The easiest way for a typical person to get a solidly above-average salary is an engineering degree or something similar like computer science, physics, or plain old math.
It is not financially rewarding to teach math, and there is no way to fix this.
Money won't fix the problem. It would, except that you can't do what you need to do with it.
Getting bright people to enter teaching would require that you convince them that they'll be able to get a solid income for decades into the future. In the cheap places that means $60000. It's probably $150000 in Silicon Valley. Even if you could offer that today, you couldn't convince them that it would always be offered.
The money has to come from somewhere. There are a lot more teachers than NFL quarterbacks. This isn't a matter of few dollars per taxpayer. It's simply massive. My rough estimate is $4000 to $10000 per family per year, rather than the $2000 to $3000 right now.
You can't pay some teachers more than others, except based on seniority. The union won't let it happen. This means you can't just pay extra for math teachers, you can't pay extra for the teachers that most help students get better test scores, and you can't pay extra to a new teacher who happens to be wonderful.
Politically, you can't purposely give better teachers to the better students. You can't do that with equipment either. You thus have to waste money where it won't do any good.
Also note that you can't discipline a student (or their parents) in any effective way, and you can't even get rid of students who are disruptive and destructive. Besides directly ruining the educational environment, this is yet another reason for highly capable people to avoid the teaching profession.
Suppose we put aside a huge chunk of money for a period of 40 to 80 years. We then decide to use it. (as you're suggesting for social security)
That works if your definition of "huge" is on a personal scale. You can invest over the years, then make use of the money.
That doesn't work if your definition of "huge" is on a national scale. The economy reacts when you suddenly want to use the money. When a bunch of unproductive (elderly) people suddenly spend money, you have a problem. You might as well just be running the printing presses to make more money.
The new new math really is a democrat/libral thing.
To hide the fact that certain unfairly favored groups of people (darker skin or non-English) frequently have trouble with math, we make math into some kind of feel-good fuzzy mess with no right answers.
We ask math questions that can be answered only with guess-and-check methods, at least without fancy PhD-level stuff. We ask kids to write about their "favorite" number. We don't teach how to really manipulate numbers. We have kids using 1st-grade tools like counting sticks in middle school or even high school. We pretend to teach advanced stuff (statistics, fractals, vector calculus) with a treatment that is superficial, incomplete, and confusing.
Typically this is called "constructionism" or "constructivism", but proponents of those ideas often try to wiggle away when this is pointed out. (claiming that the new-new-math people are doing it wrongly or are otherwise not real)
HR drones are mainly a problem prior to the interview. BTW, I generally bypass them.
At every place I've ever interviewed, my first face-to-face contact after the receptionist was a technical person. (usually the potential manager, sometimes a potential co-worker)
The receptionist asks who you are there to see, calls them, they come get you, etc.
To a certain degree, I actually want to get filtered out by the people that expect a suit. I don't think this has happened to me, but it wouldn't be a bad thing. Such places would be unpleasant to work at.
BTW, it's not a plain white Hanes T-shirt with holes and armpit stains, nor one that advertizes beer!
Once this is better than the **average** driver, it will be required. (by law or to get insurance)
The death rate for idiots will definitely go down. It's much less clear what will happen with skilled drivers. They don't have to face the idiots, but they can't avoid road problems either.
You have to reverse engineer the **encoding** of the data. The data isn't sitting there in any normal way; it is scrambled and checksummed in multiple undocumented ways.
Obviously you must remove the heads. You'd use a magnetic force microscope. This can give 10 nm resolution. It's terribly horribly awful slow.
Glass platters look just like metal ones. That have a mirror-like metal coating on them.
When a glass platter breaks, it isn't like breaking normal glass. You don't get a few big chunks and a few chips. You get zillions of needle-like shards. They fly everywhere.
6 months is about the bare minimum. 2 years would be better.
There is a clean-room effort. There is the actual reading of the platters, which is damn slow when you can't use the original head. (since oversampling is needed, this could be weeks or months!) There is the reverse engineering of the insanely complicated encoding. There is the writing of software to fly a virtual head over a virtual disk.
Yes, it can be done. No, it isn't affordable unless the drive belonged to Osama. The effort would require dozens of engineers who already have experience doing the job. You're looking at a price tag of 5 to 50 million dollars.
