortage of sodium bicarbinate mixed with distilled water. It is impossible for the US to have a shortage of this substance and yet we do.
Why? The FDA pharma regulations make it almost impossible for US suppliers to make drugs. It is very hard to be a pharma company in the US. Which is why we have so few pharma companies and why it is so common for new pharma companies to go bankrupt. The FDA kills them.
Drugs in the US are dramatically more expensive in the US than outside not because of evil corporations but because Americans cannot import foreign drugs and domestic suppliers are crushed with crippling regulations.
Again... the regulations are so crippling that we have shortages of baking soda.
This is what our government has done. These drugs are not expensive because of companies. Absent government regulation we'd have more domestic producers and/or would import from out of the country. It is the government that prevents both effective domestic production and foreign import.
And that is just ONE example of how they fuck it up.
A statistical pattern of hiring given population groups due to those skills being found statistically in those groups doesn't mean the hiring system is racist.
That is like saying that luxury car sellers are racist because they tend to sell cars to rich people.
By your logic we should hire people with no medical degrees to be doctors if the people being hired as under represented in the medical industry.
Your argument is at best a confession of insanity. At BEST. That is assuming you actually believe what you are saying. I frankly don't. I think you know you're full of crap.
Would you accept a cab driver that didn't pass a driver's test?
Holding people to standards is apparently racist.
Never mind that suggesting that is actually racist on your part at best because we're all humans and we can all achieve the same things indifferent to our race... UNLESS our race actually has a cognitive value to it. Naturally race does not have any cognitive value to it. Being a given race doesn't make you smarter or dumber or better at any given task.
So why harp on race endlessly? Many people of any race are successes or failures. And people that fail and succeed generally do so for very similar reasons indifferent to their race.
People from single parent house holds tend to have a hard time. Single mother house holds are even worse statistically. Poverty doesn't help. Drug abuse doesn't help. Certain neighborhoods have high levels of criminal activity that drags families into multi-generational dysfunction.
We can address those problems because they're not race. Race is a waste of time. It isn't why people are poor or rich.
Look at Asians in the US to prove the point. There was and is racism against asians. And yet asians if anything do better in the US than most if not all other racial groups in the US.
How does your flawed racial narrative account for asians?
Asians prove my point and prove you wrong. Because Asians despite suffering some racism do very well in the US because they have a healthy culture. They have strong family values, they get married and stay married, they raise their children as a family, they indoctrinate their children with good work ethics and good study skills... etc etc etc.
This is all known. We all know this. You know it and I know it.
Which is part of how I know you're just a liar. It isn't about race. Its about ability. And not racial ability but personal ability as developed over your life to prepare for given tasks. If you haven't prepared then you won't have the needed abilities and you won't get hired because you are not what is needed or desired.
Well, the burden of proof is on the accuser to make that argument unless you want to go full witch trial with this one.
Got anything? Or have we inverted the burden of proof in legal proceedings so we can fall entirely into tyranny and then into barbarism?
You do realize that if you invert the burden of proof it won't merely be applied where you want it to be applied.
For example, let us just say you are accused of murder or something. Can you prove you didn't do it? In many cases you won't be able to prove you didn't do it. And that will set the default position to guilty on accusation.
This is increasingly something that the PC crowd are asking for because they're short sighted fools. They think it will only be applied to a rape case or a murder case that they've gotten excited about on facebook.
It won't stop there. The burden of proof must be on the accused or nearly all of us are guilty of everything. We don't have the evidence to prove we didn't do any random thing you make up.
You know this... many deceitful people exploiting this political climate know this as well. But it will destroy our society if you don't stop.
And no one is going to do well when that happens. The people in our societies will suffer and as our societies themselves trend into tyranny and police states... they will become increasingly militaristic and hostile to their neighbors... it will mean the old wars of conquest again. And there will be no society to restrain it... because you killed it for nothing.
