From a scientific point of view, this is solved already. Scientists agree that if we stop emitting CO2 now we have a good chance of not fucking up too much.
Unfortunately, despite their awesome brain power, scientists do not control laws, cars, power plants etc and therefore can't just stop the CO2 emissions themselves. They can come up with a solution but someone else must implement it. Maybe we should all just agree to vote for a scientist in the next election?
As any good source should, they link directly to the actual articles and you can read them yourself. This is a published comment on her method: https://www.sciencedirect.com/...
In short, she's using an oversimplified model of the sun, knows nothing about the impact on climate and the implications would anyway be very small (-0.3C compared to a warming of +0.2C per decade).
I remember the solar cycle was a popular scape goat for global warming about 20 years ago but the focus shifted after a few years when there was just to much science showing the effect was minimal. Back then, the claim was that sunspots caused more solar storms and a "huge" amplifying effect due to cloud formation etc. Further research showed the effect was small and the deniers changed focus. This seems like a remake.
Yes, and according to the article DVDs were replaced by smart watches which satisfy the same need for round and shiny objects. I suppose this explains why so few people buy smart watches...
LIGO (for detecting gravitational waves) is essentially the exact same thing, just 1000x bigger. Of course it costs more and is vastly more complicated.
Besides, comparing Nobel prizes says more about the Nobel committee than it says about science. They have awarded far too many prizes to supraconductors and solid state physics if you ask me...
Yes you can. Agnosticism is a statement about knowledge in general and not just religion.
The English biologist Thomas Henry Huxley coined the word agnostic in 1869, and said "It simply means that a man shall not say he knows or believes that which he has no scientific grounds for professing to know or believe."
I've always wanted to try something like that but haven't (yet). How does it work for editing (cursor movement, backspace), keyboard shortcuts (mark text, cut text etc) and entry of symbols (brackets etc)? How many symbols are available? Do you have to release all keys before typing the next character or is it "smart" in some way?
It would also require the addition of morse code for arrow key left, control+arrow key left, shift+control+arrow key left, control+X, backspace etc. We do a lot on our keyboards which isn't just plain text.
QWERTY are two things: A physical layout of keys in rows and a logical mapping of symbols to those keys. Dvorak is only the latter.
I think physical layout is much more important than the key mapping. I have a bit of RSI on the outer side of the hands (Emacs pinkie) and would like to reduce the stress on the pinkie. As a programmer I am probably biased but I don't find the layout of the letters to be problematic. I spend a lot more time thinking than typing so typing speed isn't a huge issue. I also spend more time navigating, editing or typing weird symbols like braces and brackets than I do typing plain English text and those actions are some of the least ergonomic. Most key maps like Dvorak or Colemak optimise key placement based on the typing of english text but do nothing for symbols or placement of modifier keys (shift, control, alt). Neo is the only one I know that remaps symbols and movement by creating more modifiers/layers but all the modifiers are still along the edge of the keyboard and pressed with the pinky (and it is in German).
I have tried some ergonomic keyboards but none have caught on either because they were not radical enough (just a qwerty split in half) or because they were too big, expensive or not laptop friendly. Actually, none are laptop friendly if you do type on your lap. Some of the single hand chorded keyboards could be laptop friendly enough but they would have to support all the various brackets and braces for general use.
To scratch my own itch, I have been experimenting with my own chorded shortcuts software which distinguishes between ordinary typing and when you're Holding one key and Press-Releasing another. It basically treats the key events "Press A - Press/Release B - Release A" as a shortcut which can be mapped to anything. It helps with my particular case of RSI since I can remap all modifiers to the strong fingers (hold V or M for Ctrl, F and J for Shift, hold Space for movement and deletions etc). It works rather well but I do occasionally get some false positives so it isn't perfect yet. The code is here ( https://github.com/hopr/hopr ) if anyone is interested but it is a prototype without any polish since I am the only user. It works for me though.