Linus named it Freax. An FTP site maintainer at ftp.funet.fi chose Linux when offering Linus some space, creating an empty directory by that name. Linus put his files there, and thus the OS became Linux.
Some way for X to detect that there is no way for a window to fit on the screen and add some scrolly bars to it to make everything accessible. Perhaps it's purely the fault of the window manager or library though and not X, or maybe it's both?
The original FVWM ("Feeble Virtual Window Manager") did this. FVWM is still a rather nice window manager, assuming you don't mind editing ~/.fvwmrc to adjust it.
It was found that renaming quake.exe to quack.exe would affect performance. The reason is that the driver purposely degrades the quality for stuff that is used in benchmarks. This is dishonest, and it is a filthy hack. It's damn obvious why video drivers are a major cause of crashes; they dig around in kernel memory (totally undocumented) to enable dirty hacks.
They effectively do have governmental authority, since many parts government will refuse to buy anything not UL listed.
In any case, I was responding to this:
Studying the safety and effectiveness of drugs is far better left to the private sector. Just look at the effectiveness of the Underwriters' Laboratories, versus the useless work rules promulgated by OSHA.
As should be obvious, the UL is full of crap. They are not an example of non-government being better than government; they are at least as awful. If anything they are worse, since you can't lobby congress to fix them.
Consider the OLPC XO as an example. After much arguing, OLPC managed to talk the UL down to just the very silly warnings that appear when you shut down the laptop.
> In terms of total runtime, the common case is that the Javascript is inner loop code.
Indeed, and the idea is that once you run a loop a few times you jit it.
Ah, but you've already burned up lots of time in the interpreter. Translating is not (should not be) much slower than interpreting. If the Javascript executes once, pure JIT is nearly as fast as interpreting. If the Javascript executes more than once, pure JIT is better.
> Even if the JIT translator is slower than a few passes through the interpreter, why are > you caring so much about code that barely matters?
Because there is memory overhead to jitting everything, even things you don't hit.
Uh, why do you have any significant memory overhead? Even badly optimized code should be pretty compact.
Also, jitting evething (standard function-level jit) means you have to actually keep the function calls. Tracing effectively inlines everything.
Pretty much, yes. I'm suspecting that they made some sort of error in their JIT implementation. Having done optimization work on a JIT before, I happen to know that an interpreter can be bad. A decent JIT runs **very** fast.
JIT until the next branch, execute, repeat. Careful profiling of the JIT result can help to avoid trouble.
Tracemonkey is not optimized for the common case. Just the opposite is true, nearly.
In terms of total runtime, the common case is that the Javascript is inner loop code. You fail to immediately JIT this; instead you waste time in the interpreter. Even if the JIT translator is slower than a few passes through the interpreter, why are you caring so much about code that barely matters? Run it through the JIT, avoiding the need to even track if it is a hot spot or not. By doing this, you can totally eliminate the interpreter. Elimination of the interpreter means that you can make the JIT even faster.
Finally, a hint about the type inference issue:
Hook your JITed chunks of code up to the variables that they depend on. If the variable type changes, invalidate the chunks of code that access it. This means you only check for type change in the code that can cause type change, rather than in the code that merely reads the variables.
Kill adults who can not read. Kill adults who can not do basic math.
Prohibit birth control. Require all fertile adults to report for sperm donation or insemination if they go 3 years without a birth.
From age 5 to 12, test kids several times a year. Those who are bottom-ranked for several tests in a row get executed. You set the bar such that the total cull is well above 50%.
Slashdot Mirror
It is financially rewarding to learn math.
The easiest way for a typical person to get a
solidly above-average salary is an engineering
degree or something similar like computer science, physics, or plain old math.
It is not financially rewarding to teach math,
and there is no way to fix this.
Money won't fix the problem. It would, except
that you can't do what you need to do with it.
Getting bright people to enter teaching would
require that you convince them that they'll be
able to get a solid income for decades into the
future. In the cheap places that means $60000.
It's probably $150000 in Silicon Valley. Even
if you could offer that today, you couldn't
convince them that it would always be offered.
The money has to come from somewhere. There are
a lot more teachers than NFL quarterbacks. This
isn't a matter of few dollars per taxpayer. It's
simply massive. My rough estimate is $4000 to
$10000 per family per year, rather than the
$2000 to $3000 right now.