You might well claim that is hyperbolic... then present your evidence to support your claim or concede you have no case.
Or if you neither provide evidence nor concede... you validate my hyperbole.
You're fucked any way you go... not my fault. You're the one that fed your own credibility feet first into the wood chipper.
As to disparate impact, its a subversion of common law in that you can't show intent and in fact I don't think anyone in that case is suggesting there was intent.
If something impacts someone more than someone else without intent then that can't be racism. Racism requires a racial bias. There was no racial bias.
There was a geographic bias which just so happened to coincide with a poor community.
And really, if I wanted to address something like that... I'd suspect the way to go would be to simply not hire people that lived that far away from the job. Mission accomplished. Fire some people and your statistical argument goes away. Ironically the people you'd fire would be the people you're trying to help.
Seriously, you want a company to apologize for hiring people that have to commute a long way to work? They applied for the job. They knew where the job was in the first place. By this argument, they shouldn't have been hired.
What a win for civil rights./s
There was no racism and this lawsuit probably limited opportunities for poor people that desperately need that work. What fucking humanitarians you guys are...
Cite it if its there. You'll not be able to. You'll just look at it, be disappointed that your narrative can't be supported, and then throw out an insult or say nothing.
If its there, then it should be really easy to quote it and win. You won't because you literally can't.
If they are not programmed to find race of discriminate on a racial basis then they're not racist.
If instead their bias is against a lack of experience or education and that happens to correlate with given races then that is not racism but a product of those races statistically not having those skills.
If you limit the population pool to ONLY people with the skills cited and ignore the rest of the population that does not have the skills indifferent to race.
And then recalculate your population distribution of employees to the population that has the skills... do you think there would be a statistical variance? If not, then clearly there is no racism.
What you're doing is finding something that correlates with race that is a reasonable hiring criteria and blaming the use of a hiring criteria... like job experience and education... with racism.
That's not intellectually supportable.
Regardless, we're well into the put up or get laughed off the stage phase of the discussion. Can you show any evidence of hiring software with clear racial biases built into them. That is something that is clearly and obviously and indisputably solving for race and then discriminating on that basis?
Because if not... you have an argument about as solid as Santa Clause and the Easter Bunny. By all means, have your little ritual if it makes you happy. But don't expect anyone else to actually believe the fat man is real unless they're six and are expecting presents.
First, I don't argue that AI "cannot" be programmed to have a racial bias. My argument rather is that there is no evidence that anyone has actually implemented such a hiring system anywhere... certainly not amoungst the major multi national corporations. And those are generally the targets for legal and political agitation probably because slimy people think they can get a fat settlement simply by being annoying.
Second, racial diversity is not actually hugely helpful for innovation. This is one of the odd logical contradictions you get from PC politics. This notion that "we're all human" but "if you have different pigment then you have different ideas"... its nonsense. You can have a more monolithic set of ideas with a very racially diverse group or a very diverse set of ideas with a single race or even a single race and gender. To say otherwise is to suggest that races have different mental processes biologically which if you don't contradict the position opens the door for saying given races will be better or worse at given tasks. Since we all agree that all races are generally equally able to do the same things at the same level... you can't contradict that position and then say that mixing the races offers some kind of intellectual advantage. This is a "x=2, x=3" argument. X cannot equal 2 and 3 at the same time. How this sort of patently absurd logic gets through our cultural filters is pretty terrifying really.
Third, your point about Google is well taken and reveals the double standards that are at play here. Google is naturally openly discouraging certain races from applying to jobs whilst these people here seem convinced that phantom discrimination exists despite their inability to actually present any evidence.
Well, given that I provided something and have offered to provide more... and you've done nothing but say patently false things whilst providing zero evidence.
Your ignorance is confirmed.
This discussion is over. Do not presume to contradict people that know more than you on a subject. It merely leads to the further spread of ignorance.