I would very much like a redesigned keyboard with all movement, editing and modifier keys moved to the bottom (thumb) or center (index finger) but I want it on both laptops and desktops. Laptops won't change anytime soon so I think we are stuck with QWERTY. A more radical remapping of the qwerty keyboard would be interesting (like use all inner keys FGHJ etc as layers/modifiers like Neo) but only for some special groups like programmers or people with RSI. Neurological implants would of course be cool but so would magic wands.
Well, I'm going out on a limb here, but maybe people who type competitively aren't doing the same thing as people
That is a very good point. Most typing comparisons use the typing of already known plain english text which is not what most people write. We spend much more time pausing, thinking, re-reading and editing than most comparisons take into account.
This is statistics for emacs only (and I don't know how they got it): http://ergoemacs.org/emacs/com... But if a similar pattern holds for other types of typing then only about 50% of keystrokes are creating text. The other half is navigation and commands (up, down, save, search etc). So any realistic typing test should include editing and navigation and not just text input.
That's just dumb. On AVERAGE, emotional stories SEEM to have more weight in the USA than many (but probably not all) countries. That doesn't make ALL Americans comparable to Hannibal Lecter. It doesn't even make the most gullible person in the entire country horrible. Just gullible.
Even better, do not "make homeless" people. But that requires lots of "socialist" stuff like free education, health care and mental care which many Americans object to.
It seems to be a universal phenomenon to focus more on helping those who are visibly in need than preventing the need to arise. Emergency aid for spectacular disasters can eclipse the everyday aid of providing education or sanitary facilities. The cost effectiveness of providing emergency aid instead of prevention is also not discussed enough.
The American version of this seem to be a bit stronger than elsewhere. Politicians can be celebrated for personally helping a sick/poor child while at the same time removing social safety nets and thus creating many more. In many other countries they would have been called hypocrites but in America it seems like they get away with it (yes, it is a very anecdotal statement). It seems like emotional stories are much more important than in many other parts of the world and your personal character is more judged by those stories than by your actions.
$37 million isn't much as you point out and helping homeless doesn't help much in the long run. Spending the same amount on improving education in poor areas or even a marketing campaign for more "socialist" policies would probably be a more cost effective choice.
Perhaps a solution is to explicitly gender-bias the input. Include all the high-performing female employees in the input data and only a small number of male employees so that your sample is 50/50 -- or even biased towards women. But my guess is that the people at Amazon are plenty smart enough to have tried that idea, and that for some (perhaps inscrutable) reason it didn't work.
It would be really interesting to see the technical details of what they did and why it failed. A 50/50 set should work and if it didn't it would be really interesting to know why (unless the sample was just too small, that would not be interesting).
Again, why would a rules based system even consider those variables?
Who said it was rules based? I am a sloppy reader so it is an honest question.
Let us just cut to the chase here, is there any situation where you'd accept more men than women being hired statistically?
I don't know what that has to do with the question at hand but yes. I studied physics and about 25% of the students were female. Therefore I expect that in 50 years, the nobel prize in physics will go to men 75% of the time. Similarly in the workplace even though that can change faster. I don't see anything wrong with that. It's just the way things are and any change will take time.
Purge any submission to the system of a gender identifier...
It is not that easy. I assume they run some kind of pattern matching algorithm and it can just as easily focus on non-obvious gender differences as obvious ones. It could be that hobbies or volunteer work are just as good predictors of gender as a name (and therefore sources of bias). Maybe even the choice of words or textual layout could make a difference. It would take a lot of scrubbing and perhaps a complete rewrite of the job applications to remove all traces of personality.
When they say the result is biased, I assume they have manually reviewed the result and found examples where similar candidates were ranked differently for no relevant reason. But, I haven't read the article so I don't know how they came to that conclusion.
Scientists don't use it for the language, they use it for the libraries. Numpy is extremely fast (since it is written in C, C++, Fortran or Cython or whatever) and very convenient to use (since it is wrapped in Python).
Pandas. It is basically calculations with tables in Python. You can do most/all things Excel can (pivot tables etc) and many things Sql can (group by etc).
But, the difference from excel is the separation of data and logic. You create a script with calculations and then run it on your data. No risk of accidentally replacing a the formula in E237 with a "value",
Thanks for the links! Both the summary and the cnbc article link the wrong one.