You can't pay some teachers more than others,
except based on seniority. The union won't
let it happen. This means you can't just pay
extra for math teachers, you can't pay extra
for the teachers that most help students get
better test scores, and you can't pay extra to
a new teacher who happens to be wonderful.
Politically, you can't purposely give better
teachers to the better students. You can't do
that with equipment either. You thus have to
waste money where it won't do any good.
Also note that you can't discipline a student
(or their parents) in any effective way, and
you can't even get rid of students who are
disruptive and destructive. Besides directly
ruining the educational environment, this is
yet another reason for highly capable people
to avoid the teaching profession.
Suppose we put aside a huge chunk of money for
a period of 40 to 80 years. We then decide to
use it. (as you're suggesting for social security)
That works if your definition of "huge" is on
a personal scale. You can invest over the years,
then make use of the money.
That doesn't work if your definition of "huge"
is on a national scale. The economy reacts when
you suddenly want to use the money. When a bunch
of unproductive (elderly) people suddenly spend
money, you have a problem. You might as well
just be running the printing presses to make
more money.
The new new math really is a democrat/libral thing.
To hide the fact that certain unfairly favored
groups of people (darker skin or non-English)
frequently have trouble with math, we make math
into some kind of feel-good fuzzy mess with no
right answers.
We ask math questions that can be answered only
with guess-and-check methods, at least without
fancy PhD-level stuff. We ask kids to write
about their "favorite" number. We don't teach
how to really manipulate numbers. We have kids
using 1st-grade tools like counting sticks in
middle school or even high school. We pretend
to teach advanced stuff (statistics, fractals,
vector calculus) with a treatment that is
superficial, incomplete, and confusing.
Typically this is called "constructionism" or
"constructivism", but proponents of those ideas
often try to wiggle away when this is pointed out.
(claiming that the new-new-math people are
doing it wrongly or are otherwise not real)
abortion, gay agenda, gun confiscation, etc.
She's terrifying to republicans.
Her support for bunsiness is totally centered
around the entertainment industry (movies, music).
That's generally a democrat thing.
HR drones are mainly a problem prior to the interview. BTW, I generally bypass them.
At every place I've ever interviewed, my first face-to-face contact after the receptionist was a technical person. (usually the potential manager, sometimes a potential co-worker)
The receptionist asks who you are there to see, calls them, they come get you, etc.
To a certain degree, I actually want to get filtered out by the people that expect a suit. I don't think this has happened to me, but it wouldn't be a bad thing. Such places would be unpleasant to work at.
BTW, it's not a plain white Hanes T-shirt with holes and armpit stains, nor one that advertizes beer!
Shaving is unmanly. You want to look like a boy?
Most of the great hackers have beards. Ritchie, Stallman, Cox...
A bare chin (and suit even! WTF???) screams "marketdroid". It means somebody who does Powerpoint. (probably eats tofu even!)
Your goal should be to attain greybeard status.
I go to interviews equipped with full beard and non-dress shirt. It may even be jeans and T-shirt. It works most excellently.
Some places are sold out, but yes you can get one.
Try the Canadian site.
Once this is better than the **average** driver,
it will be required. (by law or to get insurance)
The death rate for idiots will definitely go down.
It's much less clear what will happen with skilled
drivers. They don't have to face the idiots, but
they can't avoid road problems either.
You have to reverse engineer the **encoding** of the data. The data isn't sitting there in any normal way; it is scrambled and checksummed in multiple undocumented ways.
Obviously you must remove the heads. You'd use a magnetic force microscope. This can give 10 nm resolution. It's terribly horribly awful slow.
They eat normal protien. Bad protien is both offspring and a waste product.
Glass platters look just like metal ones.
That have a mirror-like metal coating on them.
When a glass platter breaks, it isn't like
breaking normal glass. You don't get a few
big chunks and a few chips. You get zillions
of needle-like shards. They fly everywhere.
6 months is about the bare minimum.
2 years would be better.
There is a clean-room effort. There is the actual reading of the platters, which is damn slow when you can't use the original head. (since oversampling is needed, this could be weeks or months!) There is the reverse engineering of the insanely complicated encoding. There is the writing of software to fly a virtual head over a virtual disk.
Yes, it can be done. No, it isn't affordable unless the drive belonged to Osama. The effort would require dozens of engineers who already have experience doing the job. You're looking at a price tag of 5 to 50 million dollars.
