An average brain neuron takes its inputs from an _average_ of 7000 other neurons, with the max being somewhere around 10k,
Purkinje cells in the cerebellum have sum up about 100k inputs. Purkinje cells are the sole output of the cerebellar cortex. And the cerebellum has as many neurons as the rest of the brain (~10^10). So your point is very valid, just an order of magnitude or so short on the estimate.
You can similarly compare the temperature of the human brain and then observe that the machines have long bypassed it. Does it make machines smarter? I don't think so.
The brain is so insanely parallel and the neurons are not just digital gates, more like computers in themselves. The machines of today are a far cry from the brain in how they are built. But sure, you can compare them by some meaningless parameter to say that we're close. How about the clock frequency: neurons are 1kHz devices, and modern CPUs are in GHz now...
As you noticed, the reviewers are not paid a dime. They are neither paid for the original reviews, nor for re-reviewing revised manuscripts.
The editors are compensated somewhat, but this is a symbolic amount, not even close to the analogous compensation in non-scientific magazines. Besides, the authors pay page charges, somewhat like $50/page nowadays. Does it cost $50 to typeset a page for the web if you start with an already electronic text? Hell, I used to do this work for $10/hour and I would do a lot more than 1 page/hour.
But ultimately, nobody even says that it has to be charge-free. Charge what you want, enjoy subscription money for a year, then give it a free access. How long would it be fair to be collecting money on the same material?
Python may be great as a general scripting language
Scripting has nothing to do with it. Python is a great language. True, it is not compiled, but then neither is MATLAB. I tried both for quite a while. In my opinion, Python beats MATLAB hands down. It is equivalent or better in purely numerical computations, and is actually sane for other programming as well.
What good is the naked kernel? For any user, kernel is a must, but so are the applications. If applications all go GPLv3 then good luck selling naked kernel.
Indeed. The ideal process would be a taxpayer-sponsored publishing. There's some overhead in maintaining the organization, so it's not completely zero cost. However, it must be far lower that what the publishers want us to believe. One could have taxpayer-run electronic publishing, and then allow commercial publishers to print and sell the articles for those who want the nice and shiny paper version (as opposed to printing it yourself).
What gets me the most is that currently publishers make you sign the copyright waver to transfer rights to them. All such forms that I have seen start with "The copyright law requires that you transfer the copyright..." which is a complete bullshit. I could have held the copyright and just given them permission to publish it once, there's nothing in any law that requires copyright transfer for publishing.
But if I don't sign that form then I don't get published, and then I don't get funded for research because I have no publications. Catch-22.
In Wikipedia anybody and their dog can edit. In contrast, in peer-reviewed journals it's the editors who select the competent referees. Your comparison is not fair: there is definitely a bunch of loonies who would love to referee the papers, but they never get invited.
Nobody said that the publisher has to be handsomely paid to have an unpaid editors and unpaid reviewer that they have now.
I had no idea about the job and busy schedule. I thought you're looking for a nice tool that allows you to develop apps on all platforms. How would I know that you had more in mind than just that:-)
Next thing you'll say that you're proficient with the cyrillic keyboard and the programming language has to accept statements in cyrillic:-) Just kidding.
I don't want you to do anything. You said that you want to develop cross-platform and I pointed out that Python does the job. Nowhere in your original post was money an issue.
I guess 30K sucks compared to 150K, but this is irrelevant to languages themselves.
I want to be able to develop applications in both Windows and Linux. VS.Net and Mono allow me to use the same code with very little tweaking between platforms and keep using my Visual BASIC skills I learned over a decade ago.
The python + gtk allow you to do the same. Why hold on to Visual BASIC, if it's shitty in the first place?
How do you know he was torturing? Why is that that killing for the sake food is OK, but killing for the sake of new knowledge is not? Should we proibit non-vegeterian food?
The problem is, if such device is made, the invisible woman will be blind. No light will reach her eyes, because if some does then it some absorbed on the retina and cannot be "bypassed" around anymore. So instead of a powerful invisible person we get a handciapped individual, whose position is especially bad because nobody can help her, 'cause the can't be seen!
In case you haven't noticed, DMCA is fully legitimate (unfortunately), because it is the acting law (in the US at least). You cannot legitimize it any more than it already is. So pretending it does not exist or is not enforceable is not an option.
The article in sparse on details. It would definitely be nice to know the exact methodology: what else was considered besides the number of disclosed/patched vulnerabilites, how those were determined, etc.
Without it, the study is hardly different from hadnwaving.
