In the past decade, a lot of big firms with serious computing needs have been building huge off-site compute centers. Cheap power (a nearby power plant) and cheap cooling (a nearby river) tend to be the driving factors. Now and then you find a great site (mainly because the power plant will commit to providing lots of off-peak power,) and when you do, you often find a "facility" (think 100 ft underground, huge water-cooling system built) available for lease or sale nearby. Go figure, people have been doing this for a long time!
3) More products, services, and a stock split for Google. I was right about the first bits but that's like predicting sunset will come. The split didn't happen because I never realized how much cash Google was going to generate -- far more than they can even spend. So the item is wrong.
What does how much cash a company generates have to do with a stock split? Nothing (well, generating lots of cash makes a split slightly more, rather than less, likely.)
Does he think that a stock split somehow raises capital? He should stick to technology and avoid finance.
No. A stock option is worth $0 if expired. It has non-negative value if not expired, e.g. you can profit from owning an option even if the underlying price never reaches the strike price (that's called delta hedging.)
(Admittedly, the afterburners were ramjets, but that still isn't scramjet tech.)
Almost by definition, afterburners (fuel dumps after the turbine) are not ramjets (tubes with clever geometry that are engines.) Afterburners are an inefficient way to get a bit more thrust for a short time (e.g. takeoff, evasion,) ramjets are engines (that don't work until you have a fair bit of airspeed.)
You can mix ramjets and rockets (stuff the ramjet with solid fuel, use it as a booster to get to speed,) but ramjets and jets do not mix.
Largely because Americans only consider average houses to have a lifespan of 30 years or so (i.e. shorter than Britain, longer than Japan.) Obviously, we can and do build houses with much longer expected lifespans, but the typical house is essentially obsolete within 30 years: ceilings too low, rooms too small, underpowered electricity, no intranet, ugly layout due to incremental expansion, etc. The next thirty years may see a new obsolesence: energy wastage, too big, etc.
So, its mostly a cost thing combined with an abundance of resources! England gives houses names, we give them numbers:)
Basically, the programs do not beat the market, but lower the standard deviation (risk) involved in trading
Um, that IS beating the market: all investments are a combination of risk and expected reward (e.g. treasuries are low-risk and low-return, junk bonds are higher risk and higher return.) If you can reduce risk without reducing return, you can make buckets of money (people will line up outside your door wanting to give you capital.)
That said, don't trust any academic studies on this topic. There are probably only 50 people on the planet who actually deeply understand this topic, and they aren't talking. E.g. when I was running a hedge fund in the early 90s, most of the techniques described in the article were already widespread (yes, we had real-time newswire scrubbers running back then.)
As the article touches on, speed is important: a 5 millisecond advantage lets you make real money without needing to predict anything. The predictors do make excess profits, but most either don't understand the risks they are taking, or actively hide those risk from their investors. D.E.Shaw, mentioned in the article, basically blew itself up back in 1999 or so. LTCM did the same thing.
Some people (my 3 year old nephew) claim that Communists killed Jesus. This is patently false, because Jesus was killed and ressurected years before Communism was invented.
You're just cherry-picking. I'm sure there are lots of arguments that the communists killed Jesus, and you're only arguing with your 3 year-old nephew's argument.
This is reasonable description of the fundamental problem: some areas of scientific inquiry are of deep interest to some laymen because they provide support for certain cherished views (relativity, rocketry, and radiation are unimportant, but evolution, Laffer curves, genetic predisposition to homosexuality, global warming, etc, are fertile ground from which to construct a simple philosophy.)
The non-scientific pundit then takes a common sense idea (e.g. a bacteria can't give birth to a human, if taxes were 100% no one would work harder, a whirlwind in a junkyard doesn't make a 747,) grabs a few quotes from the large body of technical literature, and declares the whole field absurd.
Scientists try to respond using science, but it's mostly "he said, she said" to the layman. The pundits gleefully declare the scientists can't cite anything that refutes them (because scientists tend not to publish papers refuting batshit crazy ideas.)
