A single computer with redundancy in the OS gets killed by a single well-placed ray. On the other hand, the redundant ARM array consists of physically separate machines, even if one gets permanently fried others will keep running. And I picked ARM because it offers tiny power draw while delivering good performance. The power budget of a deep-space craft is really tight.
I'd instead go with a RAIA -- a horde of off-the-shelf ARMs. Within the power budget of a single 20MHz 80386 you can fit nine 2GHz SoCs. Have them vote -- there's no way every single of them gets hit by a ray within a time slice. Periodically, resync their memory (especially when the vote disagrees). A 2GHz machine can take quite an overhead while doing the work previously done by a 20MHz one...
This assumes the 386 was alone -- it was at least doubled or tripled. So if you don't need 18x or 27x redundancy, you can do something else with the extra power.
But let's assume you do want that 27x redundancy. It's still a two orders of magnitude speed boost, and that's assuming same speed clock-to-clock. Which is wrong, as 386 timings were downright scary. Especially in floating point, with a hundred or more clock cycles per instruction. Modern ARM on the other hand includes a vectorized FPU...
the only purpose of having a VP O' Diversity is to ensure sufficient hues of dermis and the correct ratio of penises/vaginas
Actually, they don't even care about penises/vaginas, but about whether you claim to have one. Thus, apply lipstick, get a boob job, for a better cultural fit also dye your hair purple, and you're all set.
You see, in the '30s your skin had to be white and your genitals unmutilated. The preferred skin color and shape of genitals have changed, but the whole concept stays the same.
Your flaw is assuming that an objective and unbiased ranking of candidates is possible. It's not. There are no perfect indicators of ability. Test scores, grades, connections, etc - all have flaws.
Then try adding a random noise to the scores. Unless the noise itself is drastically skewed, picking the top N values of such noisy test scores will still produce a better result (measured by the sum of our hidden ability values) than if you picked with some artificial bias.
All are influenced by the resources you and your family have.
You mean, a candidate who was raised by a family of honest programmers is not likely to give a better value for the company than one who was raised by a single permanently drunk mother who told her kids to disdain "literate faggots", and who first learned to code only in a course teached in prison?
A company can organize a charity to help kids who'd otherwise glorify "thug life" (be they blacks from a Detroit ghetto or whites from Kosciuszki street in my home town), but that's extremely unlikely to turn them into top engineers in short or medium term.
It's like the NBA scouts who go to China or Africa searching for seven footers,
Have you noticed what proportion of NBA players are black? And Chinese? In 2014, there was a _single_ person described as of asian ancestry, in most other years there was none. If the ability was indeed equal, a racial group who makes ~60% of world and 5% of US population wouldn't make only 0.2% of NBA players while a group with 12% of US population produces ~70% of players.
This also gives a crucial observation: that even though blacks have a significant genetic edge over whites, they are still far from 100% of NBA players. As James Damore mentioned, ability scores have a significant overlap, the mean doesn't matter here. A white NBA player who did manage to get above the threshold is no worse than a black player, the ratio of whites is just significantly smaller than that of the general population.
See what happens when a repetitively smart Trump voter outs themselves.
Damore is strongly left-wing, he merely dared to be not orthodox enough.
I'm outright scared by modern US-style politics (most western countries have a variant of this): you see nothing but echo chambers, both left and right wingers carefully avoid places where they could be not in majority. And both positions have became so extreme that applying even basic reason is enough to rip them to shreds -- but either side will instead consider you to be a heretic instead of entertaining the idea that perhaps their religion might be unsound.
Now add another factor to your simulation: Diverse teams are more productive than homogeneous teams.
Now this is something new to me. This is quite astonishing -- with all the shouting from both sides, I'd expect someone to have raised this argument long ago, but no, I hear nothing but buzzwords repeated like a religion. And a quick search shows some plausible-looking sources. Thank you, a rational argument that I might be wrong is always welcome.
Exact numbers are tough to find
Yeah...
What you'll find is that unless your means or variances are *very* different for the populations, the diversity "boost" dominates
Actually, it seems that the means/variances are indeed extremely different. For example, gender: among the top 1000 kernel committers, there's only 8 women (both "8" and "1000" include only those whose first names reveal the gender). Those 8 are no worse than their peers (see my explanation in this thread why worse mean is indistinguishable from rarity), but such a massive disparity can't be dismissed. It is said intelligence-by-gender is actually a matter of variance instead of mean (ie, men are more likely to be either geniuses or idiots), but that doesn't help the long tail.
