Alternative schedules is one thing, but unhealthy focus on work is another. Everyone I know who consistently works more than 40 hour weeks without alternate time compensation (10 on 5 off, long breaks, etc.) seriously compromises the quality and productivity of their work. And most don't realize it. There's only marginal gains traded for some serious psychological harms.
Alphazero might be more general than human-tuned algorithms for specific games, but it is still very much restricted to finite games (or games where the infinite component can be ignored), and relies heavily on the expert encoding of the rules to aid learning. The rules are not simply "read" by alphazero, the boardspace and legal moves are built into the network in a very specific way to facilitate learning. Then it requires truly massive computing power to train to a *specific* game, and additional massive power to play against outside opponents, during which it is no longer learning.
Alphazero is an impressive feat, but more as a technological step showing that hefty computing power can allow more generalized neural networks can be used to produce good results in these sorts of finite games. It's definitely not crossing any AI thresholds.
But cars are always safety critical systems, regardless of whether computers and software are involved. I wouldn't want to drive a car that wasn't treated as a safety critical system.
That's not right. If you drilled a bunch of holes in the table, the paper would indeed fly up and you wouldn't break the ruler. The airspace between the paper and table is clearly an integral component of the phenomenon.
Native Americans living in the Rocky Mountain area was primarily nomadic prior to the arrival of Columbus, but most other areas of the US were primarily populated by agrarian tribes (though hunter-gatherer cultures were interspersed). Many previously agrarian tribes moved to a more nomadic lifestyle after the arrival of Columbus; the nomadic lifestyle was made more attractive by the combination of civilization collapse due to depopulation by western diseases (which killed upwards of 50-80% of the peoples of North America even prior to Jamestown and Plymouth), the boom of many animal species such as buffalo and passenger pigeons after the depopulation, and the introduction of domestic horses.
Speaking of statistics, there were 21 people hired, and 17 were "non-asian" and 4 were "asian"
Now here's the problem when we go down the rabbit hole of stats, percentages, and hiring by race.
The lawsuit makes no claims about percentages by race in relation to the larger population (as you seem to imply), but only to percentage in reference to the qualified applications. 95 of the qualified applicants for those 21 QA engineering intern positions were asian, and only 35 were non-Asian. But half of the non-Asian applicants were hired while only 5% of the Asian applicants were hired, indicating differential hiring from the applicant pool.
From the article: "The likelihood that this result occurred according to chance is approximately one in a billion," said the lawsuit,
Yeah - I'm going to have to call bullshit on that - I'd love to see that math.
I'm not sure exactly which method they've used, but here's one simple option:
Assuming equivalent merit of applicants, "random chance" means choosing 21 applicants randomly from the applicant pool of 95 Asian and 35 non-Asian applicants. We want to compute the probability that only 4 or less of those 21 selected applicants are Asian. The number of Asian applicants selected follow a hypergeometric distribution. This distribution doesn't have a simple formula computing tail probabilities (your welcome to look it up), but many statistics packages will do it. This methods gives us a roughly 1 in 60 million probability of 4 or less Asian hires due to random chance. Not 1 in a billion, but I suspect that the Department of Labor's analysis incorporated additional information. And 1 in 60 million is still low enough to suggest that something is going on.
But then, according to the article The majority of Palantir's hires as engineering interns, as well as two other engineering positions, "came from an employee referral system that disproportionately excluded Asians,"
And both the comments are from the lawsuit, so I give them veracity as an integral part of the lawsuit
Well, right there is your answer. These interns and two engineering positions came from employee referrals.
In my professional life, I'm exposed to a lot of different ethnicities, but if I referred every "asian" for placement in an open position, and no non "asians" at all, and they all were hired, it might hit 20 percent, of which this lawsuit is considered racially discriminating.
And this is likely (at least part) of what's going on. Employee referral programs are definitely not "by random chance".
So what this is actually an attack against is the process of referral. Should a referral of a known non-"asian" who might have a great track record, be disregarded for an unknown "asian" person, or even more importantly, if referrals are to become verboten, should a non-"asian" be not hired and an "asian" of known flaws be hired?
So perhaps the process of employees giving assessments of people they know is what is considered racist. That's actually a little scary, because it means an employee who knows a person as a bad actor will then not be allowed to make commentary on that.
Employee referral programs can exacerbate existing demographic imbalances, and cause members of certain demographics to have a much more difficult time getting hired at your company. Just because the discrimination is structurally propagated does not get the company off the hook. Large companies that want to avoid exactly this situation closely monitor their key demographics and balance their use of employee referrals to ensure diversity.
