A famous quote from one of the project leaders, Fred Jelinek if I'm not mistaken was that for every linguist that he fired from the team, the performance of the system improved by 10%...
I work quite heavily with Linux machines and they only crashed in case of hardware errors. Some machines even continued their server tasks with a totally broken hard disk.
On the other hand, when I toyed around with Excel and some Visual Basic script under XP the other day, the bl**dy thing refused to shut down properly. Had to kill the "explorer" before it would stop. And don't get me started on NT, or Word for that matter...
Then again, XP is a whole lot friendlier than any Linux distro I've ever seen, and that probably means more to most people than total stability.
How come nobody mentions Vivisimo and Teoma in this context? They happen to present the results for different meanings of the word "apple", although they cannot read your minder neither, very much like Googe. They only make it easier for laymen and novices to see that there are other "meanings" for the same search term.
However, this discussion is also blatantly Anglo-centric. In the rest of the world, "Apple" is even more strongly connected to the computer manufacturer (or the Beatles!). Try appel, Apfel, manzana, or pomme in Google and see what you'll find then (or take a look here.
I think you forgot to mention that they can also reject the European budget in its entirety.
M-x sarcasm-mode-off
I hope you do realize that that's not the kind of power a parliament should have. A parliament should at least be able to vote off single members of the council, have influence on the "civil servants" appointed in the highest region, create and ammend laws on its own, change parts of the budget and block plans by other authorities (e.g. the expansion of the EU), be able to influence the "foreign" policies, etc.
As long as they don't have that kind of power, they are almost powerless. And what's the fuss about in relation to the new "Consitution"? Who gets to be the chair for how long!
Don't bother about people putting moral pressure on you, as I've seen in the postings before. That's quite unreasonable for two reasons.
First, if the company goes bankrupt, you'll need another job anyway, so the sooner you start looking, the better.
Second, it's not your responsability, but management's. You cannot be blamed for bringing the company to the point where it is now, so don't feel guilty about the consequences of your actions. Furthermore, somebody else fired you co-workers, and that should make you responsible? No.
There is another reasons why you might walk: it gives better job security to your current co-workers. The company will need them more than ever and will save a few bucks on your salary.
And as all people who have taken more advanced maths know: you can't plot the entire FFT in a 1000x200 pixel size window.
I can send you a file with exactly the same FFT output as the one claimed to represent the original and I can assure you it will sound distinctly different.
Here is how to do it: the height of each pixel corresponds most likely to sqrt(a[k]^2+b[k]^2), where a[k]+ib[k] is the k-th element in the fft output. So, you can try to estimate what the original value of the sqrt must have been by eyeballing the graph (just like the article poster did). Then you just guess the relation between the real and imaginary part, since it is not shown in the graph. You've got about 1000 of these points, but your song probably needs a few minutes at 44.1kHz, which translates to 8 million per channel (the stereo distribution of course also is missing from the graph, giving you some extra freedom to randomize numbers). Now distribute your first point over the first 8000 of the new song, the second over the second 8000, etc., feed it into the inverse FFT and you've got a file that's undistinguishable from the original, according to a visual spectral comparison.
But I bet you'll be able to *hear* the difference...
What's extensive? I've got my code built on template classes for lists, sets and associative arrays and the main program is 100k lines. Compile times on an old 500MHz Linux machine with gcc are completely acceptable. Compile times on my even slower mac with CodeWarrior are a little worse, but still acceptable.
Debugging? Both platforms handle templates transparently. Especially CodeWarrior does a nice job in its totally integrated environment, but gdb ain't bad neither.
Bloat? Neither. Yes, a list of ints will generate exactly the same code as a list of long ints twice, but this kind of template generates little code, so that's a small price to pay for transparency.
I don't know how many weeks I've lost to helping others debug or rewrite their code because they thought they would do something "clever" with templates and they ended up creating a maze.
That's a good point, but it simply refers to bad design and/or programming. Templates do offer an extra way of complicating code, but it doesn't have to be that way. Multiple inheritance can be a similar drag, but simple inheritance almost never fails in helping write clean code. Are you going to argue that inheritance is therefore bad? Didn't think so.
No, I think templates are really, really useful, and I like them a lot better than the inheritance mechanism. Just keep the design and programming clean, like everywhere else...
No, my point was that it was a copy of Windows 95 and Windows 95 is bad, bad, bad. User-unfriendliness disguised as a desirable gadget. Bright colors, big buttons. XP is even worse. Did you ever understand these changing menus? Horrible. So that's the way Linux should not be going. Perhaps it is a reasonable way to win existing Windows users, but it is not going to win new computer users over...