I'll engage with you in a different discussion. But in this discussion... we're done unless you provide evidence. You won't be able to... so I suspect you'll either say nothing or throw out a cheap insult with no evidence. But I'd be shocked if you even tried to show evidence at this point. You probably had a brief look at the issue, saw it all contradicted your narrative and instead of owning up to your error you're instead just lying to me here.
To weigh them in the manner you suggest they would have to have built in associations between places and these racial groups.
For your concept to work they would have to do that. Do you have any evidence what so ever that they do this or are you just making things up?
Answer... but we both already know you're full of crap.
Cite proof. The source code can be audited. Surely court cases could demand it. Where is the evidence?
What we both know is that that isn't what is happening. There have been a million studies on this issue. There have been endless fishing expeditions. They have all come up bust. You can show a hiring pattern but the hiring pattern every time is supported by the actual presence of the people with the skills they want to hire.
That means the correlating element is not race... its ability.
Provide a link to one of these programs being set up in the manner you suggest they're all being set up.
You are creating a presumption of guilt environment where the accused must prove their innocence simply on your accusation.
If this is the direction you want to go with law, then you're going to suffer literal witch trials before long. And you'll deserve them.
So your argument is that the program was configured to filter for those names?
You think there is a name look up table that associates given name types with given cultural backgrounds?
Is that your argument?
Do you HONESTLY think that the program was set up that way. Because I don't believe you honestly think that. I think you're just making up implausible garbage to save a stillborn argument.
The only thing anyone has ever actually proved was this thing from Xerox: https://www.theguardian.com/sc... ""But the most problematic correlation had to do with geography. Job applicants who lived farther from the job were more likely to churn. This makes sense: long commutes are a pain. But Xerox managers noticed another correlation. Many of the people suffering those long commutes were coming from poor neighbourhoods. So Xerox, to its credit, removed that highly correlated churn data from its model. The company sacrificed a bit of efficiency for fairness.""
Note this was not the result of racism, was not designed to be racist, and only affected "poor neighborhoods" in that the commute was longer.
Any actual evidence of this racism?
I suspect not. We're increasingly a post-evidence society at this point. Just go with your dogmatic belief and then cling to it in the face of all empirical contradiction and absent any evidence.
Such progress you people have made over your old religious past. Truly pathetic.
And if you want to rebut that... find some evidence. Absent that... its sustained.
Slashdot Mirror
I also pointed out that spending has no correlation on education quality.
Since your spending whine is entirely based on the false argument that more money would mean better education... it really doesn't matter.
Again, teachers in private schools are often paid WORSE than public teachers. And yet the educational attainment is higher.
You often have more spending in urban areas than in rural areas and yet rural students on average preform much better despite spending less money.
Often various regions will have uniform per student spending and yet performance will vary widely from school to school.
All of this makes clear that spending is not the controlling factor on literacy, graduation rates, test scores, or any other metric you might pretend to care about.
Your indifference to the lack of correlation calls into question your integrity in the discussion given your continuing fixation on who is spending more or less.
Not only is your spending argument wrong but again... EVEN IF IT WERE RIGHT you'd still have no point because there is no correlation between spending and education quality.
And keep in mind... I'm saying correlation. Even if you had correlation... which you don't... that wouldn't necessarily grant causation.
But you clearly don't care. You've got a programmed narrative you're going with and like some deranged chat bot you're going to keep repeating it no matter how irrelevant the it is.
No correlation.
Process that or fail the Turing Test.
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 only obtain a racial discrepancy by ignoring ability.
You get this number by taking the total population of race X and race Y and then looking at the hiring distribution at the company.
That is how you get your base line.
The counter position is that if you exclude everyone in the population that doesn't have the skills... such as a medical degree to become a doctor... then the comparison between the remaining population of X and Y and the hiring pattern of the company makes perfect sense.
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.
Your entire position would be laughable if there weren't enough fools in society that would attempt to virtue signal the society into destruction.