This is not to say the scientific consensus in any field is right (I am skeptical of several fields in terms of predictions being made and overall synthesis,) but all science is vulnerable to misunderatanding by laymen when public opposition is natural (e.g. from preachers, oil companies, racist politicians.)
Often true, but not always. If you know your stuff and you prove it in other ways, the offers can come flooding in
Exactly. My ideal hire:
1) Knows CS. Algorithms, complexity, etc 2) Knows Software engineering. I.e. writing good code that solves the problem. 3) Is smart. 4) Could/can work in a group. 5) Will learn the business.
I've been on one day round trips to MIT (hoping for a candidate that hits 3/5,) brunches with personal recommedations (several good hires, including one for C++ based job who didn't know C++,) and random calls to people who published beautiful code on the net (some great hires.)
A degree is useful, but only as a foot in the door: we look for the people we want, HR does the paperwork, not the filtering.
Pretty much no one commits suicide in a tactical situation - they're too busy trying to fix things. This is not to say silly planners don't provide suicide pills -- they do, but an extra 1oz of weight isn't that bad, and it makes for good PR.
Actually, v1 (hereafter referred to as 'he') didn't write that there was no DNA in fingernails; he wrote that DNA was most reliably extracted from living cells.
Let's look at his original post:
He is probably well-aware that hair (minus a follicle) and fingernail clippings are both just keratin (chitin?) and contain no biological material or DNA of any sort.
Looks like you are so wrong it's just embarassing.
Organic extraction of aged human nail material yielded a sufficient quantity of DNA for successful mtDNA sequencing; however, STR analysis was unsuccessful.
Translation: we could find traces of DNA, but not enough to identify the person. To use something for identification in a criminal case requires that it be very reliable and very unique. In a test group of 15 they could not identify the owner in all cases. How are you going to pull that off when you are searching a 10,000 person database?
DNA is most reliably harvested from living cells. (try your google on that) There are no living cells in fingernails or hair strands. Only hair follicles and maybe some toejam.
Ok, fair enough, 8 year old fingernails aren't a good source source of DNA (d'oh - how much DNA do you expect to be left in an eight year old steak, for that matter.) Let's look at the authors' abstact of the paper that you quoted out of context: In order to process nails as evidence, a validation study was performed to demonstrate that DNA could be successfully extracted from human nail material. Organic extractions of DNA were performed on fresh fingernails, fresh toenails, and aged (~eight years old) fingernails. The isolated DNA was quantified and analyzed by polymerase chain reaction (PCR)-based testing. This study demonstrated that the extraction of fresh, human nail material yielded sufficient quantities of DNA for successful short tandem repeat (STR) analysis and mitochondrial DNA (mtDNA) sequencing. A random sampling of fingernails from 22 different individuals (15 females, 6 males and 1 unknown) was tested. Successful STR profiles were obtained from all fingernail specimens collected from male individuals. Full or partial STR profiles were obtained from the fingernail specimens collected from 10 of the 15 female individuals. All nails amplified for mtDNA produced PCR product; all products sequenced produced the correct mtDNA type. Organic extraction of aged human nail material yielded a sufficient quantity of DNA for successful mtDNA sequencing; however, STR analysis was unsuccessful.
Wow! They could do mtDNA sequencing on decade old fingernails, but STR analysis was unsuccessful. Fresh nails yielded plenty of DNA. Color me stunned! Perhaps you could explain again how this has anything to do with your claims that fingernails don't have DNA in them?
Where were you camping? I hope Alaska and not Antarctica - most kerosenes freeze in the -30 to -70 centigrade range. Some at merely -20 centigrade or so, but that's still really fucking cold!
I got.0236 in under ten seconds. I work with a lot of PhD mathematicians, and know most of them could probably have gotten 1 or 2 more significant digits in those 10 seconds.
Sure, we don't hire them for their ability to divide in their heads, but it's hard to conceive of hiring a mathematician who isn't able to do basic math (much like hiring a CS PhD who can't tell me how many cycles it takes to strcmp( "apple", "apples" ).)
a) usually politically driven, and if you voice the wrong opinion you will be disliked. b) driven by the incompetents - if they could do something profitable, they would be doing that. c) out of date by the time they are finalized.