I'm not a Windows user. I am a long time Unix (since version 6 on a PDP-11), RSX, VMS, Linux and MacOS user (since system 5). My problem was precisely that they are trying to copy the interface of Windows: that interface is horrible. And I agree: XP is worse, but with bright friendly colors, so it *seems* user-friendly.
About the karma whoring you might be right: although they are legitimate concerns I voiced (I do feel that more attention should be given to the user experience), I did get lucky to post soon enough. Did you ever notice that early posts receive much more karma than late posts?
What I would like to see is not another technical feat, but an effort to bring the Linux desk top closer to the a-technical masses.
I've recently had the "pleasure" of reinstalling Red Hat Linux and neither Gnome nor KDE are user-friendly at all. Yes, they do copy the Windows 95 desk top, but no, that's not going to help my father. And don't even start about the built-in file/web/help/and-what-not browsers.
With all this high configurability that's available in both windowing systems, couldn't a group of more human-interface oriented people build a layman interface on top of either Gnome or KDE?
A boffin is "a person engaged in scientific (esp. military) research".
So, if you're a genuine pacifist scientist, you might oppose, but otherwise it's just another way of making sure that in 20 years time scientists will oppose to being described as "scientists", as it will have some negative connotations.
Imagine, this is on "FASTS"'s top ten issues for 2003. I guess one of the other issues is: what to do when you have too much spare time.
> The only way of truely solving bugs is to know exactly whats going on in the code
Spoken like a true zealot, who has never written large programs in collaboration with others. Did you ever try to solve a memory allocation bug in a program of more than 100 lines without a debugger? That's when you learn to appreciate the power of debugging statements and conditional break points. Or when you have rare conditions in your networking code and the documentation just isn't accurate (think "Linux and setsockopt" here). It's really practical to be able to stop a program as soon as some condition is broken. You could do it using assertions, but that's just another way of debugging...
Yes, you *are* too old for this sh*t, since teaching MS Word is a total waste of time. Anybody who has been taught the basics of computer operations (and I mean basics, not how to use the right mouse button to click on an icon) can figure out for themselves how to write something in Word, given a simple manual. And most of the functions in Word will be replaced by something else within 5 years, so this "knowledge" is totally useless.
However, algebra (and other branches of maths) doesn't change and can be really useful. There are so many people who cannot even interpret numbers correctly, let alone probabilities and statistics. And that is really sad.
I'm not surpised. Schools tend to take away hours from maths and physics for teaching computer "science", so that would explain enough. Pity that MS Word is considered more important than algebra.
If I were you, I would start with classes on the principles underlying computers: how does the hardware work and how does it execute software? I assume they have got quite some computer experience, so it is a good idea to make them try to figure out how you go from the basic principles to complex software (e.g. games, which always appeals).
You might show them how input is treated in a simple system and how a program can manipulate a video screen. Then you might want to make them try to program something like that themselves, but to do so you would have to create a really easy environment, where everything just looks basic, so no "#include ".
After that, you could go on to explain the need for operating systems and their functionality. Once they finished that, they should have less problem understanding the basics of Linux or Windows.
Don't waste your time talking about stupid installation procedures or some GUI thingy that isn't going to last long.
And, Linux is not the only OS around. Make them choose for themselves. Show them the power of Linux if you want (shell, IO redirection, scripting, pipes, sockets, that kind of thing), but don't start a Linux Bible School...
As much as I admire Amnesty, this sounds like nonsense to me. Are they also going to sue foreign manufacturers of pens or paper that is used by Chinese authorities?
Disappointing is what you are. I just took a look of your home page, a vanity URL if ever I saw one (http://www.erickrout.com/), and boy, even if it's an adapted copy of an Onion article (BTW, did you check the copyrights?), how do you manage to get your head so far up your own arse?
Don't even *try* to play the manager and the insightful one here, because you don't have a clue as to what the open source "movement" is about. Your comments make you look to an open source contributor as TPHB to Dilbert.
But anyway, let's make this thing more positive. What do you think "we Open Sourcers truly need"? This is your opportunity...
But I think you have formulated the essence of the Open Source "movement": it can only be kept running by people who like to do it. Otherwise, it will become work, and somebody will have to pay for it.