It's worse than that (or better, depending on perspective). So far as I can tell, Alibaba decided to sell a limited number of leftover boxes at cost through their internal sales system. Apparently the system was immediately overloaded, so employees weren't able actually purchase the boxes by clicking the "buy" button. A few employees whipped up a script to click the button faster to try to get orders through, and ended up buying 124 boxes between them. Alibaba called this a "hack" and fired them 2 hours later.
If that's actually how it went down, it sounds pretty damn stupid on Alibaba's part. Maybe there's some cultural differences, but a) selling something through a computer system where demand far outstrips supply is stupid to begin with - just do a lottery for purchase rights; b) restrict the number of boxes an individual can buy; c) maybe just retroactively restrict their purchases to certain # of boxes rather than firing them?
Sounds like power-tripping managers who don't know what a hack is and think that throwing a limited number of (apparently highly desirable) items out to a large mass of employees first-come-first-serve is a hilarious fucking game.
So state's are small and homogeneous enough that it's totally okay for them to pass such regulations? It's just the bigwigs in Washington that you want out, but the bigwigs in Sacremento/Albany/Juneau/whatever are totally fine?
Ha ha ha ha "more reliable" ha ha ha ha.
What on earth makes you think that reliability is the reason that companies put no-third-party-repair clauses in their contracts?
Yeah, no. "Making the world more fair" is ideally the whole point of government, no matter where you sit. Protecting individual freedoms is one of the mandates of government. Take away as much government as you seem to want to, and you'll find yourself with less freedoms, not more. And those marginalized individuals that you are so quick to throw under the bus? Their individual freedoms are infringed all the time!
Well, it's not a simulator for one. You're flying the drone just as much as you would fly a plane. Hell, it's probably harder if you're not getting tactile feedback. Who cares where your ass is sitting?
If you scroll down three sentences, you'll see a chart where, by golly! state police workers have the second highest rate of workplace injury, surpassed only by (state government) nursing! If you look at days away from work rate, job transfer or restriction (DART), it moves to fourth behind state and local nursing and local fire departments.
Construction makes the list - it's close to policing for injury rate and DART, but the RATE is smaller for both measures. There's just a heck of a lot more construction workers that police officers, so total injuries are very high.
A coin flip is very different from what we usually mean by a judgment. Would you trust a human soldier that makes random choices about what to do in life-or-death situations? Ideally, they should be able to explain their reasoning / training in each situation, and if the decision does not hold up to retrospective scrutiny, should be held accountable for their actions and/or should adjust their reasoning used in future situations. "Judgement" with a large component of randomness involves no reasoning, and so I'm skeptical of it's usefulness in life-or-death situations (and whether it should be called "judgement" at all).
Historically, people have had a broad tendency to be ruled by unelected tyrants. Clearly people in today's democratic societies are just yearning for some other form of tyranny to fill that gaping hole in their hearts.
"Reason" is good, but too often gets conflated with "Rational self-interest" in discussions of morality. In fact, emotions and irrationality seem to help us become more predictable agents and get around some of the sticking points of rational self-interest.
Here's a paper on the evolutionary advantage of the emotional dimension of moral agents in societies:
http://www.cass.city.ac.uk/__d...
That's well and good (well not good, bad), but Tibbit's "solution" addresses the problem with text bans that you've posted. Unless you've can point out specific problems with his solution, it seems like it's worth think about and testing. Texting while driving is a big problem. If texting while driving bans are also a big problem, then some other solution is needed. Saying "reality is messy" to dismiss proposed solutions is slightly useful to make us think about potential problems, but is really the laziest form of criticism.
I have a problem with people whining about not letting politicians and bureaucrats run roughshod over our rights while shrugging about corporations and individuals run roughshod all over our rights. There's gotta be some balance, man.
Yes I'm still waiting for my damn GE recall.
I'm amazed by how often these "MOOCs can't replace traditional classroom" arguments come up. Well no shit! The MOOC formats provide broad access and logistical ease for self-learning better than any previously existing educational technology, but online interactions are still no match for personal assistance. Most MOOC sites realize this - they're not trying to supplant traditional education, they're trying to supplement it. I fail to see how the MOOC revolution is over when there's still plenty of opportunity to fill in gaps between traditional universities and massive free sites like Coursera (which is exactly what the author's site is doing).
The Golden Fleece awards were mostly a load of anti-intellectual bullshit that had no comprehension of how mug applied science relies on basic science.
Slashdot Mirror
Alternative schedules is one thing, but unhealthy focus on work is another. Everyone I know who consistently works more than 40 hour weeks without alternate time compensation (10 on 5 off, long breaks, etc.) seriously compromises the quality and productivity of their work. And most don't realize it. There's only marginal gains traded for some serious psychological harms.