Likewise, race. The Jews make 27% of Nobel winners in scientific fields despite being only 0.2% of world population. US Black men (7% of population) commit over half of murders. I suspect both of these to be overwhelmingly caused by culture rather than genetics, but those are highly correlated.
Google having over 20% female engineers compared to only 0.8% top kernel contributors being female suggests the bias is way above any plausible diversity boost. This is actively harmful to women who are genuinely skilled -- they are to be likely to be disregarded by their peers as "affirmative action hires".
Thus, I won't concede my argument yet, but you poked quite a hole in it.
You're conflating assistance for low-scorers with picking the cream of the crop. A wheelchair-bound person is not going to win a running competition. Heck, he won't even win an endurance car race -- the first moment any routine maintenance needs to be done, he'll be stuck with a trivial malfunction that an able-bodied driver would fix in minutes.
As for a girls-only CS class, yes, it is discrimination. Gender doesn't make an individual worse, it may at most affect the average of a population. That is, even though men on the average have much higher upper body strength, I wouldn't want to pick a fight with a female lumberjack.
It is counterintuitive, but there's no or only a slight difference (depends on the distribution) between a group being less numerous, or having less skill.
We don't care about the mean, or the bulk of the population -- only about the tail end. And tail ends of distribution D(x) tend to be similar between D(x-a) vs D(x)/b -- for some distributions like exponential exactly equal, for some close enough to be hard to distinguish on real noisy data.
And we don't care about the number of graduates either, as that is affected by artificial programs. A more telling metric is eg. the number of women among top 1000 kernel contributors, a sample of Debian package maintainers, etc. I've did the legwork and counted kernel contributors with gender-obvious names (I'm familiar with western and slavic first names), among the top 1000 commiters whose first name reveals gender, there's _8_ women. Yes, only 8 out of 1000!
But those 8 are no worse than their peers. Not only rarity is indistinguishable from low avg skill, low avg skill in indistinguishable from rarity! Thus, if you apply equal fair standards, you'll get a smaller proportion of top achievers than the general population would imply, but individuals from group B who do qualify above the threshold, are no worse than individuals from group A.
How is "reverse" discrimination not discrimination?
You can make a simulation: generate a number of individuals with assigned skill scores, by a given distribution. Generate also a population B, with a same or similar distribution but a lower mean (or alternatively, same mean and lower variance, etc). Use any bell-curve distribution (such as normal) with no cap (so D&D-like 3d6 is out).
Now, pick N top scorers from the combined population. Compare the same with various kinds of racism:
only the "better" group A (exclusive traditional racism)
a bonus for group A (traditional racist preference)
a bonus for group B (affirmative action racism)
racial quotas
You'll see that any kind of racism hurts the person doing the discrimination as he gets an unoptimal result. You can also notice that affirmative action is drastically more harmful than traditional racism. Both are bad, though, and there's a big gain for being race- (and gender-, etc) blind.
No one keeps you from using a sane init (and the whole bunch of daemons that systemd craps over) on Debian. Just in case you'd wonder why Red Hat used to be the dominant distribution a decade ago but is not any more.
You forgot tablets. No, not ones with an ARM and a touchscreen. Ones written in cuneiform and baked in clay.
Because that's the oldest recorded repeat of this complaint. I do bet, though, that grandpa Uuk wasn't happy about that new-fangled "fire" thingy, either.
Has per-tab audio muting (and is muted by default).
Firefox has this covered, by dropping non-PulseAudio sound. I don't watch TV or similar crap in a browser, and when someone insists I view a YouTube vide, I can click "download" and view it in a sane player. No, that "apulse" hack doesn't work right -- it produces a loud pop every time an unrelated program starts sound -- but fortunately I can live without.
So the no sound bug ends up an accidental feature:)
/usr/bin/python is not going to ever point to python3 in Debian, at least according to the maintainer. Unfortunately, there are distros which have done this step.
Python 2 and Python 3 are similar but different languages, akin to Perl 5 and Perl 6. As they're not supposed to be compatible, changing the meaning of the hashbang is a recipe for disaster.
EOL crap like Java or.NET first. Flash is nowhere as insecure or bloated as these two. It at least tries to contain bad code, even if it does an imperfect job. So be wary of Flash programs, don't give them Internet access and provide additional containment when running ones from a dodgy source -- but a Flash game from Newsgrounds is safer than a regular.exe from a random website. And unlike AAA games, it's not certain to include malware.