As for the CNBC article, they say "They observed that when exposed to only 1 milligram per milliliter of the artificial sweeteners, the bacteria found in the digestive system became toxic" but is that really what the study shows?
The bioluminescent bacteria, which luminesce when they detect toxicants, act as a sensing model representative of the complex microbial system. Both induced luminescent signals and bacterial growth were measured. Toxic effects were found when the bacteria were exposed to certain concentrations of the artificial sweeteners. In the bioluminescence activity assay, two toxicity response patterns were observed, namely, the induction and inhibition of the bioluminescent signal. An inhibition response pattern may be observed in the response of sucralose in all the tested strains: TV1061 (MLIC = 1 mg/mL), DPD2544 (MLIC = 50 mg/mL) and DPD2794 (MLIC = 100 mg/mL). It is also observed in neotame in the DPD2544 (MLIC = 2 mg/mL) strain. On the other hand, the induction response pattern may be observed in its response in saccharin in TV1061 (MLIndC = 5 mg/mL) and DPD2794 (MLIndC = 5 mg/mL) strains, aspartame in DPD2794 (MLIndC = 4 mg/mL) strain, and ace-k in DPD2794 (MLIndC = 10 mg/mL) strain
I am not a chemist/biologist and can not decipher that but
a) Bacteria in the gut did not become toxic. It was a genetically engineered bioluminescent strain of e-coli that signals when it detects toxicants. b) There are many reported numbers ranging from 1 mg/mL to 100 mg/mL as well as different responses to different sweeteners.
So if I just assume the limit for toxicity is 1 mg/mL like they say, a can of diet coke contains 125 mg of Aspartame and is 8 oz = 240 mL which is about 0.5 mg/mL and thus not-toxic. Also, if the other fluids in the body dilute the coke it should be even more safe since it is the concentration that matters. Is that a reasonable conclusion?
I'm confused. The summary says "it would warm the surface of the continental United States by 0.24 C" but the technologyreview article suggests it is a local change. So, is 0.24 C an average over the whole country or is it the local warming around a wind farm? If it is only local then... who cares?
I was very surprised by this since I've read about the Oceanographer McCullogh before and wasn't impressed.
I can't find any official source confirming this except perhaps Univ of Plymouth which is where McCullogh works. Is there an official DARPA announcement out there somewhere?
The only information I can find is from Motherboard (an interview) and forums/McCulloughs twitter. There is some more info here: https://forum.nasaspaceflight.... search for flux_capacitors post about 1/3 down.
The information seem to be from McCulloughs twitter and the money is for him and a post doc "developing the theory" and two experiments by other people.
The first experiment is this one http://www.jbis.org.uk/paper.p... which is supposed to replicate the em drive effect with lasers instead of em radiation.
The second will test a LEM drive (whatever that is).
In short, this is all related to the EM drive which explains why DARPA might think it is worth a small investment. It's not that DARPA suddenly thought QI sounded interesting but rather practical experiments to figure out what is going with the EM drive.
I am still surprised McCullough gets money for doing theory but the experiments sound like reasonable high-risk high-reward investments.
FTFY. Even pharmaceutical patents have issues. There's suboptimal research where patentability is a more important property than efficiency. There's no motivation for doing research on old drugs or natural non-patentable substances. It is better to create something similar but just different enough to make it patentable and then do research on this "new" drug. Even if there is no improvement you can still patent it, do a bit more research until some studies show a tiny improvement by pure chance and then do some heavy marketing.
There's also the moral issue when sick people in poor countries can't afford an expensive patented drug and are not allowed to make cheap generic copies.
So, even if medical patents kind of work, they are not perfect.
Slashdot Mirror
From a scientific point of view, this is solved already. Scientists agree that if we stop emitting CO2 now we have a good chance of not fucking up too much.
Unfortunately, despite their awesome brain power, scientists do not control laws, cars, power plants etc and therefore can't just stop the CO2 emissions themselves. They can come up with a solution but someone else must implement it. Maybe we should all just agree to vote for a scientist in the next election?
Here's a summary of their claims from a more respectable source:
https://www.theguardian.com/en...