Yes, Americans have longer work weeks and get very little time off compared to most other developed countries. And that's not a good thing.
Alphazero might be more general than human-tuned algorithms for specific games, but it is still very much restricted to finite games (or games where the infinite component can be ignored), and relies heavily on the expert encoding of the rules to aid learning. The rules are not simply "read" by alphazero, the boardspace and legal moves are built into the network in a very specific way to facilitate learning. Then it requires truly massive computing power to train to a *specific* game, and additional massive power to play against outside opponents, during which it is no longer learning. Alphazero is an impressive feat, but more as a technological step showing that hefty computing power can allow more generalized neural networks can be used to produce good results in these sorts of finite games. It's definitely not crossing any AI thresholds.
But cars are always safety critical systems, regardless of whether computers and software are involved. I wouldn't want to drive a car that wasn't treated as a safety critical system.
Circumstances are very different.
That's not right. If you drilled a bunch of holes in the table, the paper would indeed fly up and you wouldn't break the ruler. The airspace between the paper and table is clearly an integral component of the phenomenon.
Native Americans living in the Rocky Mountain area was primarily nomadic prior to the arrival of Columbus, but most other areas of the US were primarily populated by agrarian tribes (though hunter-gatherer cultures were interspersed). Many previously agrarian tribes moved to a more nomadic lifestyle after the arrival of Columbus; the nomadic lifestyle was made more attractive by the combination of civilization collapse due to depopulation by western diseases (which killed upwards of 50-80% of the peoples of North America even prior to Jamestown and Plymouth), the boom of many animal species such as buffalo and passenger pigeons after the depopulation, and the introduction of domestic horses.
As well as being irrelevant, it's actually totally false. Only a small fraction of pre-contact Native American cultures were primarily nomadic.
It's for their own good, right? Wouldn't it be great if we didn't have to put up with all these "used" women and sloppy seconds?
Speaking of statistics, there were 21 people hired, and 17 were "non-asian" and 4 were "asian"
Now here's the problem when we go down the rabbit hole of stats, percentages, and hiring by race.
The lawsuit makes no claims about percentages by race in relation to the larger population (as you seem to imply), but only to percentage in reference to the qualified applications. 95 of the qualified applicants for those 21 QA engineering intern positions were asian, and only 35 were non-Asian. But half of the non-Asian applicants were hired while only 5% of the Asian applicants were hired, indicating differential hiring from the applicant pool.
From the article: "The likelihood that this result occurred according to chance is approximately one in a billion," said the lawsuit,
Yeah - I'm going to have to call bullshit on that - I'd love to see that math.
I'm not sure exactly which method they've used, but here's one simple option: Assuming equivalent merit of applicants, "random chance" means choosing 21 applicants randomly from the applicant pool of 95 Asian and 35 non-Asian applicants. We want to compute the probability that only 4 or less of those 21 selected applicants are Asian. The number of Asian applicants selected follow a hypergeometric distribution. This distribution doesn't have a simple formula computing tail probabilities (your welcome to look it up), but many statistics packages will do it. This methods gives us a roughly 1 in 60 million probability of 4 or less Asian hires due to random chance. Not 1 in a billion, but I suspect that the Department of Labor's analysis incorporated additional information. And 1 in 60 million is still low enough to suggest that something is going on.
But then, according to the article The majority of Palantir's hires as engineering interns, as well as two other engineering positions, "came from an employee referral system that disproportionately excluded Asians,"
And both the comments are from the lawsuit, so I give them veracity as an integral part of the lawsuit
Well, right there is your answer. These interns and two engineering positions came from employee referrals.
In my professional life, I'm exposed to a lot of different ethnicities, but if I referred every "asian" for placement in an open position, and no non "asians" at all, and they all were hired, it might hit 20 percent, of which this lawsuit is considered racially discriminating.
And this is likely (at least part) of what's going on. Employee referral programs are definitely not "by random chance".
So what this is actually an attack against is the process of referral. Should a referral of a known non-"asian" who might have a great track record, be disregarded for an unknown "asian" person, or even more importantly, if referrals are to become verboten, should a non-"asian" be not hired and an "asian" of known flaws be hired?
So perhaps the process of employees giving assessments of people they know is what is considered racist. That's actually a little scary, because it means an employee who knows a person as a bad actor will then not be allowed to make commentary on that.
Employee referral programs can exacerbate existing demographic imbalances, and cause members of certain demographics to have a much more difficult time getting hired at your company. Just because the discrimination is structurally propagated does not get the company off the hook. Large companies that want to avoid exactly this situation closely monitor their key demographics and balance their use of employee referrals to ensure diversity.