There is no upside here because there is no magic standard that will make things better.
How would you feel if the network guys just needed to hack a small piece of your code? Surely you wouldn't mind if they checked out your source code and fiddled around?
Sounds reasonable to me - this is the approach we use at my firm. It's a rare week when I don't get mail saying someone has fixed or enhanced my code, then pushed it to production machines. It's a rare week when I don't do the same to someone else's code.
Works well when you make the effort to only hire good people, treat them like adults, and reward them according to their contributions.
Bah. A behavior as simple as protecting your tribe's corpses from opportunistic predators quickly ensures you generally don't get attacked by eagles. Once a raptor sees it doesn't get a free meal from attacking proto-humans, it quickly gives up.
Heck, burying your dead becomes a great advantage: predators gain nothing from killing your species, and soon seek prey that actually gives them food! Maybe human death rituals (e.g. burial, burning, leaving to vultures) got started because they ensured predators didn't profit from the death of the victim.
Well, they mostly make a record of things like tapestries. So, they lay out the item (often several hundred square feet,) then build what looks like a manually operatered 2-d x-y plotter over it. The camera and light system move in one axis across a bridge-like thing, and the bridge move a step in the other axis after each row-scan. All this takes place in a windowless room with the doors closed (to keep the light calibrated.)
Yep, next they piece the thousands of images together with custom panorama software. One problem is that gravity/temperate/humidity cause the object to change shape from minute to minute, so it's really a stitch and morph job (done by the Chudnovsky brothers with their homebrew supercomputer.)
Well, I'm just an F3 amateur, so I defer to my wife who shot 8x10 until recently. Her lab (they like to record the art they are working on) has mostly switched to digital now: that means scaffolding for the camera, etc, but it does give them the gigapixel images they want with a lot less post-processing hassle. For real hi-res stuff, having digital images as the basic inputs is proving to a big win.
Good point. Farming enabled much higher population densities, but at significant cost to the individuals: diet changed from 100+ items for some hunter-gatherers to around 2-5 staples for early agricultural communities. Lifespan and adult body size also usually decreased (probably due to worse nutrition and a better environment for nasty viruses.) Average work hours per day went up (to support the new ruling class.) Culture increased, but at a cost to the individual.
Well, I could have said he makes "spread" the fundamental idea, but that would really help anyone who hadn't read the article. As noted by many posters, spread is just a function of angle.
I doubt there are any unique insights from his approach: he's basically made angle the fundamental element of trig, rather than the more usual distance.
That said, it might be an interesting way to teach/explore trig. By doing away with the trig functions (which are just the distance->angle mappings,) he gets to solve many simple problems with just algebra and a final square root. Because the sqrt is explicit, this approach might give students a better mental model for trig: as things stand, most students just treat sin, cos, etc, as black boxes, and apply the rules (SOHCAHTOA) by rote.
I'd like to see how his method stacks up when applied to, say, an entire beginner level book or classic text, on geometry/trig.
If you want to be a code monkey, take vocational courses: learn Java and C++, learn about the buzzwords du jour (XML, SOAP. Ajax, whatever.)
If you actually want to be a serious programmer/designer, get a strong grounding in CS - that means data structures + algorithms, automata, numerics, compilers, OS design, etc. Know C++, Lisp, and a functional language.
So DZ doesn't want programmers who know how to write a compiler? Great, on his next big project, he'll wind up with a system with several embedded ad-hoc languages hidden away in it. Worse, he won't see anything wrong with that.
I've hired programmers for C++ environments who didn't know C++: if they're strong enough, they'll be up to speed in a few weeks.
At my current job, we want to see core CS courses. We have to teach/retrain all our hires anyway, so we might as might hire people with a strong theoretical foundation in CS.
Slashdot Mirror
In the past decade, a lot of big firms with serious computing needs have been building huge off-site compute centers. Cheap power (a nearby power plant) and cheap cooling (a nearby river) tend to be the driving factors. Now and then you find a great site (mainly because the power plant will commit to providing lots of off-peak power,) and when you do, you often find a "facility" (think 100 ft underground, huge water-cooling system built) available for lease or sale nearby. Go figure, people have been doing this for a long time!