No, the original poster is a pointy hair type, who doesn't have a clue. Saying things like "we need..., and we needed it yesterday". What a luser. Either he's someone trying to leech of open source projects, or he's a pompous little twat, who would very much like to become an open source manager, so that his initials will be on par with this damn George Nicholas Ulysses!
They found *ONE* occurrence. On *filtered data* (filtering causes strange biases in expectations of probability). The existence of strange matter is not really generally accepted. No-one knows the size or mass of this thing (it was estimated post-hoc on the basis of this theory). And small things, even with lots of energy, cannot produce (seismic) events with that much effect, since it's in essence just a very strong needle prick: it will cut right through everything with great damage to the environment. Plus, coincidence still exists. So, your prerequisites (take into account the number of events that match the profile but...) cannot be met.
And, as the original poster said: if you're looking for just one random event, given enough data, it's likely you can find it.
Now, if they would have found a whole string of correspondences, that would be something. But this is -- at best -- a hypothesis that needs to be watched over a very long time to become more probable.
Sorry, but "speech" and "understanding" are not the words to use to describe (Open)Cyc. If you yourself would have take the trouble to read the page you're referring to, you would have read: "OpenCyc can be used as the basis of a wide variety of intelligent applications such as : * speech understanding, etc."
So it's not NLP, but a module that could be coupled to a system that has an NLP component and needs to do some reasoning over it.
And for this you got a score of 3? Lack of knowledge does clearly not interfere with your karma.
Processing mathematics is easy compared to natural language. There were calculators (think abacus, or Pascal's mechanical calculator) long before anyone made an attempt at implementing language processing. Solving equations on the other hand is difficult, and you will find that (open source) algebraic manipulation programs are much rarer than simple math programs.
Plus, most people who do NLP know quite a lot of math, whereas most people who know math, know very little about NLP. So that explains part of the bias.
Let's not compare the complexity of user interface (design/implementation) with that of a serious NLP application. GUIs are easy. Try Visual Basic if you don't believe me... The programming tools for building GUIs are (of course) more complex, but still straight-forward compared to NLP.
Slashdot Mirror
A famous quote from one of the project leaders, Fred Jelinek if I'm not mistaken was that for every linguist that he fired from the team, the performance of the system improved by 10%...
I work quite heavily with Linux machines and they only crashed in case of hardware errors. Some machines even continued their server tasks with a totally broken hard disk.
On the other hand, when I toyed around with Excel and some Visual Basic script under XP the other day, the bl**dy thing refused to shut down properly. Had to kill the "explorer" before it would stop. And don't get me started on NT, or Word for that matter...
Then again, XP is a whole lot friendlier than any Linux distro I've ever seen, and that probably means more to most people than total stability.
How come nobody mentions Vivisimo and Teoma in this context? They happen to present the results for different meanings of the word "apple", although they cannot read your minder neither, very much like Googe. They only make it easier for laymen and novices to see that there are other "meanings" for the same search term.
However, this discussion is also blatantly Anglo-centric. In the rest of the world, "Apple" is even more strongly connected to the computer manufacturer (or the Beatles!). Try appel, Apfel, manzana, or pomme in Google and see what you'll find then (or take a look here.
My God, where are your moderator points when you need them. And could somebody introduce a "moronic nerd with serious lack of humor" label?
M-x sarcasm-mode-on
I think you forgot to mention that they can also reject the European budget in its entirety.
M-x sarcasm-mode-off
I hope you do realize that that's not the kind of power a parliament should have. A parliament should at least be able to vote off single members of the council, have influence on the "civil servants" appointed in the highest region, create and ammend laws on its own, change parts of the budget and block plans by other authorities (e.g. the expansion of the EU), be able to influence the "foreign" policies, etc.
As long as they don't have that kind of power, they are almost powerless. And what's the fuss about in relation to the new "Consitution"? Who gets to be the chair for how long!
Come on!
But don't go before getting another job.
Don't bother about people putting moral pressure on you, as I've seen in the postings before. That's quite unreasonable for two reasons.
First, if the company goes bankrupt, you'll need another job anyway, so the sooner you start looking, the better.
Second, it's not your responsability, but management's. You cannot be blamed for bringing the company to the point where it is now, so don't feel guilty about the consequences of your actions. Furthermore, somebody else fired you co-workers, and that should make you responsible? No.
There is another reasons why you might walk: it gives better job security to your current co-workers. The company will need them more than ever and will save a few bucks on your salary.
Good luck to you and the people in your company.
Now at least we know where Dilbert and his PHB work...