When I look at this:
ar,pw,res = (sys.argv,lambda u:hmac.new(sys.argv[1],u).hexdigest(),re.search) I see three asignments. One is the list of the command line arguments, another is the function defined right here, and the last one is the existing function. Looks like multiple assignments to unrelated things, if you ask me:-)
Of course, when function returns a tuple which you unpack it is different matter, but the above clearly qualifies as multiple statements (carefully and smartly hidden, I must say). If he were not aiming for the most compact code, he would have written the same using three lines.
...although the python program is not putting multiple statements on one line, you can't do that in python.
A little off topic, but yes, you can -- just put the semicolon in between:
$ python
>>> import os; os.listdir(os.getcwd())
Also, the following is nothing else but the multiple statements carefully hidden by the tuple packing/unpacking:
a,b,c = (blah,lambda x:blah,blah-blah)
Correction: I saw it first in 1996, and here is the link. It's all in Russian, pictures are bad quality, and of the then-presidential-candidates. However, it makes pretty much the same point:-)
I think that, while we're not sure about the ultimate purpose of the circuitry, we have a pretty good idea on how we use face recognition behaviorally, for some milions of years now:-)
As you correctly say, the image may activate 60% and 40% of corresponding networks. My point was that, behaviorally, subjects were not likely to tell "60% Monroe + 40% Thatcher", no matter what percentage of which netwroks were activated. Simply because the behavior of face recognition is used to identify the person, that is, to make a decision of 100%, no matter the underlying mechanisms. To put it simply, when you meet somebody in the street, you need to say "Hi Maggie" or "Hi Marylin" and not any 60%-40% mixture of those.
I know that nothing is strictly obvious, but similar recognition experiments have been done in almost every sensory modality. I hate to speculate before I read the actual paper, and I have trouble getting it just now. However, I would be willing to bet that they may have found a hysteresis (going from M to T and going back from T to M results in flip in different percentage mixes) and all other things typical for recognition.
Mind you, reporting a flip has nothing to do with the fMRI -- anybody may do such an experiment with a decent morphing program. I saw this first in 1998:-)
Slashdot Mirror
An average brain neuron takes its inputs from an _average_ of 7000 other neurons, with the max being somewhere around 10k,
Purkinje cells in the cerebellum have sum up about 100k inputs. Purkinje cells are the sole output of the cerebellar cortex. And the cerebellum has as many neurons as the rest of the brain (~10^10). So your point is very valid, just an order of magnitude or so short on the estimate.
You can similarly compare the temperature of the human brain and then observe that the machines have long bypassed it. Does it make machines smarter? I don't think so.
The brain is so insanely parallel and the neurons are not just digital gates, more like computers in themselves. The machines of today are a far cry from the brain in how they are built. But sure, you can compare them by some meaningless parameter to say that we're close. How about the clock frequency: neurons are 1kHz devices, and modern CPUs are in GHz now...
As you noticed, the reviewers are not paid a dime. They are neither paid for the original reviews, nor for re-reviewing revised manuscripts.
The editors are compensated somewhat, but this is a symbolic amount, not even close to the analogous compensation in non-scientific magazines. Besides, the authors pay page charges, somewhat like $50/page nowadays. Does it cost $50 to typeset a page for the web if you start with an already electronic text? Hell, I used to do this work for $10/hour and I would do a lot more than 1 page/hour.
But ultimately, nobody even says that it has to be charge-free. Charge what you want, enjoy subscription money for a year, then give it a free access. How long would it be fair to be collecting money on the same material?
Python may be great as a general scripting language
Scripting has nothing to do with it. Python is a great language. True, it is not compiled, but then neither is MATLAB. I tried both for quite a while. In my opinion, Python beats MATLAB hands down. It is equivalent or better in purely numerical computations, and is actually sane for other programming as well.
What good is the naked kernel? For any user, kernel is a must, but so are the applications. If applications all go GPLv3 then good luck selling naked kernel.
This is the dream. The reality is that upstream never uses launchpad. So all this beatiful integration is *just another bug racker*.
+ bug/120569
Check this out: https://bugs.launchpad.net/ubuntu/+source/gramps/
It will stay open forever, until upstream of gtkspell decides to fix it.
This means that there will be no more drafts. If changes are made, they will appear in the released license. Seems pretty clear to me.
Indeed. The ideal process would be a taxpayer-sponsored publishing. There's some overhead in maintaining the organization, so it's not completely zero cost. However, it must be far lower that what the publishers want us to believe. One could have taxpayer-run electronic publishing, and then allow commercial publishers to print and sell the articles for those who want the nice and shiny paper version (as opposed to printing it yourself).