3) More products, services, and a stock split for Google. I was right about the first bits but that's like predicting sunset will come. The split didn't happen because I never realized how much cash Google was going to generate -- far more than they can even spend. So the item is wrong.
What does how much cash a company generates have to do with a stock split? Nothing (well, generating lots of cash makes a split slightly more, rather than less, likely.)
Does he think that a stock split somehow raises capital? He should stick to technology and avoid finance.
No. A stock option is worth $0 if expired. It has non-negative value if not expired, e.g. you can profit from owning an option even if the underlying price never reaches the strike price (that's called delta hedging.)
(Admittedly, the afterburners were ramjets, but that still isn't scramjet tech.)
Almost by definition, afterburners (fuel dumps after the turbine) are not ramjets (tubes with clever geometry that are engines.) Afterburners are an inefficient way to get a bit more thrust for a short time (e.g. takeoff, evasion,) ramjets are engines (that don't work until you have a fair bit of airspeed.)
You can mix ramjets and rockets (stuff the ramjet with solid fuel, use it as a booster to get to speed,) but ramjets and jets do not mix.
Largely because Americans only consider average houses to have a lifespan of 30 years or so (i.e. shorter than Britain, longer than Japan.) Obviously, we can and do build houses with much longer expected lifespans, but the typical house is essentially obsolete within 30 years: ceilings too low, rooms too small, underpowered electricity, no intranet, ugly layout due to incremental expansion, etc. The next thirty years may see a new obsolesence: energy wastage, too big, etc.
:)
So, its mostly a cost thing combined with an abundance of resources! England gives houses names, we give them numbers
Basically, the programs do not beat the market, but lower the standard deviation (risk) involved in trading
Um, that IS beating the market: all investments are a combination of risk and expected reward (e.g. treasuries are low-risk and low-return, junk bonds are higher risk and higher return.) If you can reduce risk without reducing return, you can make buckets of money (people will line up outside your door wanting to give you capital.)
That said, don't trust any academic studies on this topic. There are probably only 50 people on the planet who actually deeply understand this topic, and they aren't talking. E.g. when I was running a hedge fund in the early 90s, most of the techniques described in the article were already widespread (yes, we had real-time newswire scrubbers running back then.)
As the article touches on, speed is important: a 5 millisecond advantage lets you make real money without needing to predict anything. The predictors do make excess profits, but most either don't understand the risks they are taking, or actively hide those risk from their investors. D.E.Shaw, mentioned in the article, basically blew itself up back in 1999 or so. LTCM did the same thing.
Some people (my 3 year old nephew) claim that Communists killed Jesus. This is patently false, because Jesus was killed and ressurected years before Communism was invented.
You're just cherry-picking. I'm sure there are lots of arguments that the communists killed Jesus, and you're only arguing with your 3 year-old nephew's argument.
This is reasonable description of the fundamental problem: some areas of scientific inquiry are of deep interest to some laymen because they provide support for certain cherished views (relativity, rocketry, and radiation are unimportant, but evolution, Laffer curves, genetic predisposition to homosexuality, global warming, etc, are fertile ground from which to construct a simple philosophy.)
The non-scientific pundit then takes a common sense idea (e.g. a bacteria can't give birth to a human, if taxes were 100% no one would work harder, a whirlwind in a junkyard doesn't make a 747,) grabs a few quotes from the large body of technical literature, and declares the whole field absurd.
Scientists try to respond using science, but it's mostly "he said, she said" to the layman. The pundits gleefully declare the scientists can't cite anything that refutes them (because scientists tend not to publish papers refuting batshit crazy ideas.)
This is not to say the scientific consensus in any field is right (I am skeptical of several fields in terms of predictions being made and overall synthesis,) but all science is vulnerable to misunderatanding by laymen when public opposition is natural (e.g. from preachers, oil companies, racist politicians.)