I did and the link invalidated your position. I literally posted it above.
So you've apparently gone into a fantasy land where you're not in Check...
Fine... *takes king*
The grade school costs in the US show no correlation between spending and performance so it doesn't matter.
... you're just not looking at anything are you...
"This example is also important because of its real world consequences. After the article was published, a team of statisticians at Stanford decided to study the cost of fairness. It is possible to take the COMPAS algorithm and manipulate it to be fair. But in the process of doing this, accuracy is reduced. The Stanford team shows that if this were done, the (mostly black) high risk convicts that the manipulated algorithm would release would then commit 9 percent more violent crimes. Furthermore, 17 percent of the people in jail would be (mostly white) individuals at a very low risk of re-offending."
I linked to the actual study itself. The PDF of the study. And it basically completely contradicted your position.
You keep linking to click bait articles.
Link to your study. And I'll then quote the other study that debunked it if that is even required.
You have no case. Arguing with people like you is like arguing with flatearthers. Evidence is irrelevant. You just go to "i'm right dot com" and cite your priest as if that means anything.
I cited the ACTUAL STUDY. I provided a direct link to the PDF. I then quoted its conclusion. It concluded there was no racial bias.
Get rekt.
Try again.
You're really bad at this...
you mean "this" Propublica article:
https://assets.documentcloud.o...
"" ...
Chapter 4
Conclusion
1. PP Conclusion: âoe
Black defendants who were predicted to recidivate (i.e., given a âoeNot
Lowâ score) actually did recidivate at a higher rate (63%) than the white
defendants (59%). This finding provides evidence of predictive parity for
the GRRS for blacks and whites in the target population.
2. PP Conclusion: âoe
White defendants who were predicted not to recidivate (i.e., given a
âoeLowâ score) actually did not recidivate at a higher rate (71%) than the
We use the Positive and Negative Predictive Values for stating the conclusions instead
of their complements to follow standard practice in the field.
black defendants (65%). This finding provides evidence of predictive
parity for the GRRS for blacks and whites in the target population.
3. PP Conclusion: âoe
The black defendants were not assigned inappropriately high risk scores.
PPâ(TM)s conclusion is based on their results from their misspecified reverse
logistic regression model, which is contradicted by their results from
their correctly specified Cox survival model. PPâ(TM)s own Cox survival
analysis showed that a variable coding White (0) versus Black (1) had a
positive effect for predicting recidivism over and beyond the COMPAS
Low, Medium, and High levels. There was an interaction showing that
this effect was smaller for the Medium and High levels, but for all three
levels, the predictions from the model were higher for black defendants.
The PP authors are essentially arguing that the COMPAS risk levels
overpredict for blacks. If that were the case, then the predictions from a
regression model that included the risk levels and race (black vs. white)
designed to predict recidivism would need to be adjusted down for blacks
vs whites. However, when Black is added to the Cox survival model, the
predictions from the model for all three risk levels are adjusted in the
opposite direction. Thus, the risk levels donâ(TM)t overpredict for blacks. A
more detailed technical discussion is provided in section 3.4.
4. PP Conclusion: âoe
Black defendants who were predicted to recidivate for violent recidivism
(i.e., given a âoeNot Lowâ score) actually did recidivate at a marginally
higher rate (21%) than the white defendants (17%). And white defen-
dants who were predicted not to recidivate (i.e., given a âoeLowâ score)
actually did not recidivate at a marginally higher rate (93%) than the
black defendants (91%). These findings provide evidence of predictive
parity for the VRRS for blacks and whites in the target population.
5. PP Conclusion: âoe
The black defendants were not assigned inappropriately high risk scores
for violent recidivism. In a survival analysis, a variable coding White
(0) versus Black (1) had a positive but non-significant main effect for
predicting recidivism over and beyond the COMPAS Low, Medium, and
High levels. There was no interaction of this variable with levels as
reported by PP. This finding indicates that the COMPAS risk levels
neither underpredict or overpredict for black defendants.