It's worse than that (or better, depending on perspective). So far as I can tell, Alibaba decided to sell a limited number of leftover boxes at cost through their internal sales system. Apparently the system was immediately overloaded, so employees weren't able actually purchase the boxes by clicking the "buy" button. A few employees whipped up a script to click the button faster to try to get orders through, and ended up buying 124 boxes between them. Alibaba called this a "hack" and fired them 2 hours later. If that's actually how it went down, it sounds pretty damn stupid on Alibaba's part. Maybe there's some cultural differences, but a) selling something through a computer system where demand far outstrips supply is stupid to begin with - just do a lottery for purchase rights; b) restrict the number of boxes an individual can buy; c) maybe just retroactively restrict their purchases to certain # of boxes rather than firing them? Sounds like power-tripping managers who don't know what a hack is and think that throwing a limited number of (apparently highly desirable) items out to a large mass of employees first-come-first-serve is a hilarious fucking game.
So state's are small and homogeneous enough that it's totally okay for them to pass such regulations? It's just the bigwigs in Washington that you want out, but the bigwigs in Sacremento/Albany/Juneau/whatever are totally fine?
Ha ha ha ha "more reliable" ha ha ha ha. What on earth makes you think that reliability is the reason that companies put no-third-party-repair clauses in their contracts?
Yeah, no. "Making the world more fair" is ideally the whole point of government, no matter where you sit. Protecting individual freedoms is one of the mandates of government. Take away as much government as you seem to want to, and you'll find yourself with less freedoms, not more. And those marginalized individuals that you are so quick to throw under the bus? Their individual freedoms are infringed all the time!
Well, it's not a simulator for one. You're flying the drone just as much as you would fly a plane. Hell, it's probably harder if you're not getting tactile feedback. Who cares where your ass is sitting?
If you scroll down three sentences, you'll see a chart where, by golly! state police workers have the second highest rate of workplace injury, surpassed only by (state government) nursing! If you look at days away from work rate, job transfer or restriction (DART), it moves to fourth behind state and local nursing and local fire departments. Construction makes the list - it's close to policing for injury rate and DART, but the RATE is smaller for both measures. There's just a heck of a lot more construction workers that police officers, so total injuries are very high.
Television, video games, beer, and anything else potentially distracting to poorly performing students should be illegal too!
A coin flip is very different from what we usually mean by a judgment. Would you trust a human soldier that makes random choices about what to do in life-or-death situations? Ideally, they should be able to explain their reasoning / training in each situation, and if the decision does not hold up to retrospective scrutiny, should be held accountable for their actions and/or should adjust their reasoning used in future situations. "Judgement" with a large component of randomness involves no reasoning, and so I'm skeptical of it's usefulness in life-or-death situations (and whether it should be called "judgement" at all).
Pretty sure that's not the kind of "judgement" that we want our military using, whether human or robot!
Historically, people have had a broad tendency to be ruled by unelected tyrants. Clearly people in today's democratic societies are just yearning for some other form of tyranny to fill that gaping hole in their hearts.
"Reason" is good, but too often gets conflated with "Rational self-interest" in discussions of morality. In fact, emotions and irrationality seem to help us become more predictable agents and get around some of the sticking points of rational self-interest. Here's a paper on the evolutionary advantage of the emotional dimension of moral agents in societies: http://www.cass.city.ac.uk/__d...
That's well and good (well not good, bad), but Tibbit's "solution" addresses the problem with text bans that you've posted. Unless you've can point out specific problems with his solution, it seems like it's worth think about and testing. Texting while driving is a big problem. If texting while driving bans are also a big problem, then some other solution is needed. Saying "reality is messy" to dismiss proposed solutions is slightly useful to make us think about potential problems, but is really the laziest form of criticism.
I have a problem with people whining about not letting politicians and bureaucrats run roughshod over our rights while shrugging about corporations and individuals run roughshod all over our rights. There's gotta be some balance, man. Yes I'm still waiting for my damn GE recall.
I'm amazed by how often these "MOOCs can't replace traditional classroom" arguments come up. Well no shit! The MOOC formats provide broad access and logistical ease for self-learning better than any previously existing educational technology, but online interactions are still no match for personal assistance. Most MOOC sites realize this - they're not trying to supplant traditional education, they're trying to supplement it. I fail to see how the MOOC revolution is over when there's still plenty of opportunity to fill in gaps between traditional universities and massive free sites like Coursera (which is exactly what the author's site is doing).
The Golden Fleece awards were mostly a load of anti-intellectual bullshit that had no comprehension of how mug applied science relies on basic science.