Often true, but not always. If you know your stuff and you prove it in other ways, the offers can come flooding in
Exactly. My ideal hire:
1) Knows CS. Algorithms, complexity, etc
2) Knows Software engineering. I.e. writing good code that solves the problem.
3) Is smart.
4) Could/can work in a group.
5) Will learn the business.
I've been on one day round trips to MIT (hoping for a candidate that hits 3/5,) brunches with personal recommedations (several good hires, including one for C++ based job who didn't know C++,) and random calls to people who published beautiful code on the net (some great hires.)
A degree is useful, but only as a foot in the door: we look for the people we want, HR does the paperwork, not the filtering.
Parent is a voice of sanity. Mod up.
Pretty much no one commits suicide in a tactical situation - they're too busy trying to fix things. This is not to say silly planners don't provide suicide pills -- they do, but an extra 1oz of weight isn't that bad, and it makes for good PR.
Actually, v1 (hereafter referred to as 'he') didn't write that there was no DNA in fingernails; he wrote that DNA was most reliably extracted from living cells.
Let's look at his original post:
He is probably well-aware that hair (minus a follicle) and fingernail clippings are both just keratin (chitin?) and contain no biological material or DNA of any sort.
Looks like you are so wrong it's just embarassing.
Organic extraction of aged human nail material yielded a sufficient quantity of DNA for successful mtDNA sequencing; however, STR analysis was unsuccessful.
Translation: we could find traces of DNA, but not enough to identify the person. To use something for identification in a criminal case requires that it be very reliable and very unique. In a test group of 15 they could not identify the owner in all cases. How are you going to pull that off when you are searching a 10,000 person database?
DNA is most reliably harvested from living cells. (try your google on that) There are no living cells in fingernails or hair strands. Only hair follicles and maybe some toejam.
Ok, fair enough, 8 year old fingernails aren't a good source source of DNA (d'oh - how much DNA do you expect to be left in an eight year old steak, for that matter.) Let's look at the authors' abstact of the paper that you quoted out of context:
In order to process nails as evidence, a validation study was performed to demonstrate that DNA could be successfully extracted from human nail material. Organic extractions of DNA were performed on fresh fingernails, fresh toenails, and aged (~eight years old) fingernails. The isolated DNA was quantified and analyzed by polymerase chain reaction (PCR)-based testing. This study demonstrated that the extraction of fresh, human nail material yielded sufficient quantities of DNA for successful short tandem repeat (STR) analysis and mitochondrial DNA (mtDNA) sequencing. A random sampling of fingernails from 22 different individuals (15 females, 6 males and 1 unknown) was tested. Successful STR profiles were obtained from all fingernail specimens collected from male individuals. Full or partial STR profiles were obtained from the fingernail specimens collected from 10 of the 15 female individuals. All nails amplified for mtDNA produced PCR product; all products sequenced produced the correct mtDNA type. Organic extraction of aged human nail material yielded a sufficient quantity of DNA for successful mtDNA sequencing; however, STR analysis was unsuccessful.
Wow! They could do mtDNA sequencing on decade old fingernails, but STR analysis was unsuccessful. Fresh nails yielded plenty of DNA. Color me stunned! Perhaps you could explain again how this has anything to do with your claims that fingernails don't have DNA in them?
ffs. Hair and nails are not pure keratin and are both rich in DNA. A simple Google search shows you are 100% wrong.
Where were you camping? I hope Alaska and not Antarctica - most kerosenes freeze in the -30 to -70 centigrade range. Some at merely -20 centigrade or so, but that's still really fucking cold!
I'd consider your answer worth an A. It's in the ballpark, explains what's going on in the algorithm, and touches on the right larger issues.
:)
I'd offer you a job, but I doubt prisoner number 652 needs one
I got .0236 in under ten seconds. I work with a lot of PhD mathematicians, and know most of them could probably have gotten 1 or 2 more significant digits in those 10 seconds.
Sure, we don't hire them for their ability to divide in their heads, but it's hard to conceive of hiring a mathematician who isn't able to do basic math (much like hiring a CS PhD who can't tell me how many cycles it takes to strcmp( "apple", "apples" ).)