""
To recap from the conclusion:
1. The "group" predicted to commit further crimes did commit further crimes at a higher rate.
2. The "group" predicted to not commit further crimes did commit fewer crimes.
3. The algorithm did not over predict.
4. Largely a repeat of 1 and 2 with an added cavot that it applies to violent crime as well.
5. The models are mathematically sound.
As to neural nets, this is totally irrelevant... why would the neural net be racist if it were simply instructed to hire people that had certain job experience and educational backgrounds?
You're just making things up at this point. Self teaching neural nets? Come on. Who do you think you're fooling with this crap? This is pathetic.
As to your link, the Guardian article wasn't addressing hiring software but rather a language association system. It wasn't actually biasing against anyone. It merely saw certain words used in certain contexts and made associations on that basis. Hiring software doesn't do that.
Strike one.
The New scientist article was talking about stop and frisk in New York as administered by HUMAN BEINGS and not computers. There's no citation in that article of evidence of actual racial bias on the part of a computer in that article. What is more, the methodology of how they determined something was racist by the NYPD was not specified. You clearly didn't read your citations.
Strike two.
The article from the Atlantic is referring to accusations made against a court reporter software... the issue was investigated...
https://www.washingtonpost.com...
There are two or three papers I could send you that would explain why the accusation didn't made sense. Fuck it, here they are:
https://www.documentcloud.org/...
http://www.crj.org/assets/2017...
Strike three.
As to neural networks etc... why would the neural network be consistently biased against race X? I mean, if it is unpredictable then that shouldn't be consistent.
Think about it. You're making an argument that relies on some randomness and chaotic behavior and yet the results are very consistent. That makes your position indefensible. You can't argue that these black boxes programmed by different people, given different data sets, and given different critera are all going to be racist against the same group. That makes no sense. Such systems would be as likely to be biased against different groups or fixate on things entirely unrelated to race. I see no justification for the argument that hiring AIs would be racially obsessed and specifically biased against a given ethnic or racial group absent instruction to do so.
This is a translation of the whole "subconscious racism" argument that was spun up when the civil rights vultures ran out of actual racists to attack. So they had to start attacking non-racists otherwise who would need them? Now that we're moving on to machines you have to claim the robots are racists too.
Its sad.
no correlation between teacher pay and student performance.
Most private school teachers make less money than do public school teachers and yet their students have a much higher academic achievement.
There is no correlation between paying them more and getting better results.
that is not to say that you can pay them zero or that if you offered better compensation you couldn't get better people to teach.
However, that isn't what happens when pay is increased.
When I increase pay at a business the concept would be that I have a hiring shortfall or I want to improve the quality of the type of person I am hiring. that means firing bad teachers and replacing them with teachers you couldn't get before you raised the pay.
What they do instead is keep the existing bad teachers, don't improve standards to be a teacher, and just raise their wages.
if you have a bad employee, paying them more doesn't make them a good employee. If you have an ignorant person, paying them more doesn't cause the to become educated. If you have a lazy person or an immoral person or an unethical person working for you... paying them more doesn't change anything but what they are getting paid.
The concept is very very simple. Consider if you have a hospital with bad doctors. Will paying the doctors more cause them to be better doctors? No.
Offering more pay might help you REPLACE the bad doctors but you have to actually do that.
The teacher's unions that endlessly advocate for more pay never accept mass firings of bad teachers. But if you want to go the "pay them more" route then you MUST fire the bad teachers and replace them with higher quality individuals that could not be attracted to the industry prior to the pay increase. If you keep the existing work force you're just pissing money down the drain.
My victory is self evident. Its not arrogant to say 1+1=2. Rather it is idiotic to contradict the statement.
Our systems were more efficient before the existing government intrusions.