As any good source should, they link directly to the actual articles and you can read them yourself. This is a published comment on her method:
https://www.sciencedirect.com/...
In short, she's using an oversimplified model of the sun, knows nothing about the impact on climate and the implications would anyway be very small (-0.3C compared to a warming of +0.2C per decade).
I remember the solar cycle was a popular scape goat for global warming about 20 years ago but the focus shifted after a few years when there was just to much science showing the effect was minimal. Back then, the claim was that sunspots caused more solar storms and a "huge" amplifying effect due to cloud formation etc. Further research showed the effect was small and the deniers changed focus. This seems like a remake.
Yes, and according to the article DVDs were replaced by smart watches which satisfy the same need for round and shiny objects. I suppose this explains why so few people buy smart watches...
Each advance is increasingly more complex and costly than the last.
Yes, this. And here is proof. The Michelson-Morley experiment which proved the ether didn't exist was a few meters in size.
https://en.wikipedia.org/wiki/...
LIGO (for detecting gravitational waves) is essentially the exact same thing, just 1000x bigger.
Of course it costs more and is vastly more complicated.
https://en.wikipedia.org/wiki/...
Besides, comparing Nobel prizes says more about the Nobel committee than it says about science. They have awarded far too many prizes to supraconductors and solid state physics if you ask me...
Thanks! I will look into it when I get the time.
Heck you can be a theist agnostic.
No you can't.
Yes you can. Agnosticism is a statement about knowledge in general and not just religion.
The English biologist Thomas Henry Huxley coined the word agnostic in 1869, and said "It simply means that a man shall not say he knows or believes that which he has no scientific grounds for professing to know or believe."
https://en.wikipedia.org/wiki/...
Knowing when you have sufficient evidence to ”know or believe" is not an easy question though.
You are right about gnosticism though. It is unrelated.
It is the reverse Turing test. If you can't tell a stupid machine from a human, then the human must be stupid.
Could you please elaborate?
I've always wanted to try something like that but haven't (yet). How does it work for editing (cursor movement, backspace), keyboard shortcuts (mark text, cut text etc) and entry of symbols (brackets etc)? How many symbols are available? Do you have to release all keys before typing the next character or is it "smart" in some way?
It would also require the addition of morse code for arrow key left, control+arrow key left, shift+control+arrow key left, control+X, backspace etc. We do a lot on our keyboards which isn't just plain text.
QWERTY are two things: A physical layout of keys in rows and a logical mapping of symbols to those keys. Dvorak is only the latter.
I think physical layout is much more important than the key mapping. I have a bit of RSI on the outer side of the hands (Emacs pinkie) and would like to reduce the stress on the pinkie. As a programmer I am probably biased but I don't find the layout of the letters to be problematic. I spend a lot more time thinking than typing so typing speed isn't a huge issue. I also spend more time navigating, editing or typing weird symbols like braces and brackets than I do typing plain English text and those actions are some of the least ergonomic. Most key maps like Dvorak or Colemak optimise key placement based on the typing of english text but do nothing for symbols or placement of modifier keys (shift, control, alt). Neo is the only one I know that remaps symbols and movement by creating more modifiers/layers but all the modifiers are still along the edge of the keyboard and pressed with the pinky (and it is in German).
I have tried some ergonomic keyboards but none have caught on either because they were not radical enough (just a qwerty split in half) or because they were too big, expensive or not laptop friendly. Actually, none are laptop friendly if you do type on your lap. Some of the single hand chorded keyboards could be laptop friendly enough but they would have to support all the various brackets and braces for general use.
To scratch my own itch, I have been experimenting with my own chorded shortcuts software which distinguishes between ordinary typing and when you're Holding one key and Press-Releasing another. It basically treats the key events "Press A - Press/Release B - Release A" as a shortcut which can be mapped to anything. It helps with my particular case of RSI since I can remap all modifiers to the strong fingers (hold V or M for Ctrl, F and J for Shift, hold Space for movement and deletions etc). It works rather well but I do occasionally get some false positives so it isn't perfect yet. The code is here ( https://github.com/hopr/hopr ) if anyone is interested but it is a prototype without any polish since I am the only user. It works for me though.