If you can, avoid the discussion/meetings/emails.
Firm-wide standardization drives are
a) usually politically driven, and if you voice the wrong opinion you will be disliked.
b) driven by the incompetents - if they could do something profitable, they would be doing that.
c) out of date by the time they are finalized.
There is no upside here because there is no magic standard that will make things better.
Nearly 1000 developers. We're Fortune 500.
How would you feel if the network guys just needed to hack a small piece of your code? Surely you wouldn't mind if they checked out your source code and fiddled around?
Sounds reasonable to me - this is the approach we use at my firm. It's a rare week when I don't get mail saying someone has fixed or enhanced my code, then pushed it to production machines. It's a rare week when I don't do the same to someone else's code.
Works well when you make the effort to only hire good people, treat them like adults, and reward them according to their contributions.
Bah. A behavior as simple as protecting your tribe's corpses from opportunistic predators quickly ensures you generally don't get attacked by eagles. Once a raptor sees it doesn't get a free meal from attacking proto-humans, it quickly gives up.
Heck, burying your dead becomes a great advantage: predators gain nothing from killing your species, and soon seek prey that actually gives them food! Maybe human death rituals (e.g. burial, burning, leaving to vultures) got started because they ensured predators didn't profit from the death of the victim.
Well, they mostly make a record of things like tapestries. So, they lay out the item (often several hundred square feet,) then build what looks like a manually operatered 2-d x-y plotter over it. The camera and light system move in one axis across a bridge-like thing, and the bridge move a step in the other axis after each row-scan. All this takes place in a windowless room with the doors closed (to keep the light calibrated.)
Yep, next they piece the thousands of images together with custom panorama software. One problem is that gravity/temperate/humidity cause the object to change shape from minute to minute, so it's really a stitch and morph job (done by the Chudnovsky brothers with their homebrew supercomputer.)
Well, I'm just an F3 amateur, so I defer to my wife who shot 8x10 until recently. Her lab (they like to record the art they are working on) has mostly switched to digital now: that means scaffolding for the camera, etc, but it does give them the gigapixel images they want with a lot less post-processing hassle. For real hi-res stuff, having digital images as the basic inputs is proving to a big win.
Good point. Farming enabled much higher population densities, but at significant cost to the individuals: diet changed from 100+ items for some hunter-gatherers to around 2-5 staples for early agricultural communities. Lifespan and adult body size also usually decreased (probably due to worse nutrition and a better environment for nasty viruses.) Average work hours per day went up (to support the new ruling class.) Culture increased, but at a cost to the individual.
Well, I could have said he makes "spread" the fundamental idea, but that would really help anyone who hadn't read the article. As noted by many posters, spread is just a function of angle.
I doubt there are any unique insights from his approach: he's basically made angle the fundamental element of trig, rather than the more usual distance.
That said, it might be an interesting way to teach/explore trig. By doing away with the trig functions (which are just the distance->angle mappings,) he gets to solve many simple problems with just algebra and a final square root. Because the sqrt is explicit, this approach might give students a better mental model for trig: as things stand, most students just treat sin, cos, etc, as black boxes, and apply the rules (SOHCAHTOA) by rote.
I'd like to see how his method stacks up when applied to, say, an entire beginner level book or classic text, on geometry/trig.
If you want to be a code monkey, take vocational courses: learn Java and C++, learn about the buzzwords du jour (XML, SOAP. Ajax, whatever.)
If you actually want to be a serious programmer/designer, get a strong grounding in CS - that means data structures + algorithms, automata, numerics, compilers, OS design, etc. Know C++, Lisp, and a functional language.
So DZ doesn't want programmers who know how to write a compiler? Great, on his next big project, he'll wind up with a system with several embedded ad-hoc languages hidden away in it. Worse, he won't see anything wrong with that.
I've hired programmers for C++ environments who didn't know C++: if they're strong enough, they'll be up to speed in a few weeks.
At my current job, we want to see core CS courses. We have to teach/retrain all our hires anyway, so we might as might hire people with a strong theoretical foundation in CS.