Our healthcare costs are doubling as are our education costs etc as a result of feedback subsidies.
Where in the government offers money to help people buy something, the market responds by raising prices, the government responds by increasing subsidies, and the market responds by increasing prices again... over and over.
It happened with the US housing market, our education, and our healthcare.
large portions of the US population are economically illiterate. This is not unusual, economic illiteracy is actually very common throughout the world and last I checked europeans score worse than Americans on it.
The point is that people don't understand basic economic concepts and so don't understand why given courses of action are obviously stupid.
To prove that it is government mismanagement and incompetence... We have a medical baking soda shortage in the US.
https://www.nytimes.com/2017/0...
To be very clear here, this is a shortage of sodium bicarbinate mixed with distilled water. It is impossible for the US to have a shortage of this substance and yet we do.
Why? The FDA pharma regulations make it almost impossible for US suppliers to make drugs. It is very hard to be a pharma company in the US. Which is why we have so few pharma companies and why it is so common for new pharma companies to go bankrupt. The FDA kills them.
Drugs in the US are dramatically more expensive in the US than outside not because of evil corporations but because Americans cannot import foreign drugs and domestic suppliers are crushed with crippling regulations.
Again... the regulations are so crippling that we have shortages of baking soda.
here is a list of drug shortages in the US:
https://www.accessdata.fda.gov...
This is what our government has done. These drugs are not expensive because of companies. Absent government regulation we'd have more domestic producers and/or would import from out of the country. It is the government that prevents both effective domestic production and foreign import.
And that is just ONE example of how they fuck it up.
It goes on and on and on and on.
A statistical pattern of hiring given population groups due to those skills being found statistically in those groups doesn't mean the hiring system is racist.
That is like saying that luxury car sellers are racist because they tend to sell cars to rich people.
By your logic we should hire people with no medical degrees to be doctors if the people being hired as under represented in the medical industry.
Your argument is at best a confession of insanity. At BEST. That is assuming you actually believe what you are saying. I frankly don't. I think you know you're full of crap.
Would you accept a cab driver that didn't pass a driver's test?
Holding people to standards is apparently racist.
Never mind that suggesting that is actually racist on your part at best because we're all humans and we can all achieve the same things indifferent to our race... UNLESS our race actually has a cognitive value to it. Naturally race does not have any cognitive value to it. Being a given race doesn't make you smarter or dumber or better at any given task.
So why harp on race endlessly? Many people of any race are successes or failures. And people that fail and succeed generally do so for very similar reasons indifferent to their race.
People from single parent house holds tend to have a hard time. Single mother house holds are even worse statistically. Poverty doesn't help. Drug abuse doesn't help. Certain neighborhoods have high levels of criminal activity that drags families into multi-generational dysfunction.
We can address those problems because they're not race. Race is a waste of time. It isn't why people are poor or rich.
Look at Asians in the US to prove the point. There was and is racism against asians. And yet asians if anything do better in the US than most if not all other racial groups in the US.
How does your flawed racial narrative account for asians?
Asians prove my point and prove you wrong. Because Asians despite suffering some racism do very well in the US because they have a healthy culture. They have strong family values, they get married and stay married, they raise their children as a family, they indoctrinate their children with good work ethics and good study skills... etc etc etc.
This is all known. We all know this. You know it and I know it.
Which is part of how I know you're just a liar. It isn't about race. Its about ability. And not racial ability but personal ability as developed over your life to prepare for given tasks. If you haven't prepared then you won't have the needed abilities and you won't get hired because you are not what is needed or desired.
It has nothing to do with race.
... oh come on...
What I obviously meant:
If they are not programmed to find discrimination on a racial basis then they're not racist.
Well, the burden of proof is on the accuser to make that argument unless you want to go full witch trial with this one.
Got anything? Or have we inverted the burden of proof in legal proceedings so we can fall entirely into tyranny and then into barbarism?