This default position is easily observable, what you propose runs contrary to obvious observations. This doesn't mean the latter is wrong (continental drift, Newton vs Einstein), but the burden of proof is on the side making extraordinary claims.
At this point, it looks like I'm wasting my time responding, but I'll bite.
First, extraordinary claims require extraordinary proof. Since what you propose is contrary to what an ordinary person can easily perceive (and you don't claim something that requires specialized tools, like a big telescope), the burden of proof is on you. So here, please provide some proof why there would be more than two genders, or why the number of genders would be different than the number of sexes. There's some (dubious) evidence that the feeling of a gender might be flipped during prenatal development, but that still conserves the number at two
Second, for a theory to be scientific, it needs to provide testable predictions. If there are more than two genders, then what? What should be the effects, how a purported third gender would change the person's behaviour?
Third, a field whose papers have an extremely high bullshit quotient can be dismissed wholesale. The definition of "bullshit quotient" is typically: "flip a number of statements with a yes/no value to the opposite, and give such a modified paper to a bunch of experts in the field. If they fail to spot that something is wrong, that paper's bullshit quotient is at least the number of statements we flipped divided by the number of statements flippable". Glowing reviews by multiple reviewers of the "Conceptual Penis", a paper that was partially machine-generated and partially hand-edited for maximum nonsense, plus their suggestion to promote that paper to a more prestigious journal, which was then accepted, suggests a quite overwhelming value of the bullshit quotient.
Thus, "gender studies" are not science, they're a religion.
even he seems to suggest no edits for purely cosmetic reasons. The safe money is that they specifically ban any edits that aren't correcting life-threatening conditions
And even safer money says a huge industry of cosmetic edits will pop up before life-threatening conditions even start regulatory approval trials. You'll just need to fly out of the country to do so.
Slashdot Mirror
A single computer with redundancy in the OS gets killed by a single well-placed ray. On the other hand, the redundant ARM array consists of physically separate machines, even if one gets permanently fried others will keep running. And I picked ARM because it offers tiny power draw while delivering good performance. The power budget of a deep-space craft is really tight.
I'd instead go with a RAIA -- a horde of off-the-shelf ARMs. Within the power budget of a single 20MHz 80386 you can fit nine 2GHz SoCs. Have them vote -- there's no way every single of them gets hit by a ray within a time slice. Periodically, resync their memory (especially when the vote disagrees). A 2GHz machine can take quite an overhead while doing the work previously done by a 20MHz one...
This assumes the 386 was alone -- it was at least doubled or tripled. So if you don't need 18x or 27x redundancy, you can do something else with the extra power.
But let's assume you do want that 27x redundancy. It's still a two orders of magnitude speed boost, and that's assuming same speed clock-to-clock. Which is wrong, as 386 timings were downright scary. Especially in floating point, with a hundred or more clock cycles per instruction. Modern ARM on the other hand includes a vectorized FPU...
the only purpose of having a VP O' Diversity is to ensure sufficient hues of dermis and the correct ratio of penises/vaginas
Actually, they don't even care about penises/vaginas, but about whether you claim to have one. Thus, apply lipstick, get a boob job, for a better cultural fit also dye your hair purple, and you're all set.
You see, in the '30s your skin had to be white and your genitals unmutilated. The preferred skin color and shape of genitals have changed, but the whole concept stays the same.
Your flaw is assuming that an objective and unbiased ranking of candidates is possible. It's not. There are no perfect indicators of ability. Test scores, grades, connections, etc - all have flaws.
Then try adding a random noise to the scores. Unless the noise itself is drastically skewed, picking the top N values of such noisy test scores will still produce a better result (measured by the sum of our hidden ability values) than if you picked with some artificial bias.
All are influenced by the resources you and your family have.
You mean, a candidate who was raised by a family of honest programmers is not likely to give a better value for the company than one who was raised by a single permanently drunk mother who told her kids to disdain "literate faggots", and who first learned to code only in a course teached in prison?
A company can organize a charity to help kids who'd otherwise glorify "thug life" (be they blacks from a Detroit ghetto or whites from Kosciuszki street in my home town), but that's extremely unlikely to turn them into top engineers in short or medium term.
It's like the NBA scouts who go to China or Africa searching for seven footers,
Have you noticed what proportion of NBA players are black? And Chinese? In 2014, there was a _single_ person described as of asian ancestry, in most other years there was none. If the ability was indeed equal, a racial group who makes ~60% of world and 5% of US population wouldn't make only 0.2% of NBA players while a group with 12% of US population produces ~70% of players.