What gets me the most is that currently publishers make you sign the copyright waver to transfer rights to them. All such forms that I have seen start with "The copyright law requires that you transfer the copyright..." which is a complete bullshit. I could have held the copyright and just given them permission to publish it once, there's nothing in any law that requires copyright transfer for publishing.
But if I don't sign that form then I don't get published, and then I don't get funded for research because I have no publications. Catch-22.
In Wikipedia anybody and their dog can edit. In contrast, in peer-reviewed journals it's the editors who select the competent referees. Your comparison is not fair: there is definitely a bunch of loonies who would love to referee the papers, but they never get invited.
Nobody said that the publisher has to be handsomely paid to have an unpaid editors and unpaid reviewer that they have now.
Fine with me. I just hate the "import string" statement :-)
Sheesh, don't learn python :-)
:-)
:-) Just kidding.
I had no idea about the job and busy schedule. I thought you're looking for a nice tool that allows you to develop apps on all platforms. How would I know that you had more in mind than just that
Next thing you'll say that you're proficient with the cyrillic keyboard and the programming language has to accept statements in cyrillic
I don't want you to do anything. You said that you want to develop cross-platform and I pointed out that Python does the job. Nowhere in your original post was money an issue.
I guess 30K sucks compared to 150K, but this is irrelevant to languages themselves.
Says who? and by what criteria? BTW, I didn't call you names. I guess I should have.
I want to be able to develop applications in both Windows and Linux. VS.Net and Mono allow me to use the same code with very little tweaking between platforms and keep using my Visual BASIC skills I learned over a decade ago.
The python + gtk allow you to do the same. Why hold on to Visual BASIC, if it's shitty in the first place?
How do you know he was torturing? Why is that that killing for the sake food is OK, but killing for the sake of new knowledge is not? Should we proibit non-vegeterian food?
The problem is, if such device is made, the invisible woman will be blind. No light will reach her eyes, because if some does then it some absorbed on the retina and cannot be "bypassed" around anymore. So instead of a powerful invisible person we get a handciapped individual, whose position is especially bad because nobody can help her, 'cause the can't be seen!
In case you haven't noticed, DMCA is fully legitimate (unfortunately), because it is the acting law (in the US at least). You cannot legitimize it any more than it already is. So pretending it does not exist or is not enforceable is not an option.
It takes Word file and spits out plain text. It can also do some more tricks.
The article in sparse on details. It would definitely be nice to know the exact methodology: what else was considered besides the number of disclosed/patched vulnerabilites, how those were determined, etc. Without it, the study is hardly different from hadnwaving.
Nice! You must be an obfuscation expert :-)
When I look at this:
I see three asignments. One is the list of the command line arguments, another is the function defined right here, and the last one is the existing function. Looks like multiple assignments to unrelated things, if you ask me :-)
:-)
ar,pw,res = (sys.argv,lambda u:hmac.new(sys.argv[1],u).hexdigest(),re.search)
Of course, when function returns a tuple which you unpack it is different matter, but the above clearly qualifies as multiple statements (carefully and smartly hidden, I must say). If he were not aiming for the most compact code, he would have written the same using three lines.
Not that it matters, really
...although the python program is not putting multiple statements on one line, you can't do that in python.
A little off topic, but yes, you can -- just put the semicolon in between:
$ python
>>> import os; os.listdir(os.getcwd())
Also, the following is nothing else but the multiple statements carefully hidden by the tuple packing/unpacking:
a,b,c = (blah,lambda x:blah,blah-blah)
I agree with the main point, though.
Correction: I saw it first in 1996, and here is the link. It's all in Russian, pictures are bad quality, and of the then-presidential-candidates. However, it makes pretty much the same point :-)
I think that, while we're not sure about the ultimate purpose of the circuitry, we have a pretty good idea on how we use face recognition behaviorally, for some milions of years now :-)
As you correctly say, the image may activate 60% and 40% of corresponding networks. My point was that, behaviorally, subjects were not likely to tell "60% Monroe + 40% Thatcher", no matter what percentage of which netwroks were activated. Simply because the behavior of face recognition is used to identify the person, that is, to make a decision of 100%, no matter the underlying mechanisms. To put it simply, when you meet somebody in the street, you need to say "Hi Maggie" or "Hi Marylin" and not any 60%-40% mixture of those.
I know that nothing is strictly obvious, but similar recognition experiments have been done in almost every sensory modality. I hate to speculate before I read the actual paper, and I have trouble getting it just now. However, I would be willing to bet that they may have found a hysteresis (going from M to T and going back from T to M results in flip in different percentage mixes) and all other things typical for recognition.
Mind you, reporting a flip has nothing to do with the fMRI -- anybody may do such an experiment with a decent morphing program. I saw this first in 1998 :-)