You do realize that if you invert the burden of proof it won't merely be applied where you want it to be applied.
For example, let us just say you are accused of murder or something. Can you prove you didn't do it? In many cases you won't be able to prove you didn't do it. And that will set the default position to guilty on accusation.
This is increasingly something that the PC crowd are asking for because they're short sighted fools. They think it will only be applied to a rape case or a murder case that they've gotten excited about on facebook.
It won't stop there. The burden of proof must be on the accused or nearly all of us are guilty of everything. We don't have the evidence to prove we didn't do any random thing you make up.
You know this... many deceitful people exploiting this political climate know this as well. But it will destroy our society if you don't stop.
And no one is going to do well when that happens. The people in our societies will suffer and as our societies themselves trend into tyranny and police states... they will become increasingly militaristic and hostile to their neighbors... it will mean the old wars of conquest again. And there will be no society to restrain it... because you killed it for nothing.
You might well claim that is hyperbolic... then present your evidence to support your claim or concede you have no case.
Or if you neither provide evidence nor concede... you validate my hyperbole.
You're fucked any way you go... not my fault. You're the one that fed your own credibility feet first into the wood chipper.
As to disparate impact, its a subversion of common law in that you can't show intent and in fact I don't think anyone in that case is suggesting there was intent.
If something impacts someone more than someone else without intent then that can't be racism. Racism requires a racial bias. There was no racial bias.
There was a geographic bias which just so happened to coincide with a poor community.
And really, if I wanted to address something like that... I'd suspect the way to go would be to simply not hire people that lived that far away from the job. Mission accomplished. Fire some people and your statistical argument goes away. Ironically the people you'd fire would be the people you're trying to help.
Seriously, you want a company to apologize for hiring people that have to commute a long way to work? They applied for the job. They knew where the job was in the first place. By this argument, they shouldn't have been hired.
What a win for civil rights. /s
There was no racism and this lawsuit probably limited opportunities for poor people that desperately need that work. What fucking humanitarians you guys are...
No, there was no evidence of anything.
Cite it if its there. You'll not be able to. You'll just look at it, be disappointed that your narrative can't be supported, and then throw out an insult or say nothing.
If its there, then it should be really easy to quote it and win. You won't because you literally can't.
We've seen arguments lately that the scientific method is racist and sexist.
So yeah... PC culture is basically going full anti intellectual... probably because it is logically indefensible.
If the program that controls the operation of the system contains no racist instructions then the program is not racist.
Game over.
Insert coin?
If they are not programmed to find race of discriminate on a racial basis then they're not racist.
If instead their bias is against a lack of experience or education and that happens to correlate with given races then that is not racism but a product of those races statistically not having those skills.
If you limit the population pool to ONLY people with the skills cited and ignore the rest of the population that does not have the skills indifferent to race.
And then recalculate your population distribution of employees to the population that has the skills... do you think there would be a statistical variance? If not, then clearly there is no racism.
What you're doing is finding something that correlates with race that is a reasonable hiring criteria and blaming the use of a hiring criteria... like job experience and education... with racism.
That's not intellectually supportable.
Regardless, we're well into the put up or get laughed off the stage phase of the discussion. Can you show any evidence of hiring software with clear racial biases built into them. That is something that is clearly and obviously and indisputably solving for race and then discriminating on that basis?
Because if not... you have an argument about as solid as Santa Clause and the Easter Bunny. By all means, have your little ritual if it makes you happy. But don't expect anyone else to actually believe the fat man is real unless they're six and are expecting presents.
First, I don't argue that AI "cannot" be programmed to have a racial bias. My argument rather is that there is no evidence that anyone has actually implemented such a hiring system anywhere... certainly not amoungst the major multi national corporations. And those are generally the targets for legal and political agitation probably because slimy people think they can get a fat settlement simply by being annoying.