I would very much like a redesigned keyboard with all movement, editing and modifier keys moved to the bottom (thumb) or center (index finger) but I want it on both laptops and desktops. Laptops won't change anytime soon so I think we are stuck with QWERTY. A more radical remapping of the qwerty keyboard would be interesting (like use all inner keys FGHJ etc as layers/modifiers like Neo) but only for some special groups like programmers or people with RSI. Neurological implants would of course be cool but so would magic wands.
Well, I'm going out on a limb here, but maybe people who type competitively aren't doing the same thing as people
That is a very good point. Most typing comparisons use the typing of already known plain english text which is not what most people write. We spend much more time pausing, thinking, re-reading and editing than most comparisons take into account.
This is statistics for emacs only (and I don't know how they got it):
http://ergoemacs.org/emacs/com...
But if a similar pattern holds for other types of typing then only about 50% of keystrokes are creating text. The other half is navigation and commands (up, down, save, search etc). So any realistic typing test should include editing and navigation and not just text input.
That's just dumb. On AVERAGE, emotional stories SEEM to have more weight in the USA than many (but probably not all) countries. That doesn't make ALL Americans comparable to Hannibal Lecter. It doesn't even make the most gullible person in the entire country horrible. Just gullible.
Even better, do not "make homeless" people. But that requires lots of "socialist" stuff like free education, health care and mental care which many Americans object to.
It seems to be a universal phenomenon to focus more on helping those who are visibly in need than preventing the need to arise. Emergency aid for spectacular disasters can eclipse the everyday aid of providing education or sanitary facilities. The cost effectiveness of providing emergency aid instead of prevention is also not discussed enough.
The American version of this seem to be a bit stronger than elsewhere. Politicians can be celebrated for personally helping a sick/poor child while at the same time removing social safety nets and thus creating many more. In many other countries they would have been called hypocrites but in America it seems like they get away with it (yes, it is a very anecdotal statement). It seems like emotional stories are much more important than in many other parts of the world and your personal character is more judged by those stories than by your actions.
$37 million isn't much as you point out and helping homeless doesn't help much in the long run. Spending the same amount on improving education in poor areas or even a marketing campaign for more "socialist" policies would probably be a more cost effective choice.
Meanwhile Judge Kavanaugh is out there feeling the homeless.
FTFY
Perhaps a solution is to explicitly gender-bias the input. Include all the high-performing female employees in the input data and only a small number of male employees so that your sample is 50/50 -- or even biased towards women. But my guess is that the people at Amazon are plenty smart enough to have tried that idea, and that for some (perhaps inscrutable) reason it didn't work.
It would be really interesting to see the technical details of what they did and why it failed. A 50/50 set should work and if it didn't it would be really interesting to know why (unless the sample was just too small, that would not be interesting).
Again, why would a rules based system even consider those variables?
Who said it was rules based? I am a sloppy reader so it is an honest question.
Let us just cut to the chase here, is there any situation where you'd accept more men than women being hired statistically?
I don't know what that has to do with the question at hand but yes. I studied physics and about 25% of the students were female. Therefore I expect that in 50 years, the nobel prize in physics will go to men 75% of the time. Similarly in the workplace even though that can change faster. I don't see anything wrong with that. It's just the way things are and any change will take time.
Purge any submission to the system of a gender identifier...
It is not that easy. I assume they run some kind of pattern matching algorithm and it can just as easily focus on non-obvious gender differences as obvious ones. It could be that hobbies or volunteer work are just as good predictors of gender as a name (and therefore sources of bias). Maybe even the choice of words or textual layout could make a difference. It would take a lot of scrubbing and perhaps a complete rewrite of the job applications to remove all traces of personality.
When they say the result is biased, I assume they have manually reviewed the result and found examples where similar candidates were ranked differently for no relevant reason. But, I haven't read the article so I don't know how they came to that conclusion.
Scientists don't use it for the language, they use it for the libraries. Numpy is extremely fast (since it is written in C, C++, Fortran or Cython or whatever) and very convenient to use (since it is wrapped in Python).