Linus named it Freax. An FTP site maintainer
at ftp.funet.fi chose Linux when offering Linus
some space, creating an empty directory by that
name. Linus put his files there, and thus the OS
became Linux.
Racist black people can lie to pollsters
about race-related opinions.
Some way for X to detect that there is no way for a window to fit on the screen and add some scrolly bars to it to make everything accessible. Perhaps it's purely the fault of the window manager or library though and not X, or maybe it's both?
The original FVWM ("Feeble Virtual Window Manager") did this. FVWM is still a rather nice window manager, assuming you don't mind editing ~/.fvwmrc to adjust it.
It was found that renaming quake.exe to quack.exe
would affect performance. The reason is that the
driver purposely degrades the quality for stuff
that is used in benchmarks. This is dishonest, and
it is a filthy hack. It's damn obvious why video
drivers are a major cause of crashes; they dig
around in kernel memory (totally undocumented) to
enable dirty hacks.
Open Source fixes this problem automatically.
They effectively do have governmental authority,
since many parts government will refuse to buy
anything not UL listed.
In any case, I was responding to this:
Studying the safety and effectiveness of drugs is far better left to the private sector. Just look at the effectiveness of the Underwriters' Laboratories, versus the useless work rules promulgated by OSHA.
As should be obvious, the UL is full of crap. They are not an example of non-government being better than government; they are at least as awful. If anything they are worse, since you can't lobby congress to fix them.
Those people are why we have the idiot warnings.
Consider the OLPC XO as an example. After much
arguing, OLPC managed to talk the UL down to
just the very silly warnings that appear when
you shut down the laptop.
Here it is, with "fixed" wording:
http://dev.laptop.org/~cscott/ul_warning.png
No.
1. have a vet remove some meat from the cow
2. put the meat in a freezer for 8 years
3. test the cow (now 8 to 10 years old)
4. if the test passes, ship the meat
> In terms of total runtime, the common case is that the Javascript is inner loop code.
Indeed, and the idea is that once you run a loop a few times you jit it.
Ah, but you've already burned up lots of time in the interpreter. Translating is not (should not be) much slower than interpreting. If the Javascript executes once, pure JIT is nearly as fast as interpreting. If the Javascript executes more than once, pure JIT is better.
> Even if the JIT translator is slower than a few passes through the interpreter, why are
> you caring so much about code that barely matters?
Because there is memory overhead to jitting everything, even things you don't hit.
Uh, why do you have any significant memory overhead? Even badly optimized code should be pretty compact.
Also, jitting evething (standard function-level jit) means you have to actually keep the function calls. Tracing effectively inlines everything.
True, but modern processors handle call/ret well.
Pretty much, yes. I'm suspecting that they made
some sort of error in their JIT implementation.
Having done optimization work on a JIT before,
I happen to know that an interpreter can be bad.
A decent JIT runs **very** fast.
JIT until the next branch, execute, repeat.
Careful profiling of the JIT result can help
to avoid trouble.
Having worked on emulators a bit...
Tracemonkey is not optimized for the common
case. Just the opposite is true, nearly.
In terms of total runtime, the common case
is that the Javascript is inner loop code.
You fail to immediately JIT this; instead
you waste time in the interpreter. Even if
the JIT translator is slower than a few passes
through the interpreter, why are you caring
so much about code that barely matters? Run
it through the JIT, avoiding the need to even
track if it is a hot spot or not. By doing
this, you can totally eliminate the interpreter.
Elimination of the interpreter means that you
can make the JIT even faster.
Finally, a hint about the type inference issue:
Hook your JITed chunks of code up to the variables
that they depend on. If the variable type changes,
invalidate the chunks of code that access it.
This means you only check for type change in the
code that can cause type change, rather than in
the code that merely reads the variables.
Human stupidity can be fixed. Simple:
Kill adults who can not read. Kill adults who
can not do basic math.
Prohibit birth control. Require all fertile adults
to report for sperm donation or insemination if
they go 3 years without a birth.
From age 5 to 12, test kids several times a year.
Those who are bottom-ranked for several tests in
a row get executed. You set the bar such that the
total cull is well above 50%.
Average family: 8 to 16 kids born, 2 to 4 survive
Perhaps it will have to ship with Vista.