Second, racial diversity is not actually hugely helpful for innovation. This is one of the odd logical contradictions you get from PC politics. This notion that "we're all human" but "if you have different pigment then you have different ideas"... its nonsense. You can have a more monolithic set of ideas with a very racially diverse group or a very diverse set of ideas with a single race or even a single race and gender. To say otherwise is to suggest that races have different mental processes biologically which if you don't contradict the position opens the door for saying given races will be better or worse at given tasks. Since we all agree that all races are generally equally able to do the same things at the same level... you can't contradict that position and then say that mixing the races offers some kind of intellectual advantage. This is a "x=2, x=3" argument. X cannot equal 2 and 3 at the same time. How this sort of patently absurd logic gets through our cultural filters is pretty terrifying really.
Third, your point about Google is well taken and reveals the double standards that are at play here. Google is naturally openly discouraging certain races from applying to jobs whilst these people here seem convinced that phantom discrimination exists despite their inability to actually present any evidence.
I can throw rhetorical rotten fruit at mental cripples if I want. Its a free country.
So no link. What a surprise.
Well, given that I provided something and have offered to provide more... and you've done nothing but say patently false things whilst providing zero evidence.
Your ignorance is confirmed.
This discussion is over. Do not presume to contradict people that know more than you on a subject. It merely leads to the further spread of ignorance.
I'll engage with you in a different discussion. But in this discussion... we're done unless you provide evidence. You won't be able to... so I suspect you'll either say nothing or throw out a cheap insult with no evidence. But I'd be shocked if you even tried to show evidence at this point. You probably had a brief look at the issue, saw it all contradicted your narrative and instead of owning up to your error you're instead just lying to me here.
Pity.
To weigh them in the manner you suggest they would have to have built in associations between places and these racial groups.
For your concept to work they would have to do that. Do you have any evidence what so ever that they do this or are you just making things up?
Answer... but we both already know you're full of crap.
Cite proof. The source code can be audited. Surely court cases could demand it. Where is the evidence?
What we both know is that that isn't what is happening. There have been a million studies on this issue. There have been endless fishing expeditions. They have all come up bust. You can show a hiring pattern but the hiring pattern every time is supported by the actual presence of the people with the skills they want to hire.
That means the correlating element is not race... its ability.
Provide a link to one of these programs being set up in the manner you suggest they're all being set up.
You are creating a presumption of guilt environment where the accused must prove their innocence simply on your accusation.
If this is the direction you want to go with law, then you're going to suffer literal witch trials before long. And you'll deserve them.
So your argument is that the program was configured to filter for those names?
You think there is a name look up table that associates given name types with given cultural backgrounds?
Is that your argument?
Do you HONESTLY think that the program was set up that way. Because I don't believe you honestly think that. I think you're just making up implausible garbage to save a stillborn argument.
The only thing anyone has ever actually proved was this thing from Xerox:
https://www.theguardian.com/sc...
""But the most problematic correlation had to do with geography. Job applicants who lived farther from the job were more likely to churn. This makes sense: long commutes are a pain. But Xerox managers noticed another correlation. Many of the people suffering those long commutes were coming from poor neighbourhoods. So Xerox, to its credit, removed that highly correlated churn data from its model. The company sacrificed a bit of efficiency for fairness.""
Note this was not the result of racism, was not designed to be racist, and only affected "poor neighborhoods" in that the commute was longer.
Any actual evidence of this racism?
I suspect not. We're increasingly a post-evidence society at this point. Just go with your dogmatic belief and then cling to it in the face of all empirical contradiction and absent any evidence.
Such progress you people have made over your old religious past. Truly pathetic.
And if you want to rebut that... find some evidence. Absent that... its sustained.
What have I said specifically that you feel is a supposition.
Give an example please.
And keep in mind that if you start attempting to bait me into the tired internet quote war.. I'm going to expect quid pro quo from you.
I find it an especially tedious form of rhetoric that I warn you I am familiar with and shall not be rhetorically effective.