And as all people who have taken more advanced maths know: you can't plot the entire FFT in a 1000x200 pixel size window.
I can send you a file with exactly the same FFT output as the one claimed to represent the original and I can assure you it will sound distinctly different.
Here is how to do it: the height of each pixel corresponds most likely to sqrt(a[k]^2+b[k]^2), where a[k]+ib[k] is the k-th element in the fft output. So, you can try to estimate what the original value of the sqrt must have been by eyeballing the graph (just like the article poster did). Then you just guess the relation between the real and imaginary part, since it is not shown in the graph. You've got about 1000 of these points, but your song probably needs a few minutes at 44.1kHz, which translates to 8 million per channel (the stereo distribution of course also is missing from the graph, giving you some extra freedom to randomize numbers). Now distribute your first point over the first 8000 of the new song, the second over the second 8000, etc., feed it into the inverse FFT and you've got a file that's undistinguishable from the original, according to a visual spectral comparison.
But I bet you'll be able to *hear* the difference...
What's extensive? I've got my code built on template classes for lists, sets and associative arrays and the main program is 100k lines. Compile times on an old 500MHz Linux machine with gcc are completely acceptable. Compile times on my even slower mac with CodeWarrior are a little worse, but still acceptable.
Debugging? Both platforms handle templates transparently. Especially CodeWarrior does a nice job in its totally integrated environment, but gdb ain't bad neither.
Bloat? Neither. Yes, a list of ints will generate exactly the same code as a list of long ints twice, but this kind of template generates little code, so that's a small price to pay for transparency.
I don't know how many weeks I've lost to helping others debug or rewrite their code because they thought they would do something "clever" with templates and they ended up creating a maze.
That's a good point, but it simply refers to bad design and/or programming. Templates do offer an extra way of complicating code, but it doesn't have to be that way. Multiple inheritance can be a similar drag, but simple inheritance almost never fails in helping write clean code. Are you going to argue that inheritance is therefore bad? Didn't think so.
No, I think templates are really, really useful, and I like them a lot better than the inheritance mechanism. Just keep the design and programming clean, like everywhere else...
No, my point was that it was a copy of Windows 95 and Windows 95 is bad, bad, bad. User-unfriendliness disguised as a desirable gadget. Bright colors, big buttons. XP is even worse. Did you ever understand these changing menus? Horrible. So that's the way Linux should not be going. Perhaps it is a reasonable way to win existing Windows users, but it is not going to win new computer users over...
No Anonymous Coward,
I'm not a Windows user. I am a long time Unix (since version 6 on a PDP-11), RSX, VMS, Linux and MacOS user (since system 5). My problem was precisely that they are trying to copy the interface of Windows: that interface is horrible. And I agree: XP is worse, but with bright friendly colors, so it *seems* user-friendly.
About the karma whoring you might be right: although they are legitimate concerns I voiced (I do feel that more attention should be given to the user experience), I did get lucky to post soon enough. Did you ever notice that early posts receive much more karma than late posts?
What I would like to see is not another technical feat, but an effort to bring the Linux desk top closer to the a-technical masses.
I've recently had the "pleasure" of reinstalling Red Hat Linux and neither Gnome nor KDE are user-friendly at all. Yes, they do copy the Windows 95 desk top, but no, that's not going to help my father. And don't even start about the built-in file/web/help/and-what-not browsers.
With all this high configurability that's available in both windowing systems, couldn't a group of more human-interface oriented people build a layman interface on top of either Gnome or KDE?
A boffin is "a person engaged in scientific (esp. military) research".
So, if you're a genuine pacifist scientist, you might oppose, but otherwise it's just another way of making sure that in 20 years time scientists will oppose to being described as "scientists", as it will have some negative connotations.
Imagine, this is on "FASTS"'s top ten issues for 2003. I guess one of the other issues is: what to do when you have too much spare time.
> The only way of truely solving bugs is to know exactly whats going on in the code
Spoken like a true zealot, who has never written large programs in collaboration with others. Did you ever try to solve a memory allocation bug in a program of more than 100 lines without a debugger? That's when you learn to appreciate the power of debugging statements and conditional break points. Or when you have rare conditions in your networking code and the documentation just isn't accurate (think "Linux and setsockopt" here). It's really practical to be able to stop a program as soon as some condition is broken. You could do it using assertions, but that's just another way of debugging...