This also gives a crucial observation: that even though blacks have a significant genetic edge over whites, they are still far from 100% of NBA players. As James Damore mentioned, ability scores have a significant overlap, the mean doesn't matter here. A white NBA player who did manage to get above the threshold is no worse than a black player, the ratio of whites is just significantly smaller than that of the general population.
See what happens when a repetitively smart Trump voter outs themselves.
Damore is strongly left-wing, he merely dared to be not orthodox enough.
I'm outright scared by modern US-style politics (most western countries have a variant of this): you see nothing but echo chambers, both left and right wingers carefully avoid places where they could be not in majority. And both positions have became so extreme that applying even basic reason is enough to rip them to shreds -- but either side will instead consider you to be a heretic instead of entertaining the idea that perhaps their religion might be unsound.
Now add another factor to your simulation: Diverse teams are more productive than homogeneous teams.
Now this is something new to me. This is quite astonishing -- with all the shouting from both sides, I'd expect someone to have raised this argument long ago, but no, I hear nothing but buzzwords repeated like a religion. And a quick search shows some plausible-looking sources. Thank you, a rational argument that I might be wrong is always welcome.
Exact numbers are tough to find
Yeah...
What you'll find is that unless your means or variances are *very* different for the populations, the diversity "boost" dominates
Actually, it seems that the means/variances are indeed extremely different. For example, gender: among the top 1000 kernel committers, there's only 8 women (both "8" and "1000" include only those whose first names reveal the gender). Those 8 are no worse than their peers (see my explanation in this thread why worse mean is indistinguishable from rarity), but such a massive disparity can't be dismissed. It is said intelligence-by-gender is actually a matter of variance instead of mean (ie, men are more likely to be either geniuses or idiots), but that doesn't help the long tail.
Likewise, race. The Jews make 27% of Nobel winners in scientific fields despite being only 0.2% of world population. US Black men (7% of population) commit over half of murders. I suspect both of these to be overwhelmingly caused by culture rather than genetics, but those are highly correlated.
Google having over 20% female engineers compared to only 0.8% top kernel contributors being female suggests the bias is way above any plausible diversity boost. This is actively harmful to women who are genuinely skilled -- they are to be likely to be disregarded by their peers as "affirmative action hires".
Thus, I won't concede my argument yet, but you poked quite a hole in it.
Definitely, more data is needed. More data!
You're conflating assistance for low-scorers with picking the cream of the crop. A wheelchair-bound person is not going to win a running competition. Heck, he won't even win an endurance car race -- the first moment any routine maintenance needs to be done, he'll be stuck with a trivial malfunction that an able-bodied driver would fix in minutes.
As for a girls-only CS class, yes, it is discrimination. Gender doesn't make an individual worse, it may at most affect the average of a population. That is, even though men on the average have much higher upper body strength, I wouldn't want to pick a fight with a female lumberjack.
It is counterintuitive, but there's no or only a slight difference (depends on the distribution) between a group being less numerous, or having less skill.
We don't care about the mean, or the bulk of the population -- only about the tail end. And tail ends of distribution D(x) tend to be similar between D(x-a) vs D(x)/b -- for some distributions like exponential exactly equal, for some close enough to be hard to distinguish on real noisy data.
And we don't care about the number of graduates either, as that is affected by artificial programs. A more telling metric is eg. the number of women among top 1000 kernel contributors, a sample of Debian package maintainers, etc. I've did the legwork and counted kernel contributors with gender-obvious names (I'm familiar with western and slavic first names), among the top 1000 commiters whose first name reveals gender, there's _8_ women. Yes, only 8 out of 1000!
But those 8 are no worse than their peers. Not only rarity is indistinguishable from low avg skill, low avg skill in indistinguishable from rarity! Thus, if you apply equal fair standards, you'll get a smaller proportion of top achievers than the general population would imply, but individuals from group B who do qualify above the threshold, are no worse than individuals from group A.
How is "reverse" discrimination not discrimination?
You can make a simulation: generate a number of individuals with assigned skill scores, by a given distribution. Generate also a population B, with a same or similar distribution but a lower mean (or alternatively, same mean and lower variance, etc). Use any bell-curve distribution (such as normal) with no cap (so D&D-like 3d6 is out).