Pandas. It is basically calculations with tables in Python. You can do most/all things Excel can (pivot tables etc) and many things Sql can (group by etc).
But, the difference from excel is the separation of data and logic. You create a script with calculations and then run it on your data. No risk of accidentally replacing a the formula in E237 with a "value",
Thanks for the links! Both the summary and the cnbc article link the wrong one.
As for the CNBC article, they say "They observed that when exposed to only 1 milligram per milliliter of the artificial sweeteners, the bacteria found in the digestive system became toxic" but is that really what the study shows?
The bioluminescent bacteria, which luminesce when they detect toxicants, act as a sensing model representative of the complex microbial system. Both induced luminescent signals and bacterial growth were measured. Toxic effects were found when the bacteria were exposed to certain concentrations of the artificial sweeteners. In the bioluminescence activity assay, two toxicity response patterns were observed, namely, the induction and inhibition of the bioluminescent signal. An inhibition response pattern may be observed in the response of sucralose in all the tested strains: TV1061 (MLIC = 1 mg/mL), DPD2544 (MLIC = 50 mg/mL) and DPD2794 (MLIC = 100 mg/mL). It is also observed in neotame in the DPD2544 (MLIC = 2 mg/mL) strain. On the other hand, the induction response pattern may be observed in its response in saccharin in TV1061 (MLIndC = 5 mg/mL) and DPD2794 (MLIndC = 5 mg/mL) strains, aspartame in DPD2794 (MLIndC = 4 mg/mL) strain, and ace-k in DPD2794 (MLIndC = 10 mg/mL) strain
I am not a chemist/biologist and can not decipher that but
a) Bacteria in the gut did not become toxic. It was a genetically engineered bioluminescent strain of e-coli that signals when it detects toxicants.
b) There are many reported numbers ranging from 1 mg/mL to 100 mg/mL as well as different responses to different sweeteners.
So if I just assume the limit for toxicity is 1 mg/mL like they say, a can of diet coke contains 125 mg of Aspartame and is 8 oz = 240 mL which is about 0.5 mg/mL and thus not-toxic. Also, if the other fluids in the body dilute the coke it should be even more safe since it is the concentration that matters. Is that a reasonable conclusion?
I'm confused. The summary says "it would warm the surface of the continental United States by 0.24 C" but the technologyreview article suggests it is a local change. So, is 0.24 C an average over the whole country or is it the local warming around a wind farm? If it is only local then... who cares?
All software is automating someone else's job.
Grand Theft Auto?
I was very surprised by this since I've read about the Oceanographer McCullogh before and wasn't impressed.
I can't find any official source confirming this except perhaps Univ of Plymouth which is where McCullogh works. Is there an official DARPA announcement out there somewhere?
The only information I can find is from Motherboard (an interview) and forums/McCulloughs twitter. There is some more info here:
https://forum.nasaspaceflight....
search for flux_capacitors post about 1/3 down.
The information seem to be from McCulloughs twitter and the money is for him and a post doc "developing the theory" and two experiments by other people.
The first experiment is this one
http://www.jbis.org.uk/paper.p...
which is supposed to replicate the em drive effect with lasers instead of em radiation.
The second will test a LEM drive (whatever that is).
In short, this is all related to the EM drive which explains why DARPA might think it is worth a small investment. It's not that DARPA suddenly thought QI sounded interesting but rather practical experiments to figure out what is going with the EM drive.
I am still surprised McCullough gets money for doing theory but the experiments sound like reasonable high-risk high-reward investments.
"Passes mustard"
Oww... that just got to hurt...
pharmaceutical patents sort of work.
FTFY. Even pharmaceutical patents have issues. There's suboptimal research where patentability is a more important property than efficiency. There's no motivation for doing research on old drugs or natural non-patentable substances. It is better to create something similar but just different enough to make it patentable and then do research on this "new" drug. Even if there is no improvement you can still patent it, do a bit more research until some studies show a tiny improvement by pure chance and then do some heavy marketing.
There's also the moral issue when sick people in poor countries can't afford an expensive patented drug and are not allowed to make cheap generic copies.
So, even if medical patents kind of work, they are not perfect.