Yes, you *are* too old for this sh*t, since teaching MS Word is a total waste of time. Anybody who has been taught the basics of computer operations (and I mean basics, not how to use the right mouse button to click on an icon) can figure out for themselves how to write something in Word, given a simple manual. And most of the functions in Word will be replaced by something else within 5 years, so this "knowledge" is totally useless.
However, algebra (and other branches of maths) doesn't change and can be really useful. There are so many people who cannot even interpret numbers correctly, let alone probabilities and statistics. And that is really sad.
I'm not surpised. Schools tend to take away hours from maths and physics for teaching computer "science", so that would explain enough. Pity that MS Word is considered more important than algebra.
If I were you, I would start with classes on the principles underlying computers: how does the hardware work and how does it execute software? I assume they have got quite some computer experience, so it is a good idea to make them try to figure out how you go from the basic principles to complex software (e.g. games, which always appeals).
You might show them how input is treated in a simple system and how a program can manipulate a video screen. Then you might want to make them try to program something like that themselves, but to do so you would have to create a really easy environment, where everything just looks basic, so no "#include ".
After that, you could go on to explain the need for operating systems and their functionality. Once they finished that, they should have less problem understanding the basics of Linux or Windows.
Don't waste your time talking about stupid installation procedures or some GUI thingy that isn't going to last long.
And, Linux is not the only OS around. Make them choose for themselves. Show them the power of Linux if you want (shell, IO redirection, scripting, pipes, sockets, that kind of thing), but don't start a Linux Bible School...
As much as I admire Amnesty, this sounds like nonsense to me. Are they also going to sue foreign manufacturers of pens or paper that is used by Chinese authorities?
Disappointing is what you are. I just took a look of your home page, a vanity URL if ever I saw one (http://www.erickrout.com/), and boy, even if it's an adapted copy of an Onion article (BTW, did you check the copyrights?), how do you manage to get your head so far up your own arse?
Don't even *try* to play the manager and the insightful one here, because you don't have a clue as to what the open source "movement" is about. Your comments make you look to an open source contributor as TPHB to Dilbert.
But anyway, let's make this thing more positive. What do you think "we Open Sourcers truly need"? This is your opportunity...
Hear, hear.
..., and we needed it yesterday". What a luser. Either he's someone trying to leech of open source projects, or he's a pompous little twat, who would very much like to become an open source manager, so that his initials will be on par with this damn George Nicholas Ulysses!
But I think you have formulated the essence of the Open Source "movement": it can only be kept running by people who like to do it. Otherwise, it will become work, and somebody will have to pay for it.
No, the original poster is a pointy hair type, who doesn't have a clue. Saying things like "we need
Why didn't I use the preview button???
They found *ONE* occurrence. On *filtered data* (filtering causes strange biases in expectations of probability). The existence of strange matter is not really generally accepted. No-one knows the size or mass of this thing (it was estimated post-hoc on the basis of this theory). And small things, even with lots of energy, cannot produce (seismic) events with that much effect, since it's in essence just a very strong needle prick: it will cut right through everything with great damage to the environment. Plus, coincidence still exists. So, your prerequisites (take into account the number of events that match the profile but ...) cannot be met.
And, as the original poster said: if you're looking for just one random event, given enough data, it's likely you can find it.
Now, if they would have found a whole string of correspondences, that would be something. But this is -- at best -- a hypothesis that needs to be watched over a very long time to become more probable.
Sorry, but "speech" and "understanding" are not the words to use to describe (Open)Cyc. If you yourself would have take the trouble to read the page you're referring to, you would have read: "OpenCyc can be used as the basis of a wide variety of intelligent applications such as : * speech understanding, etc."
So it's not NLP, but a module that could be coupled to a system that has an NLP component and needs to do some reasoning over it.
And for this you got a score of 3? Lack of knowledge does clearly not interfere with your karma.
Precisely.
Processing mathematics is easy compared to natural language. There were calculators (think abacus, or Pascal's mechanical calculator) long before anyone made an attempt at implementing language processing. Solving equations on the other hand is difficult, and you will find that (open source) algebraic manipulation programs are much rarer than simple math programs.
Plus, most people who do NLP know quite a lot of math, whereas most people who know math, know very little about NLP. So that explains part of the bias.
Let's not compare the complexity of user interface (design/implementation) with that of a serious NLP application. GUIs are easy. Try Visual Basic if you don't believe me... The programming tools for building GUIs are (of course) more complex, but still straight-forward compared to NLP.
I suppose you came fully dressed as Dobby?