Now, pick N top scorers from the combined population. Compare the same with various kinds of racism:
You'll see that any kind of racism hurts the person doing the discrimination as he gets an unoptimal result. You can also notice that affirmative action is drastically more harmful than traditional racism. Both are bad, though, and there's a big gain for being race- (and gender-, etc) blind.
So you're not using Debian
No one keeps you from using a sane init (and the whole bunch of daemons that systemd craps over) on Debian. Just in case you'd wonder why Red Hat used to be the dominant distribution a decade ago but is not any more.
You forgot tablets. No, not ones with an ARM and a touchscreen. Ones written in cuneiform and baked in clay.
Because that's the oldest recorded repeat of this complaint. I do bet, though, that grandpa Uuk wasn't happy about that new-fangled "fire" thingy, either.
Alas, old versions don't get security support, and browsers are among (if not THE) most sensitive to security vulnerabilities pieces of software.
Has per-tab audio muting (and is muted by default).
Firefox has this covered, by dropping non-PulseAudio sound. I don't watch TV or similar crap in a browser, and when someone insists I view a YouTube vide, I can click "download" and view it in a sane player. No, that "apulse" hack doesn't work right -- it produces a loud pop every time an unrelated program starts sound -- but fortunately I can live without.
So the no sound bug ends up an accidental feature :)
How long until this is offered through an API, to detect whether you're paying attention to ads?
Day 0, duh. That's the primary purpose of this feature.
You're forgetting the one quality without which an ad can't be unobtrusive.
Except that Python's upstream concurs.
/usr/bin/python is not going to ever point to python3 in Debian, at least according to the maintainer. Unfortunately, there are distros which have done this step.
Python 2 and Python 3 are similar but different languages, akin to Perl 5 and Perl 6. As they're not supposed to be compatible, changing the meaning of the hashbang is a recipe for disaster.
Java is being used on the server, not in the browser. Nobody uses applets anymore, and they are **already** being EOLed.
Yeah, we're only 1/3 of the way here. Applets are gone, but Dalvik/ART and Java-on-server still need to be purged off the face of the Earth.
Also, Flash being moribund is kind of a blessing here -- a game from 2010 or 2005 is pretty unlikely to use an exploit announced in 2017.
EOL crap like Java or .NET first. Flash is nowhere as insecure or bloated as these two. It at least tries to contain bad code, even if it does an imperfect job. So be wary of Flash programs, don't give them Internet access and provide additional containment when running ones from a dodgy source -- but a Flash game from Newsgrounds is safer than a regular .exe from a random website. And unlike AAA games, it's not certain to include malware.
... "professional" brewers do know exactly what's in their beer?
Technologies to analyze urine, even a horse's, are far more advanced and routine than those for beer.
This default position is easily observable, what you propose runs contrary to obvious observations. This doesn't mean the latter is wrong (continental drift, Newton vs Einstein), but the burden of proof is on the side making extraordinary claims.
At this point, it looks like I'm wasting my time responding, but I'll bite.
First, extraordinary claims require extraordinary proof. Since what you propose is contrary to what an ordinary person can easily perceive (and you don't claim something that requires specialized tools, like a big telescope), the burden of proof is on you. So here, please provide some proof why there would be more than two genders, or why the number of genders would be different than the number of sexes. There's some (dubious) evidence that the feeling of a gender might be flipped during prenatal development, but that still conserves the number at two
Second, for a theory to be scientific, it needs to provide testable predictions. If there are more than two genders, then what? What should be the effects, how a purported third gender would change the person's behaviour?
Third, a field whose papers have an extremely high bullshit quotient can be dismissed wholesale. The definition of "bullshit quotient" is typically: "flip a number of statements with a yes/no value to the opposite, and give such a modified paper to a bunch of experts in the field. If they fail to spot that something is wrong, that paper's bullshit quotient is at least the number of statements we flipped divided by the number of statements flippable". Glowing reviews by multiple reviewers of the "Conceptual Penis", a paper that was partially machine-generated and partially hand-edited for maximum nonsense, plus their suggestion to promote that paper to a more prestigious journal, which was then accepted, suggests a quite overwhelming value of the bullshit quotient.
Thus, "gender studies" are not science, they're a religion.
even he seems to suggest no edits for purely cosmetic reasons. The safe money is that they specifically ban any edits that aren't correcting life-threatening conditions
And even safer money says a huge industry of cosmetic edits will pop up before life-threatening conditions even start regulatory approval trials. You'll just need to fly out of the country to do so.
Poland. Have yet to meet a single person who uses WhatsApp.